ZSHALL(1)

📦

version 5.9.2 · 7 versions

💅 ZSHALL(1) — the Z shell meta-man page

okay so this is zshall(1), from the General Commands Manual, and it's the Z shell meta-man page — like, the whole meta situation. This page bundles all the separate zsh manual pages, and bestie, it's a lot. 💖

🎀 NAME 🎀

zshall — the Z shell meta-man page

✨ OVERVIEW ✨

Because zsh contains many features, the zsh manual has been split into a number of sections. This manual page includes all the separate manual pages in the following order:

zsh — Zsh overview
zshroadmap — Informal introduction to the manual
zshmisc — Anything not fitting into the other sections
zshexpn — Zsh command and parameter expansion
zshparam — Zsh parameters
zshoptions — Zsh options
zshbuiltins — Zsh built-in functions
zshzle — Zsh command line editing
zshcompwid — Zsh completion widgets
zshcompsys — Zsh completion system
zshcompctl — Zsh completion control
zshmodules — Zsh loadable modules
zshcalsys — Zsh built-in calendar functions
zshtcpsys — Zsh built-in TCP functions
zshzftpsys — Zsh built-in FTP client
zshcontrib — Additional zsh functions and utilities

💖 DESCRIPTION 💖

Zsh is a UNIX command interpreter (shell) usable as an interactive login shell and as a shell script command processor. Of the standard shells, zsh most closely resembles ksh but includes many enhancements. It does not provide compatibility with POSIX or other shells in its default operating mode: see the section `Compatibility' below.

Zsh has command line editing, builtin spelling correction, programmable command completion, shell functions (with autoloading), a history mechanism, and a host of other features.

🫶 AUTHOR 🫶

Zsh was originally written by Paul Falstad. Zsh is now maintained by the members of the zsh-workers mailing list <zsh-workers@zsh.org>. The development is currently coordinated by Peter Stephenson <pws@zsh.org>. The coordinator can be contacted at <coordinator@zsh.org>, but matters relating to the code should generally go to the mailing list.

🌈 AVAILABILITY 🌈

Zsh is available from the following HTTP and anonymous FTP site.

ftp://ftp.zsh.org/pub/
https://www.zsh.org/pub/

The up-to-date source code is available via Git from Sourceforge. See https://sourceforge.net/projects/zsh/ for details. A summary of instructions for the archive can be found at https://zsh.sourceforge.io/.

🎀 MAILING LISTS 🎀

Zsh has several mailing lists:

<zsh-announce@zsh.org> — Announcements about releases, major changes in the shell and the monthly posting of the Zsh FAQ. (moderated)
<zsh-users@zsh.org> — User discussions.
<zsh-workers@zsh.org> — Hacking, development, bug reports and patches.
<zsh-security@zsh.org> — Private mailing list (the general public cannot subscribe to it) for discussing bug reports with security implications, i.e., potential vulnerabilities. If you find a security problem in zsh itself, please mail this address.

To subscribe or unsubscribe, send mail to the associated administrative address for the mailing list.

<zsh-announce-subscribe@zsh.org>
<zsh-users-subscribe@zsh.org>
<zsh-workers-subscribe@zsh.org>
<zsh-announce-unsubscribe@zsh.org>
<zsh-users-unsubscribe@zsh.org>
<zsh-workers-unsubscribe@zsh.org>

YOU ONLY NEED TO JOIN ONE OF THE MAILING LISTS AS THEY ARE NESTED. All submissions to zsh-announce are automatically forwarded to zsh-users. All submissions to zsh-users are automatically forwarded to zsh-workers.

If you have problems subscribing/unsubscribing to any of the mailing lists, send mail to <listmaster@zsh.org>.

The mailing lists are archived; the archives can be accessed via the administrative addresses listed above. There is also a hypertext archive available at https://www.zsh.org/mla/.

🌸 THE ZSH FAQ 🌸

Zsh has a list of Frequently Asked Questions (FAQ), maintained by Peter Stephenson <pws@zsh.org>. It is regularly posted to the newsgroup comp.unix.shell and the zsh-announce mailing list. The latest version can be found at any of the Zsh FTP sites, or at https://www.zsh.org/FAQ/. The contact address for FAQ-related matters is <faqmaster@zsh.org>.

🦋 THE ZSH WEB PAGE 🦋

Zsh has a web page which is located at https://www.zsh.org/. The contact address for web-related matters is <webmaster@zsh.org>.

🧚 THE ZSH USERGUIDE 🧚

A userguide is currently in preparation. It is intended to complement the manual, with explanations and hints on issues where the manual can be cabbalistic, hierographic, or downright mystifying (for example, the word hierographic does not exist). It can be viewed in its current state at https://zsh.sourceforge.io/Guide/. At the time of writing, chapters dealing with startup files and their contents and the new completion system were essentially complete.

⭐ INVOCATION ⭐

The following flags are interpreted by the shell when invoked to determine where the shell will read commands from:

-c — Take the first argument as a command to execute, rather than reading commands from a script or standard input. If any further arguments are given, the first one is assigned to $0, rather than being used as a positional parameter.
-i — Force shell to be interactive. It is still possible to specify a script to execute.
-s — Force shell to read commands from the standard input. If the -s flag is not present and an argument is given, the first argument is taken to be the pathname of a script to execute.

If there are any remaining arguments after option processing, and neither of the options -c or -s was supplied, the first argument is taken as the file name of a script containing shell commands to be executed. If the option PATH_SCRIPT is set, and the file name does not contain a directory path (i.e. there is no / in the name), first the current directory and then the command path given by the variable PATH are searched for the script. If the option is not set or the file name contains a / it is used directly.

After the first one or two arguments have been appropriated as described above, the remaining arguments are assigned to the positional parameters.

For further options, which are common to invocation and the set builtin, see zshoptions(1).

The long option --emulate followed (in a separate word) by an emulation mode may be passed to the shell. The emulation modes are those described for the emulate builtin, see zshbuiltins(1). The --emulate option must precede any other options (which might otherwise be overridden), but following options are honoured, so may be used to modify the requested emulation mode. Note that certain extra steps are taken to ensure a smooth emulation when this option is used compared with the emulate command within the shell: for example, variables that conflict with POSIX usage such as path are not defined within the shell.

Options may be specified by name using the -o option. -o acts like a single-letter option, but takes a following string as the option name. For example,

zsh -x -o shwordsplit scr

runs the script scr, setting the XTRACE option by the corresponding letter -x and the SH_WORD_SPLIT option by name. Options may be turned off by name by using +o instead of -o. -o can be stacked up with preceding single-letter options, so for example -xo shwordsplit or -xoshwordsplit is equivalent to -x -o shwordsplit.

Options may also be specified by name in GNU long option style, --option-name. When this is done, - characters in the option name are permitted: they are translated into _, and thus ignored. So, for example, zsh --sh-word-split invokes zsh with the SH_WORD_SPLIT option turned on. Like other option syntaxes, options can be turned off by replacing the initial - with a +; thus +-sh-word-split is equivalent to --no-sh-word-split. Unlike other option syntaxes, GNU-style long options cannot be stacked with any other options, so for example -x-shwordsplit is an error, rather than being treated like -x --shwordsplit.

The special GNU-style option --version is handled; it sends to standard output the shell's version information, then exits successfully. --help is also handled; it sends to standard output a list of options that can be used when invoking the shell, then exits successfully.

Option processing may be finished, allowing following arguments that start with - or + to be treated as normal arguments, in two ways. Firstly, a lone - (or +) as an argument by itself ends option processing. Secondly, a special option -- (or +-), which may be specified on its own (which is the standard POSIX usage) or may be stacked with preceding options (so -x- is equivalent to -x --). Options are not permitted to be stacked after -- (so -x-f is an error), but note the GNU-style option form discussed above, where --shwordsplit is permitted and does not end option processing.

Except when the sh/ksh emulation single-letter options are in effect, the option -b (or +b) ends option processing. -b is like --, except that further single-letter options can be stacked after the -b and will take effect as normal.

🌀 COMPATIBILITY 🌀

Okay bestie, so here's the tea: zsh is a total people-pleaser and will try to cosplay as sh or ksh when you invoke it as sh or ksh. More precisely — and this is the fun bit — it peeks at the first letter of the name it was invoked by, skipping any initial r (which is assumed to stand for restricted), and if that letter is b, s or k, it'll emulate sh or ksh.

And plot twist: if you invoke it as su (which happens on certain systems when the shell is executed by the su command), the shell will try to find an alternative name from the SHELL environment variable and do its emulation based on that. Sneaky but iconic.

In sh and ksh compatibility modes, these parameters are not special and not initialized by the shell:

ARGC, argv, cdpath, fignore, fpath, HISTCHARS, mailpath, MANPATH, manpath, path, prompt, PROMPT, PROMPT2, PROMPT3, PROMPT4, psvar, status

The usual zsh startup/shutdown scripts are not executed. Login shells source /etc/profile followed by $HOME/.profile. If the ENV environment variable is set on invocation, $ENV is sourced after the profile scripts. And listen — the value of ENV is subjected to parameter expansion, command substitution, and arithmetic expansion before being interpreted as a pathname. Also note that the PRIVILEGED option affects the execution of startup files too. 👑

The following options get set if the shell is invoked as sh or ksh:

NO_BAD_PATTERN
NO_BANG_HIST
NO_BG_NICE
NO_EQUALS
NO_FUNCTION_ARGZERO
GLOB_SUBST
NO_GLOBAL_EXPORT
NO_HUP
INTERACTIVE_COMMENTS
KSH_ARRAYS
NO_MULTIOS
NO_NOMATCH
NO_NOTIFY
POSIX_BUILTINS
NO_PROMPT_PERCENT
RM_STAR_SILENT
SH_FILE_EXPANSION
SH_GLOB
SH_OPTION_LETTERS
SH_WORD_SPLIT

Additionally, the BSD_ECHO and IGNORE_BRACES options are set if zsh is invoked as sh. And get this — the KSH_OPTION_PRINT, LOCAL_OPTIONS, PROMPT_BANG, PROMPT_SUBST and SINGLE_LINE_ZLE options are set if zsh is invoked as ksh. ✨

Please note that, whilst reasonable efforts are taken to address incompatibilities when they arise, zsh does not guarantee complete emulation of other shells, nor POSIX compliance. For more information on the differences between zsh and other shells, please refer to chapter 2 of the shell FAQ, https://www.zsh.org/FAQ/.

🔒 RESTRICTED SHELL 🔒

When the basename of the command used to invoke zsh starts with the letter r, or the -r command line option is supplied at invocation, the shell becomes restricted. Emulation mode is determined after stripping the letter r from the invocation name. Ngl, the following things are straight-up disabled in restricted mode:

changing directories with the cd builtin
changing or unsetting the EGID, EUID, GID, HISTFILE, HISTSIZE, IFS, LD_AOUT_LIBRARY_PATH, LD_AOUT_PRELOAD, LD_LIBRARY_PATH, LD_PRELOAD, MODULE_PATH, module_path, PATH, path, SHELL, UID and USERNAME parameters
specifying command names containing /
specifying command pathnames using hash
redirecting output to files
using the exec builtin command to replace the shell with another command
using jobs -Z to overwrite the shell process' argument and environment space
using the ARGV0 parameter to override argv[0] for external commands
turning off restricted mode with set +r or unsetopt RESTRICTED

These restrictions are enforced after processing the startup files. Your startup files should set up PATH to point to a directory of commands which can be safely invoked in the restricted environment. They may also add further restrictions by disabling selected builtins.

Restricted mode can also be activated any time by setting the RESTRICTED option. That immediately enables all the restrictions described above, even if the shell still has not processed all startup files. ⚡

A shell Restricted Mode is an outdated way to restrict what users may do: modern systems have better, safer and more reliable ways to confine user actions, such as chroot jails, containers and zones.

A restricted shell is very difficult to implement safely. The feature may be removed in a future version of zsh.

It is important to realise that the restrictions only apply to the shell, not to the commands it runs (except for some shell builtins). While a restricted shell can only run the restricted list of commands accessible via the predefined PATH variable, it does not prevent those commands from running any other command.

As an example, if env is among the list of allowed commands, then it allows the user to run any command — as env is not a shell builtin command and can run arbitrary executables. 💀

So when implementing a restricted shell framework it's important to be fully aware of what actions each of the allowed commands or features (which may be regarded as modules) can perform.

Many commands can have their behaviour affected by environment variables. Except for the few listed above, zsh does not restrict the setting of environment variables.

If a perl, python, bash, or other general purpose interpreted script is treated as a restricted command, the user can work around the restriction by setting specially crafted PERL5LIB, PYTHONPATH, BASHENV (etc.) environment variables. On GNU systems, any command can be made to run arbitrary code when performing character set conversion (including zsh itself) by setting a GCONV_PATH environment variable. Those are only a few examples. 🙈

Bear in mind that, contrary to some other shells, readonly is not a security feature in zsh as it can be undone, and so cannot be used to mitigate the above.

A restricted shell only works if the allowed commands are few and carefully written so as not to grant more access to users than intended. It's also important to restrict what zsh module the user may load, as some of them — such as zsh/system, zsh/mapfile and zsh/files — allow bypassing most of the restrictions. 🫶

🌸 STARTUP/SHUTDOWN FILES 🌸

Commands are first read from /etc/zsh/zshenv — and bestie, this one cannot be overridden. After that, subsequent behaviour is modified by the RCS and GLOBAL_RCS options. The former affects all startup files, while the second only affects global startup files (those shown here with a path starting with a /). If one of the options is unset at any point, any subsequent startup file(s) of the corresponding type will not be read. It's also possible for a file in $ZDOTDIR to re-enable GLOBAL_RCS. Both RCS and GLOBAL_RCS are set by default. ✨

Then commands are read from $ZDOTDIR/.zshenv. If the shell is a login shell, commands are read from /etc/zsh/zprofile and then $ZDOTDIR/.zprofile. Then, if the shell is interactive, commands are read from /etc/zsh/zshrc and then $ZDOTDIR/.zshrc. Finally, if the shell is a login shell, /etc/zsh/zlogin and $ZDOTDIR/.zlogin are read.

When a login shell exits, the files $ZDOTDIR/.zlogout and then /etc/zsh/zlogout are read. This happens with either an explicit exit via the exit or logout commands, or an implicit exit by reading end-of-file from the terminal. However — and this is the important bit — if the shell terminates due to exec'ing another process, the logout files are not read. These are also affected by the RCS and GLOBAL_RCS options. Note also that the RCS option affects the saving of history files, i.e. if RCS is unset when the shell exits, no history file will be saved. 😭

If ZDOTDIR is unset, HOME is used instead. Files listed above as being in /etc may be in another directory, depending on the installation.

As /etc/zsh/zshenv is run for all instances of zsh, it's important that it be kept as small as possible. In particular, it's a good idea to put code that does not need to be run for every single shell behind a test of the form if [[ -o rcs ]]; then ... so that it won't be executed when zsh is invoked with the -f option. 🧚

Any of these files may be pre-compiled with the zcompile builtin command (see zshbuiltins(1)). If a compiled file exists (named for the original file plus the .zwc extension) and it is newer than the original file, the compiled file will be used instead. Slay.

zsh 5.9.2                        July 12, 2026                        ZSHALL(1)

ZSHROADMAP(1)               General Commands Manual               ZSHROADMAP(1)

🗺️ NAME 🗺️

zshroadmap — informal introduction to the zsh manual. The Zsh Manual, like the shell itself, is large and often complicated. This section of the manual provides some pointers to areas of the shell that are likely to be of particular interest to new users, and indicates where in the rest of the manual the documentation is to be found. 💖

🌅 WHEN THE SHELL STARTS 🌅

When it starts, the shell reads commands from various files. These can be created or edited to customize the shell. See the section Startup/Shutdown Files in zsh(1).

If no personal initialization files exist for the current user, a function is run to help you change some of the most common settings. It won't appear if your administrator has disabled the zsh/newuser module. The function is designed to be self-explanatory. You can run it by hand with:

autoload -Uz zsh-newuser-install; zsh-newuser-install -f

See also the section User Configuration Functions in zshcontrib(1). 🎀

🦋 INTERACTIVE USE 🦋

okay so talking to the shell? that's all the builtin Zsh Line Editor, aka ZLE. it's fully documented in zshzle(1) if you wanna go deep 💅

the very first thing you gotta decide is Emacs vs Vi editing mode, because the keys are honestly so different. ngl Emacs editing mode is probably more natural for beginners, and you can pick it explicitly with bindkey -e ✨

there's a history mechanism for pulling back lines you typed before (easiest way: just hit Up or Down arrow). BUT — and this is important, bestie — unlike other shells, zsh will not save those lines when the shell exits unless you set the appropriate variables. and the number of history lines kept by default is kinda tiny (like, 30 lines 🥲). go read about the shell variables (the docs call them parameters) HISTFILE, HISTSIZE and SAVEHIST in zshparam(1). also note: right now it's only possible to read and write the files that save history when the shell is interactive, so it doesn't work from scripts 🙈

the shell now supports the UTF-8 character set (and others too, if your OS supports them). this is (mostly) handled transparently by the shell, but how well terminal emulators support it varies a lot. there's some discussion of this in the shell FAQ, https://www.zsh.org/FAQ/. specifically note that for combining characters to be handled, the option COMBINING_CHARS needs to be set 👑. and because the shell is now more sensitive to how the character set is defined, if you're upgrading from an older version you should make sure the appropriate variable is set to an appropriate value — either LANG (affects all aspects of the shell's operation) or LC_CTYPE (affects only character set handling). this applies even if you're using a single-byte character set including ASCII extensions like ISO-8859-1 or ISO-8859-15 💖. see the description of LC_CTYPE in zshparam(1).

✨ Completion ✨

completion is a thing lots of shells have. it lets you type only part of a word (usually the prefix) and the shell fills in the rest. but zsh's completion system is programmable, which is the slay part 🎀. like, you can set the shell to complete email addresses in the arguments to the mail command from your ~/.abook/addressbook; usernames, hostnames, and even remote paths in arguments to scp, and on and on. literally anything you can write in zsh or glue together with zsh can be the source of what the line editor offers up as possible completions 💅

zsh has two completion systems:

an old one, the so-called compctl completion (named after the builtin command that serves as its complete and only user interface), and
a new one, called compsys, organized as a library of builtin and user-defined functions.

the two differ in how you specify completion behavior. the new system is more customizable and ships with completions for tons of commonly used commands, so it's the one to prefer, fr fr 🌟

the completion system has to be enabled explicitly when the shell starts. more info in zshcompsys(1) 🫶

🧚 Extending the line editor 🧚

besides completion, the line editor is super extensible through shell functions. some useful ones come with the shell, giving you stuff like:

insert-composed-char — composing characters that aren't on your keyboard
match-words-by-style — configuring what the line editor considers a word when you move or delete by word
history-beginning-search-backward-end, etc. — alternative ways of searching the shell history
replace-string, replace-pattern — functions for replacing strings or patterns globally in the command line
edit-command-line — edit the command line with an external editor.

see the section ZLE Functions' in zshcontrib(1)` for the descriptions of these 💜

🌈 OPTIONS 🌈

the shell has a lot of options for changing how it behaves. they cover every aspect of the shell, and honestly just browsing the full documentation is the only good way to get acquainted with all the possibilities. see zshoptions(1) 📚

🍓 PATTERN MATCHING 🍓

the shell has a rich set of patterns available for file matching (the docs call it filename generation, and for historical reasons it's also known as globbing) and for use when programming. these are described in the section Filename Generation' in zshexpn(1)`. ✨

the ones that are especially interesting — because other pattern matching systems don't commonly support them — are:

** — for matching over multiple directories
| — for matching either of two alternatives
~, ^ — the ability to exclude patterns from matching when the EXTENDED_GLOB option is set
(...) — glob qualifiers, included in parentheses at the end of the pattern, which select files by type (such as directories) or attribute (such as size).

🦋 GENERAL COMMENTS ON SYNTAX 🦋

so zsh's syntax is in some ways similar to the Korn shell, and therefore more remotely to the original UNIX shell, the Bourne shell — but its default behaviour doesn't entirely correspond to those shells 👀. general shell syntax is introduced in the section Shell Grammar' in zshmisc(1)`.

one difference you'll run into a lot: variables substituted onto the command line are not split into words. see the description of the shell option SH_WORD_SPLIT in the section Parameter Expansion' in zshexpn(1). in zsh you can either explicitly request the splitting (e.g. ${=foo}) or use an array when you want a variable to expand to more than one word. see the section Array Parameters' in zshparam(1) 🫶

👑 PROGRAMMING 👑

the most convenient way to add enhancements to the shell is usually to write a shell function and arrange for it to be autoloaded. functions are described in the section Functions' in zshmisc(1)`. and if you're coming from the C shell and its relatives: heads up, aliases are used less in zsh because they don't perform argument substitution, only simple text replacement 💀

a few general functions, other than the line editor ones described above, come with the shell and are described in zshcontrib(1). features include:

promptinit — a prompt theme system for changing prompts easily, see the section `Prompt Themes'
zsh-mime-setup — a MIME-handling system which dispatches commands according to the suffix of a file as done by graphical file managers
zcalc — a calculator
zargs — a version of xargs that makes the find command redundant
zmv — a command for renaming files by means of shell patterns.

zsh 5.9.2                        July 12, 2026                    ZSHROADMAP(1)
────────────────────────────────────────────────────────────────────────────────
ZSHMISC(1)                  General Commands Manual                  ZSHMISC(1)

🌸 NAME 🌸

zshmisc - everything and then some 💅

⭐ SIMPLE COMMANDS & PIPELINES ⭐

a simple command is a sequence of optional parameter assignments followed by blank-separated words, with optional redirections sprinkled in between. for a description of assignment, see the beginning of zshparam(1) ✨

the first word is the command to be executed, and the rest of the words, if there are any, are arguments to that command. if a command name is given, the parameter assignments modify the environment of the command when it's executed. the value of a simple command is its exit status, or 128 plus the signal number if it got terminated by a signal 🔥. for example,

echo foo

is a simple command with arguments. easy!

a pipeline is either a simple command, or a sequence of two or more simple commands where each command is separated from the next by | or |&. when commands are separated by |, the standard output of the first command is connected to the standard input of the next. |& is shorthand for 2>&1 |, which connects both the standard output and the standard error of the command to the standard input of the next one 🎀. the value of a pipeline is the value of the last command — unless the pipeline is preceded by !, in which case the value is the logical inverse of the value of the last command. for example,

echo foo | sed 's/foo/bar/'

is a pipeline, where the output (foo plus a newline) of the first command gets passed to the input of the second 😌

if a pipeline is preceded by coproc, it's executed as a coprocess; a two-way pipe is established between it and the parent shell. the shell can read from or write to the coprocess using the >&p and <&p redirection operators, or with print -p and read -p. a pipeline cannot be preceded by both coproc and ! — pick one, bestie 💀. if job control is active, the coprocess can be treated as an ordinary background job for things other than input and output.

a sublist is either a single pipeline, or a sequence of two or more pipelines separated by && or ||. if two pipelines are separated by &&, the second pipeline is executed only if the first succeeds (returns a zero status). if two pipelines are separated by ||, the second is executed only if the first fails (returns a nonzero status). both operators have equal precedence and are left associative 👑. the value of the sublist is the value of the last pipeline executed. for example,

dmesg | grep panic && print yes

is a sublist consisting of two pipelines — the second is just a simple command that will be executed if and only if the grep command returns a zero status. if it doesn't, the value of the sublist is that return status; otherwise it's the status returned by the print (almost certainly zero).

a list is a sequence of zero or more sublists, where each sublist is terminated by ;, &, &|, &!, or a newline 🌟. that terminator may optionally be left off the last sublist in the list when the list appears as a complex command inside (...) or {...}. when a sublist is terminated by ; or a newline, the shell waits for it to finish before executing the next sublist. if a sublist is terminated by a &, &|, or &!, the shell executes the last pipeline in it in the background and does not wait for it to finish — and note the difference from other shells, which execute the whole sublist in the background 👀. a backgrounded pipeline returns a status of zero.

more generally, a list can be seen as a set of any shell commands whatsoever, including the complex commands below; that's implied wherever the word list shows up in later descriptions. for example, the commands in a shell function form a special sort of list 🫶

✨ PRECOMMAND MODIFIERS ✨

Okay so bestie, listen 💖 — sometimes you wanna tweak the vibe of a command before it even runs. That's what a precommand modifier is for. You stick one of these bad boys in front of a simple command and it changes how the whole thing gets interpreted. They're shell builtin commands, with one iconic exception: nocorrect, which is a reserved word (yes, it's special like that 💅).

- — slips a - onto the front of the command's argv[0] string before running it. Just a little dash to set the mood ✨
builtin — tells the shell "bestie, this word is a builtin," so it does NOT go looking for a shell function or an external command. It's giving "I know exactly what I want" energy 👑
command [ -pvV ] — says "treat this command word as an external command," not a shell function or builtin. But wait, there's a plot twist: if the POSIX_BUILTINS option is set, builtins will still run — though some of their special little properties get suppressed 🙈. Let's break the flags down:
- -p — searches a default path instead of whatever's in $path. Fresh start vibes 🌸
- -v — makes command act like whence
- -V — literally equivalent to whence -v
exec [ -cl ] [ -a argv0 ] — kind of dramatic, ngl 😭. The next command (plus all its args) runs in place of the current process instead of as a sub-process. The shell does not fork — it gets replaced. That's main character energy to the max. Heads up:
- The shell does not invoke TRAPEXIT
- It also does not source zlogout files
- The options here exist purely for compatibility with other shells

Now the flags themselves:

-c — clears the environment. Clean slate, periodt ✨
-l — same deal as the - precommand modifier: it treats the replacement command as a login shell, and runs it with a - prepended to its argv[0] string. ⚠️ This flag does nothing if you also use -a, so don't bother, bestie.
-a — lets you explicitly set the argv[0] string (that's the name of the command as the process itself sees it) used by the replacement command. It's directly equivalent to setting a value for the ARGV0 environment variable. Very "I pick my own destiny" of it 🦋

nocorrect — turns off spelling correction for every single word in the command. 🚨 Important: this one has to come before any other precommand modifier, because it gets interpreted immediately, before any parsing even happens. Also, it does nothing in non-interactive shells. Sad but true 😔
noglob — filename generation (a.k.a. globbing) is skipped for all the words. No wildcards expanding on your watch, diva 🌈

💅 COMPLEX COMMANDS 💅

okay so a "complex command" in zsh is basically any of these big-girl structures. buckle up bestie, there's a bunch 🎀

🦋 `if`

if list then list [ elif list then list ] ... [ else list ] fi

the if list runs first, and if it comes back with a zero exit status (i.e. truthy, yes i know it's backwards, stay with me), the then list runs. otherwise the elif list runs, and if its status is zero, the then list runs. and if every single elif list returns nonzero? then the else list gets its moment. slay. ✨

🌸 `for` (the classic word-list one)

for name ... [ in word ... ] term do list done

expand that list of words, set the parameter name to each one in turn, and run list every time. if you leave out the in word part, it uses the positional parameters instead of the words. 🫶

the term is one or more newline or ; — those terminate the words, and they're optional when the in word is omitted.

here's the iconic part: you can put more than one parameter name before the list of words. if you give N names, then on each go-round of the loop the next N words get assigned to the matching parameters. if there are more names than remaining words, those leftover parameters each get set to the empty string. the loop ends when there's no remaining word to assign to the first name. 👑

also, in can literally only appear as the first name in the list — anywhere else and it gets treated as marking the end of the list. iykyk.

🧚 `for` (the arithmetic one)

for (( [expr1] ; [expr2] ; [expr3] )) do list done

the arithmetic expression expr1 is evaluated first (peep the Arithmetic Evaluation section). then expr2 gets evaluated over and over until it evaluates to zero — and while it's non-zero, list runs and expr3 gets evaluated. if you omit any of the expressions, it behaves as if it evaluated to 1. no cap. 💜

⭐ `while`

while list do list done

run the do list for as long as the while list returns a zero exit status. that's it, that's the vibe check.

🌈 `until`

until list do list done

run the do list for as long as the until list returns a nonzero exit status. literally the inverse queen of while.

🍓 `repeat`

repeat word do list done

word gets expanded and treated as an arithmetic expression, which must evaluate to a number n. then list runs n times. cute, efficient, iconic. ✨

⚠️ okay important: the repeat syntax is disabled by default when the shell starts in a mode emulating another shell. you can turn it on with the command enable -r repeat.

🎀 `case`

case word in [ [(] pattern [ | pattern ] ... ) list (;;|;&|;|) ] ...
esac

runs the list associated with the first pattern that matches word, if any match at all. the patterns are built the same way as filename generation — go read the Filename Generation section if you need the deets.

now the fussy bits, and yes they matter 👀: unless the SH_GLOB option is set, the shell treats the whole pattern-with-alternatives as equivalent to a group of patterns inside parentheses — although white space may appear around the parentheses and the vertical bar, and it'll be stripped from the pattern at those points. white space may appear elsewhere in the pattern, and that is not stripped. if SH_GLOB is set, so that an opening parenthesis can be unambiguously treated as part of the case syntax, the expression gets parsed into separate words and those are treated as strict alternatives (like in other shells).

and here's the terminator drama 💀:

if the list that ran is terminated with ;& rather than ;;, the following list is also executed. the rule for the terminator of that following list — ;;, ;& or ;| — applies unless the esac is reached.
if the list that ran is terminated with ;|, the shell keeps scanning the patterns looking for the next match, executes the corresponding list, and applies the rule for that list's terminator — ;;, ;& or ;|. important: word is not re-expanded; every applicable pattern gets tested against the same word.

🔥 `select`

select name [ in word ... term ] do list done

where term is one or more newline or ; to terminate the words. it prints the set of words, each one preceded by a number. if the in word is omitted, it uses the positional parameters. the PROMPT3 prompt gets printed and a line is read — from the line editor if the shell is interactive and that's active, otherwise from standard input. 🌸

if that line is the number of one of the listed words, then parameter name is set to the word corresponding to that number. if the line is empty, the selection list gets printed again (round two bestie). otherwise, the value of parameter name is set to null. the contents of the line read from standard input get saved in the parameter REPLY. list runs for each selection until a break or end-of-file shows up.

💜 `( list )`

run list in a subshell. traps set by the trap builtin get reset to their default values while list is executing — with one exception: ignored signals will continue to be ignored if the option POSIXTRAPS is set. 💅

🫶 `{ list }`

run list. yes that's the whole thing. main character minimalism.

👑 `{ try-list } always { always-list }`

_first_ execute try-list. and regardless of errors, or break or continue commands encountered inside try-list, execute always-list. then execution continues from the result of executing try-list — in other words, any error, or break or continue, is treated in the normal way, exactly as if always-list weren't even there. the two chunks are called the try block and the always block. 🎀

⚠️ optional newlines or semicolons may appear after the always — but note, they may not appear between the preceding closing brace and the always.

an error in this context means a condition like a syntax error that causes the shell to abort execution of the current function, script, or list. syntax errors encountered while the shell is parsing the code do not cause the always-list to be executed. for example, a badly constructed if block inside try-list would make the shell abort during parsing, so always-list would not run — whereas a bad substitution like ${*foo*} would cause a run-time error, after which always-list would run. 😭

an error condition can be tested and reset with the special integer variable TRY_BLOCK_ERROR. outside an always-list its value is irrelevant, but it's initialised to -1. inside always-list, the value is 1 if an error occurred in the try-list, else 0. if TRY_BLOCK_ERROR is set to 0 during the always-list, the error condition caused by the try-list is reset, and shell execution continues normally after the end of always-list. altering the value during the try-list isn't useful (unless it's part of an enclosing always block). 🧚

regardless of TRY_BLOCK_ERROR, after the end of always-list the normal shell status $? is the value returned from try-list. it'll be non-zero if there was an error — even if TRY_BLOCK_ERROR was set to zero. iconic but rude.

the following executes the given code while ignoring any errors it causes. it's an alternative to the usual convention of protecting code by running it in a subshell:

{
    # code which may cause an error
  } always {
    # This code is executed regardless of the error.
    (( TRY_BLOCK_ERROR = 0 ))
}
# The error condition has been reset.

when a try block occurs outside of any function, a return or an exit encountered in try-list does not cause the always-list to run. instead the shell exits immediately after any EXIT trap has been executed. otherwise, a return command encountered in try-list will cause always-list to run, just like break and continue. ⭐

🌟 defining functions

function [ -T ] word ... [ () ] [ term ] { list }
word ... () [ term ] { list }
word ... () [ term ] command

where term is one or more newline or ;. defines a function that's referenced by any one of word. normally you only give one word; multiple words are usually only useful for setting traps. the body of the function is the list between the { and }. see the Functions section. 🦋

the options of function mean this:

-T — enable tracing for this function, as though with functions -T. see the docs for the -f option to the typeset builtin, in zshbuiltins(1).

if the option SH_GLOB is set for compatibility with other shells, then whitespace may appear between the left and right parentheses when there's a single word — otherwise the parentheses get treated as forming a globbing pattern in that case. 🙈

in any of the forms above, a redirection may appear outside the function body, like this:

func() { ... } 2>&1

that redirection is stored with the function and applied whenever the function is executed. any variables in the redirection are expanded at the point the function is executed, but outside the function scope. 💫

🎬 `time`

time [ pipeline ]

the pipeline is executed, and timing statistics get reported on the standard error in the form specified by the TIMEFMT parameter. if pipeline is omitted, it prints statistics about the shell process and its children. ⏱️

💅 `[[ exp ]]`

evaluates the conditional expression exp and returns a zero exit status if it's true. see the Conditional Expressions section for a description of exp. go off. 👑

🦋 ALTERNATE FORMS FOR COMPLEX COMMANDS 🦋

okay bestie, so a bunch of zsh's complex commands have alternate forms — like the secret alt routes of shell syntax ✨ These are non-standard and highkey not obvious even to seasoned shell programmers, so do not use them anywhere portability of shell code is a concern. Periodt. 💅

The short versions below only work if sublist is of the form { list } or if the SHORT_LOOPS option is set. For the if, while and until commands, in both these cases the test part of the loop must also be suitably delimited, such as by [[ ... ]] or (( ... )), else the end of the test will not be recognized 😬 For the for, repeat, case and select commands no such special form for the arguments is necessary, but the other condition (the special form of sublist or use of the SHORT_LOOPS option) still applies. The SHORT_REPEAT option is available to enable the short version only for the repeat command.

if list { list } [ elif list { list } ] ... [ else { list } ] — an alternate form of if. The rules mean that

if [[ -o ignorebraces ]] {
  print yes
}

works, but

if true {  # Does not work!
  print yes
}

does not, since the test is not suitably delimited 🙈

if list sublist — a short form of the alternate if. The same limitations on the form of list apply as for the previous form.
for name ... ( word ... ) sublist — a short form of for.
for name ... [ in word ... ] term sublist — where term is at least one newline or ;. Another short form of for.
for (( [expr1] ; [expr2] ; [expr3] )) sublist — a short form of the arithmetic for command.
foreach name ... ( word ... ) list end — another form of for.
while list { list } — an alternative form of while. Note the limitations on the form of list mentioned above 👆
until list { list } — an alternative form of until. Note the limitations on the form of list mentioned above too 👆
repeat word sublist — this is a short form of repeat.
case word { [ [(] pattern [ | pattern ] ... ) list (;;|;&|;|) ] ... } — an alternative form of case.
select name [ in word ... term ] sublist — where term is at least one newline or ;. A short form of select.
function word ... [ () ] [ term ] sublist — this is a short form of function.

👑 RESERVED WORDS 👑

The following words are recognized as reserved words when used as the first word of a command — unless they're quoted or disabled using disable -r:

do done esac then elif else fi for case if while function repeat time
until select coproc nocorrect foreach end ! [[ { } declare export float
integer local readonly typeset

Additionally, } is recognized in any position if neither the IGNORE_BRACES option nor the IGNORE_CLOSE_BRACES option is set. 🫶

💀 ERRORS 💀

Certain errors are treated as fatal by the shell: in an interactive shell, they cause control to return to the command line, and in a non-interactive shell they cause the shell to be aborted 😭 In older versions of zsh, a non-interactive shell running a script would not abort completely, but would resume execution at the next command to be read from the script, skipping the remainder of any functions or shell constructs such as loops or conditions; this somewhat illogical behaviour can be recovered by setting the option CONTINUE_ON_ERROR.

Fatal errors found in non-interactive shells include:

Failure to parse shell options passed when invoking the shell
Failure to change options with the set builtin
Parse errors of all sorts, including failures to parse mathematical expressions
Failures to set or modify variable behaviour with typeset, local, declare, export, integer, float
Execution of incorrectly positioned loop control structures (continue, break)
Attempts to use regular expression with no regular expression module available
Disallowed operations when the RESTRICTED options is set
Failure to create a pipe needed for a pipeline
Failure to create a multio
Failure to autoload a module needed for a declared shell feature
Errors creating command or process substitutions
Syntax errors in glob qualifiers
File generation errors where not caught by the option BAD_PATTERN
All bad patterns used for matching within case statements
File generation failures where not caused by NOMATCH or similar options
All file generation errors where the pattern was used to create a multio
Memory errors where detected by the shell
Invalid subscripts to shell variables
Attempts to assign read-only variables
Logical errors with variables such as assignment to the wrong type
Use of invalid variable names
Errors in variable substitution syntax
Failure to convert characters in $'...' expressions

If the POSIX_BUILTINS option is set, more errors associated with shell builtin commands are treated as fatal, as specified by the POSIX standard.

🎀 COMMENTS 🎀

In non-interactive shells, or in interactive shells with the INTERACTIVE_COMMENTS option set, a word beginning with the third character of the histchars parameter (# by default) causes that word and all the following characters up to a newline to be ignored. So cute, so simple 💖

🌸 ALIASING 🌸

Every eligible word in the shell input is checked to see if there's an alias defined for it 👀 If so, it's replaced by the text of the alias if it's in command position (if it could be the first word of a simple command), or if the alias is global. And if the replacement text ends with a space, the next word in the shell input is always eligible for purposes of alias expansion. Slay.

It is an error for the function name, word, in the sh-compatible function definition syntax word () ... to be a word that resulted from alias expansion, unless the ALIAS_FUNC_DEF option is set. 🚨

An alias is defined using the alias builtin; global aliases may be defined using the -g option to that builtin.

A word is defined as:

Any plain string or glob pattern
Any quoted string, using any quoting method (note that the quotes must be part of the alias definition for this to be eligible)
Any parameter reference or command substitution
Any series of the foregoing, concatenated without whitespace or other tokens between them
Any reserved word (case, do, else, etc.)
With global aliasing, any command separator, any redirection operator, and ( or ) when not part of a glob pattern

Alias expansion is done on the shell input before any other expansion except history expansion. Therefore, if an alias is defined for the word foo, alias expansion may be avoided by quoting part of the word, e.g. \foo. Any form of quoting works, although there's nothing to prevent an alias being defined for the quoted form such as \foo as well 😅

In particular, note that quoting must be used when using unalias to remove global aliases:

% alias -g foo=bar
% unalias foo
unalias: no such hash table element: bar
% unalias \foo
%

When POSIX_ALIASES is set, only plain unquoted strings are eligible for aliasing. The alias builtin does not reject ineligible aliases, but they are not expanded. Ngl, rude.

For use with completion, which would remove an initial backslash followed by a character that isn't special, it may be more convenient to quote the word by starting with a single quote, i.e. 'foo; completion will automatically add the trailing single quote. iykyk ✨

🧚 Alias difficulties 🧚

Although aliases can be used in ways that bend normal shell syntax, not every string of non-white-space characters can be used as an alias. Highkey tragic.

Any set of characters not listed as a word above is not a word, hence no attempt is made to expand it as an alias, no matter how it's defined (i.e. via the builtin or the special parameter aliases described in the section THE ZSH/PARAMETER MODULE in zshmodules(1)). However, as noted in the case of POSIX_ALIASES above, the shell does not attempt to deduce whether the string corresponds to a word at the time the alias is created. 🙃

For example, an expression containing an = at the start of a command line is an assignment and cannot be expanded as an alias; a lone = is not an assignment but can only be set as an alias using the parameter, as otherwise the = is taken part of the syntax of the builtin command.

It is not presently possible to alias the (( token that introduces arithmetic expressions, because until a full statement has been parsed, it cannot be distinguished from two consecutive ( tokens introducing nested subshells. Also, if a separator such as && is aliased, \&& turns into the two tokens \& and &, each of which may have been aliased separately. Similarly for \<<, \>|, etc. 💀

There is a commonly encountered problem with aliases illustrated by the following code:

alias echobar='echo bar'; echobar

This prints a message that the command echobar could not be found. This happens because aliases are expanded when the code is read in; the entire line is read in one go, so that when echobar is executed it's too late to expand the newly defined alias. This is often a problem in shell scripts, functions, and code executed with source or .. Consequently, use of functions rather than aliases is recommended in non-interactive code. 🫶

💜 QUOTING 💜

A character may be quoted (that is, made to stand for itself) by preceding it with a \. \ followed by a newline is ignored. Iconic.

A string enclosed between $' and ' is processed the same way as the string arguments of the print builtin, and the resulting string is considered to be entirely quoted. A literal ' character can be included in the string by using the \' escape. 🌈

All characters enclosed between a pair of single quotes ('') that is not preceded by a $ are quoted. A single quote cannot appear within single quotes unless the RC_QUOTES option is set, in which case a pair of single quotes are turned into a single quote. For example,

print ''''

outputs nothing apart from a newline if RC_QUOTES is not set, but one single quote if it is set. The way that one tiny option changes everything 😭

Inside double quotes (""), parameter and command substitution occur, and \ quotes the characters \, `, ", $, and the first character of $histchars (default !). ⭐

🔀 REDIRECTION 🔀

Okay bestie, this is the part where we talk about where your output goes and where your input comes from 💅 Redirecting is literally the glow-up of file descriptors.

So first things first: if a command is followed by & and job control is not active, then the default standard input for that command is the empty file /dev/null. Otherwise, the environment for the execution of a command contains the file descriptors of the invoking shell as modified by input/output specifications.

The following may appear anywhere in a simple command, or may precede or follow a complex command. Expansion occurs before word or digit is used, except as noted below. And if the result of substitution on word produces more than one filename, redirection occurs for each separate filename in turn. (yes, all of them, one at a time, she's thorough ✨)

The actual operators 🫶

< word — opens word for reading as standard input. Opening a file this way when it doesn't exist is an error, no cap 😭
<> word — opens word for reading and writing as standard input. If the file doesn't exist, it gets created 🎀
> word — opens word for writing as standard output. If it doesn't exist, it's created. If it does exist and the CLOBBER option is unset, this is an error; otherwise it gets truncated to zero length 💀
>| word / >! word — same as >, except the file is truncated to zero length if it exists, regardless of CLOBBER. She does what she wants 🔥
>> word — opens word for writing in append mode as standard output. If the file doesn't exist and both CLOBBER and APPEND_CREATE options are unset, that's an error; otherwise the file gets created 🌱
>>| word / >>! word — same as >>, except the file is created if it doesn't exist, regardless of CLOBBER and APPEND_CREATE 👑

Here-documents, here-strings, the whole family 🧚

<<[-] word — the shell input is read up to a line that is the same as word, or to an end-of-file. No parameter expansion, command substitution or filename generation is performed on word. The resulting document, called a here-document, becomes the standard input.

If any character of word is quoted with single or double quotes or a \, no interpretation is placed upon the characters of the document. Otherwise, parameter and command substitution occurs, \ followed by a newline is removed, and \ must be used to quote the characters \, $, ` and the first character of word.

Note that word itself does not undergo shell expansion. Backquotes in word don't have their usual effect; instead they behave similarly to double quotes, except that the backquotes themselves are passed through unchanged. (This info is given for completeness and it is not recommended that backquotes be used 🙈) Quotes in the form $'...' have their standard effect of expanding backslashed references to special characters.

If <<- is used, then all leading tabs are stripped from word and from the document. Tabs only, bestie, don't come for me about spaces ✋

<<< word — performs shell expansion on word and passes the result to standard input. This is known as a here-string 🎀 Compare the use of word in here-documents above, where word does not undergo shell expansion. The result will have a trailing newline after it, always 💖

Duplicating, closing, coprocess shenanigans 🦋

<& number / >& number — standard input/output is duplicated from file descriptor number (see dup2(2)).
<& - / >& - — close the standard input/output. bye bestie 👋
<& p / >& p — the input/output from/to the coprocess is moved to the standard input/output.

Both stdout AND stderr at once, because you're extra 💅

>& word / &> word — (except where >& word matches one of the above syntaxes; &> can always be used to avoid this ambiguity). Redirects both standard output and standard error (file descriptor 2) in the manner of > word. Note that this does not have the same effect as > word 2>&1 in the presence of multios (see the section below). lowkey important, don't skip it 👀
>&| word / >&! word / &>| word / &>! word — redirects both standard output and standard error (file descriptor 2) in the manner of >| word 🔥
>>& word / &>> word — redirects both standard output and standard error (file descriptor 2) in the manner of >> word 🌱
>>&| word / >>&! word / &>>| word / &>>! word — redirects both standard output and standard error (file descriptor 2) in the manner of >>| word 👑

Digits, order, and why it matters 😤

If one of the above is preceded by a digit, then the file descriptor referred to is the one specified by that digit instead of the default 0 or 1.

The order in which redirections are specified is significant. The shell evaluates each redirection in terms of the (file descriptor, file) association at the time of evaluation. For example:

... 1>fname 2>&1

first associates file descriptor 1 with file fname. It then associates file descriptor 2 with the file associated with file descriptor 1 (that is, fname). If the order of redirections were reversed, file descriptor 2 would be associated with the terminal (assuming file descriptor 1 had been) and then file descriptor 1 would be associated with file fname. It's giving "sequence is the whole plot" 🎬

The |& command separator described in Simple Commands & Pipelines in zshmisc(1) is a shorthand for 2>&1 |.

The various forms of process substitution, <(list) and =(list) for input and >(list) for output, are often used together with redirection. For example, if word in an output redirection is of the form >(list) then the output is piped to the command represented by list. See Process Substitution in zshexpn(1). iconic combo honestly 🌈

🎀 OPENING FILE DESCRIPTORS USING PARAMETERS 🎀

Okay this part is genuinely so slay. When the shell is parsing arguments to a command, and the shell option IGNORE_BRACES is not set, a different form of redirection is allowed: instead of a digit before the operator, there's a valid shell identifier enclosed in braces.

The shell will open a new file descriptor that is guaranteed to be at least 10, and set the parameter named by the identifier to the file descriptor opened. No whitespace is allowed between the closing brace and the redirection character 🚫 For example:

... {myfd}>&1

This opens a new file descriptor that is a duplicate of file descriptor 1 and sets the parameter myfd to the number of the file descriptor, which will be at least 10. The new file descriptor can be written to using the syntax >&$myfd. The file descriptor remains open in subshells and forked external executables 💖

The syntax {varid}>&-, for example {myfd}>&-, may be used to close a file descriptor opened in this fashion. Note that the parameter given by varid must previously be set to a file descriptor in this case.

It is an error to open or close a file descriptor in this fashion when the parameter is readonly. However, it is not an error to read or write a file descriptor using <&$param or >&$param if param is readonly. Read-only ≠ off-limits for reading, iykyk 💅

If the option CLOBBER is unset, it is an error to open a file descriptor using a parameter that is already set to an open file descriptor previously allocated by this mechanism. Unsetting the parameter before using it for allocating a file descriptor avoids the error. (little pro move right there ✨)

Note that this mechanism merely allocates or closes a file descriptor; it does not perform any redirections from or to it. It's usually convenient to allocate a file descriptor prior to use as an argument to exec. The syntax does not in any case work when used around complex commands such as parenthesised subshells or loops, where the opening brace is interpreted as part of a command list to be executed in the current shell. so don't try it there bestie 😭

The following shows a typical sequence of allocation, use, and closing of a file descriptor:

integer myfd
exec {myfd}>~/logs/mylogfile.txt
print This is a log message. >&$myfd
exec {myfd}>#-

Wait — correction, the closing line is exactly:

exec {myfd}>&-

Note that the expansion of the variable in the expression >&$myfd occurs at the point the redirection is opened. This is after the expansion of command arguments and after any redirections to the left on the command line have been processed. Timing is everything, periodt ⏰

🦋 MULTIOS 🦋

ok so bestie, here's the tea on multios — if you try to open a file descriptor for writing more than once, the shell doesn't just give up, it opens that file descriptor as a pipe to a process that copies its input to all the outputs you specified. it's giving tee energy, fr. this happens provided the MULTIOS option is set — and it is, by default, so you're already living that life. thus:

date >foo >bar

writes the date to two whole files, named foo and bar. iconic. ✨ and note — a pipe is an implicit redirection, so:

date >foo | cat

writes the date into the file foo, and also pipes it to cat. both. at once. main character energy.

ngl, quick heads up: the shell opens all the files to be used in the multio process immediately, not at the point they're about to be written. so like, no lazy-loading here 🙈

also note that redirections are always expanded in order. this is true regardless of what MULTIOS is set to, but with the option in effect there are extra consequences, iykyk. for example, the meaning of >&1 changes after a previous redirection:

date >&1 >output

in the case above, the >&1 refers to the standard output at the start of the line; the result is similar to the tee command. slay. but consider:

date >output >&1

since redirections are evaluated in order, when the >&1 is encountered the standard output is already set to the file output, so another copy of the output gets sent to that file. lowkey this is unlikely to be what you intended 💀

if the MULTIOS option is set, the word after a redirection operator is also subjected to filename generation (globbing). thus:

: > *

will truncate all files in the current directory, assuming there's at least one. 😭 (without the MULTIOS option, it would create an empty file called *.) so like... maybe don't run that one casually, bestie. similarly, you can do:

echo exit 0 >> *.sh

now flip it: if you try to open a file descriptor for reading more than once, and MULTIOS is set, the shell opens it as a pipe to a process that copies all the specified inputs to its output in the order specified. 📚

it should be noted that each file is opened immediately, not at the point where it's about to be read — this behaviour differs from cat, so if strictly standard behaviour is needed, cat should be used instead. no cap, that's the actual rule.

thus:

sort <foo <fubar

or even:

sort <f{oo,ubar}

is equivalent to cat foo fubar | sort. cute.

expansion of the redirection argument occurs at the point the redirection is opened, at the point described above for the expansion of the variable in >&$myfd.

note that a pipe is an implicit redirection; thus:

cat bar | sort <foo

is equivalent to cat bar foo | sort — and yes, mind the order of the inputs 👑

if the MULTIOS option is unset, each redirection replaces the previous redirection for that file descriptor. however, all files redirected to are actually opened, so:

echo Hello > bar > baz

when MULTIOS is unset will truncate bar, and write Hello into baz. sneaky.

⚠️ okay here's a genuine bug-shaped problem, pay attention: there's a problem when an output multio is attached to an external program. a simple example shows this:

cat file >file1 >file2
cat file1 file2

here, it's possible that the second cat will not display the full contents of file1 and file2 (i.e. the original contents of file repeated twice). 💀

the reason? the multios are spawned after the cat process is forked from the parent shell, so the parent shell does not wait for the multios to finish writing data. this means the command as shown can exit before file1 and file2 are completely written. as a workaround, it's possible to run the cat process as part of a job in the current shell:

{ cat file } >file >file2

here, the {...} job will pause to wait for both files to be written. problem solved, we love a fix 🫶

🌸 REDIRECTIONS WITH NO COMMAND 🌸

when a simple command consists of one or more redirection operators and zero or more parameter assignments, but no command name, zsh can behave in several ways. pick your fighter:

if the parameter NULLCMD is not set or the option CSH_NULLCMD is set, an error is caused. this is the csh behavior and CSH_NULLCMD is set by default when emulating csh. 😬
if the option SH_NULLCMD is set, the builtin : is inserted as a command with the given redirections. this is the default when emulating sh or ksh.
otherwise, if the parameter NULLCMD is set, its value will be used as a command with the given redirections. and if both NULLCMD and READNULLCMD are set, then the value of the latter will be used instead of that of the former when the redirection is an input. 👀

the default for NULLCMD is cat and for READNULLCMD is more. thus:

< file

shows the contents of file on standard output, with paging if that's a terminal. cute! and yes, NULLCMD and READNULLCMD may refer to shell functions. flexible queen.

👑 COMMAND EXECUTION 👑

if a command name contains no slashes, the shell attempts to locate it. here's the whole search order, bestie:

if there exists a shell function by that name, the function is invoked as described in the section Functions. 🎀
if there exists a shell builtin by that name, the builtin is invoked.
otherwise, the shell searches each element of $path for a directory containing an executable file by that name. 🔍

if execution fails: an error message is printed, and one of the following values is returned.

127 — the search was unsuccessful. the error message is command not found: cmd. 😭
126 — the executable file has insufficient permissions, is a directory or special file, or is not a script and is in a format unrecognized by the operating system. the exact conditions and error message are operating system-dependent; see execve(2).

if execution fails because the file is not in executable format, and the file is not a directory, it's assumed to be a shell script. /bin/sh is spawned to execute it. and if the program is a file beginning with #!, the remainder of the first line specifies an interpreter for the program. the shell will execute the specified interpreter on operating systems that do not handle this executable format in the kernel. 🐚

if no external command is found but a function command_not_found_handler exists, the shell executes this function with all command line arguments. the return status of the function becomes the status of the command. ✨

heads up, this one's important: the handler is executed in a subshell forked to execute an external command, hence changes to directories, shell parameters, etc. have no effect on the main shell. so no sneaky state changes from in there, bestie 🙈

💜 FUNCTIONS 💜

shell functions are defined with the function reserved word or the special syntax funcname (). shell functions are read in and stored internally. alias names are resolved when the function is read. functions are executed like commands with the arguments passed as positional parameters. (see the section Command Execution.) 🫶

functions execute in the same process as the caller and share all files and present working directory with the caller. and a trap on EXIT set inside a function is executed after the function completes in the environment of the caller. the way that works is honestly so clean 🌈

the return builtin is used to return from function calls.

function identifiers can be listed with the functions builtin. functions can be undefined with the unfunction builtin. periodt. 💅

🦋 AUTOLOADING FUNCTIONS 🦋

Okay bestie so picture this: you can mark a function as undefined using the autoload builtin — or, same energy, functions -u, or typeset -fu. What does "undefined" mean? It means the function literally has no body. Nada. Empty. It's giving ghost.

Then the very first time you execute that function, the shell goes hunting for its actual definition using the elements of the fpath variable. So the classic setup for autoloading your lil' guys looks like this:

fpath=(~/myfuncs $fpath)
autoload myfunc1 myfunc2 ...

Now, the usual alias expansion that happens while reading get suppressed if the autoload builtin (or its equivalent) is given the -U option. This is recommended for functions that ship with the zsh distribution, fr fr. 💅 And heads up: for functions precompiled with the zcompile builtin command, the flag -U must be provided when the .zwc file is created, because that corresponding info gets compiled right into it. No cap.

For each element in fpath, the shell checks three possible files, and the newest one wins to load the definition:

element.zwc — a file made with the zcompile builtin command, expected to contain the definitions for all functions in the directory named element. It's treated exactly like a directory containing files for functions, and it gets searched for the definition of the function. If the definition isn't found there, the search keeps going with the other two possibilities below. Also: if element already includes a .zwc extension (aka you explicitly gave that extension yourself), then element is searched for the function's definition without comparing its age to other files — in fact, there doesn't even need to be a directory named element without the suffix. So dropping an element like /usr/local/funcs.zwc into fpath speeds up the search, with the glow-up cost that any functions inside must be explicitly recompiled by hand before the shell notices changes. 🙈
element/function.zwc — a file created with zcompile, expected to contain the definition for function. It may include other function definitions too, but those are neither loaded nor executed — a file found this way is searched only for the definition of function. Side characters stay side characters. 💀
element/function — a plain file of zsh command text, taken to be the definition for function.

So in summary bestie, the search order is: first, in the parents of directories in fpath, for the newer of either a compiled directory or a directory in fpath; second, if more than one of these contains a definition for the function you're seeking, the leftmost one in fpath gets chosen; and third, within a directory, the newer of either a compiled function or an ordinary function definition is used. 👑

Now for the ✨behaviour✨. If the KSH_AUTOLOAD option is set, or the file contains only a simple definition of the function, the file's contents get executed. This will normally define the function in question, but it may also do some initialization — and that initialization runs in the context of the function execution, so it may therefore define local parameters. It is an error if the function isn't defined by loading the file. 😭

Otherwise, the function body (with no surrounding funcname() {...}) is taken to be the complete contents of the file. This form is iconic because it lets the file be used directly as an executable shell script. 🫶 If processing the file results in the function being re-defined, the function itself is not re-executed. To force the shell to do initialization and then call the defined function, the file should contain initialization code (which gets executed and then discarded) in addition to a complete function definition (which gets retained for subsequent calls to the function), plus a call to the shell function — including any arguments — at the end.

For example, suppose the autoload file func contains:

func() { print This is func; }
print func is initialized

Then func; func with KSH_AUTOLOAD set produces both messages on the first call, but only the message This is func on the second and subsequent calls. Without KSH_AUTOLOAD set, it produces the initialization message on the first call, and the other message on the second and subsequent calls. Sneaky lil' difference, iykyk. 🎀

It's also possible to create a function that isn't marked as autoloaded but loads its own definition by searching fpath, by using autoload -X inside a shell function. For example, these two are equivalent:

myfunc() {
  autoload -X
}
myfunc args...

and

unfunction myfunc   # if myfunc was defined
autoload myfunc
myfunc args...

In fact, the functions command outputs builtin autoload -X as the body of an autoloaded function. Why? So that this:

eval "$(functions)"

produces a reasonable result. 🌈 A true autoloaded function can be identified by the presence of the comment # undefined in the body, because all comments get discarded from defined functions. Little forensic detail for you there. 🔍

And if you want to load the definition of an autoloaded function myfunc without executing myfunc, use:

autoload +X myfunc

🌸 ANONYMOUS FUNCTIONS 🌸

Okay so if no name is given for a function, it's anonymous and gets handled specially. Either form of function definition works: a () with no preceding name, or a function with an immediately following open brace. The function is executed immediately at the point of definition and is not stored for future use. Its name gets set to (anon). Main character energy, zero attachment. 💅

Arguments to the function may be specified as words following the closing brace that defines the function — hence if there are none, no arguments (other than $0) are set. And this is a difference from how other functions are parsed: normal function definitions may be followed by certain keywords such as else or fi, and those will be treated as arguments to anonymous functions, so you need a newline or semicolon to force keyword interpretation. 🙈

Also note that the argument list of any enclosing script or function is hidden (as would be the case for any other function called at this point).

Redirections may be applied to the anonymous function the same way as to a current-shell structure enclosed in braces. The main use of anonymous functions is to provide a scope for local variables — which is particularly convenient in start-up files, since those don't provide their own local variable scope. 🎀

For example,

variable=outside
function {
  local variable=inside
  print "I am $variable with arguments $*"
} this and that
print "I am $variable"

outputs the following:

I am inside with arguments this and that
I am outside

Watch out bestie: function definitions with arguments that expand to nothing — for example name=; function $name { ... } — are not treated as anonymous functions. Instead, they're treated as normal function definitions where the definition is silently discarded. 💀 Poof. Gone.

💅 SPECIAL FUNCTIONS 💅

Okay so bestie, listen — some functions are just built different. Like, if you define them, the shell literally treats them as having special meaning. Main character functions only. 👑

🪝 Hook Functions

So here's the vibe: for each of the functions below, you can also define an array that has the same name as the function with _functions glued onto the end. Every element in that array gets taken as the name of a function to execute — and it runs in the same context, with the same arguments, and the same initial value of $? as the basic function itself. No cap.

Example time: if $chpwd_functions is an array containing the values mychpwd and chpwd_save_dirstack, then the shell tries to execute the functions chpwd, mychpwd and chpwd_save_dirstack, in that order. Slay.

Any function that doesn't exist? Silently ignored. No drama, no error, it just moves on. A function found by this mechanism is what the rest of the docs call a hook function. ✨

⚠️ Careful though: an error in any function causes the subsequent functions not to be run. And heads up — an error in a precmd hook causes an immediately following periodic function not to run (though it may run at the next opportunity). So like, don't be messy in your hooks, bestie. 😭

chpwd — Executed whenever the current working directory is changed. New dir, who dis? 🦋
periodic — If the parameter PERIOD is set, this function is executed every $PERIOD seconds, just before a prompt. Note that if multiple functions are defined using the array periodic_functions, only one period is applied to the complete set of functions, and the scheduled time is not reset if the list of functions is altered. Hence the set of functions is always called together. Group project energy — everybody shows up at once. 👯‍♀️
precmd — Executed before each prompt. Note that precommand functions are not re-executed simply because the command line is redrawn, as happens, for example, when a notification about an exiting job is displayed. So it's not re-running just 'cause your screen refreshed, iykyk. 🌸
preexec — Executed just after a command has been read and is about to be executed. If the history mechanism is active (regardless of whether the line was discarded from the history buffer), the string that the user typed is passed as the first argument — otherwise it's an empty string. The actual command that will be executed (including expanded aliases) is passed in two different forms:
- the second argument is a single-line, size-limited version of the command (with things like function bodies elided)
- the third argument contains the full text that is being executed

Three arguments, three vibes. 💅

zshaddhistory — Executed when a history line has been read interactively, but before it is executed. The sole argument is the complete history line (so that any terminating newline will still be present). Don't forget that newline is right there. 🎀

If any of the hook functions returns status 1 (or any non-zero value other than 2, though this is not guaranteed for future versions of the shell), the history line will not be saved — although it lingers in the history until the next line is executed, allowing you to reuse or edit it immediately. Ghosting the history file but still in the group chat, basically. 👻

If any of the hook functions returns status 2, the history line will be saved on the internal history list, but not written to the history file. In case of a conflict, the first non-zero status value is taken. First come, first served, periodt.

A hook function may call fc -p ... to switch the history context so that the history is saved in a different file from that in the global HISTFILE parameter. This is handled specially: the history context is automatically restored after the processing of the history line is finished. Very demure, very mindful, cleans up after itself. 💖

The following example function works with one of the options INC_APPEND_HISTORY or SHARE_HISTORY set, in order that the line is written out immediately after the history entry is added. It first adds the history line to the normal history with the newline stripped, which is usually the correct behaviour. Then it switches the history context so that the line will be written to a history file in the current directory.

zshaddhistory() {
  print -sr -- ${1%%$'\n'}
  fc -p .zsh_local_history
}

Iconic. Copy-paste it, bestie. 🍓

zshexit — Executed at the point where the main shell is about to exit normally. This is not called by exiting subshells, nor when the exec precommand modifier is used before an external command. Also, unlike TRAPEXIT, it is not called when functions exit. So it's main-character-only, no understudies. 🎭

🪤 Trap Functions

The functions below are treated specially but — plot twist — they do not have corresponding hook arrays. No _functions array for these babies. 🙈

TRAPNAL — If defined and non-null, this function will be executed whenever the shell catches a signal SIGNAL, where NAL is a signal name as specified for the kill builtin. The signal number will be passed as the first parameter to the function. 📟

If a function of this form is defined and null, the shell and processes spawned by it will ignore SIGNAL. Null = "we don't see it, it's not there." 💀

The return status from the function is handled specially:

If it is zero, the signal is assumed to have been handled, and execution continues normally. ✨
Otherwise, the shell will behave as interrupted except that the return status of the trap is retained.

Programs terminated by uncaught signals typically return the status 128 plus the signal number. Hence the following causes the handler for SIGINT to print a message, then mimic the usual effect of the signal:

TRAPINT() {
  print "Caught SIGINT, aborting."
  return $(( 128 + $1 ))
}

The functions TRAPZERR, TRAPDEBUG and TRAPEXIT are never executed inside other traps. Nested traps? Not on their watch. 🚫

TRAPDEBUG — If the option DEBUG_BEFORE_CMD is set (as it is by default), executed before each command; otherwise executed after each command. See the description of the trap builtin in zshbuiltins(1) for details of additional features provided in debug traps. Underrated function ngl, the debugging queen. 🔥
TRAPEXIT — Executed when the shell exits, or when the current function exits if defined inside a function. The value of $? at the start of execution is the exit status of the shell or the return status of the function exiting. It sees everything on the way out. 👑
TRAPZERR — Executed whenever a command has a non-zero exit status. However, the function is not executed if the command occurred in a sublist followed by && or || — only the final command in a sublist of this type causes the trap to be executed. So && and || chains keep it quiet until the very last one, bestie. 🤫

The function TRAPERR acts the same as TRAPZERR on systems where there is no SIGERR (this is the usual case).

The functions beginning TRAP may alternatively be defined with the trap builtin — this may be preferable for some uses. Setting a trap with one form removes any trap of the other form for the same signal; removing a trap in either form removes all traps for the same signal. One in, all out. 💅

The forms

TRAPNAL() {
 # code
}

(known as 'function traps') and

trap '
 # code
' NAL

(known as 'list traps') are equivalent in most ways — the exceptions being the following:

Function traps have all the properties of normal functions, appearing in the list of functions and being called with their own function context rather than the context where the trap was triggered. They get their own little world. 🌈
The return status from function traps is special, whereas a return from a list trap causes the surrounding context to return with the given status. Different rules for different traps, iykyk. ⭐
Function traps are not reset within subshells, in accordance with zsh behaviour; list traps are reset, in accordance with POSIX behaviour. Zsh stays zsh-ing, POSIX stays POSIX-ing. 🧚

👯 JOBS 👯

okay so if the MONITOR option is set, an interactive shell pairs a job with each pipeline. it keeps a whole table of current jobs — that's the one jobs prints for you — and hands out cute little integer numbers to each one. so when a job starts asynchronously with &, the shell prints a line to standard error that looks like this:

[1] 1234

which is literally saying: that asynchronously-started job is job number 1, and it had one (top-level) process, whose process ID was 1234. iconic little receipt 🧾✨

here's the tea though — if a job is started with &| or &!, that job is immediately disowned. bestie it does not even get a seat at the table: after startup it has no place in the job table, and it's not subject to any of the job control features described in this section. it's giving ghost 👻

🫶 the whole ^Z situation

if you're running a job and you wanna do something else, you may hit the key ^Z (control-Z), which sends a TSTP signal to the current job. (this key may be redefined by the susp option of the external stty command, fyi.) the shell will then normally indicate that the job has been suspended, and print another prompt. slay, you have your terminal back 💅

from there you can manipulate the state of this job — put it in the background with the bg command, or run some other commands and then eventually bring the job back into the foreground with the foreground command fg. a ^Z takes effect immediately and is like an interrupt in that pending output and unread input are discarded when it is typed. no take-backsies 🗑️

also: a job being run in the background will suspend if it tries to read from the terminal. lowkey rude of it but okay.

😭 the shell-function fork drama (this one matters)

note that if the job running in the foreground is a shell function, then suspending it will have the effect of causing the shell to fork. this is necessary to separate the function's state from that of the parent shell performing the job control, so that the latter can return to the command line prompt. it's a whole custody arrangement fr fr 🧸

as a result, even if fg is used to continue the job, the function will no longer be part of the parent shell, and any variables set by the function will not be visible in the parent shell. thus the behaviour is different from the case where the function was never suspended. and ngl, zsh is different from many other shells in this regard — that's not a bug bestie, that's the vibe.

one additional side effect: use of disown with a job created by suspending shell code in this fashion is delayed — the job can only be disowned once any process started from the parent shell has terminated. at that point, the disowned job disappears silently from the job list. poof. no ceremony. 🎀

the same behaviour is found when the shell is executing code as the right hand side of a pipeline or any complex shell construct such as if, for, etc., in order that the entire block of code can be managed as a single job. main character block energy 👑

🌸 background output n stuff

background jobs are normally allowed to produce output, but this can be disabled by giving the command stty tostop. if you set this tty option, then background jobs will suspend when they try to produce output — like they do when they try to read input. equal opportunity suspending 💀

when a command is suspended and continued later with the fg or wait builtins, zsh restores tty modes that were in effect when it was suspended. this (intentionally) does not apply if the command is continued via kill -CONT, nor when it is continued with bg. read that again, it's a real distinction 🔥

🎀 referring to jobs (the cheat sheet)

there are several ways to refer to jobs in the shell. a job can be referred to by the process ID of any process of the job, or by one of the following:

%number — the job with the given number.
%string — the last job whose command line begins with string.
%?string — the last job whose command line contains string.
%% — current job.
%+ — equivalent to %%.
%- — previous job.

memorise these, they're your besties 🫶✨

⭐ notifications & traps

the shell learns immediately whenever a process changes state. it normally informs you whenever a job becomes blocked so that no further progress is possible. if the NOTIFY option is not set, it waits until just before it prints a prompt before it informs you. all such notifications are sent directly to the terminal — not to the standard output or standard error. sneaky but correct 📬

when the monitor mode is on, each background job that completes triggers any trap set for CHLD. so like... your trap is gonna fire, that's the whole thing.

💀 trying to leave with jobs still running

when you try to leave the shell while jobs are running or suspended, you will be warned that:

You have suspended (running) jobs

you may use the jobs command to see what they are. if you do this, or immediately try to exit again, the shell will not warn you a second time — the suspended jobs will be terminated, and the running jobs will be sent a SIGHUP signal, if the HUP option is set. so like, that second exit is a real one, no cap 😭

to avoid having the shell terminate the running jobs, either use the nohup(1) command or the disown builtin. protect your children 🙏

🚨 SIGNALS 🚨

the INT and QUIT signals for an invoked command are ignored if the command is followed by & and the MONITOR option is not active. the shell itself always ignores the QUIT signal — that's just its personality. otherwise, signals have the values inherited by the shell from its parent (but see the TRAPNAL special functions in the section Functions — go look, iykyk).

certain jobs are run asynchronously by the shell other than those explicitly put into the background; even in cases where the shell would usually wait for such jobs, an explicit exit command or exit due to the option ERR_EXIT will cause the shell to exit without waiting. examples of such asynchronous jobs are process substitution (see the section PROCESS SUBSTITUTION in the zshexpn(1) manual page) and the handler processes for multios (see the section MULTIOS in the zshmisc(1) manual page). so yeah, sometimes the shell just dips, and that's valid ✌️✨

✨ ARITHMETIC EVALUATION ✨

okay so bestie, your shell can literally do math 💅 Integers and floating point, no cap. You've got two ways in: the builtin let, or a substitution shaped like $((...)). Pick your fighter 🫶

For integers, the shell is usually compiled to use 8-byte precision when that's available — otherwise it's 4 bytes. Wanna test it? Run this:

print - $(( 12345678901 ))

If the number comes back looking exactly the same, congrats, your precision is at least 8 bytes 💖 Floating point arithmetic always uses the double type, with whatever corresponding precision your compiler and library happen to provide.

The let builtin command takes arithmetic expressions as arguments, and each one is evaluated separately. Now, since many of the arithmetic operators (plus spaces!) require quoting, there's a second form on offer: for any command that begins with a ((, every character up until a matching )) is treated as a double-quoted expression, and arithmetic expansion is performed just like it would be for an argument of let. To be precise — ((...)) is equivalent to let "...". The return status is 0 if the arithmetic value of the expression is non-zero, 1 if it's zero, and 2 if an error went down 🙈

So like, this statement:

(( val = 2 + 1 ))

is equivalent to:

let "val = 2 + 1"

and both of them assign the value 3 to the shell variable val and return a zero status. slay ✨

Integers can live in bases other than 10, bestie 🌈 A leading 0x or 0X means hexadecimal, and a leading 0b or 0B means binary. Integers may also take the form base#n, where base is a decimal number between two and thirty-six representing the arithmetic base, and n is a number in that base — for example, 16#ff is 255 in hexadecimal. The base# part may also be omitted, in which case base 10 is used. And for backwards compatibility, the form [base]n is also accepted 👑

An integer expression, or a base given in the form base#n, may contain underscores (_) after the leading digit for visual guidance — these are just ignored in computation. Examples: 1_000_000 or 0xffff_ffff, which are equivalent to 1000000 and 0xffffffff respectively. cute AND readable, we love that 🎀

You can also specify a base to be used for output, in the form [#base] — like [#16] for example. This gets used when outputting arithmetical substitutions or when assigning to scalar parameters, but an explicitly defined integer or floating point parameter will not be affected. If an integer variable is implicitly defined by an arithmetic expression, any base specified this way gets set as that variable's output arithmetic base, exactly as if the option -i base to the typeset builtin had been used. Heads up: the expression has no precedence, and if it occurs more than once in a mathematical expression, the last one encountered wins. For clarity it's recommended you put it at the beginning of an expression 💜

As an example:

typeset -i 16 y
print $(( [#8] x = 32, y = 32 ))
print $x $y

outputs first 8#40 — the rightmost value in the given output base — and then 8#40 16#20, because y has been explicitly declared to have output base 16, while x (assuming it doesn't already exist) is implicitly typed by the arithmetic evaluation, where it acquires the output base 8. iykyk 🧚

The base may be replaced or followed by an underscore, which may itself be followed by a positive integer (if it's missing, the value 3 is used). This indicates that underscores should be inserted into the output string, grouping the number for visual clarity. That following integer specifies how many digits to group together. Like so:

setopt cbases
print $(( [#16_4] 65536 ** 2 ))

outputs 0x1_0000_0000 🔥

This feature also works with floating point numbers, in which case the base must be omitted — grouping goes away from the decimal point. For example:

zmodload zsh/mathfunc
print $(( [#_] sqrt(1e7) ))

outputs 3_162.277_660_168_379_5 (the number of decimal places shown may vary, ngl).

If the C_BASES option is set, hexadecimal numbers are output in the standard C format — for example 0xFF instead of the usual 16#FF. And if the option OCTAL_ZEROES is also set (it is not by default, jsyk), octal numbers get treated similarly and hence appear as 077 instead of 8#77. This option has no effect on the output of bases other than hexadecimal and octal, and these formats are always understood on input 👑

When an output base is specified using the [#base] syntax, an appropriate base prefix will be output if necessary, so that the value output is valid syntax for input. If the # is doubled — for example [##16] — then no base prefix is output. noted 📝

Floating point constants are recognized by the presence of a decimal point or an exponent. The decimal point may be the first character of the constant, but the exponent character e or E may not be, because it'll get taken for a parameter name 😭 All numeric parts (before and after the decimal point and in the exponent) may contain underscores after the leading digit for visual guidance — those are ignored in computation.

An arithmetic expression uses nearly the same syntax and associativity of expressions as in C. periodt 💅

🌸 the operators (native mode) 🌸

In the native mode of operation, the following operators are supported, listed in decreasing order of precedence:

+ - ! ~ ++ -- — unary plus/minus, logical NOT, complement, `{pre,post}{in,de}crement
<< >> — bitwise shift left, right
& — bitwise AND
^ — bitwise XOR
| — bitwise OR
** — exponentiation
* / % — multiplication, division, modulus (remainder)
+ - — addition, subtraction
< > <= >= — comparison
== != — equality and inequality
&& — logical AND
|| ^^ — logical OR, XOR
? : — ternary operator
= += -= *= /= %= &= ^= |= <<= >>= &&= ||= ^^= **= — assignment
, — comma operator

The operators &&, ||, &&=, and ||= are short-circuiting, and only one of the latter two expressions in a ternary operator is evaluated. Also note the precedence of the bitwise AND, OR, and XOR operators 👀

🦋 with `C_PRECEDENCES` set 🦋

With the option C_PRECEDENCES, the precedences (but no other properties) of the operators are altered to be the same as those in most other languages that support the relevant operators:

+ - ! ~ ++ -- — unary plus/minus, logical NOT, complement, `{pre,post}{in,de}crement
** — exponentiation
* / % — multiplication, division, modulus (remainder)
+ - — addition, subtraction
<< >> — bitwise shift left, right
< > <= >= — comparison
== != — equality and inequality
& — bitwise AND
^ — bitwise XOR
| — bitwise OR
&& — logical AND
^^ — logical XOR
|| — logical OR
? : — ternary operator
= += -= *= /= %= &= ^= |= <<= >>= &&= ||= ^^= **= — assignment
, — comma operator

Note that the precedence of exponentiation in both cases is below that of unary operators, hence -3**2 evaluates as 9, not -9 💀 Use parentheses where necessary: -(3**2). This is for compatibility with other shells, so don't come for me, come for them.

🧮 math functions 🧮

Mathematical functions can be called with the syntax func(args), where the function decides if the args is used as a string or as a comma-separated list of arithmetic expressions. The shell currently defines no mathematical functions by default, but the module zsh/mathfunc may be loaded with the zmodload builtin to provide standard floating point mathematical functions 🫶

An expression of the form ##x, where x is any character sequence such as a, ^A, or \M-\C-x, gives the value of that character. And an expression of the form #name gives the value of the first character of the contents of the parameter name. Character values are according to the character set used in the current locale; for multibyte character handling the option MULTIBYTE must be set. Note that this form is different from $#name, the standard parameter substitution which gives the length of the parameter name. Also, #' is accepted instead of ##`, but its use is deprecated (so like, don't 🙈).

Named parameters and subscripted arrays can be referenced by name inside an arithmetic expression without using the parameter expansion syntax. For example:

((val2 = val1 * 2))

assigns twice the value of $val1 to the parameter named val2 ✨

An internal integer representation of a named parameter can be specified with the integer builtin. Arithmetic evaluation is performed on the value of each assignment to a named parameter declared integer in this manner. And assigning a floating point number to an integer results in rounding towards zero 👑

Likewise, floating point numbers can be declared with the float builtin. There are two types, differing only in their output format, as described for the typeset builtin. The output format can be bypassed by using arithmetic substitution instead of parameter substitution — i.e. ${float} uses the defined format, but $((float)) uses a generic floating point format 💅

Promotion of integer to floating point values is performed where necessary. In addition, if any operator which requires an integer (&, |, ^, <<, >> and their equivalents with assignment) is given a floating point argument, it will be silently rounded towards zero — except for ~, which rounds down. sneaky but ok 😭

💀 the warning bit, pls read 💀

Users should beware that, in common with many other programming languages but not software designed for calculation, the evaluation of an expression in zsh is taken a term at a time, and promotion of integers to floating point does not occur in terms only containing integers. A typical result of this is that a division such as 6/8 is truncated — in this case being rounded towards 0. The FORCE_FLOAT shell option can be used in scripts or functions where floating point evaluation is required throughout.

Scalar variables can hold integer or floating point values at different times; there's no memory of the numeric type in this case.

If a variable is first assigned in a numeric context without previously being declared, it will be implicitly typed as integer or float and retain that type either until the type is explicitly changed or until the end of the scope. This can have unforeseen consequences 🚨 For example, in the loop:

for (( f = 0; f < 1; f += 0.1 )); do
# use $f
done

if f has not already been declared, the first assignment will cause it to be created as an integer, and consequently the operation f += 0.1 will always cause the result to be truncated to zero, so that the loop will fail 😭 A simple fix would be to turn the initialization into f = 0.0. It's therefore best to declare numeric variables with explicit types, bestie. trust me on this one 🫶

💅 CONDITIONAL EXPRESSIONS 💅

So like, a conditional expression is what you use with the [[ compound command to check out file attributes and compare strings — it's giving "detective mode" fr fr. Every expression gets built out of one or more of these unary or binary cuties:

-a file — true if file exists, periodt.
-b file — true if file exists AND is a block special file.
-c file — true if file exists AND is a character special file.
-d file — true if file exists AND is a directory.
-e file — true if file exists. (yeah, same vibe as -a, iconic of them honestly)
-f file — true if file exists AND is a regular file.
-g file — true if file exists AND has its setgid bit set 👑
-h file — true if file exists AND is a symbolic link.
-k file — true if file exists AND has its sticky bit set 🍓
-n string — true if the length of string is non-zero.
-o option — true if the option named option is on. option can be a single character, in which case it's a single letter option name. (peep the section Specifying Options for the deets.)
- When no option named option exists, and the POSIX_BUILTINS option hasn't been set, it returns 3 with a warning 🙈. If that option is set, it returns 1 with no warning. Different vibes, same result-ish.
-p file — true if file exists AND is a FIFO special file (aka a named pipe).
-r file — true if file exists AND is readable by the current process.
-s file — true if file exists AND has size greater than zero.
-t fd — true if file descriptor number fd is open and associated with a terminal device. ⚠️ note: fd is not optional, bestie, don't skip it.
-u file — true if file exists AND has its setuid bit set.
-v varname — true if the shell variable varname is set.
-w file — true if file exists AND is writable by the current process.
-x file — true if file exists AND is executable by the current process. Special lil case: if file exists and is a directory, then the current process has permission to search in the directory. iykyk.
-z string — true if the length of string is zero.
-L file — true if file exists AND is a symbolic link.
-O file — true if file exists AND is owned by the effective user ID of this process.
-G file — true if file exists AND its group matches the effective group ID of this process.
-S file — true if file exists AND is a socket.
-N file — true if file exists AND its access time is not newer than its modification time.
file1 -nt file2 — true if file1 exists AND is newer than file2 ✨
file1 -ot file2 — true if file1 exists AND is older than file2.
file1 -ef file2 — true if file1 and file2 exist AND refer to the same file.
string = pattern / string == pattern — true if string matches pattern. These two forms are exactly equivalent 💯. The = form is the traditional shell syntax (and hence the only one generally used with the test and [ builtins); the == form gives you compatibility with other sorts of computer language. Both slay.
string != pattern — true if string does not match pattern.
string =~ regexp — true if string matches the regular expression regexp. If the option RE_MATCH_PCRE is set, regexp is tested as a PCRE regular expression using the zsh/pcre module, else it's tested as a POSIX extended regular expression using the zsh/regex module. Upon successful match, some variables get updated; no variables are changed if the matching fails. 😤
- If the option BASH_REMATCH is not set, the scalar parameter MATCH is set to the substring that matched the pattern, and the integer parameters MBEGIN and MEND to the index of the start and end, respectively, of the match in string — such that if string is contained in variable var, the expression `${var[$MBEGIN,$MEND]}' is identical to `$MATCH'. The setting of the option KSH_ARRAYS is respected. Likewise, the array match is set to the substrings that matched parenthesised subexpressions, and the arrays mbegin and mend to the indices of the start and end positions, respectively, of the substrings within string. The arrays are not set if there were no parenthesised subexpressions. For example, if the string a short string is matched against the regular expression s(...)t, then (assuming the option KSH_ARRAYS is not set) MATCH, MBEGIN and MEND are short, 3 and 7, respectively, while match, mbegin and mend are single entry arrays containing the strings hor, 4 and 6, respectively. Obsessed with how precise this is ngl.
- If the option BASH_REMATCH is set, the array BASH_REMATCH is set to the substring that matched the pattern followed by the substrings that matched parenthesised subexpressions within the pattern.
string1 < string2 — true if string1 comes before string2 based on the ASCII value of their characters.
string1 > string2 — true if string1 comes after string2 based on the ASCII value of their characters.
exp1 -eq exp2 — true if exp1 is numerically equal to exp2. 💡 note: for purely numeric comparisons, using the ((...)) builtin described in the section ARITHMETIC EVALUATION is more convenient than conditional expressions. Just saying!
exp1 -ne exp2 — true if exp1 is numerically not equal to exp2.
exp1 -lt exp2 — true if exp1 is numerically less than exp2.
exp1 -gt exp2 — true if exp1 is numerically greater than exp2.
exp1 -le exp2 — true if exp1 is numerically less than or equal to exp2.
exp1 -ge exp2 — true if exp1 is numerically greater than or equal to exp2.
( exp ) — true if exp is true. Parentheses are your besties for grouping 🫶
! exp — true if exp is false. The hater operator, love that for us.
exp1 && exp2 — true if exp1 and exp2 are both true.
exp1 || exp2 — true if either exp1 or exp2 is true.

For compatibility, if there's a single argument that is not syntactically significant — typically a variable — the condition is treated as a test for whether the expression expands as a string of non-zero length. In other words, [[ $var ]] is the same as [[ -n $var ]]. But like, it's recommended that the second, explicit form be used where possible. Be clear, bestie 💅

Normal shell expansion is performed on the file, string and pattern arguments, but the result of each expansion is constrained to be a single word — similar to the effect of double quotes.

Filename generation is not performed on any form of argument to conditions. However! It can be forced in any case where normal shell expansion is valid, and when the option EXTENDED_GLOB is in effect, by using an explicit glob qualifier of the form (#q) at the end of the string. A normal glob qualifier expression may appear between the q and the closing parenthesis; if none appears, the expression has no effect beyond causing filename generation. The results of filename generation are joined together to form a single word, as with the results of other forms of expansion. 🦋

This special use of filename generation is only available with the [[ syntax. If the condition occurs within the [ or test builtin commands, then globbing occurs instead as part of normal command line expansion before the condition is evaluated. In this case it may generate multiple words, which are likely to confuse the syntax of the test command. 💀 Consider yourself warned.

For example,

[[ -n file*(#qN) ]]

produces status zero if and only if there's at least one file in the current directory beginning with the string file. The globbing qualifier N ensures that the expression is empty if there's no matching file. Iconic.

Pattern metacharacters are active for the pattern arguments; the patterns are the same as those used for filename generation — see zshexpn(1) — but there's no special behaviour of / nor initial dots, and no glob qualifiers are allowed.

In each of the above expressions, if file is of the form /dev/fd/n, where n is an integer, then the test is applied to the open file whose descriptor number is n, even if the underlying system doesn't support the /dev/fd directory. Sneaky but slay.

In the forms which do numeric comparison, the expressions exp undergo arithmetic expansion as if they were enclosed in $((...)).

For example, the following:

[[ ( -f foo || -f bar ) && $report = y* ]] && print File exists.

tests if either file foo or file bar exists, and if so, if the value of the parameter report begins with y; if the complete condition is true, the message File exists. is printed. Chef's kiss 🤌

✨ EXPANSION OF PROMPT SEQUENCES ✨

Prompt sequences undergo a special form of expansion. This type of expansion is also available using the -P option to the print builtin. 🎀

If the PROMPT_SUBST option is set, the prompt string is first subjected to parameter expansion, command substitution and arithmetic expansion. See zshexpn(1).

Certain escape sequences may be recognised in the prompt string.

If the PROMPT_BANG option is set, a ! in the prompt is replaced by the current history event number. A literal ! may then be represented as !!. 🫶

If the PROMPT_PERCENT option is set, certain escape sequences that start with % are expanded. Many escapes are followed by a single character, although some of these take an optional integer argument that should appear between the % and the next character of the sequence. More complicated escape sequences are available to provide conditional expansion. Main character energy, honestly. 🌈

✨ SIMPLE PROMPT ESCAPES ✨

okay bestie, this is the part where we learn all the little %-thingies you can drop into your prompt 💅 every single one of them, no skipping, let's go 🎀

💖 Special characters

%% — a literal %, that's it, that's the whole vibe. 🌸
%) — a literal ). iconic, tiny, perfect.

%l — the line (tty) the user is logged in on, without the /dev/ prefix. and like, if the name starts with /dev/tty, that prefix gets stripped too. 🫶
%M — the full machine hostname. the whole thing, no cuts.
%m — the hostname up to the first .. and listen, you can put an integer after the % to say how many components of the hostname you want ✨ with a negative integer, you get the trailing components instead. love that for us.
%n — $USERNAME. hi bestie 👋
%y — the line (tty) the user is logged in on, without the /dev/ prefix — but this one does not treat /dev/tty names specially. so yeah, slightly different energy from %l, keep that in mind 😌

💜 Shell state

%# — a # if the shell is running with privileges, a % if not. it's equivalent to %(!.#.%%). and the definition of "privileged" here? either the effective user ID is zero, or — if POSIX.1e capabilities are supported — at least one capability is raised in either the Effective or Inheritable capability vectors. yes that's a mouthful, no I'm not paraphrasing it away 🧚
%? — the return status of the last command executed just before the prompt. your little status gossip reporter.
%_ — the status of the parser, i.e. the shell constructs (like if and for) that have been started on the command line. if you give it an integer number, that many strings get printed; zero or negative or no integer means print as many as there are. most useful in prompts PS2 for continuation lines and PS4 for debugging with the XTRACE option — and in that latter case it'll also work non-interactively. slay 🔥
%^ — the status of the parser in reverse. same as %_ other than the order of strings. it's often used in RPS2. 🦋
%d, %/ — current working directory. if an integer follows the %, it specifies a number of trailing components of the current working directory to show; zero means the whole path. a negative integer specifies leading components, i.e. %-1d specifies the first component. ✨
%~ — same as %d and %/, but if the current working directory starts with $HOME, that part gets replaced by a ~. furthermore, if it has a named directory as its prefix, that part is replaced by a ~ followed by the name of the directory — but only if the result is shorter than the full path. see Dynamic and Static named directories in zshexpn(1). 🎀
%e — evaluation depth of the current sourced file, shell function, or eval. it gets incremented or decremented every time the value of %N is set or reverted to a previous value, respectively. most useful for debugging as part of $PS4. 🫶
%h, %! — current history event number. both of them, same deal.
%i — the line number currently being executed in the script, sourced file, or shell function given by %N. most useful for debugging as part of $PS4. 💀
%I — the line number currently being executed in the file %x. similar to %i, but the line number is always a line number in the file where the code was defined — even if the code is a shell function. no cap.
%j — the number of jobs.
%L — the current value of $SHLVL.
%N — the name of the script, sourced file, or shell function that zsh is currently executing, whichever was started most recently. if there is none, this is equivalent to the parameter $0. an integer may follow the % to specify a number of trailing path components to show; zero means the full path. a negative integer specifies leading components. 👑
%x — the name of the file containing the source code currently being executed. behaves as %N except that function and eval command names are not shown — instead you get the file where they were defined. 😌
%c, %., %C — trailing component of the current working directory. an integer may follow the % to get more than one component. unless %C is used, tilde contraction is performed first. > ⚠️ these are deprecated — %c and %C are equivalent to %1~ and %1/, respectively, while explicit positive integers have the same effect as for the latter two sequences.

🌸 Date and time

%D — the date in yy-mm-dd format. 🗓️
%T — current time of day, in 24-hour format.
%t, %@ — current time of day, in 12-hour, am/pm format.
%* — current time of day in 24-hour format, with seconds.
%w — the date in day-dd format.
%W — the date in mm/dd/yy format.
%D{string} — string is formatted using the strftime function. see strftime(3) for more details. various zsh extensions provide numbers with no leading zero or space if the number is a single digit:
- %f — a day of the month
- %K — the hour of the day on the 24-hour clock
- %L — the hour of the day on the 12-hour clock

in addition, if the system supports the POSIX gettimeofday system call, %. provides decimal fractions of a second since the epoch with leading zeroes. by default three decimal places are provided, but a number of digits up to 9 may be given following the % — hence %6. outputs microseconds, and %9. outputs nanoseconds. (the latter requires a nanosecond-precision clock_gettime; systems lacking this will return a value multiplied by the appropriate power of 10.) a typical example of this is the format %D{%H:%M:%S.%.}. 🔥

the GNU extension %N is handled as a synonym for %9.. 🧚

additionally, the GNU extension that a - between the % and the format character causes a leading zero or space to be stripped is handled directly by the shell for the format characters d, f, H, k, l, m, M, S and y; any other format characters are provided to the system's strftime(3) with any leading - present, so the handling is system dependent. further GNU (or other) extensions are also passed to strftime(3) and may work if the system supports them. ⭐

🎀 Visual effects

%B (%b) — start (stop) boldface mode. 💅
%E — clear to end of line.
%U (%u) — start (stop) underline mode.
%S (%s) — start (stop) standout mode.
%F (%f) — start (stop) using a different foreground colour, if supported by the terminal. the colour may be specified two ways: either as a numeric argument, as normal, or by a sequence in braces following the %F, for example %F{red}. in that latter case the values allowed are as described for the fg zle_highlight attribute; see Character Highlighting in zshzle(1). this means numeric colours are allowed in the second format also. 🌈
%K (%k) — start (stop) using a different bacKground colour. syntax is identical to that for %F and %f. 🦋
%{...%} — include a string as a literal escape sequence. the string within the braces should not change the cursor position. brace pairs can nest. 🫶

a positive numeric argument between the % and the { is treated as described for %G below.

%G — within a %{...%} sequence, include a "glitch": that is, assume that a single character width will be output. this is useful when outputting characters that otherwise cannot be correctly handled by the shell, such as the alternate character set on some terminals. the characters in question can be included within a %{...%} sequence together with the appropriate number of %G sequences to indicate the correct width. an integer between the % and G indicates a character width other than one. hence %{seq%2G%} outputs seq and assumes it takes up the width of two standard characters. ✨

multiple uses of %G accumulate in the obvious fashion; the position of the %G is unimportant. negative integers are not handled. 💀

note: when prompt truncation is in use it's advisable to divide up output into single characters within each %{...%} group so that the correct truncation point can be found. 🙈

✨ CONDITIONAL SUBSTRINGS IN PROMPTS ✨

%v — the value of the first element of the psvar array parameter. Put an integer right after the % and you get that element of the array instead, bestie. Negative integers count from the end of the array. 💅
%(x.true-text.false-text) — okay so this is a whole ternary expression, like an if/else but for your prompt. The character that follows the x is totally arbitrary; that same character is what separates the text for the true result from the text for the false result. 🎀

Here's the catch: that separator can't show up in true-text, unless it's part of a %-escape sequence. And a ) can appear in false-text — you just write it as %). Both true-text and false-text may contain arbitrarily-nested escape sequences, including even more ternary expressions. It's ternaries all the way down, no cap. 🌀

The left parenthesis may be preceded or followed by a positive integer n, which defaults to zero. A negative integer gets multiplied by -1 — except as noted below for l. The test character x can be any of these:

! — True if the shell is running with privileges.
# — True if the effective uid of the current process is n.
? — True if the exit status of the last command was n.
_ — True if at least n shell constructs were started.
C, / — True if the current absolute path has at least n elements relative to the root directory, hence / is counted as 0 elements.
c, ., ~ — True if the current path, with prefix replacement, has at least n elements relative to the root directory, hence / is counted as 0 elements.
D — True if the month is equal to n (January = 0).
d — True if the day of the month is equal to n.
e — True if the evaluation depth is at least n.
g — True if the effective gid of the current process is n.
j — True if the number of jobs is at least n.
L — True if the SHLVL parameter is at least n.
l — True if at least n characters have already been printed on the current line. When n is negative, it's true if at least abs(n) characters remain before the opposite margin (so, the left margin for RPROMPT). 🫶
S — True if the SECONDS parameter is at least n.
T — True if the time in hours is equal to n.
t — True if the time in minutes is equal to n.
v — True if the array psvar has at least n elements.
V — True if element n of the array psvar is set and non-empty.
w — True if the day of the week is equal to n (Sunday = 0).

%<string<, %>string>, %[xstring] — this trio sets the truncation behaviour for the rest of the prompt string. Heads up: the third form is deprecated, and it's equivalent to %xstringx, i.e. x may be < or >. The string gets displayed in place of whatever portion of the string got truncated — quick note, that string does not undergo prompt expansion. 🙈

Okay so the numeric argument: in the third form it may appear immediately after the [, and it specifies the maximum permitted length of the various strings that can be shown in the prompt. In the first two forms, that numeric argument may be negative — and if it is, the truncation length is worked out by subtracting the absolute value of the numeric argument from the number of character positions remaining on the current prompt line. If that gives you a zero or negative length, a length of 1 is used. In other words, a negative argument arranges things so that after truncation at least n characters remain before the right margin (left margin for RPROMPT). 💖

The forms with < truncate at the left of the string, and the forms with > truncate at the right. Example time: if the current directory is /home/pike, then the prompt %8<..<%/ will expand to ..e/pike. Iconic. ✨

In this string, the terminating character (<, > or ]) — or, like, literally any character — may be quoted by a preceding \. But careful with print -P though: it has to be doubled, because the string is also subject to standard print processing on top of any backslashes removed by a double quoted string. The worst case is therefore print -P "%<\\\\<<...". 😭

If the string is longer than the specified truncation length, it appears in full, completely replacing the truncated string.

The part of the prompt string that's up for truncation runs to the end of the string, or to the end of the next enclosing group of the %( construct, or to the next truncation encountered at the same grouping level (so truncations inside a %( are separate) — whichever comes first. In particular, a truncation with argument zero (e.g. %<<) marks the end of the range of the string to be truncated and turns truncation off from there on. For example, the prompt %10<...<%~%<<%# will print a truncated representation of the current directory, followed by a % or #, followed by a space. Without the %<<, those two characters would have been included in the string to be truncated. 🎯

Also note: %-0<< is not equivalent to %<< — it specifies that the prompt is truncated at the right margin. Different vibe entirely.

Truncation applies only within each individual line of the prompt, as delimited by embedded newlines (if any). And here's the scary part:

If the total length of any line of the prompt after truncation is greater than the terminal width, or if the part to be truncated contains embedded newlines, truncation behavior is undefined and may change in a future version of the shell. 💀

So instead, use %-n(l.true-text.false-text) to remove parts of the prompt when the available space is less than n. Smarter, safer, slay. 👑

zsh 5.9.2                        July 12, 2026                       ZSHMISC(1)
────────────────────────────────────────────────────────────────────────────────
ZSHEXPN(1)                  General Commands Manual                  ZSHEXPN(1)

🌸 NAME 🌸

zshexpn — zsh expansion and substitution

🦋 DESCRIPTION 🦋

Bestie, the following types of expansions are performed in the indicated order, in five steps:

History Expansion — this is performed only in interactive shells.
Alias Expansion — aliases are expanded immediately before the command line is parsed, as explained under Aliasing in zshmisc(1).
Process Substitution, Parameter Expansion, Command Substitution, Arithmetic Expansion, Brace Expansion — these five are performed in left-to-right fashion. On each argument, any of the five steps that are needed are performed one after the other. Hence, for example, all the parts of parameter expansion are completed before command substitution is started. After these expansions, all unquoted occurrences of the characters `\`, `'` and `"` are removed. 🧹
Filename Expansion — if the SH_FILE_EXPANSION option is set, the order of expansion is modified for compatibility with sh and ksh. In that case filename expansion is performed immediately after alias expansion, preceding the set of five expansions mentioned above.
Filename Generation — this expansion, commonly referred to as globbing, is always done last. Periodt. 💅

The following sections explain the types of expansion in detail.

🔥 HISTORY EXPANSION 🔥

History expansion lets you use words from previous command lines in the command line you're currently typing. It simplifies spelling corrections and the repetition of complicated commands. Truly a life-saver, ngl. 😭

Immediately before execution, each command is saved in the history list, the size of which is controlled by the HISTSIZE parameter. The one most recent command is always retained in any case. Each saved command in the history list is called a history event and is assigned a number, beginning with 1 (one) when the shell starts up. The history number that you may see in your prompt (see EXPANSION OF PROMPT SEQUENCES in zshmisc(1)) is the number that is to be assigned to the next command. 👀

⭐ Overview ⭐

A history expansion begins with the first character of the histchars parameter, which is ! by default, and may occur anywhere on the command line — including inside double quotes (but not inside single quotes '...' or C-style quotes $'...', nor when escaped with a backslash). 🚫

The first character is followed by an optional event designator (see the section Event Designators) and then an optional word designator (the section Word Designators); if neither of these designators is present, no history expansion occurs.

Input lines containing history expansions are echoed after being expanded, but before any other expansions take place and before the command is executed. It is this expanded form that is recorded as the history event for later references. 📝

History expansions do not nest.

By default, a history reference with no event designator refers to the same event as any preceding history reference on that command line; if it's the only history reference in a command, it refers to the previous command. However, if the option CSH_JUNKIE_HISTORY is set, then every history reference with no event specification always refers to the previous command. 🙃

For example, ! is the event designator for the previous command, so !!:1 always refers to the first word of the previous command, and !!$ always refers to the last word of the previous command. With CSH_JUNKIE_HISTORY set, then !:1 and !$ function in the same manner as !!:1 and !!$, respectively. Conversely, if CSH_JUNKIE_HISTORY is unset, then !:1 and !$ refer to the first and last words, respectively, of the same event referenced by the nearest other history reference preceding them on the current command line — or to the previous command if there is no preceding reference. 🤯

The character sequence ^foo^bar (where ^ is actually the second character of the histchars parameter) repeats the last command, replacing the string foo with bar. More precisely, the sequence ^foo^bar^ is synonymous with !!:s^foo^bar^, hence other modifiers (see the section Modifiers) may follow the final ^. In particular, ^foo^bar^:G performs a global substitution. 💫

If the shell encounters the character sequence !" in the input, the history mechanism is temporarily disabled until the current list (see zshmisc(1)) is fully parsed. The !" is removed from the input, and any subsequent ! characters have no special significance. 🎀

Okay so, like, there's also a less convenient but more comprehensible way to get at your command history — it's the fc builtin 💅. Not the vibe for everyone, but it's there!

🕰️ Event Designators 🕰️

An event designator is just a lil pointer to an entry in the history list, bestie. In the list below, remember that the initial ! in each item can be swapped out for another character by setting the histchars parameter ✨.

! — starts a history expansion, except when it's followed by a blank, newline, = or (. If a word designator comes right after it (peep the Word Designators section), you get a history reference with no event designator at all (see the Overview section).
!! — the previous command. All by itself, this expansion literally just repeats the previous command 🔁.
!n — command-line n.
!-n — the current command-line minus n.
!str — the most recent command starting with str.
!?str[?] — the most recent command containing str. That trailing ? is necessary if this reference is followed by a modifier or by any text that shouldn't count as part of str.
!# — the current command line typed in so far. The line is treated as if it were complete up to and including the word before the one with the !# reference.
!{...} — insulates a history reference from adjacent characters (if necessary). It's giving personal space 🫶.

🔤 Word Designators 🔤

A word designator says which word (or words) of a given command line get included in a history reference. A : usually separates the event spec from the word designator — and you can only skip it if the word designator starts with a ^, $, *, - or %. The designators include:

0 — the first input word (the command).
n — the *n*th argument.
^ — the first argument. That is, 1.
$ — the last argument.
% — the word matched by (the most recent) ?str search.
x-y — a range of words; x defaults to 0.
* — all the arguments, or a null value if there are none.
x* — abbreviates x-$.
x- — like x* but omitting word $.

Heads up bestie: a % word designator only works when it's used in one of !%, !:% or !?str?:%, and only when used after a !? expansion (possibly in an earlier command). Anything else = an error, although ngl the error may not be the most obvious one 😭.

🎀 Modifiers 🎀

After the optional word designator, you can add a sequence of one or more of these modifiers, each one preceded by a :. These modifiers also work on the result of filename generation and parameter expansion — except where noted, iykyk 💅.

a — turns a file name into an absolute path: prepends the current directory if necessary, removes . path segments, and removes .. path segments plus the segments immediately preceding them.

This transformation is agnostic about what's actually in the filesystem, i.e. it works on the logical, not the physical directory. It happens the same way as when you change directories with neither the CHASE_DOTS nor CHASE_LINKS option set. For example, /before/here/../after is always transformed to /before/after, no matter whether /before/here even exists or what kind of object (dir, file, symlink, etc.) it is 👑

A — turns a file name into an absolute path like the a modifier does, and then passes the result through the realpath(3) library function to resolve symbolic links.

Note: on systems that don't have a realpath(3) library function, symbolic links are not resolved, so on those systems a and A are equivalent.

Note: foo:A and realpath(foo) are different on some inputs. For realpath(foo) semantics, see the P modifier.

c — resolves a command name into an absolute path by searching the command path given by the PATH variable. This does not work for commands containing directory parts. Also note it doesn't usually work as a glob qualifier unless a file of the same name is found in the current directory.
e — removes all but the part of the filename extension following the .; see the definition of the filename extension in the description of the r modifier below. Note that according to that definition the result will be empty if the string ends with a ..
h [ digits ] — removes a trailing pathname component, shortening the path by one directory level: this is the head of the pathname. Works like dirname. If the h is followed immediately (no spaces or other separator) by any number of decimal digits, and the value of the resulting number is non-zero, that number of leading components is preserved instead of the final component being removed. In an absolute path the leading / is the first component, so, for example, if var=/my/path/to/something, then ${var:h3} substitutes /my/path. Consecutive /s are treated the same as a single /. In parameter substitution, digits may only be used if the expression is in braces, so for example the short form substitution $var:h2 is treated as ${var:h}2, not as ${var:h2}. No restriction applies to the use of digits in history substitution or globbing qualifiers. If more components are requested than are present, the entire path is substituted (so this doesn't trigger a failed modifier error in history expansion) 🫡.
l — converts the words to all lowercase.
p — prints the new command but does not execute it. Only works with history expansion.
P — turns a file name into an absolute path, like realpath(3). The resulting path will be absolute, will refer to the same directory entry as the input filename, and none of its components will be symbolic links or equal to . or ...

Unlike realpath(3), non-existent trailing components are permitted and preserved ✨.

q — quotes the substituted words, escaping further substitutions. Works with history expansion and parameter expansion, though for parameters it's only useful if the resulting text is going to be re-evaluated, like by eval.
Q — removes one level of quotes from the substituted words.
r — removes a filename extension leaving the root name. Strings with no filename extension are not altered. A filename extension is a . followed by any number of characters (including zero) that are neither . nor / and that continue to the end of the string. For example, the extension of foo.orig.c is .c, and dir.c/foo has no extension.
s/l/r/ — substitutes r for l as described below. The substitution is done only for the first string that matches l. For arrays and for filename generation, this applies to each word of the expanded text. See below for further notes on substitutions.

The forms gs/l/r and s/l/r/:G perform global substitution, i.e. substitute every occurrence of r for l. Note that the g or :G must appear in exactly the position shown 👀.

See further notes on this form of substitution below.

& — repeats the previous s substitution. Like s, may be preceded immediately by a g. In parameter expansion the & must appear inside braces, and in filename generation it must be quoted with a backslash.
t [ digits ] — removes all leading pathname components, leaving the final component (tail). Works like basename. Any trailing slashes are first removed. Decimal digits are handled as described above for h, but in this case that number of trailing components is preserved instead of the default 1; 0 is treated the same as 1.
u — converts the words to all uppercase.
x — like q, but breaks into words at whitespace. Does not work with parameter expansion 💀.

The s/l/r/ substitution works like this, bestie. By default the left-hand side of substitutions aren't patterns, they're character strings. Any character can be used as the delimiter in place of /. A backslash quotes the delimiter character. The character &, in the right-hand-side r, gets replaced by the text from the left-hand-side l. The & can be quoted with a backslash. A null l uses the previous string, either from the previous l or from the contextual scan string s from !?s. You can omit the rightmost delimiter if a newline immediately follows r; the rightmost ? in a context scan can similarly be omitted. Note that the same record of the last l and r is maintained across all forms of expansion 🧚.

Note that if a & is used within glob qualifiers, an extra backslash is needed, because & is a special character in this case.

Also note that the order of expansions affects how l and r get interpreted. When used in a history expansion — which happens before any other expansions — l and r are treated as literal strings (except as explained for HIST_SUBST_PATTERN below). When used in parameter expansion, the replacement of r into the parameter's value is done first, and then any additional process, parameter, command, arithmetic, or brace references are applied, which may evaluate those substitutions and expansions more than once if l appears more than once in the starting value 😵‍💫. When used in a glob qualifier, any substitutions or expansions are performed once at the time the qualifier is parsed, even before the :s expression itself is divided into l and r sides.

If the option HIST_SUBST_PATTERN is set, l is treated as a pattern of the usual form described in the section FILENAME GENERATION below. This can be used in all the places where modifiers are available; note, however, that in globbing qualifiers parameter substitution has already taken place, so parameters in the replacement string should be quoted to make sure they're replaced at the correct time. Note also that complicated patterns used in globbing qualifiers may need the extended glob qualifier notation (#q:s/.../.../) in order for the shell to recognize the expression as a glob qualifier. Further, note that bad patterns in the substitution are not subject to the NO_BAD_PATTERN option, so they will cause an error 🚨.

When HIST_SUBST_PATTERN is set, l may start with a # to indicate that the pattern must match at the start of the string to be substituted, and a % may appear at the start or after an # to indicate that the pattern must match at the end of the string to be substituted. The % or # may be quoted with two backslashes.

For example, the following piece of filename generation code with the EXTENDED_GLOB option:

okay so quick example so it actually clicks: this bad boy right here —

print -r -- *.c(#q:s/#%(#b)s(*).c/'S${match[1]}.C'/)

— takes the expansion of *.c and applies the glob qualifiers inside the (#q...) expression, which is just a substitution modifier anchored to the start and end of each word (#%). That anchoring turns on backreferences ((#b)), so the parenthesised subexpression becomes available in the replacement string as ${match[1]}. And the replacement string is quoted so that the parameter isn't substituted before filename generation even starts. 💅

The following f, F, w and W modifiers work only with parameter expansion and filename generation, bestie — they're listed here purely so there's one single place to look up all the modifiers. iykyk. 🎀

f — repeats the immediately following modifier (no colon in between, okay?) until the resulting word stops changing. like, literally stops. no more changes. done. 🦋
F:expr: — like f, but only repeats n times, where the expression expr evaluates to n. Any character can be used instead of the :; if you open with (, [, or {, then close with ), ], or } respectively. match your brackets, babe. ✨
w — makes the immediately following modifier work on each word in the string. every single one. 🫶
W:sep: — like w, except "words" means the parts of the string separated by sep. Any character can be used instead of the :, and opening parentheses are handled specially (see above, bestie). 💖

🌈 PROCESS SUBSTITUTION 🌈

Every part of a command argument shaped like <(list), >(list) or =(list) gets the process-substitution treatment. The expression can have other strings before or after it, except — and this is the important bit — to avoid clashing with strings and patterns that show up all the time, the = form must occur at the start of a command argument. Also, these forms only get expanded when command or assignment arguments are first parsed. You can use process substitutions after redirection operators too; in that case the substitution has to appear with no trailing string. periodt. 🎯

heads up: <<(list) is not some special syntax. It's literally equivalent to < <(list) — redirecting standard input from the result of the process substitution. So all the docs below apply to it anyway. The second form (with the space) is the recommended one for clarity. 🙈

For the < or > forms, the shell runs the commands in list as a subprocess of the job that's executing the shell command line. If your system supports the /dev/fd mechanism, the command argument is the name of the device file matching a file descriptor; otherwise, if your system supports named pipes (FIFOs), the command argument is a named pipe. If you picked the > form, then writing on that special file gives input to list. If you used <, then the file passed as an argument gets connected to the output of the list process. example time, bestie:

paste <(cut -f1 file1) <(cut -f3 file2) |
tee >(process1) >(process2) >/dev/null

That cuts field 1 and field 3 out of file1 and file2 respectively, pastes the results together, and sends it to the processes process1 and process2. iconic. ✨

If you use =(...) instead of <(...), then the file passed as an argument is the name of a temporary file containing the output of the list process. Use this instead of the < form for a program that expects to lseek (see lseek(2)) on the input file. 👑

There's an optimisation for substitutions of the form =(<<<arg), where arg is a single-word argument to the here-string redirection <<<. This form produces a file name containing the value of arg after any substitutions have been performed, and it's handled entirely within the current shell. It's effectively the reverse of the special form $(<arg), which treats arg as a file name and replaces it with the file's contents. 🧚

The = form is useful because both the /dev/fd implementation and the named pipe implementation of <(...) have drawbacks, and ngl they're real:

In the /dev/fd case, some programmes may automatically close the file descriptor in question before they even examine the file on the command line — particularly if that's necessary for security reasons, like when the programme is running setuid. 💀
In the named pipe case, if the programme doesn't actually open the file, the subshell trying to read from or write to the pipe will (in a typical implementation — different operating systems may behave differently) block for ever and have to be killed explicitly. 😭
In both cases the shell is actually supplying the information using a pipe, so programmes that expect to lseek (see lseek(2)) on the file just won't work.

Also! That earlier example can be written more compactly and efficiently (as long as the MULTIOS option is set) as:

paste <(cut -f1 file1) <(cut -f3 file2) > >(process1) > >(process2)

The shell uses pipes instead of FIFOs to implement the latter two process substitutions in that example. 🫶

There's an extra problem with >(process) though: when it's attached to an external command, the parent shell does not wait for process to finish, and therefore a command immediately following it can't rely on the results being complete. Same problem, same solution as the section MULTIOS in zshmisc(1). So in a simplified version of the example above:

paste <(cut -f1 file1) <(cut -f3 file2) > >(process)

(note that no MULTIOS are involved here), process runs asynchronously as far as the parent shell is concerned. The workaround is:

{ paste <(cut -f1 file1) <(cut -f3 file2) } > >(process)

The extra processes here are spawned from the parent shell, which will wait for their completion. slay. 💅

Another problem pops up any time a job with a substitution that requires a temporary file gets disowned by the shell — including the case where &! or &| appears at the end of a command containing a substitution. In that case the temporary file will not get cleaned up, because the shell has no memory of the job any more. 😭 Workaround: use a subshell, like

(mycmd =(myoutput)) &!

since the forked subshell will wait for the command to finish and then remove the temporary file. 🌸

A general workaround that makes sure a process substitution sticks around for an appropriate length of time is to pass it as a parameter to an anonymous shell function (that's a piece of shell code that runs immediately with function scope). for example, this code:

() {
   print File $1:
   cat $1
} =(print This be the verse)

outputs something resembling this:

File /tmp/zsh6nU0kS:
This be the verse

The temporary file created by the process substitution gets deleted when the function exits. chef's kiss. 🎀

💖 PARAMETER EXPANSION 💖

The character $ is what introduces parameter expansions, bestie. See zshparam(1) for a description of parameters — arrays, associative arrays, subscript notation for grabbing individual array elements, all of it. 📚

Note in particular that words of unquoted parameters are not automatically split on whitespace unless the option SH_WORD_SPLIT is set; see the references to that option below for more. This is an important difference from other shells, fr fr. However — same as in other shells — null words are elided from unquoted parameters' expansions. 🦋

With default options, after these assignments:

array=("first word" "" "third word")
scalar="only word"

then $array substitutes two words — first word and third word — and $scalar substitutes a single word, only word. Notice that the second element of array was elided. 👀 Scalar parameters can be elided too if their value is null (empty). To avoid elision, quote like this: "$scalar" for scalars, and "${array[@]}" or "${(@)array}" for arrays. (Those last two forms are equivalent.) 💜

Parameter expansions can involve flags, like ${(@kv)aliases}, and other operators, like ${PREFIX:-"/usr/local"}. They can be nested too! These topics get introduced below — the full rules are complicated, and they're noted at the end. 😅

In the expansions discussed below that need a pattern, the pattern has the same form as the one used for filename generation; see the section Filename Generation. Note that these patterns, along with the replacement text of any substitutions, are themselves subject to parameter expansion, command substitution, and arithmetic expansion. And on top of the operations below, the colon modifiers described in the section Modifiers inside the section History Expansion can be applied — for example, ${i:s/foo/bar/} performs string substitution on the expansion of parameter $i. ✨

In the following descriptions, word refers to a single word substituted on the command line — not necessarily a space-delimited word. got it? 💅

${name} — substitutes the value, if any, of the parameter name. The braces are required if the expansion is going to be followed by a letter, digit, or underscore that shouldn't be interpreted as part of name. Also, more complicated forms of substitution usually need the braces present; the exceptions (which only apply if the option KSH_ARRAYS is not set) are a single subscript, any colon modifiers appearing after name, or any of the characters ^, =, ~, # or + appearing before name — all of those work with or without braces. 🫶 If name is an array parameter and KSH_ARRAYS is not set, then the value of each element of name is substituted, one element per word. Otherwise the expansion results in one word only; with KSH_ARRAYS, that's the first element of an array. No field splitting is done on the result unless SH_WORD_SPLIT is set. See also the flags = and s:string:. 🌈
${+name} — if name is the name of a set parameter, 1 is substituted; otherwise 0 is substituted. yes, it's a vibe check for parameters. ✅
${name-word}, ${name:-word} — if name is set (or, in the second form, is non-null), substitute its value; otherwise substitute word. In the second form name may be omitted, in which case word is always substituted. 🎀
${name+word}, ${name:+word} — if name is set (or, in the second form, is non-null), substitute word; otherwise substitute nothing. literally nothing. 🫥
${name=word}, ${name:=word}, ${name::=word} — in the first form, if name is unset then set it to word; in the second form, if name is unset or null then set it to word; and in the third form, unconditionally set name to word. In all forms, the value of name is then substituted. 💖
${name?word}, ${name:?word} — bestie, this is the "are you SURE" check 💅. First form: if name is set, substitute its value. Second form: only if name is both set and non-null. Otherwise it prints word and exits the shell 😭 (interactive shells instead just bounce you back to the prompt, cute). If word is omitted, a standard message gets printed. And heads up — word is expanded even though its value never lands on the command line. Wild, I know.

In any of the expressions above that test a variable and substitute an alternate word, note that you can use standard shell quoting in the word value to selectively override the splitting done by the SH_WORD_SPLIT option and the = flag 🎀 — but not splitting by the s:string: flag. No cap.

In the following expressions, when name is an array and the substitution is not quoted, or if the (@) flag or the name[@] syntax is used, matching and replacement is performed on each array element separately ✨.

${name#pattern}, ${name##pattern} — if the pattern matches the beginning of the value of name, substitute the value of name with the matched portion deleted; otherwise just substitute the value of name as-is. First form = the smallest matching pattern wins 🐜; second form = the largest matching pattern wins 🐘.
${name%pattern}, ${name%%pattern} — same vibes but from the other end 🌈: if the pattern matches the end of the value of name, substitute the value of name with the matched portion deleted; otherwise just substitute the value of name. First form = smallest match preferred, second form = largest match preferred.
${name:#pattern} — if the pattern matches the value of name, substitute the empty string 🫥; otherwise just substitute the value of name. If name is an array, the matching array elements get removed (use the (M) flag if you want the opposite — remove the ones that didn't match 🙈).
${name:|arrayname} — if arrayname is the name (N.B., not the contents 👀) of an array variable, then any elements contained in arrayname are removed from the substitution of name. If the substitution is scalar — either because name is a scalar variable or the expression is quoted — the elements of arrayname are instead tested against the entire expression.
${name:*arrayname} — similar to the one right above but the opposite sense 🔄, so entries present in both the original substitution and as elements of arrayname are retained and the others get removed. It's giving intersection energy.
${name:^arrayname}, ${name:^^arrayname} — zips two arrays 🧚, such that the output array is twice as long as the shortest (longest for :^^) of name and arrayname, with the elements alternatingly being picked from them. For :^, if one of the input arrays is longer, the output stops when the end of the shorter array is reached. Thus,

a=(1 2 3 4); b=(a b); print ${a:^b}

will output 1 a 2 b. For :^^, the input is repeated until all of the longer array has been used up, and the above will output 1 a 2 b 3 a 4 b ✨.

Either or both inputs may be a scalar — they get treated as an array of length 1 with the scalar as the only element. If either array is empty, the other array is output with no extra elements inserted 🫶.

Currently the following code will output a b and 1 as two separate elements, which can be unexpected 😭. The second print provides a workaround which should continue to work if this is changed.

a=(a b); b=(1 2); print -l "${a:^b}"; print -l "${${a:^b}}"

${name:offset}, ${name:offset:length} — this syntax gives effects similar to parameter subscripting in the form $name[start,end], but it's compatible with other shells 🐚; note that both offset and length are interpreted differently from the components of a subscript. Ikyk.

If offset is non-negative, then if the variable name is a scalar, substitute the contents starting offset characters from the first character of the string; and if name is an array, substitute elements starting offset elements from the first element. If length is given, substitute that many characters or elements — otherwise the entire rest of the scalar or array 💜.

A positive offset is always treated as the offset of a character or element in name from the first character or element of the array (this is different from native zsh subscript notation 👀). Hence 0 refers to the first character or element regardless of the setting of the option KSH_ARRAYS.

A negative offset counts backwards from the end of the scalar or array, so that -1 corresponds to the last character or element, and so on 🦋.

When positive, length counts from the offset position toward the end of the scalar or array. When negative, length counts back from the end. If this results in a position smaller than offset, a diagnostic is printed and nothing is substituted 💀.

The option MULTIBYTE is obeyed, i.e. the offset and length count multibyte characters where appropriate 🌸.

offset and length undergo the same set of shell substitutions as for scalar assignment; in addition, they are then subject to arithmetic evaluation. Hence, for example

print ${foo:3}
print ${foo: 1 + 2}
print ${foo:$(( 1 + 2))}
print ${foo:$(echo 1 + 2)}

all have the same effect 🔥, extracting the string starting at the fourth character of $foo if the substitution would otherwise return a scalar, or the array starting at the fourth element if $foo would return an array. Note that with the option KSH_ARRAYS, $foo always returns a scalar (regardless of the use of the offset syntax) and a form such as ${foo[*]:3} is required to extract elements of an array named foo.

If offset is negative, the - may not appear immediately after the : as this indicates the ${name:-word} form of substitution 😱. Instead, a space may be inserted before the -. Furthermore, neither offset nor length may begin with an alphabetic character or & as these are used to indicate history-style modifiers. To substitute a value from a variable, the recommended approach is to precede it with a $ as this signifies the intention (parameter substitution can easily be rendered unreadable 💀); however, as arithmetic substitution is performed, the expression ${var: offs} does work, retrieving the offset from $offs.

For further compatibility with other shells there is a special case for array offset 0 🎀. This usually accesses the first element of the array. However, if the substitution refers to the positional parameter array, e.g. $@ or $*, then offset 0 instead refers to $0, offset 1 refers to $1, and so on. In other words, the positional parameter array is effectively extended by prepending $0. Hence ${*:0:1} substitutes $0 and ${*:1:1} substitutes $1 ✨.

${name/pattern/repl}, ${name//pattern/repl}, ${name:/pattern/repl} — replace the longest possible match of pattern in the expansion of parameter name by string repl. First form replaces just the first occurrence, second form all occurrences, and third form replaces only if pattern matches the entire string 💅. Both pattern and repl are subject to double-quoted substitution, so expressions like ${name/$opat/$npat} will work — but they obey the usual rule that pattern characters in $opat are not treated specially unless either the option GLOB_SUBST is set, or $opat is instead substituted as ${~opat} 👑

The pattern may begin with a #, in which case the pattern must match at the start of the string; or %, in which case it must match at the end of the string; or #%, in which case the pattern must match the entire string 🌈. The repl may be an empty string, in which case the final / may also be omitted. To quote the final / in other cases it should be preceded by a single backslash; this is not necessary if the / occurs inside a substituted parameter. Note also that the #, % and #% are not active if they occur inside a substituted parameter, even at the start 🙈.

If, after quoting rules apply, ${name} expands to an array, the replacements act on each element individually. Note also the effect of the I and S parameter expansion flags below 🧚; however, the flags M, R, B, E and N are not useful.

For example,

foo="twinkle twinkle little star" sub="t*e" rep="spy"
print ${foo//${~sub}/$rep}
print ${(S)foo//${~sub}/$rep}

Here, the ~ ensures that the text of $sub is treated as a pattern rather than a plain string. In the first case, the longest match for t*e is substituted and the result is spy star, while in the second case, the shortest matches are taken and the result is spy spy lispy star 🔥.

${#spec} — if spec is one of the above substitutions, substitute the length in characters of the result instead of the result itself 📏. If spec is an array expression, substitute the number of elements of the result. This has the side-effect that joining is skipped even in quoted forms, which may affect other sub-expressions in spec 👀. Note that ^, =, and ~, below, must appear to the left of # when these forms are combined.

If the option POSIX_IDENTIFIERS is not set, and spec is a simple name, then the braces are optional 🎀; this is true even for special parameters so e.g. $#- and $#* take the length of the string $- and the array $* respectively. If POSIX_IDENTIFIERS is set, then braces are required for the # to be treated in this fashion.

${^spec}, ${^^spec} — turn on the RC_EXPAND_PARAM option for the evaluation of spec; if the ^ is doubled, turn it off 🔄. When this option is set, array expansions of the form foo${xx}bar, where the parameter xx is set to (a b c), are substituted with fooabar foobbar foocbar instead of the default fooa b cbar. Note that an empty array will therefore cause all arguments to be removed 💀 — main character energy, honestly.

Internally, each such expansion is converted into the equivalent list for brace expansion. Like, ${^var} becomes {$var[1],$var[2],...}, and it gets processed exactly as described in the Brace Expansion section below 🫶 — but heads up, the expansion happens immediately, and any explicit brace expansion happens later. And if word splitting is also in effect, the $var[N] may themselves get split into different list elements. iykyk.

${=spec}, ${==spec} — do word splitting using the rules for SH_WORD_SPLIT while spec is being evaluated, but regardless of whether the parameter appears in double quotes. If the = is doubled, it turns that off. This forces parameter expansions to be split into separate words before substitution, using IFS as a delimiter — which is literally the default in most other shells. 💅

Also note: splitting is applied to word in the assignment forms of spec before the assignment to name is performed. Which affects the result of array assignments with the A flag. Keep that in mind bestie.

${~spec}, ${~~spec} — turn on the GLOB_SUBST option for the evaluation of spec; if the ~ is doubled, turn it off. When this option is set, the string resulting from the expansion gets interpreted as a pattern anywhere that's possible, like in filename expansion and filename generation, and in pattern-matching contexts like the right hand side of the = and != operators in conditions. ✨

In nested substitutions, note that the effect of the ~ applies to the result of the current level of substitution. A surrounding pattern operation on the result may cancel it. So for example, if the parameter foo is set to *, then ${~foo//\*/*.c} is substituted by the pattern *.c, which may be expanded by filename generation — but ${${~foo}//\*/*.c} substitutes to the string *.c, which will not be further expanded. Sneaky, right? 😭

If a ${...} type parameter expression or a $(...) type command substitution is used in place of name above, it gets expanded first and the result is used as if it were the value of name. So nested operations are totally a thing: ${${foo#head}%tail} substitutes the value of $foo with both head and tail deleted. 👑

The form with $(...) is often useful in combination with the flags described next — see the examples below. Each name or nested ${...} in a parameter expansion may also be followed by a subscript expression, as described in Array Parameters in zshparam(1).

Note that double quotes may appear around nested expressions, in which case only the part inside is treated as quoted. For example, ${(f)"$(foo)"} quotes the result of $(foo), but the flag (f) (see below) is applied using the rules for unquoted expansions. And note further that quotes are themselves nested in this context — for example, in "${(@f)"$(foo)"}", there are two sets of quotes: one surrounding the whole expression, and the other (redundant) surrounding the $(foo) as before. Quotes within quotes within quotes, it's giving inception 🌀

✨ Parameter Expansion Flags ✨

If the opening brace is directly followed by an opening parenthesis, the string up to the matching closing parenthesis will be taken as a list of flags. In cases where repeating a flag is meaningful, the repetitions don't need to be consecutive — for example, (q%q%q) means the same thing as the more readable (%%qqq). Slay. The following flags are supported:

# — evaluate the resulting words as numeric expressions and interpret these as character codes. Output the corresponding characters. Note that this form is entirely distinct from use of the # without parentheses. If the MULTIBYTE option is set and the number is greater than 127 (i.e. not an ASCII character) it's treated as a Unicode character. 🌈
% — expand all % escapes in the resulting words the same way as in prompts (see EXPANSION OF PROMPT SEQUENCES in zshmisc(1)). If this flag is given twice, full prompt expansion is done on the resulting words, depending on the setting of the PROMPT_PERCENT, PROMPT_SUBST and PROMPT_BANG options.
@ — in double quotes, array elements are put into separate words. E.g., "${(@)foo}" is equivalent to "${foo[@]}" and "${(@)foo[1,2]}" is the same as "$foo[1]" "$foo[2]". This is distinct from field splitting by the f, s or z flags, which still applies within each array element. ✨
A — convert the substitution into an array expression, even if it otherwise would be scalar. This has lower precedence than subscripting, so one level of nested expansion is required in order for subscripts to apply to array elements. Thus ${${(A)name}[1]} yields the full value of name when name is scalar. 🎀

This assigns an array parameter with ${...=...}, ${...:=...} or ${...::=...}. If this flag is repeated (as in AA), it assigns an associative array parameter. Assignment is made before sorting or padding; if field splitting is active, the word part is split before assignment. The name part may be a subscripted range for ordinary arrays; when assigning an associative array, the word part must be converted to an array, for example by using ${(AA)=name=...} to activate field splitting. Surrounding context such as additional nesting or use of the value in a scalar assignment may cause the array to be joined back into a single string again. (Iconic behaviour tbh.)

a — sort in array index order; when combined with O, sort in reverse array index order. Note that a is therefore equivalent to the default, but Oa is useful for obtaining an array's elements in reverse order. 🔥
b — quote with backslashes only characters that are special to pattern matching. This is useful when the contents of the variable are to be tested using GLOB_SUBST, including the ${~...} switch.

Quoting using one of the q family of flags does not work for this purpose, since quotes are not stripped from non-pattern characters by GLOB_SUBST. In other words,

pattern=${(q)str}
[[ $str = ${~pattern} ]]

works if $str is a*b but not if it is a b — whereas

pattern=${(b)str}
[[ $str = ${~pattern} ]]

is always true for any possible value of $str. No cap. 💅

c — with ${#name}, count the total number of characters in an array, as if the elements were concatenated with spaces between them. This is not a true join of the array, so other expressions used with this flag may have an effect on the elements of the array before it's counted. 🦋
C — capitalize the resulting words. "Words" in this case refers to sequences of alphanumeric characters separated by non-alphanumerics — not to words that result from field splitting. Capitalize slay 💖
D — assume the string or array elements contain directories and attempt to substitute the leading part of these by names. The remainder of the path (the whole of it, if the leading part wasn't substituted) is then quoted so that the whole string can be used as a shell argument. This is the reverse of ~ substitution — see the section FILENAME EXPANSION below.
e — perform single word shell expansions, namely parameter expansion, command substitution and arithmetic expansion, on the result. Such expansions can be nested, but too deep recursion may have unpredictable effects. ⚠️ You've been warned bestie.
f — split the result of the expansion at newlines. This is a shorthand for ps:\n:.
F — join the words of arrays together using newline as a separator. This is a shorthand for pj:\n:.
g:opts: — process escape sequences like the echo builtin when no options are given (g::). With the o option, octal escapes don't take a leading zero. With the c option, sequences like ^X are also processed. With the e option, it processes \M-t and similar sequences like the print builtin. With both of the o and e options, it behaves like the print builtin — except that in none of these modes is \c interpreted. 🙈
i — sort case-insensitively. May be combined with n or O.
k — if name refers to an associative array, substitute the keys (element names) rather than the values of the elements. Used with subscripts (including ordinary arrays), it forces indices or keys to be substituted even if the subscript form refers to values. However, this flag may not be combined with subscript ranges. With the KSH_ARRAYS option, a subscript [*] or [@] is needed to operate on the whole array, as usual. 💀
L — convert all letters in the result to lower case.
n — sort decimal integers numerically; if the first differing characters of two test strings are not digits, sorting is lexical. + and - are not treated specially — they're treated as any other non-digit. Integers with more initial zeroes are sorted before those with fewer or none. Hence the array foo+24 foo1 foo02 foo2 foo3 foo20 foo23 is sorted into the order shown. May be combined with i or O. 🔢
- — as n, but a leading minus sign indicates a negative decimal integer. A leading minus sign not followed by an integer does not trigger numeric sorting. Note that + signs are not handled specially (this may change in the future).
o — sort the resulting words in ascending order; if this appears on its own, the sorting is lexical and case-sensitive (unless the locale renders it case-insensitive). Sorting in ascending order is the default for other forms of sorting, so this is ignored if combined with a, i, n or -. 💜
O — sort the resulting words in descending order; O without a, i, n or - sorts in reverse lexical order. May be combined with a, i, n or - to reverse the order of sorting. Flip it and reverse it 🔄
P — this forces the value of the parameter name to be interpreted as a further parameter name, whose value will be used where appropriate. Note that flags set with one of the typeset family of commands (in particular case transformations) are not applied to the value of name used in this fashion. 🧚

Like, if you use that with a nested parameter or command substitution, the result of that is taken as a parameter name the same way. So if you've got foo=bar and bar=baz, then ${(P)foo}, ${(P)${foo}}, and ${(P)$(echo bar)} all expand to baz — yes bestie, all three of them, it's the same energy 💅

And likewise, if the reference is itself nested, the expression with the flag gets treated as if it were directly replaced by the parameter name. It's an error if this nested substitution produces an array with more than one word — like, it will simply not tolerate that 🫠. For example, if name=assoc where the parameter assoc is an associative array, then ${${(P)name}[elt]} refers to the element of the associative subscripted elt.

q — Quote characters that are special to the shell in the resulting words with backslashes; unprintable or invalid characters get quoted using the $'\NNN' form, with separate quotes for each octet.

Give this flag twice and the resulting words are quoted in single quotes; give it three times and they're quoted in double quotes — and in those forms there's no special handling of unprintable or invalid characters attempted at all. Give it four times and the words are quoted in single quotes preceded by a $. Note (important!!) that in all three of these forms the quoting is done unconditionally, even if it doesn't change the way the resulting string would be interpreted by the shell. No vibes check, just quotes ✨

If a q- is given (only a single q may appear), a minimal form of single quoting is used that only quotes the string if it's needed to protect special characters. Typically this form gives the most readable output — lowkey the icon of the family 🎀

If a q+ is given, an extended form of minimal quoting is used that causes unprintable characters to be rendered using $'...'. This quoting is similar to that used by the output of values by the typeset family of commands.

Q — Remove one level of quotes from the resulting words. Undressing the string, basically 🙈
t — Use a string describing the type of the parameter where the value of the parameter would usually appear. This string is made of keywords separated by hyphens (-). The first keyword in the string describes the main type, and it can be one of scalar, array, integer, float or association. The other keywords describe the type in more detail:
- local — for local parameters
- left — for left justified parameters
- right_blanks — for right justified parameters with leading blanks
- right_zeros — for right justified parameters with leading zeros
- lower — for parameters whose value is converted to all lower case when it is expanded
- upper — for parameters whose value is converted to all upper case when it is expanded
- readonly — for readonly parameters
- tag — for tagged parameters
- tied — for parameters tied to another parameter in the manner of PATH (colon-separated list) and path (array), whether these are special parameters or user-defined with typeset -T
- export — for exported parameters
- unique — for arrays which keep only the first occurrence of duplicated values
- hide — for parameters with the hide flag
- hideval — for parameters with the hideval flag
- special — for special parameters defined by the shell
u — Expand only the first occurrence of each unique word. Dedupe era 💅
U — Convert all letters in the result to upper case. SCREAMING, but make it fashion 🔥
v — Used with k, substitute (as two consecutive words) both the key and the value of each associative array element. Used with subscripts, force values to be substituted even if the subscript form refers to indices or keys.
V — Make any special characters in the resulting words visible. Spilling the invisible tea 👀
w — With ${#name}, count words in arrays or strings; the s flag may be used to set a word delimiter.
W — Similar to w with the difference that empty words between repeated delimiters are also counted. No word left behind 🫶
X — With this flag, parsing errors occurring with the Q, e and # flags or the pattern matching forms such as ${name#pattern} are reported. Without the flag, errors are silently ignored. Snitches get... reported errors, actually 💀
z — Split the result of the expansion into words using shell parsing to find the words, i.e. taking into account any quoting in the value. Comments are not treated specially but as ordinary strings, similar to interactive shells with the INTERACTIVE_COMMENTS option unset (however, see the Z flag below for related options).

Note that this is done very late — even later than the (s) flag. So to access single words in the result, use nested expansions as in ${${(z)foo}[2]}. Likewise, to remove the quotes in the resulting words use ${(Q)${(z)foo}}.

0 — Split the result of the expansion on null bytes. This is a shorthand for ps:\0:.

The following flags (except p) are followed by one or more arguments as shown. Any character, or the matching pairs (...), {...}, [...], or <...>, may be used in place of a colon as delimiters — but heads up bestie, when a flag takes more than one argument, a matched pair of delimiters must surround each argument 🎀

p — Recognize the same escape sequences as the print builtin in string arguments to any of the flags described below that follow this argument.

Alternatively, with this option string arguments may be in the form $var in which case the value of the variable is substituted. Note this form is strict; the string argument does not undergo general parameter expansion.

For example,

sep=:
val=a:b:c
print ${(ps.$sep.)val}

splits the variable on a :.

~ — Strings inserted into the expansion by any of the flags below are to be treated as patterns. This applies to the string arguments of flags that follow ~ within the same set of parentheses. Compare with ~ outside parentheses, which forces the entire substituted string to be treated as a pattern. Hence, for example,

[[ "?" = ${(~j.|.)array} ]]

treats | as a pattern and succeeds if and only if $array contains the string ? as an element. The ~ may be repeated to toggle the behaviour; its effect only lasts to the end of the parenthesised group. Toggle toggle ✨

j:string: — Join the words of arrays together using string as a separator. Note that this occurs before field splitting by the s:string: flag or the SH_WORD_SPLIT option.
l:expr::string1::string2: — Pad the resulting words on the left. Each word will be truncated if required and placed in a field expr characters wide.

The arguments :string1: and :string2: are optional; neither, the first, or both may be given. Note that the same pairs of delimiters must be used for each of the three arguments. The space to the left will be filled with string1 (concatenated as often as needed) or spaces if string1 is not given. If both string1 and string2 are given, string2 is inserted once directly to the left of each word, truncated if necessary, before string1 is used to produce any remaining padding.

If either of string1 or string2 is present but empty, i.e. there are two delimiters together at that point, the first character of $IFS is used instead.

If the MULTIBYTE option is in effect, the flag m may also be given, in which case widths will be used for the calculation of padding; otherwise individual multibyte characters are treated as occupying one unit of width.

If the MULTIBYTE option is not in effect, each byte in the string is treated as occupying one unit of width.

Control characters are always assumed to be one unit wide; this allows the mechanism to be used for generating repetitions of control characters. Very niche, very slay 💅

m — Only useful together with one of the flags l or r or with the # length operator when the MULTIBYTE option is in effect. Use the character width reported by the system in calculating how much of the string it occupies or the overall length of the string. Most printable characters have a width of one unit, however certain Asian character sets and certain special effects use wider characters; combining characters have zero width. Non-printable characters are arbitrarily counted as zero width; how they would actually be displayed will vary.

If the m is repeated, the character either counts zero (if it has zero width), else one. For printable character strings this has the effect of counting the number of glyphs (visibly separate characters), except for the case where combining characters themselves have non-zero width (true in certain alphabets). Glyph counting, we love to see it 🦋

r:expr::string1::string2: — As l, but pad the words on the right and insert string2 immediately to the right of the string to be padded.

Left and right padding may be used together. In this case the strategy is to apply left padding to the first half width of each of the resulting words, and right padding to the second half. If the string to be padded has odd width the extra padding is applied on the left. Math, but make it cute 🎀

s:string: — Force field splitting at the separator string. Note that a string of two or more characters means that all of them must match in sequence; this differs from the treatment of two or more characters in the IFS parameter. See also the = flag and the SH_WORD_SPLIT option. An empty string may also be given, in which case every character will be a separate element.

For historical reasons, the usual behaviour that empty array elements are retained inside double quotes is disabled for arrays generated by splitting; hence the following:

line="one::three"
print -l "${(s.:.)line}"

produces two lines of output for one and three and elides the empty field. To override this behaviour, supply the (@) flag as well, i.e. "${(@s.:.)line}". The empty field said "not me being erased" 😭

Z:opts: — literally just z but it also takes a whole combo of option letters sandwiched between a matching pair of delimiter characters 🎀 no options at all? then it behaves exactly like z, no notes. Here's what you can actually put in there:
- (Z+c+) — makes comments get parsed as a string and kept around; any field in the resulting array that starts with an unquoted comment character is a comment, periodt 💅
- (Z+C+) — parses comments and then yeets them. The rule for what counts as a comment is the standard one: anything between a word starting with the third character of $HISTCHARS (default #) up to the next newline is a comment 🗑️
- (Z+n+) — makes unquoted newlines get treated as regular ol' whitespace; otherwise they're treated like shell code delimiters and converted into semicolons 😤

You can stack these options inside the same pair of delimiters, like (Z+Cn+) — go off ✨

_:flags: — bestie, this one is reserved for future use 🙈 as of this revision of zsh there are literally no valid flags; anything after an underscore (other than an empty pair of delimiters) is treated as an error, and the flag itself does absolutely nothing 💀

So the following flags are the ones that actually mean something with the ${...#...} or ${...%...} forms. And the S, I, and * flags can ALSO be used with the ${.../...} forms 🫶

S — with # or ##, it searches for the match that starts closest to the start of the string (a "substring match" 😌). Of all the matches sitting at one particular position, # picks the shortest and ## picks the longest:

% str="aXbXc"
% echo ${(S)str#X*}
abXc
% echo ${(S)str##X*}
a
%

With % or %%, it searches for the match that starts closest to the end of the string:

% str="aXbXc"
% echo ${(S)str%X*}
aXbc
% echo ${(S)str%%X*}
aXb
%

heads up: % and %% do not search for the match that ends closest to the end of the string, even though that's what you'd expect 🙈

With substitution via ${.../...} or ${...//...}, S specifies non-greedy matching, i.e. the shortest match gets replaced instead of the longest:

% str="abab"
% echo ${str/*b/_}
_
% echo ${(S)str/*b/_}
_ab
%

I:expr: — searches for the exprth match (where expr evaluates to a number 🔢). This only applies when you're searching for substrings, either with the S flag, or with ${.../...} (only the *expr*th match gets substituted) or ${...//...} (all matches from the *expr*th one onwards get substituted). Default is to take the first match, obviously 💅

The *expr*th match gets counted so that there's either one or zero matches from each starting position in the string — though for global substitution, matches that overlap previous replacements are ignored. With the ${...%...} and ${...%%...} forms, the starting position for the match moves backwards from the end as the index increases, while with the other forms it moves forward from the start.

So with the string

which switch is the right switch for Ipswich?

substitutions of the form ${(SI:N:)string#w*ch} as N increases from 1 will match and remove which, witch, witch and wich; the form using ## will match and remove which switch is the right switch for Ipswich, witch is the right switch for Ipswich, witch for Ipswich and wich. The form using % will remove the same matches as for #, but in reverse order, and the form using %% will remove the same matches as for ##, also in reverse order 🔥

* — enables EXTENDED_GLOB for substitution via ${.../...} or ${...//...}. Note that ** does not disable extendedglob 👀
B — includes the index of the beginning of the match in the result ⭐
E — includes the index one character past the end of the match in the result (yes, note that this is inconsistent with other uses of parameter index 🙃)
M — includes the matched portion in the result 💖
N — includes the length of the match in the result 📏
R — includes the unmatched portion in the result (the *R*est) 🍓

Rules 💅

Here's the summary of the rules for substitution — this assumes braces are present around the substitution, i.e. ${...}. Some particular examples are given below.

⚠️ note that the Zsh Development Group accepts no responsibility for any brain damage which may occur during the reading of the following rules 😭💀

Nested substitution If multiple nested ${...} forms are present, substitution happens from the inside outwards 🌀 At each level, the substitution takes account of whether the current value is a scalar or an array, whether the whole substitution is in double quotes, and what flags are supplied to the current level of substitution — just as if the nested substitution were the outermost one. The flags do not propagate up to enclosing substitutions; the nested substitution will return either a scalar or an array as determined by the flags, possibly adjusted for quoting. All the following steps take place, where applicable, at all levels of substitution.

Note that, unless the (P) flag is present, the flags and any subscripts apply directly to the value of the nested substitution; for example, the expansion ${${foo}} behaves exactly the same as ${foo}. When the (P) flag is present in a nested substitution, the other substitution rules get applied to the value before it's interpreted as a name, so ${${(P)foo}} may differ from ${(P)foo} 👑

At each nested level of substitution, the substituted words undergo all forms of single-word substitution (i.e. not filename generation), including command substitution, arithmetic expansion and filename expansion (i.e. leading ~ and =). So for example, ${${:-=cat}:h} expands to the directory where the cat program resides. (Explanation, because yes you need it: the internal substitution has no parameter but a default value =cat, which gets expanded by filename expansion to a full path; the outer substitution then applies the modifier :h and takes the directory part of the path 🧚)

Internal parameter flags Any parameter flags set by one of the typeset family of commands — in particular the -L, -R, -Z, -u and -l options for padding and capitalization — are applied directly to the parameter value. Note these flags are options to the command, e.g. typeset -Z; they are not the same as the flags used within parameter substitutions 🙅

At the outermost level of substitution, the (P) flag (rule 4.) ignores these transformations and uses the unmodified value of the parameter as the name to be replaced. This is usually the behavior you want, because padding may make the value syntactically illegal as a parameter name — but if you do want capitalization changes, use the ${${(P)foo}} form (rule 25.) ✨

Parameter subscripting If the value is a raw parameter reference with a subscript, such as ${var[3]}, the effect of subscripting gets applied directly to the parameter. Subscripts are evaluated left to right; subsequent subscripts apply to the scalar or array value yielded by the previous subscript. So if var is an array, ${var[1][2]} is the second character of the first word, but ${var[2,4][2]} is the entire third word (the second word of the range of words two through four of the original array 😵‍💫). Any number of subscripts may appear. Flags such as (k) and (v) which alter the result of subscripting are applied.
Parameter name replacement At the outermost level of nesting only, the (P) flag gets applied. This treats the value so far as a parameter name (which may include a subscript expression) and replaces that with the corresponding value. This replacement happens later if the (P) flag appears in a nested substitution ⏳

If the value so far names a parameter that has internal flags (rule 2.), those internal flags are applied to the new value after replacement.

Double-quoted joining If the value after this process is an array, and the substitution appears in double quotes, and neither an (@) flag nor a # length operator is present at the current level — then the words of the value get joined with the first character of the parameter $IFS, by default a space, between each word (single word arrays are not modified 🫶). If the (j) flag is present, that's used for joining instead of $IFS.
Nested subscripting Any remaining subscripts (i.e. of a nested substitution) get evaluated at this point, based on whether the value is an array or a scalar. As with 3., multiple subscripts can appear. Note that ${foo[2,4][2]} is thus equivalent to ${${foo[2,4]}[2]} and also to "${${(@)foo[2,4]}[2]}" (the nested substitution returns an array in both cases), but not to "${${foo[2,4]}[2]}" (the nested substitution returns a scalar because of the quotes 👀).
Modifiers Any modifiers — as specified by a trailing #, %, / (possibly doubled) or by a set of modifiers of the form :... (see the section Modifiers in the section History Expansion) — are applied to the words of the value at this level 🌸
Character evaluation Any (#) flag is applied, evaluating the result so far numerically as a character 🔤
Length Any initial # modifier, i.e. in the form ${#var}, is used to evaluate the length of the expression so far 📐
Forced joining If the (j) flag is present, or no (j) flag is present but the string is to be split as given by rule 11., and joining did not take place at rule 5., any words in the value are joined together using the given string or the first character of $IFS if none. Note that the (F) flag implicitly supplies a string for joining in this manner 🎀
Simple word splitting — if one of the (s) or (f) flags is in the mix, or the = specifier was there (like ${=var}), the word gets split on occurrences of whatever string you named — or, for = with neither of those two flags present, on any of the characters in $IFS. ✂️

And if there's no (s), (f) or = at all, but the word isn't quoted and the option SH_WORD_SPLIT is set? Then it splits on any of the characters in $IFS. Heads up: this step, too, happens at all levels of a nested substitution. 💅

Case modification — whatever case flip you asked for with (L), (U) or (C) gets applied here. 🔠
Escape sequence replacement — first all the replacements from the (g) flag go down, then any prompt-style formatting from the (%) family of flags lands. ✨
Quote application — any quoting or unquoting via (q) and (Q) and their related flags is applied. 🎀
Directory naming — any directory name substitution using the (D) flag is applied. 📁
Visibility enhancement — any modifications to make characters visible using the (V) flag are applied. 👀
Lexical word splitting — if the (z) flag, or one of the forms of the (Z) flag, is present, the word is split like it was a shell command line, so quotation marks and other metacharacters decide what counts as a word. Note that this kind of splitting is completely separate from the one in rule 11: it does not use $IFS, and it does not cause forced joining. 🧠
Uniqueness — if the result is an array and the (u) flag was present, duplicate elements get yeeted from the array. 🫧
Ordering — if the result is still an array and one of the (o) or (O) flags was present, the array gets reordered. 🔄
RC_EXPAND_PARAM — right here is where the decision happens about whether any resulting array elements get combined element by element with the surrounding text, as given by either the RC_EXPAND_PARAM option or the ^ flag. 🧩
Re-evaluation — any (e) flag is applied to the value, forcing it to be re-examined for new parameter substitutions, but also for command and arithmetic substitutions. 🔁
Padding — any padding of the value by the (l.fill.) or (r.fill.) flags gets applied. ⬜
Semantic joining — in contexts where the expansion semantics require a single word to come out, all the words get rejoined with the first character of IFS between them. So in ${(P)${(f)lines}} the value of ${lines} is split at newlines, but then it has to be joined back together again before the (P) flag can be applied. 🫶

If a single word is not required, this rule gets skipped entirely. 🙈

Empty argument removal — if the substitution doesn't appear in double quotes, any resulting zero-length argument — whether it came from a scalar or from an element of an array — is elided from the list of arguments inserted into the command line. 🚮

Strictly speaking, the removal actually happens later, since the same thing happens with other forms of substitution; the point to note here is simply that it occurs after any of the above parameter operations. 📝

Nested parameter name replacement — if the (P) flag is present and rule 4. has not applied, the value so far is treated as a parameter name (which may include a subscript expression) and replaced with the corresponding value, with the internal flags (rule 2.) applied to the new value. 🪄

🎀 Examples 🎀

The flag f is a whole vibe for splitting a double-quoted substitution line by line. For example, ${(f)"$(<file)"} substitutes the contents of file divided so that each line is an element of the resulting array. Compare that with the effect of $(<file) on its own, which divides the file up by words — or the same thing inside double quotes, which makes the entire content of the file one single string. 📄✨

The following illustrates the rules for nested parameter expansions. Suppose that $foo contains the array (bar baz):

"${(@)${foo}[1]}" — this produces the result b. First, the inner substitution "${foo}", which has no array (@) flag, produces a single word result "bar baz". The outer substitution "${(@)...[1]}" detects that this is a scalar, so that (despite the (@) flag) the subscript picks the first character. 😭 iconic behaviour honestly
"${${(@)foo}[1]}" — this produces the result bar. In this case, the inner substitution "${(@)foo}" produces the array (bar baz). The outer substitution "${...[1]}" detects that this is an array and picks the first word. This is similar to the simple case "${foo[1]}". 💅

As an example of the rules for word splitting and joining, suppose $foo contains the array (ax1 bx1). Then:

${(s/x/)foo} — produces the words a, 1 b and 1.
${(j/x/s/x/)foo} — produces a, 1, b and 1.
${(s/x/)foo%%1*} — produces a and b (note the extra space 👀). Because substitution occurs before either joining or splitting, the operation first generates the modified array (ax bx), which is joined to give "ax bx", and then split to give a, b and ''. That final empty string then gets elided, since it's not in double quotes. 🫧

💸 COMMAND SUBSTITUTION 💸

A command enclosed in parentheses preceded by a dollar sign, like $(...), or quoted with grave accents, like `...`, is replaced with its standard output, with any trailing newlines deleted. If the substitution isn't enclosed in double quotes, the output gets broken into words using the IFS parameter. 🔨

The substitution $(cat foo) may be replaced by the faster $(<foo). In this case foo undergoes single word shell expansions (parameter expansion, command substitution and arithmetic expansion) — but not filename generation. 🚀

If the option GLOB_SUBST is set, the result of any unquoted command substitution, including the special form just mentioned, is eligible for filename generation. 🌈

🔢 ARITHMETIC EXPANSION 🔢

A string of the form $[exp] or $((exp)) is substituted with the value of the arithmetic expression exp. exp is subjected to parameter expansion, command substitution and arithmetic expansion before it is evaluated. See the section Arithmetic Evaluation. 🧮

🎁 BRACE EXPANSION 🎁

A string of the form foo{xx,yy,zz}bar is expanded to the individual words fooxxbar, fooyybar and foozzbar. Left-to-right order is preserved. This construct may be nested. Commas may be quoted in order to include them literally in a word. 🎀

An expression of the form {n1..n2}, where n1 and n2 are integers, is expanded to every number between n1 and n2 inclusive. If either number begins with a zero, all the resulting numbers will be padded with leading zeroes to that minimum width, but for negative numbers the - character is also included in the width. If the numbers are in decreasing order the resulting sequence will also be in decreasing order. 🔢

An expression of the form {n1..n2..n3}, where n1, n2, and n3 are integers, is expanded as above, but only every n3th number starting from n1 is output. If n3 is negative the numbers are output in reverse order — this is slightly different from simply swapping n1 and n2 in the case that the step n3 doesn't evenly divide the range. Zero padding can be specified in any of the three numbers, and specifying it in the third can be useful to pad for example {-99..100..01} which is not possible to specify by putting a 0 on either of the first two numbers (i.e. pad to two characters). 🧚

An expression of the form {c1..c2}, where c1 and c2 are single characters (which may be multibyte characters), is expanded to every character in the range from c1 to c2 in whatever character sequence is used internally. For characters with code points below 128 this is US ASCII (this is the only case most users will need). If any intervening character is not printable, appropriate quotation is used to render it printable. If the character sequence is reversed, the output is in reverse order, e.g. {d..a} is substituted as d c b a. 🔤

If a brace expression matches none of the above forms, it's left unchanged — unless the option BRACE_CCL (an abbreviation for brace character class) is set. In that case, it gets expanded to a list of the individual characters between the braces sorted into the order of the characters in the ASCII character set (multibyte characters are not currently handled 🙈). The syntax is similar to a [...] expression in filename generation: - is treated specially to denote a range of characters, but ^ or ! as the first character is treated normally. For example, {abcdef0-9} expands to 16 words 0 1 2 3 4 5 6 7 8 9 a b c d e f. 🎉

Note that brace expansion is not part of filename generation (globbing); an expression such as */{foo,bar} is split into two separate words */foo and */bar before filename generation takes place. In particular, note that this is liable to produce a no match error if either of the two expressions does not match; this is to be contrasted with */(foo|bar), which is treated as a single pattern but otherwise has similar effects. ⚠️

To combine brace expansion with array expansion, see the ${^spec} form described in the section Parameter Expansion above. 🫶

💖 FILENAME EXPANSION 💖

okay so bestie, every single word gets a little vibe check: does it start with an unquoted ~? 👀 If it does, then the word up to a / (or the whole word if there's no /) gets checked to see if it can be swapped out in one of the ways we're about to describe. If the answer is yes, then the ~ and the checked portion get replaced with the appropriate substitute value. slay. ✨

A ~ all by itself gets replaced by the value of $HOME. A ~ followed by a + or a - gets replaced by the current or previous working directory, respectively. cute. 🏠

A ~ followed by a number gets replaced by the directory at that position in the directory stack. ~0 is equivalent to ~+, and ~1 is the top of the stack. ~+ followed by a number gets replaced by the directory at that position in the directory stack. ~+0 is equivalent to ~+, and ~+1 is the top of the stack. ~- followed by a number gets replaced by the directory that many positions from the bottom of the stack. ~-0 is the bottom of the stack. and just so you know, the PUSHD_MINUS option exchanges the effects of ~+ and ~- where they are followed by a number. iconic behavior tbh. 💅

🌸 Dynamic named directories 🌸

If the function zsh_directory_name exists, or the shell variable zsh_directory_name_functions exists and contains an array of function names, then those functions are used to implement dynamic directory naming. The functions are tried in order until one returns status zero, so it is important that functions test whether they can handle the case in question and return an appropriate status. like, do your vibe check first, bestie. 🙈

A ~ followed by a string namstr in unquoted square brackets is treated specially as a dynamic directory name. Note that the first unquoted closing square bracket always terminates namstr. The shell function is passed two arguments: the string n (for name) and namstr. It should either set the array reply to a single element which is the directory corresponding to the name and return status zero (executing an assignment as the last statement is usually sufficient), or it should return status non-zero. In the former case the element of reply is used as the directory; in the latter case the substitution is deemed to have failed. If all functions fail and the option NOMATCH is set, an error results. 😭

The functions defined as above are also used to see if a directory can be turned into a name, for example when printing the directory stack or when expanding %~ in prompts. In this case each function is passed two arguments: the string d (for directory) and the candidate for dynamic naming. The function should either return non-zero status, if the directory cannot be named by the function, or it should set the array reply to consist of two elements: the first is the dynamic name for the directory (as would appear within ~[...]), and the second is the prefix length of the directory to be replaced. For example, if the trial directory is /home/myname/src/zsh and the dynamic name for /home/myname/src (which has 16 characters) is s, then the function sets

reply=(s 16)

The directory name so returned is compared with possible static names for parts of the directory path, as described below; it is used if the prefix length matched (16 in the example) is longer than that matched by any static name. longest match wins, periodt. 👑

It is not a requirement that a function implements both n and d calls; for example, it might be appropriate for certain dynamic forms of expansion not to be contracted to names. In that case any call with the first argument d should cause a non-zero status to be returned. no pressure. 🫶

The completion system calls zsh_directory_name c followed by equivalent calls to elements of the array zsh_directory_name_functions, if it exists, in order to complete dynamic names for directories. The code for this should be as for any other completion function as described in zshcompsys(1). 📚

As a working example, here is a function that expands any dynamic names beginning with the string p: to directories below /home/pws/perforce. In this simple case a static name for the directory would be just as effective.

zsh_directory_name() {
  emulate -L zsh
  setopt extendedglob
  local -a match mbegin mend
  if [[ $1 = d ]]; then
    # turn the directory into a name
    if [[ $2 = (#b)(/home/pws/perforce/)([^/]##)* ]]; then
      typeset -ga reply
      reply=(p:$match[2] $(( ${#match[1]} + ${#match[2]} )) )
    else
      return 1
    fi
  elif [[ $1 = n ]]; then
    # turn the name into a directory
    [[ $2 != (#b)p:(?*) ]] && return 1
    typeset -ga reply
    reply=(/home/pws/perforce/$match[1])
  elif [[ $1 = c ]]; then
    # complete names
    local expl
    local -a dirs
    dirs=(/home/pws/perforce/*(/:t))
    dirs=(p:${^dirs})
    _wanted dynamic-dirs expl 'dynamic directory' compadd -S\] -a dirs
    return
  else
    return 1
  fi
  return 0
}

🦋 Static named directories 🦋

A ~ followed by anything not already covered — consisting of any number of alphanumeric characters or underscore (_), hyphen (-), or dot (.) — is looked up as a named directory, and replaced by the value of that named directory if found. Named directories are typically home directories for users on the system. They may also be defined if the text after the ~ is the name of a string shell parameter whose value begins with a /. Note that trailing slashes will be removed from the path to the directory (though the original parameter is not modified). lowkey important detail. 🌈

It is also possible to define directory names using the -d option to the hash builtin. 🎀

When the shell prints a path (e.g. when expanding %~ in prompts or when printing the directory stack), the path is checked to see if it has a named directory as its prefix. If so, then the prefix portion is replaced with a ~ followed by the name of the directory. The shorter of the two ways of referring to the directory is used, i.e. either the directory name or the full path; the name is used if they are the same length. The parameters $PWD and $OLDPWD are never abbreviated in this fashion. they said "not me, we're staying full path." 💀

✨ `=` expansion ✨

If a word begins with an unquoted = and the EQUALS option is set, the remainder of the word is taken as the name of a command. If a command exists by that name, the word is replaced by the full pathname of the command. easy. ⭐

💅 Notes 💅

Filename expansion is performed on the right hand side of a parameter assignment, including those appearing after commands of the typeset family. In this case, the right hand side will be treated as a colon-separated list in the manner of the PATH parameter, so that a ~ or an = following a : is eligible for expansion. All such behaviour can be disabled by quoting the ~, the =, or the whole expression (but not simply the colon); the EQUALS option is also respected. 🫶

If the option MAGIC_EQUAL_SUBST is set, any unquoted shell argument in the form identifier=expression becomes eligible for file expansion as described in the previous paragraph. Quoting the first = also inhibits this. okay so keep that first = quoted if you don't want the magic. 🙈

🔥 FILENAME GENERATION 🔥

If a word contains an unquoted instance of one of the characters *, (, |, <, [, or ?, it is regarded as a pattern for filename generation, unless the GLOB option is unset. If the EXTENDED_GLOB option is set, the ^ and # characters also denote a pattern; otherwise they are not treated specially by the shell. 👀

The word is replaced with a list of sorted filenames that match the pattern. If no matching pattern is found, the shell gives an error message, unless the NULL_GLOB option is set, in which case the word is deleted; or unless the NOMATCH option is unset, in which case the word is left unchanged. so dramatic, honestly. 😭

In filename generation, the character / must be matched explicitly; also, a . must be matched explicitly at the beginning of a pattern or after a /, unless the GLOB_DOTS option is set. No filename generation pattern matches the files . or ... In other instances of pattern matching, the / and . are not treated specially. periodt. 💅

🎀 Glob Operators 🎀

* — Matches any string, including the null string.
? — Matches any character.
[...] — Matches any of the enclosed characters. Ranges of characters can be specified by separating two characters by a -. A - or ] may be matched by including it as the first character in the list. There are also several named classes of characters, in the form [:name:] with the following meanings. The first set use the macros provided by the operating system to test for the given character combinations, including any modifications due to local language settings, see ctype(3):
- [:alnum:] — The character is alphanumeric
- [:alpha:] — The character is alphabetic
- [:ascii:] — The character is 7-bit, i.e. is a single-byte character without the top bit set.
- [:blank:] — The character is a blank character
- [:cntrl:] — The character is a control character
- [:digit:] — The character is a decimal digit
- [:graph:] — The character is a printable character other than whitespace
- [:lower:] — The character is a lowercase letter
- [:print:] — The character is printable
- [:punct:] — The character is printable but neither alphanumeric nor whitespace
- [:space:] — The character is whitespace
- [:upper:] — The character is an uppercase letter
- [:xdigit:] — The character is a hexadecimal digit

Another set of named classes is handled internally by the shell and is not sensitive to the locale:

[:IDENT:] — The character is allowed to form part of a shell identifier, such as a parameter name; this test respects the POSIX_IDENTIFIERS option
[:IFS:] — The character is used as an input field separator, i.e. is contained in the IFS parameter
[:IFSSPACE:] — The character is an IFS white space character; see the documentation for IFS in the zshparam(1) manual page. 🧚
[:INCOMPLETE:] — matches a byte that starts an incomplete multibyte character 🌈. Heads up though: there can be a whole sequence of more than one byte that, taken together, forms the prefix of a multibyte character. To test for a potentially incomplete byte sequence, use the pattern [[:INCOMPLETE:]]*. This will never match a sequence starting with a valid multibyte character. Slay 💅.
[:INVALID:] — matches a byte that does not start a valid multibyte character 💀. Note this may be a continuation byte of an incomplete multibyte character, since any part of a multibyte string consisting of invalid and incomplete multibyte characters is treated as single bytes. Fr fr, that's the whole vibe.
[:WORD:] — the character is treated as part of a word; this test is sensitive to the value of the WORDCHARS parameter 👑

Okay so the square brackets are additional to the ones enclosing the whole set of characters — so to test for a single alphanumeric character you need [[:alnum:]] 🫶. Named character sets can be used alongside other types too, e.g. [[:alpha:]0-9] ✨.

[^...], [!...] — just like [...], except that it matches any character which is not in the given set 🙈.
<[x]-[y]> — matches any number in the range x to y, inclusive 🔥. Either of the numbers may be omitted to make the range open-ended; hence <-> matches any number. To match individual digits, the [...] form is more efficient.

Be careful when using other wildcards adjacent to patterns of this form 😭 — for example, <0-9>* will actually match any number whatsoever at the start of the string, since the <0-9> will match the first digit, and the * will match any others. This is a trap for the unwary, but is in fact an inevitable consequence of the rule that the longest possible match always succeeds ⭐. Expressions such as <0-9>[^[:digit:]]* can be used instead. iykyk.

(...) — matches the enclosed pattern. This is used for grouping 🎀. If the KSH_GLOB option is set, then a @, *, +, ? or ! immediately preceding the ( is treated specially, as detailed below 👇. The option SH_GLOB prevents bare parentheses from being used in this way, though the KSH_GLOB option is still available.

Note that grouping cannot extend over multiple directories: it is an error to have a / within a group (this only applies for patterns used in filename generation) 🚨. There is one exception: a group of the form (pat/)# appearing as a complete path segment can match a sequence of directories. For example, foo/(a*/)#bar matches foo/bar, foo/any/bar, foo/any/anyother/bar, and so on. Iconic behaviour honestly 👑.

x|y — matches either x or y. This operator has lower precedence than any other 🦋. The | character must be within parentheses, to avoid interpretation as a pipeline. The alternatives are tried in order from left to right.
^x — (requires EXTENDED_GLOB to be set.) Matches anything except the pattern x 💜. This has a higher precedence than /, so ^foo/bar will search directories in . except ./foo for a file named bar.
x~y — (requires EXTENDED_GLOB to be set.) Match anything that matches the pattern x but does not match y. This has lower precedence than any operator except |, so */*~foo/bar will search for all files in all directories in . and then exclude foo/bar if there was such a match 🌸. Multiple patterns can be excluded by foo~bar~baz. In the exclusion pattern (y), / and . are not treated specially the way they usually are in globbing.
x# — (requires EXTENDED_GLOB to be set.) Matches zero or more occurrences of the pattern x ✨. This operator has high precedence; 12# is equivalent to 1(2#), rather than (12)#. It is an error for an unquoted # to follow something which cannot be repeated 💀; this includes an empty string, a pattern already followed by ##, or parentheses when part of a KSH_GLOB pattern (for example, !(foo)# is invalid and must be replaced by *(!(foo))).
x## — (requires EXTENDED_GLOB to be set.) Matches one or more occurrences of the pattern x 🔥. This operator has high precedence; 12## is equivalent to 1(2##), rather than (12)##. No more than two active # characters may appear together 🙈. (Note the potential clash with glob qualifiers in the form 1(2##) which should therefore be avoided.)

🎀 ksh-like Glob Operators 🎀

If the KSH_GLOB option is set, the effects of parentheses can be modified by a preceding @, *, +, ? or ! 👑. This character need not be unquoted to have special effects, but the ( must be.

@(...) — match the pattern in the parentheses. (Like (...).)
*(...) — match any number of occurrences. (Like (...)#, except that recursive directory searching is not supported.)
+(...) — match at least one occurrence. (Like (...)##, except that recursive directory searching is not supported.)
?(...) — match zero or one occurrence. (Like (|...).)
!(...) — match anything but the expression in parentheses. (Like (^(...)).)

⭐ Precedence ⭐

The precedence of the operators given above is (highest) ^, /, ~, | (lowest) 🦋; the remaining operators are simply treated from left to right as part of a string, with # and ## applying to the shortest possible preceding unit (i.e. a character, ?, [...], <...>, or a parenthesised expression) 💅.

As mentioned above, a / used as a directory separator may not appear inside parentheses, while a | must do so 😭; in patterns used in other contexts than filename generation (for example, in case statements and tests within [[...]]), a / is not special; and / is also not special after a ~ appearing outside parentheses in a filename pattern. No cap, that's just how it is 🫶.

💖 Globbing Flags 💖

There are various flags which affect any text to their right up to the end of the enclosing group or to the end of the pattern 🌸; they require the EXTENDED_GLOB option. All take the form (#X) where X may have one of the following forms:

i — case insensitive: upper or lower case characters in the pattern match upper or lower case characters 🎀.
l — lower case characters in the pattern match upper or lower case characters; upper case characters in the pattern still only match upper case characters.
I — case sensitive: locally negates the effect of i or l from that point on 💀.
b — activate backreferences for parenthesised groups in the pattern 🦋; this does not work in filename generation. When a pattern with a set of active parentheses is matched, the strings matched by the groups are stored in the array $match, the indices of the beginning of the matched parentheses in the array $mbegin, and the indices of the end in the array $mend, with the first element of each array corresponding to the first parenthesised group, and so on 👑. These arrays are not otherwise special to the shell. The indices use the same convention as does parameter substitution, so that elements of $mend and $mbegin may be used in subscripts; the KSH_ARRAYS option is respected. Sets of globbing flags are not considered parenthesised groups; only the first nine active parentheses can be referenced ✨.

For example,

foo="a_string_with_a_message"
if [[ $foo = (a|an)_(#b)(*) ]]; then
  print ${foo[$mbegin[1],$mend[1]]}
fi

prints string_with_a_message. Note that the first set of parentheses is before the (#b) and does not create a backreference 🙈.

Backreferences work with all forms of pattern matching other than filename generation, but note that when performing matches on an entire array, such as ${array#pattern}, or a global substitution, such as ${param//pat/repl}, only the data for the last match remains available 😭. In the case of global replacements this may still be useful. See the example for the m flag below.

The numbering of backreferences strictly follows the order of the opening parentheses from left to right in the pattern string, although sets of parentheses may be nested 🧚. There are special rules for parentheses followed by # or ##. Only the last match of the parenthesis is remembered: for example, in [[ abab = (#b)([ab])# ]], only the final b is stored in match[1] 💀. Thus extra parentheses may be necessary to match the complete segment: for example, use X((ab|cd)#)Y to match a whole string of either ab or cd between X and Y, using the value of $match[1] rather than $match[2].

If the match fails none of the parameters is altered, so in some cases it may be necessary to initialise them beforehand 🌸. If some of the backreferences fail to match — which happens if they are in an alternate branch which fails to match, or if they are followed by # and matched zero times — then the matched string is set to the empty string, and the start and end indices are set to -1 😭.

Pattern matching with backreferences is slightly slower than without ⭐.

B — deactivate backreferences, negating the effect of the b flag from that point on 💅.
cN,M — the flag (#cN,M) can be used anywhere that the # or ## operators can be used except in the expressions (*/)# and (*/)## in filename generation, where / has special meaning 🚨; it cannot be combined with other globbing flags and a bad pattern error occurs if it is misplaced. It is equivalent to the form {N,M} in regular expressions. The previous character or group is required to match between N and M times, inclusive 🔥. The form (#cN) requires exactly N matches; (#c,M) is equivalent to specifying N as 0; (#cN,) specifies that there is no maximum limit on the number of matches 👑.

okay bestie, we're still deep in the globbing flags and honestly these little guys are doing so much 💅 let's keep it rolling:

m — sets references to the match data for the entire string matched. It's giving backreferencing energy, but it does not work in filename generation. The flag has to be in effect at the end of the pattern, i.e. not local to a group. The parameters $MATCH, $MBEGIN, and $MEND get set to the string matched and to the indices of the beginning and end of the string, respectively. This is most useful in parameter substitutions, because otherwise the string matched is obvious.

For example:

arr=(veldt jynx grimps waqf zho buck)
print ${arr//(#m)[aeiou]/${(U)MATCH}}

forces all the matches (i.e. all vowels) into uppercase, printing vEldt jynx grImps wAqf zhO bUck.

Unlike backreferences, there is no speed penalty for using match references, other than the extra substitutions required for the replacement strings in cases such as the example shown. ✨

M — deactivates the m flag, hence no references to match data will be created.
anum — approximate matching: num errors are allowed in the string matched by the pattern. The rules for this are described in the next subsection. 🧚
s, e — unlike the other flags, these have only a local effect, and each must appear on its own: (#s) and (#e) are the only valid forms. The (#s) flag succeeds only at the start of the test string, and the (#e) flag succeeds only at the end of the test string; they correspond to ^ and $ in standard regular expressions. They're useful for matching path segments in patterns other than those in filename generation (where path segments are in any case treated separately). For example, *((#s)|/)test((#e)|/)* matches a path segment test in any of the following strings: test, test/at/start, at/end/test, in/test/middle.

Another use is in parameter substitution; for example, ${array/(#s)A*Z(#e)} will remove only elements of an array which match the complete pattern A*Z. There are other ways of performing many operations of this type, however the combination of the substitution operations / and // with the (#s) and (#e) flags provides a single simple and memorable method. 🎀

Note that assertions of the form (^(#s)) also work, i.e. match anywhere except at the start of the string, although this actually means anything except a zero-length portion at the start of the string; you need to use (""~(#s)) to match a zero-length portion of the string not at the start. important detail bestie 💀

q — a q and everything up to the closing parenthesis of the globbing flags are ignored by the pattern matching code. This is intended to support the use of glob qualifiers, see below. The result is that the pattern (#b)(*).c(#q.) can be used both for globbing and for matching against a string. In the former case, the (#q.) will be treated as a glob qualifier and the (#b) will not be useful, while in the latter case the (#b) is useful for backreferences and the (#q.) will be ignored. Note that colon modifiers in the glob qualifiers are also not applied in ordinary pattern matching. 🦋
u — respect the current locale in determining the presence of multibyte characters in a pattern, provided the shell was compiled with MULTIBYTE_SUPPORT. This overrides the MULTIBYTE option; the default behaviour is taken from the option. Compare U. (Mnemonic: typically multibyte characters are from Unicode in the UTF-8 encoding, although any extension of ASCII supported by the system library may be used.) 🌈
U — all characters are considered to be a single byte long. The opposite of u. This overrides the MULTIBYTE option.

For example, the test string fooxx can be matched by the pattern (#i)FOOXX, but not by (#l)FOOXX, (#i)FOO(#I)XX or ((#i)FOOX)X. The string (#ia2)readme specifies case-insensitive matching of readme with up to two errors.

When using the ksh syntax for grouping both KSH_GLOB and EXTENDED_GLOB must be set and the left parenthesis should be preceded by @. Note also that the flags do not affect letters inside [...] groups, in other words (#i)[a-z] still matches only lowercase letters. Finally, note that when examining whole paths case-insensitively every directory must be searched for all files which match, so that a pattern of the form (#i)/foo/bar/... is potentially slow. 😭

✨ Approximate Matching ✨

When matching approximately, the shell keeps a count of the errors found, which cannot exceed the number specified in the (#anum) flags. Four types of error are recognised:

Different characters, as in fooxbar and fooybar.
Transposition of characters, as in banana and abnana.
A character missing in the target string, as with the pattern road and target string rod.
An extra character appearing in the target string, as with stove and strove.

Thus, the pattern (#a3)abcd matches dcba, with the errors occurring by using the first rule twice and the second once, grouping the string as [d][cb][a] and [a][bc][d]. 🔥

Non-literal parts of the pattern must match exactly, including characters in character ranges: hence (#a1)??? matches strings of length four, by applying rule 4 to an empty part of the pattern, but not strings of length two, since all the ? must match. Other characters which must match exactly are initial dots in filenames (unless the GLOB_DOTS option is set), and all slashes in filenames, so that a/bc is two errors from ab/c (the slash cannot be transposed with another character). Similarly, errors are counted separately for non-contiguous strings in the pattern, so that (ab|cd)ef is two errors from aebf.

When using exclusion via the ~ operator, approximate matching is treated entirely separately for the excluded part and must be activated separately. Thus, (#a1)README~READ_ME matches READ.ME but not READ_ME, as the trailing READ_ME is matched without approximation. However, (#a1)README~(#a1)READ_ME does not match any pattern of the form READ?ME as all such forms are now excluded. the exclusion really said not today 🙈

Apart from exclusions, there is only one overall error count; however, the maximum errors allowed may be altered locally, and this can be delimited by grouping. For example, (#a1)cat((#a0)dog)fox allows one error in total, which may not occur in the dog section, and the pattern (#a1)cat(#a0)dog(#a1)fox is equivalent. Note that the point at which an error is first found is the crucial one for establishing whether to use approximation; for example, (#a1)abc(#a0)xyz will not match abcdxyz, because the error occurs at the x, where approximation is turned off.

Entire path segments may be matched approximately, so that (#a1)/foo/d/is/available/at/the/bar allows one error in any path segment. This is much less efficient than without the (#a1), however, since every directory in the path must be scanned for a possible approximate match. It is best to place the (#a1) after any path segments which are known to be correct. performance queen advice 👑

🌈 Recursive Globbing 🌈

A pathname component of the form (foo/)# matches a path consisting of zero or more directories matching the pattern foo.

As a shorthand, **/ is equivalent to (*/)#; note that this therefore matches files in the current directory as well as subdirectories. Thus:

ls -ld -- (*/)#bar

ls -ld -- **/bar

does a recursive directory search for files named bar (potentially including the file bar in the current directory). This form does not follow symbolic links; the alternative form ***/ does, but is otherwise identical. Neither of these can be combined with other forms of globbing within the same path segment; in that case, the * operators revert to their usual effect. 💀

Even shorter forms are available when the option GLOB_STAR_SHORT is set. In that case if no / immediately follows a ** or *** they are treated as if both a / plus a further * are present. Hence:

setopt GLOBSTARSHORT
ls -ld -- **.c

is equivalent to

ls -ld -- **/*.c

💅 Glob Qualifiers 💅

Patterns used for filename generation may end in a list of qualifiers enclosed in parentheses. The qualifiers specify which filenames that otherwise match the given pattern will be inserted in the argument list. this is where it gets cute and specific 🎀

If the option BARE_GLOB_QUAL is set, then a trailing set of parentheses containing no | or ( characters (or ~ if it is special) is taken as a set of glob qualifiers. A glob subexpression that would normally be taken as glob qualifiers, for example (^x), can be forced to be treated as part of the glob pattern by doubling the parentheses, in this case producing ((^x)).

If the option EXTENDED_GLOB is set, a different syntax for glob qualifiers is available, namely (#qx) where x is any of the same glob qualifiers used in the other format. The qualifiers must still appear at the end of the pattern. However, with this syntax multiple glob qualifiers may be chained together. They are treated as a logical AND of the individual sets of flags. Also, as the syntax is unambiguous, the expression will be treated as glob qualifiers just as long any parentheses contained within it are balanced; appearance of |, ( or ~ does not negate the effect. Note that qualifiers will be recognised in this form even if a bare glob qualifier exists at the end of the pattern, for example *(#q*)(.) will recognise executable regular files if both options are set; however, mixed syntax should probably be avoided for the sake of clarity. Note that within conditions using the [[ form the presence of a parenthesised expression (#q...) at the end of a string indicates that globbing should be performed; the expression may include glob qualifiers, but it is also valid if it is simply (#q). This does not apply to the right hand side of pattern match operators as the syntax already has special significance. ✨

A qualifier may be any one of the following:

/ — directories
F — full (i.e. non-empty) directories. Note that the opposite sense (^F) expands to empty directories and all non-directories. Use (/^F) for empty directories.
. — plain files
@ — symbolic links
= — sockets
p — named pipes (FIFOs)
* — executable plain files (0100 or 0010 or 0001)
% — device files (character or block special)
%b — block special files
%c — character special files
r — owner-readable files (0400)
w — owner-writable files (0200)
x — owner-executable files (0100)
A — group-readable files, a.k.a. 0040 🎀
I — group-writable files, a.k.a. 0020 ✨
E — group-executable files, a.k.a. 0010 💅
R — world-readable files, a.k.a. 0004 🌸
W — world-writable files, a.k.a. 0002 🦋
X — world-executable files, a.k.a. 0001 ⭐
s — the setuid files, a.k.a. 04000 👑
S — the setgid files, a.k.a. 02000 💜
t — files with the sticky bit, a.k.a. 01000 🍓
fspec — files whose access rights match spec. So here's the tea: spec can be an octal number, and it can optionally have a =, a +, or a - in front of it. If none of those characters is given, it behaves the exact same as =. The octal number describes the mode bits you're expecting — and how it's checked depends on the prefix:
- with =, the value you gave must match the file-modes exactly 😤
- with +, at least the bits in the given number must be set in the file-modes ✅
- with -, the bits in the number must not be set ❌

If you drop a ? instead of an octal digit anywhere in the number, that means the corresponding bits in the file-modes just don't get checked at all — and this is only useful in combination with = btw, don't come for me.

Now, if the qualifier f is followed by any other character, everything up to the next matching character is taken as a list of comma-separated sub-specs. And matching here means: [, {, and < match ], }, and > respectively, while any other character matches itself. Each sub-spec is either an octal number like described above, or a list of any of the characters u, g, o, and a, followed by a =, a +, or a -, followed by a list of any of the characters r, w, x, s, and t, or an octal digit. The first list of characters says which access rights are being checked:

u → use the ones for the owner of the file
g → check those of the group
o → test those of other users
a → test all three groups, slay

The =, +, and - again say how the modes get checked and mean the same thing as described in the first form above. Then the second list of characters says which access rights are expected: r for read access, w for write access, x for the right to execute the file (or to search a directory), s for the setuid and setgid bits, and t for the sticky bit. Iconic.

So like, *(f70?) gives you the files for which the owner has read, write, and execute permission, and for which other group members have no rights — independent of the permissions for other users. The pattern *(f-100) gives all files for which the owner does not have execute permission. And *(f:gu+w,o-rx:) gives the files for which the owner and the other members of the group have at least write permission, and for which other users don't have read or execute permission. Obsessed with her fr fr 💖

estring, +cmd — the string gets executed as shell code 🧚 The filename will be included in the list if and only if the code returns a zero status (usually the status of the last command).

In the first form, the first character after the e is used as a separator, and anything up to the next matching separator is taken as the string. Matching works the same way as before — [, {, and < match ], }, and >, respectively, while any other character matches itself. Note that expansions must be quoted in the string to prevent them from being expanded before globbing is done, no cap. string is then executed as shell code, and the string globqual is appended to the array zsh_eval_context for the duration of execution.

During the execution of string, the filename currently being tested is available in the parameter REPLY; you may alter that parameter to a string that gets inserted into the list instead of the original filename. Also, the parameter reply may be set to an array or a string, which overrides the value of REPLY. If you set it to an array, the latter is inserted into the command line word by word. Big main character energy 💅

For example, suppose a directory contains a single file lonely. Then the expression *(e:'reply=(${REPLY}{1,2})':) will cause the words lonely1 and lonely2 to be inserted into the command line. Note the quoting of string, bestie — it matters.

The form +cmd has the same effect, but there are no delimiters around cmd. Instead, cmd is taken as the longest sequence of characters following the + that are alphanumeric or underscore. Typically cmd will be the name of a shell function that contains the appropriate test. For example:

nt() { [[ $REPLY -nt $NTREF ]] }
NTREF=reffile
ls -ld -- *(+nt)

...lists all files in the directory that have been modified more recently than reffile. Chef's kiss 🔥

ddev — files on the device dev 🦋
l[-|+]ct — files having a link count less than ct (-), greater than ct (+), or equal to ct ⭐
U — files owned by the effective user ID 👑
G — files owned by the effective group ID 💜
uid — files owned by user ID id, if that's a number. Otherwise, id specifies a user name: the character after the u gets taken as a separator and the string between it and the next matching separator gets taken as a user name. The starting separators [, {, and < match the final separators ], }, and >, respectively, and any other character matches itself. The selected files are those owned by this user. For example, u:foo: or u[foo] selects files owned by user foo. Cute AND functional 🌸
gid — like uid but with group IDs or names 🎀
a[Mwhms][-|+]n — files accessed exactly n days ago.
- Files accessed within the last n days are selected using a negative value for n (-n).
- Files accessed more than n days ago are selected by a positive n value (+n).

Optional unit specifiers M, w, h, m or s (e.g. ah5) cause the check to be performed with months (of 30 days), weeks, hours, minutes or seconds instead of days, respectively. An explicit d for days is also allowed, king behavior.

Any fractional part of the difference between the access time and the current part in the appropriate units gets ignored in the comparison. For instance, echo *(ah-5) would echo files accessed within the last five hours, while echo *(ah+5) would echo files accessed at least six hours ago, as times strictly between five and six hours are treated as five hours. Math girlie hours 🧮✨

m[Mwhms][-|+]n — like the file access qualifier, except it uses the file modification time 🕐
c[Mwhms][-|+]n — like the file access qualifier, except it uses the file inode change time 🕑
L[+|-]n — files less than n bytes (-), more than n bytes (+), or exactly n bytes in length 📏

If this flag is directly followed by a size specifier k (K), m (M), or p (P) (e.g. Lk-50) the check is performed with kilobytes, megabytes, or blocks (of 512 bytes) instead. (On some systems additional specifiers are available for gigabytes, g or G, and terabytes, t or T.) If a size specifier is used, a file is regarded as "exactly" the size if the file size rounded up to the next unit is equal to the test size. Hence *(Lm1) matches files from 1 byte up to 1 Megabyte inclusive. Note also that the set of files "less than" the test size only includes files that would not match the equality test; hence *(Lm-1) only matches files of zero size. The way this is lowkey kinda unhinged but it works 💀

^ — negates all qualifiers following it 🙅‍♀️
- — toggles between making the qualifiers work on symbolic links (that's the default) and on the files they point to, if any; any symbolic link for whose target the stat system call fails (whatever the cause of the failure) is treated as a file in its own right 🪞
M — sets the MARK_DIRS option for the current pattern 🏷️
T — appends a trailing qualifier mark to the filenames, analogous to the LIST_TYPES option, for the current pattern (overrides M) ✨
N — sets the NULL_GLOB option for the current pattern 🫥
D — sets the GLOB_DOTS option for the current pattern 👀
n — sets the NUMERIC_GLOB_SORT option for the current pattern 🔢
Yn — enables short-circuit mode: the pattern will expand to at most n filenames. If more than n matches exist, only the first n matches in directory traversal order will be considered 🏃‍♀️💨

Implies oN when no oc qualifier is used.

oc — specifies how the names of the files should be sorted, bestie. The following values of c sort in the following ways:
- n — by name.
- L — by the size (length) of the files.
- l — by number of links.
- a — by time of last access, youngest first.
- m — by time of last modification, youngest first.
- c — by time of last inode change, youngest first.
- d — by directories: files in subdirectories appear before those in the current directory at each level of the search — this is best combined with other criteria, for example odon to sort on names for files within the same directory.
- N — no sorting is performed. Chaos mode, honestly.
- estring
- +cmd — sort by shell code (see below).

Note that the modifiers ^ and - are used, so *(^-oL) gives a list of all files sorted by file size in descending order, following any symbolic links. Unless oN is used, multiple order specifiers may occur to resolve ties 🌈

The default sorting is n (by name) unless the Y glob qualifier is used, in which case it's N (unsorted).

oe and o+ are special cases; they're each followed by shell code, delimited as for the e glob qualifier and the + glob qualifier respectively (see above). The code is executed for each matched file with the parameter REPLY set to the name of the file on entry and globsort appended to zsh_eval_context. The code should modify the parameter REPLY in some fashion. On return, the value of the parameter is used instead of the file name as the string on which to sort. Unlike other sort operators, oe and o+ may be repeated — but note that the maximum number of sort operators of any kind that may appear in any glob expression is 12. Twelve!! That's the limit, don't push her 😭

Oc — like o, but sorts in descending order; i.e. *(^oc) is the same as *(Oc) and *(^Oc) is the same as *(oc); Od puts files in the current directory before those in subdirectories at each level of the search. Flip it and reverse it 💅🔁
[beg[,end]] — the lil slice-y guy: it says which of the matched filenames actually make it into the returned list. Syntax is literally the same as array subscripts, bestie. beg and the optional end can be mathematical expressions, and just like in parameter subscripting they can go negative to count backward from the last match. Example: *(-OL[1,3]) hands you a list of the names of the three largest files. 💅
Pstring — ooh this one's cute. string gets prepended to each glob match as a separate word. string is delimited the exact same way as arguments to the e glob qualifier we talked about above. The qualifier can be repeated, and the words get prepended separately, so the resulting command line contains the words in the same order you gave them in the list of glob qualifiers.

A typical use (and honestly the main character use) is prepending an option before all occurrences of a file name — for example, the pattern *(P:-f:) produces the command line arguments -f file1 -f file2 ... 🫶

If the modifier ^ is active, then string gets appended instead of prepended. Prepending and appending are done independently, so both can be used on the same glob expression — for example by writing *(P:foo:^P:bar:^P:baz:), which produces the command line arguments foo baz file1 bar ... ✨

More than one of these lists can be combined, separated by commas. The whole list matches if at least one of the sublists matches (they're or'ed — the qualifiers inside the sublists are and'ed). Some qualifiers, though, affect all matches generated, no matter which sublist they showed up in. Those divas are: M, T, N, D, n, o, O, and the subscripts given in brackets ([...]). 👑

If a : appears in a qualifier list, the rest of the expression in the parenthesis gets interpreted as a modifier (see the section Modifiers in the section History Expansion). Each modifier has to be introduced by its own separate :. Also note: the result after modification does not have to be an existing file. The name of any existing file can be followed by a modifier of the form (:...) even if no actual filename generation is performed — although heads up, the presence of the parentheses causes the entire expression to be subjected to any global pattern matching options such as NULL_GLOB. 🙈

So like:

ls -ld -- *(-/)

lists all directories and symbolic links that point to directories, and

ls -ld -- *(-@)

lists all broken symbolic links, and

ls -ld -- *(%W)

lists all world-writable device files in the current directory, and

ls -ld -- *(W,X)

lists all files in the current directory that are world-writable or world-executable, and

print -rC1 /tmp/foo*(u0^@:t)

outputs the basename of all root-owned files beginning with the string foo in /tmp, ignoring symlinks, and

ls -ld -- *.*~(lex|parse).[ch](^D^l1)

lists all files having a link count of one whose names contain a dot (but not those starting with a dot, since GLOB_DOTS is explicitly switched off) except for lex.c, lex.h, parse.c and parse.h. 💀

print -rC1 b*.pro(#q:s/pro/shmo/)(#q.:s/builtin/shmiltin/)

demonstrates how colon modifiers and other qualifiers may be chained together. The ordinary qualifier . is applied first, then the colon modifiers in order from left to right. So if EXTENDED_GLOB is set and the base pattern matches the regular file builtin.pro, the shell will print shmiltin.shmo. iconic behavior, no notes. 🌈

zsh 5.9.2                        July 12, 2026                       ZSHEXPN(1)
────────────────────────────────────────────────────────────────────────────────
ZSHPARAM(1)                 General Commands Manual                 ZSHPARAM(1)

✨ NAME ✨

zshparam — zsh parameters 🎀

💖 DESCRIPTION 💖

So a parameter has a name, a value, and a bunch of attributes. A name can be any sequence of alphanumeric characters and underscores, or the single characters *, @, #, ?, -, $, or !. And a parameter whose name begins with an alphanumeric or underscore is also called a variable. that's the whole vibe check right there.

The attributes of a parameter decide the type of its value — often called the parameter type or variable type — and they also control other processing that may get applied to the value when you reference it. The value type can be:

a scalar (a string, an integer, or a floating point number),
an array (indexed numerically), or
an associative array (an unordered set of name-value pairs, indexed by name — also called a hash).

Named scalar parameters may have the exported, -x, attribute, which copies them into the process environment, and that environment then gets passed from the shell to any new processes it starts. Exported parameters are called environment variables. The shell also imports environment variables at startup time and automatically marks the corresponding parameters as exported. Some environment variables are not imported — for reasons of security, or because they'd interfere with the correct operation of other shell features. 🛡️

Parameters may also be special — meaning they have a predetermined meaning to the shell. Special parameters cannot have their type changed, and their readonly attribute can't be turned off. And if a special parameter is unset and then later recreated, the special properties will be retained. no cap, they come back with their powers intact. ⭐

To declare the type of a parameter, or to assign a string or numeric value to a scalar parameter, use the typeset builtin.

You can also assign the value of a scalar parameter by writing:

name=value

In scalar assignment, value is expanded as a single string, in which the elements of arrays are joined together; filename expansion is not performed unless the option GLOB_ASSIGN is set.

When the integer attribute -i, or a floating point attribute -E or -F, is set for name, the value is subject to arithmetic evaluation. And on top of that: by replacing = with +=, a parameter can be incremented or appended to. See the section Array Parameters and Arithmetic Evaluation (in zshmisc(1)) for additional forms of assignment. 🧚

Note that assignment may implicitly change the attributes of a parameter. For example, assigning a number to a variable in arithmetic evaluation may change its type to integer or float, and with GLOB_ASSIGN assigning a pattern to a variable may change its type to an array. surprise type change, lowkey rude but it's canon.

To reference the value of a parameter, write $name or ${name}. See Parameter Expansion in zshexpn(1) for complete details. That section also explains the effect of the difference between scalar and array assignment on parameter expansion. 💜

🦋 ARRAY PARAMETERS 🦋

To assign an array value, write one of:

set -A name value ...
name=(value ...)
name=([key]=value ...)

If no parameter name exists, an ordinary array parameter is created. If the parameter name exists and is a scalar, it gets replaced by a new array. bye bestie, hello array. 👋

In the third form, key is an expression that will be evaluated in arithmetic context (in its simplest form, an integer) that gives the index of the element to be assigned with value. In this form, any elements not explicitly mentioned that come before the largest index to which a value is assigned get assigned an empty string. The indices may be in any order. Note that this syntax is strict: [ and ]= must not be quoted, and key may not consist of the unquoted string ]=, but is otherwise treated as a simple string. The enhanced forms of subscript expression that you can use when directly subscripting a variable name — described in the section Array Subscripts below — are not available here. 💀

The syntaxes with and without the explicit key may be mixed. An implicit key is deduced by incrementing the index from the previously assigned element. Note that it is not treated as an error if later assignments in this form overwrite earlier assignments. live your life, overwrite things.

For example, assuming the option KSH_ARRAYS is not set, the following:

array=(one [3]=three four)

causes the array variable array to contain four elements one, an empty string, three and four, in that order. look at her go ✨

In the forms where only value is specified, full command line expansion is performed.

In the [key]=value form, both key and value undergo all forms of expansion allowed for single word shell expansions (this does not include filename generation); these are as performed by the parameter expansion flag (e) as described in zshexpn(1). Nested parentheses may surround value and are included as part of the value, which is joined into a plain string — this differs from ksh, which allows the values themselves to be arrays. A future version of zsh may support that. 🫶 To cause the brackets to be interpreted as a character class for filename generation, and therefore to treat the resulting list of files as a set of values, quote the equal sign using any form of quoting. Example:

name=([a-z]'='*)

To append to an array without changing the existing values, use one of the following:

name+=(value ...)
name+=([key]=value ...)

In the second form, key may specify an existing index as well as an index off the end of the old array; any existing value is overwritten by value. Also, it is possible to use [key]+=value to append to the existing value at that index. 💅

Within the parentheses on the right hand side of either form of the assignment, newlines and semicolons are treated the same as white space, separating individual values. Any consecutive sequence of such characters has the same effect. so go ahead, format it however, it's giving flexible. 🌸

Ordinary array parameters may also be explicitly declared with:

typeset -a name

Associative arrays must be declared before assignment, by using:

typeset -A name

When name refers to an associative array, the list in an assignment is interpreted as alternating keys and values:

set -A name key value ...
name=(key value ...)
name=([key]=value ...)

Note that only one of the two syntaxes above may be used in any given assignment; the forms may not be mixed. This is unlike the case of numerically indexed arrays. 🙈

Every key must have a value in this case. Note that this assigns to the entire array, deleting any elements that do not appear in the list. The append syntax may also be used with an associative array:

name+=(key value ...)
name+=([key]=value ...)

This adds a new key/value pair if the key isn't already there — and if it is already there, it cheerfully replaces the value for that key. In the second form you can also do [key]+=value to append to whatever value is already sitting at that key. The expansion works exactly the same as the corresponding forms for normal arrays, as described above. Consistent, iconic, no notes.

To create an empty array (associative arrays included, bestie), use one of these:

set -A name
name=()

🎀 Array Subscripts

Individual elements of an array can be picked out with a subscript. A subscript of the form [exp] selects the single element exp, where exp is an arithmetic expression that gets arithmetic expansion applied to it as if it were wrapped in $((...)). Elements are numbered starting at 1 — unless the KSH_ARRAYS option is set, in which case they start from zero. (Yes, this is the chaos. Yes, you get used to it. 💅)

Subscripts can live inside the braces that delimit a parameter name, so ${foo[2]} is equivalent to $foo[2]. But if KSH_ARRAYS is set, the braced form is the only one that works — brackets outside braces just aren't treated as subscripts anymore.

If KSH_ARRAYS is not set, then by default, reading an array element with a subscript that evaluates to zero gives you an empty string, while trying to write such an element is treated as an error. For backward compatibility, the KSH_ZERO_SUBSCRIPT option can be set to make subscript values 0 and 1 equivalent — see the description of the option in zshoptions(1).

Same subscripting syntax for associative arrays, except that no arithmetic expansion is applied to exp. The parsing rules for arithmetic expressions still apply though, which totally affects how you have to protect certain special characters. See Subscript Parsing below for the tea. 🫖

A subscript of the form [*] or [@] evaluates to all elements of an array — and there's no difference between the two except when they show up inside double quotes. "$foo[*]" evaluates to "$foo[1] $foo[2] ...", whereas "$foo[@]" evaluates to "$foo[1]" "$foo[2]" .... For associative arrays, [*] or [@] evaluate to all the values, in no particular order. Note that this does not substitute the keys — see the docs for the k flag under Parameter Expansion Flags in zshexpn(1) for the full rundown. When an array parameter is referenced as $name (no subscript at all), it evaluates to $name[*], unless KSH_ARRAYS is set, in which case it evaluates to ${name[0]} — and for an associative array that means the value of the key 0, which might not even exist even if other keys have values. 😭

A subscript of the form [exp1,exp2] selects all elements from exp1 to exp2, inclusive. (Associative arrays are unordered, so they don't support ranges — sorry, them's the rules.) If one of the subscripts evaluates to a negative number, say -n, then the *n*th element from the end of the array is used. So $foo[-3] is the third element from the end of the array foo, and $foo[1,-1] is the same as $foo[*]. Slay.

Subscripting also works on non-array values, in which case the subscripts specify a substring to extract. For example, if FOO is set to foobar, then echo $FOO[2,5] prints ooba. Note that some forms of subscripting described below do pattern matching, and in that case the substring extends from the start of the match of the first subscript to the end of the match of the second subscript. Like:

string="abcdefghijklm"
print ${string[(r)d?,(r)h?]}

prints defghi. That's an obvious generalisation of the rule for single-character matches. For a single subscript, only a single character is referenced — not the range of characters covered by the match.

Also note that in substring operations the second subscript is handled differently by the r and R subscript flags: the former takes the shortest match as the length, the latter the longest. So in the former case a * at the end is redundant, while in the latter case it matches the whole remainder of the string. This doesn't affect the result of the single-subscript case, since the length of the match is irrelevant there anyway.

💖 Array Element Assignment

A subscript can go on the left side of an assignment, like this:

name[exp]=value

In this form of assignment, the element or range specified by exp gets replaced by the expression on the right side. An array (but not an associative array) can be created by assigning to a range or element. Arrays don't nest, so assigning a parenthesized list of values to an element or range changes the number of elements in the array — the other elements shift over to make room for the new values. (This is not supported for associative arrays, no cap.)

This syntax also works as an argument to the typeset command:

typeset "name[exp]"=value

The value may not be a parenthesized list in this case — only single-element assignments can be made with typeset. Note that the quotes are necessary here to stop the brackets from being interpreted as filename generation operators. The noglob precommand modifier could be used instead.

To delete an element of an ordinary array, assign () to that element. To delete an element of an associative array, use the unset command:

unset "name[exp]"

✨ Subscript Flags

If the opening bracket — or the comma in a range — in any subscript expression is directly followed by an opening parenthesis, the string up to the matching closing one is treated as a list of flags, as in name[(flags)exp].

The flags s, n and b take an argument; the delimiter is shown below as : but any character — or the matching pairs (...), {...}, [...], or <...> — may be used. Big caveat though: <...> can only be used if the subscript is inside a double-quoted expression or a parameter substitution enclosed in braces, because otherwise the whole thing gets interpreted as a redirection. 💀

The flags currently understood are:

w — If the parameter you're subscripting is a scalar, this flag makes subscripting work on words instead of characters. The default word separator is whitespace. When combined with the i or I flag, the effect is to produce the index of the first character of the first/last word which matches the given pattern — and note that a failed match in this case always yields 0.
s:string: — This gives the string that separates words (for use with the w flag). The delimiter character : is arbitrary; see above.
p — Recognize the same escape sequences as the print builtin in the string argument of a subsequent s flag.
f — If the parameter you're subscripting is a scalar, this flag makes subscripting work on lines instead of characters, i.e. with elements separated by newlines. It's a shorthand for pws:\n:.
r — Reverse subscripting: if this flag is given, exp is taken as a pattern and the result is the first matching array element, substring or word (if the parameter is an array, if it's a scalar, or if it's a scalar and the w flag is given, respectively). The subscript used is the number of the matching element, so pairs of subscripts like $foo[(r)??,3] and $foo[(r)??,(r)f*] are possible — as long as the parameter isn't an associative array. If the parameter is an associative array, only the value part of each pair is compared to the pattern, and the result is that value.

If a search through an ordinary array failed, the search sets the subscript to one past the end of the array, and hence ${array[(r)pattern]} will substitute the empty string. So you can test whether a search succeeded by using the (i) flag, like this (assuming the option KSH_ARRAYS is not in effect):

[[ ${array[(i)pattern]} -le ${#array} ]]

If KSH_ARRAYS is in effect, replace -le with -lt. 💅

R — Like r, but gives the last match. For associative arrays, gives all possible matches. May be used for assigning to ordinary array elements, but not for assigning to associative arrays. On failure, for normal arrays this has the effect of returning the element corresponding to subscript 0; that's empty unless one of the options KSH_ARRAYS or KSH_ZERO_SUBSCRIPT is in effect.

Note that in subscripts with both r and R, pattern characters are active even if they were substituted for a parameter — regardless of the setting of GLOB_SUBST, which controls this feature in normal pattern matching. The flag e can be added to inhibit pattern matching. But as this flag doesn't inhibit other forms of substitution, care is still required — using a parameter to hold the key gives you the result you actually want:

key2='original key'
print ${array[(Re)$key2]}

i — Like r, but gives the index of the match instead; this may not be combined with a second argument. On the left side of an assignment, behaves like r. For associative arrays, the key part of each pair is compared to the pattern, and the first matching key found is the result. On failure it substitutes the length of the array plus one, as discussed under the description of r — or the empty string for an associative array.

Note: Although i may be applied to a scalar substitution to find the offset of a substring, the results are likely to be misleading when searching within substitutions that yield an empty string, or when searching for the empty substring. 🙈

I — Like i, but gives the index of the last match, or all possible matching keys in an associative array. On failure it substitutes 0, or the empty string for an associative array. This flag is best when testing for values or keys that don't exist.

Note: If the option KSH_ARRAYS is in effect and no match is found, the result is indistinguishable from the case when the first element of the array matches. 😬

k — If used in a subscript on an associative array, this flag causes the keys to be interpreted as patterns, and returns the value for the first key found where exp is matched by the key. Note this could be any such key, since no ordering of associative arrays is defined. This flag does not work on the left side of an assignment to an associative array element. If used on another type of parameter, this behaves like r.
K — On an associative array this is like k but returns all values where exp is matched by the keys. On other types of parameters, this has the same effect as R.
n:expr: — pair it with r, R, i or I and it makes them hand you the nth match (or nth-from-the-end match) whenever expr evaluates to n. If your array is associative this flag is literally ignored, no cap. And bestie, the delimiter : is arbitrary — see above 💅
b:expr: — same deal but for beginnings: with r, R, i or I it makes them start at the nth (or nth-last) element, word or character when expr evaluates to n. Also ignored for associative arrays. The delimiter : is arbitrary, see above again 🫶
e — makes any pattern matching that would happen on the subscript use plain string matching instead. So ${array[(re)*]} matches only the array element whose value is literally *. Note that other forms of substitution, like parameter substitution, are not inhibited. This flag can also force * or @ to be read as a single key instead of a reference to all values — it works for either purpose on the left side of an assignment ✨

See Parameter Expansion Flags (zshexpn(1)) for even more ways to manipulate the results of array subscripting 👑

🌀 Subscript Parsing

Okay so this whole discussion mostly applies to associative array key strings and to patterns used for reverse subscripting (the r, R, i, etc. flags), but it can also affect parameter substitutions that show up as part of an arithmetic expression in an ordinary subscript. Buckle up 🎢

To dodge subscript parsing limitations when assigning to associative array elements, use the append syntax:

aa+=('key with "*strange*" characters' 'value string')

The one rule to remember when writing a subscript expression: all text between the opening [ and the closing ] is interpreted as if it were inside double quotes (see zshmisc(1)). But unlike double quotes — which normally can't nest — subscript expressions may appear inside double-quoted strings or inside other subscript expressions (or both!), so the rules have two big differences 👀

Difference number one: brackets ([ and ]) must appear as balanced pairs in a subscript expression unless they're preceded by a backslash (\). So inside a subscript expression (and unlike true double-quoting) the sequence \[ becomes [, and likewise \] becomes ]. This holds even when a backslash isn't normally required — for example, the pattern [^[] (matching any character other than an open bracket) should be written [^\[] in a reverse-subscript pattern. BUT note that \[^\[\] and even \[^[] mean the same thing, because backslashes are always stripped when they appear before brackets! 😭

The same rule applies to parentheses (( and )) and braces ({ and }): they must appear either in balanced pairs or preceded by a backslash, and backslashes that protect parens or braces get removed during parsing. Why? Because parameter expansions may be surrounded by balanced braces, and subscript flags are introduced by balanced parentheses. It's giving structural integrity 💀

Difference number two: a double-quote (") may appear as part of a subscript expression without being preceded by a backslash, and therefore the two characters \" stay as two characters in the subscript (in true double-quoting, \" becomes "). However, thanks to standard shell quoting rules, any double-quotes that appear must occur in balanced pairs unless preceded by a backslash. That makes it harder to write a subscript expression with an odd number of double-quote characters, but the reason for the difference is so that when a subscript expression appears inside true double-quotes, you can still write \" (instead of \\\") for ". iykyk 🌸

To use an odd number of double quotes as a key in an assignment, use the typeset builtin and an enclosing pair of double quotes; to refer to that key's value, again use double quotes:

typeset -A aa
typeset "aa[one\"two\"three\"quotes]"=QQQ
print "$aa[one\"two\"three\"quotes]"

Super important to note: the quoting rules do not change when a parameter expansion with a subscript is nested inside another subscript expression. Meaning you don't need extra backslashes in the inner subscript expression — they get removed only once, from the innermost subscript outwards. Parameters are also expanded innermost-subscript-first, as each expansion is encountered left to right in the outer expression 🧚

A further complication comes from a way subscript parsing is not different from double-quote parsing. Just like in true double-quoting, the sequences \* and \@ remain as two characters when they appear in a subscript expression. To use a literal * or @ as an associative array key, you gotta bust out the e flag:

typeset -A aa
aa[(e)*]=star
print $aa[(e)*]

One last detail for when reverse subscripting happens: parameters appearing in the subscript expression are first expanded, and then the complete expression is interpreted as a pattern. Two effects from that — first, parameters behave as if GLOB_SUBST were on (and it cannot be turned off); second, backslashes are interpreted twice, once when parsing the array subscript and again when parsing the pattern. So in a reverse subscript you need four backslashes to make a single backslash match literally in the pattern. For complex patterns it's usually easiest to assign the desired pattern to a parameter and then refer to that parameter in the subscript, because then the backslashes, brackets, parentheses, etc. are only seen once, when the complete expression gets converted to a pattern. To match a parameter's value literally in a reverse subscript rather than as a pattern, use ${(q)name} (see zshexpn(1)) to quote the expanded value 💖

Heads up: the k and K flags are reverse subscripting for an ordinary array, but they are not reverse subscripting for an associative array! (For an associative array, the keys in the array itself are interpreted as patterns by those flags; the subscript is a plain string in that case.) 🙈

One final note, not directly about subscripting: the numeric names of positional parameters (described below) are parsed specially, so for example $2foo is equivalent to ${2}foo. Therefore, to use subscript syntax to pull a substring out of a positional parameter, the expansion must be surrounded by braces. For example, ${2[3,5]} evaluates to the third through fifth characters of the second positional parameter, but $2[3,5] is the entire second parameter concatenated with the filename generation pattern [3,5]. Main character energy vs. chaos, honestly 😭

🎀 POSITIONAL PARAMETERS

The positional parameters are your access pass to the command-line arguments of a shell function, shell script, or the shell itself — see the section Invocation, and also the section Functions. The parameter n, where n is a number, is the *n*th positional parameter. The parameter $0 is a special case, see the section Parameters Set By The Shell ⭐

The parameters *, @ and argv are arrays containing all the positional parameters; so $argv[n], etc., is equivalent to just $n. Note that the options KSH_ARRAYS or KSH_ZERO_SUBSCRIPT apply to these arrays too, so with either of those set, ${argv[0]} is equivalent to $1 and so on 🦋

Positional parameters can be changed after the shell or function starts by using the set builtin, by assigning to the argv array, or by direct assignment of the form n=value where n is the number of the positional parameter you want to change. That also creates (with empty values) any positions from 1 to n that don't already have values. And because the positional parameters form an array, an array assignment of the form n=(value ...) is allowed — it shifts all the values at positions greater than n by however many positions are needed to fit the new values in. Sliding into place 💅

🌈 LOCAL PARAMETERS

Shell function executions delimit scopes for shell parameters. (Parameters are dynamically scoped, FYI.) The typeset builtin, and its alternative forms declare, integer, local and readonly (but not export), can declare a parameter as local to the innermost scope ✨

When a parameter is read or assigned to, the innermost existing parameter of that name gets used. (That is, the local parameter hides any less-local parameter.) However, assigning to a non-existent parameter, or declaring a new parameter with export, causes it to be created in the outermost scope 👑

Local parameters disappear when their scope ends. unset can delete a parameter while it's still in scope; any outer parameter of the same name stays hidden 🙈

Special parameters can be made local too; they keep their special attributes unless either the existing or the newly-created parameter has the -h (hide) attribute. This may have unexpected effects: there's no default value, so if there's no assignment at the point the variable is made local, it'll be set to an empty value (or zero in the case of integers). The following:

typeset PATH=/new/directory:$PATH

is valid for temporarily letting the shell — or programs called from it — find the programs in /new/directory inside a function 🫶

Note that the restriction in older versions of zsh, where local parameters were never exported, has been removed. Glow-up achieved 💖

👑 PARAMETERS SET BY THE SHELL

In the parameter lists that follow, the mark <S> means the parameter is special. <Z> means the parameter doesn't exist when the shell initializes in sh or ksh emulation mode ⭐

The parameters !, #, *, -, ?, @, $, ARGC, HISTCMD, LINENO, PPID, status, TTYIDLE, zsh_eval_context, ZSH_EVAL_CONTEXT, and ZSH_SUBSHELL are read-only and therefore cannot be restored by the user, so they're not output by typeset -p. This also applies to many read-only parameters loaded from modules 🔒

The following parameters are automatically set by the shell:

! <S> — the process ID of the last command started in the background with &, put into the background with the bg builtin, or spawned with coproc.
# <S> — the number of positional parameters in decimal. Note that some confusion may occur with the syntax $#param, which substitutes the length of param. Use ${#} to resolve ambiguities. In particular, the sequence $#-... in an arithmetic expression is interpreted as the length of the parameter -, q.v. 😅
ARGC <S> <Z> — same as #.
$ <S> — bestie, this is the process ID of this shell, set right when the shell initializes. And here's the tea 🫖: processes forked from the shell without executing a new program, like command substitutions and commands grouped with (...), are subshells that duplicate the current shell, and thus they substitute the same value for $$ as their parent shell. Same ID, no cap.
- <S> — the flags handed to the shell on invocation, or by the set or setopt commands.
* <S> — an array holding the positional parameters. iconic, honestly.
argv <S> <Z> — same as *. Assigning to argv changes the local positional parameters, but plot twist 🌀: argv is not itself a local parameter. Deleting argv with unset in any function deletes it everywhere, although only the innermost positional parameter array is deleted (so * and @ in other scopes are not affected).
@ <S> — same as argv[@], even when argv is not set. yes she works even when her friend is missing 💅
? <S> — the exit status returned by the last command. your daily vibe check on whether things went well.
0 <S> — the name used to invoke the current shell, or as set by the -c command line option upon invocation. And if the FUNCTION_ARGZERO option is set, $0 gets set upon entry to a shell function to the name of the function, and upon entry to a sourced script to the name of the script, then reset to its previous value when the function or script returns. shapeshifter behavior ✨
status <S> <Z> — same as ?.
pipestatus <S> <Z> — an array containing the exit statuses returned by all commands in the last pipeline. every single one, we love a thorough queen 👑
_ <S> — the last argument of the previous command. Also, this parameter is set in the environment of every command executed to the full pathname of the command. multitasking legend.
CPUTYPE — the machine type (microprocessor class or machine model), as determined at run time.
EGID <S> — the effective group ID of the shell process. If you have sufficient privileges, you may change the effective group ID of the shell process by assigning to this parameter. Also (assuming sufficient privileges), you may start a single command with a different effective group ID by (EGID=gid; command) 💪

If this is made local, it is not implicitly set to 0, but may be explicitly set locally.

EUID <S> — the effective user ID of the shell process. If you have sufficient privileges, you may change the effective user ID of the shell process by assigning to this parameter. Also (assuming sufficient privileges), you may start a single command with a different effective user ID by (EUID=uid; command) 💪

If this is made local, it is not implicitly set to 0, but may be explicitly set locally.

ERRNO <S> — the value of errno (see errno(3)) as set by the most recently failed system call. This value is system dependent and is intended for debugging purposes. It's also useful with the zsh/system module, which allows the number to be turned into a name or message. 🔍

To use this parameter, it must first be assigned a value (typically 0 (zero)). It is initially unset for scripting compatibility. so assign first, bestie, or she won't show up.

FUNCNEST <S> — integer. If greater than or equal to zero, it's the maximum nesting depth of shell functions. When it's exceeded, an error is raised at the point where a function is called. The default value is determined when the shell is configured, but is typically 500. Increasing the value increases the danger of a runaway function recursion causing the shell to crash 😭. Setting a negative value turns off the check.
GID <S> — the real group ID of the shell process. If you have sufficient privileges, you may change the group ID of the shell process by assigning to this parameter. Also (assuming sufficient privileges), you may start a single command under a different group ID by (GID=gid; command) 💪

If this is made local, it is not implicitly set to 0, but may be explicitly set locally.

HISTCMD — the current history event number in an interactive shell — in other words, the event number for the command that caused $HISTCMD to be read. If the current history event modifies the history, HISTCMD changes to the new maximum history event number. she keeps up, always 🏃‍♀️
HOST — the current hostname.
LINENO <S> — the line number of the current line within the current script, sourced file, or shell function being executed, whichever was started most recently. Note that in the case of shell functions, the line number refers to the function as it appeared in the original definition, not necessarily as displayed by the functions builtin. so don't @ me if it looks different in the builtin 🙈
LOGNAME — if the corresponding variable is not set in the environment of the shell, it's initialized to the login name corresponding to the current login session. This parameter is exported by default, but this can be disabled using the typeset builtin. The value is set to the string returned by the getlogin(3) system call if that's available. ✨
MACHTYPE — the machine type (microprocessor class or machine model), as determined at compile time.
OLDPWD — the previous working directory. This is set when the shell initializes and whenever the directory changes. your little "wait where was I" bookmark 🎀
OPTARG <S> — the value of the last option argument processed by the getopts command.
OPTIND <S> — the index of the last option argument processed by the getopts command.
OSTYPE — the operating system, as determined at compile time.
PPID <S> — the process ID of the parent of the shell, set when the shell initializes. As with $$, the value does not change in subshells created as a duplicate of the current shell. she stays loyal to her parent 🫶
PWD — the present working directory. This is set when the shell initializes and whenever the directory changes.
RANDOM <S> — a pseudo-random integer from 0 to 32767, newly generated each time this parameter is referenced. The random number generator can be seeded by assigning a numeric value to RANDOM 🎲

The values of RANDOM form an intentionally-repeatable pseudo-random sequence; subshells that reference RANDOM will result in identical pseudo-random values unless the value of RANDOM is referenced or seeded in the parent shell in between subshell invocations. so it's not actually chaos, it's ✨curated✨ chaos.

SECONDS <S> — the number of seconds since shell invocation. If this parameter is assigned a value, then the value returned upon reference will be the value that was assigned plus the number of seconds since the assignment. ⏱️

Unlike other special parameters, the type of the SECONDS parameter can be changed using the typeset command. The type may be changed only to one of the floating point types or back to integer. For example, typeset -F SECONDS causes the value to be reported as a floating point number. The value is available to microsecond accuracy, although the shell may show more or fewer digits depending on the use of typeset. See the documentation for the builtin typeset in zshbuiltins(1) for more details.

SHLVL <S> — incremented by one each time a new shell is started. nesting doll realness 🪆
signals — an array containing the names of the signals. Note that with the standard zsh numbering of array indices, where the first element has index 1, the signals are offset by 1 from the signal number used by the operating system. For example, on typical Unix-like systems HUP is signal number 1, but is referred to as $signals[2]. This is because of EXIT at position 1 in the array, which is used internally by zsh but is not known to the operating system. On many systems there is a block of reserved or unused signal numbers before the POSIX real-time signals, so the array index can't be used as an accurate indicator of their signal number. Use, for example, kill -l SIGRTMIN instead. 💀
TRY_BLOCK_ERROR <S> — in an always block, indicates whether the preceding list of code caused an error. The value is 1 to indicate an error, 0 otherwise. It may be reset, clearing the error condition. See Complex Commands in zshmisc(1).
TRY_BLOCK_INTERRUPT <S> — this variable works in a similar way to TRY_BLOCK_ERROR, but represents the status of an interrupt from the signal SIGINT, which typically comes from the keyboard when the user types ^C. If set to 0, any such interrupt will be reset; otherwise, the interrupt is propagated after the always block. 🛑

Note that it's possible that an interrupt arrives during the execution of the always block; this interrupt is also propagated.

TTY — the name of the tty associated with the shell, if any.
TTYIDLE <S> — the idle time of the tty associated with the shell in seconds, or -1 if there's no such tty. 😴
UID <S> — the real user ID of the shell process. If you have sufficient privileges, you may change the user ID of the shell by assigning to this parameter. Also (assuming sufficient privileges), you may start a single command under a different user ID by (UID=uid; command) 💪

If this is made local, it is not implicitly set to 0, but may be explicitly set locally.

USERNAME <S> — the username corresponding to the real user ID of the shell process. If you have sufficient privileges, you may change the username (and also the user ID and group ID) of the shell by assigning to this parameter. Also (assuming sufficient privileges), you may start a single command under a different username (and user ID and group ID) by (USERNAME=username; command) 👑
VENDOR — the vendor, as determined at compile time.
zsh_eval_context <S> <Z> (ZSH_EVAL_CONTEXT <S>) — an array (colon-separated list) indicating the context of shell code that's being run. Each time a piece of shell code that is stored within the shell is executed, a string is temporarily appended to the array to indicate the type of operation being performed. Read in order, the array gives an indication of the stack of operations being performed, with the most immediate context last. 🥞

Note that the variable does not give information on syntactic context such as pipelines or subshells. Use $ZSH_SUBSHELL to detect subshells. different tool, different job, iykyk 💅

so the context can literally be any one of these, bestie — here's the full rundown 👇

cmdarg — code that came in through the -c option on the command line that invoked the shell.
cmdsubst — command substitution, like the `...` or $(...) situation.
equalsubst — file substitution using the =(...) thing.
eval — code that the eval builtin ran.
evalautofunc — code executed with the KSH_AUTOLOAD mechanism so an autoloaded function could get defined.
fc — code from the shell history that the -e option to the fc builtin ran.
file — lines of code being read straight out of a file, like when source does its thing.
filecode — lines of code read from a .zwc file instead of straight from the source file.
globqual — code run by the e or + glob qualifier.
globsort — code run to order files by the o glob qualifier.
insubst — file substitution using the <(...) construct.
loadautofunc — code read directly from a file to define an autoloaded function.
outsubst — file substitution using the >(...) construct.
sched — code the sched builtin ran.
shfunc — a shell function.
stty — code passed to stty via the STTY environment variable. Normally this goes straight to the system's stty command, so honestly you're unlikely to ever see this value in the wild 🙈.
style — code executed as part of a style that the zstyle builtin pulled from the zsh/zutil module.
toplevel — the highest execution level of a script or interactive shell.
trap — code run as a trap defined by the trap builtin. Traps defined as functions get the context shfunc instead. And since traps are asynchronous, they may have a totally different hierarchy from other code — vibe check passed, but keep that in mind 💅.
zpty — code the zpty builtin ran from the zsh/zpty module.
zregexparse-guard — code executed as a guard by the zregexparse command from the zsh/zutil module.
zregexparse-action — code executed as an action by the zregexparse command from the zsh/zutil module.

✨ `ZSH_ARGZERO` ✨

If zsh was invoked to run a script, this is the name of the script. Otherwise it's the name used to invoke the current shell. It's the same as the value of $0 when the POSIX_ARGZERO option is set — but this one is always available. iconic behaviour honestly 🫶

🦋 `ZSH_EXECUTION_STRING` 🦋

If the shell was started with the option -c, this holds the argument you passed to that option. Otherwise? Not set. sorry bestie 😭

🌸 `ZSH_NAME` 🌸

Expands to the basename of the command used to invoke this particular instance of zsh.

🎀 `ZSH_PATCHLEVEL` 🎀

The output of git describe --tags --long for the zsh repository used to build the shell. This is most useful for keeping track of versions of the shell during development between releases — so like, most users should not use it and should instead rely on $ZSH_VERSION. it's giving dev-only, iykyk 💀

⭐ `zsh_scheduled_events` ⭐

See the section The zsh/sched Module in zshmodules(1). that's it, that's the whole thing.

💜 `ZSH_SCRIPT` 💜

If zsh was invoked to run a script, this is the name of the script; otherwise it's unset.

🔥 `ZSH_SUBSHELL` 🔥

Readonly integer. Starts at zero, and gets incremented every single time the shell forks to create a subshell for executing code. So (print $ZSH_SUBSHELL) and print $(print $ZSH_SUBSHELL) both output 1, while ( (print $ZSH_SUBSHELL) ) outputs 2. nested subshells said we're going deeper 🌈

💖 `ZSH_VERSION` 💖

The version number of the release of zsh. the classic.

👑 PARAMETERS USED BY THE SHELL 👑

The following parameters are used by the shell. Again, <S> means the parameter is special, and <Z> means the parameter does not exist when the shell initializes in sh or ksh emulation mode. noted 📝

Okay so when there are two parameters with an upper- and lowercase form of the same name — like path and PATH — the lowercase one is an array and the uppercase one is a scalar with the elements of the array joined together by colons. These are similar to tied parameters created via typeset -T. The normal use for the colon-separated form is exporting to the environment, while the array form is easier to manipulate inside the shell. And lowkey the important bit: unsetting either of the pair will unset the other, they keep their special properties when recreated, and recreating one of the pair will recreate the other. they're literally inseparable, soulmate energy 🫶

ARGV0 — If exported, its value is used as the argv[0] of external commands. Usually used in constructs like ARGV0=emacs nethack.
BAUD — The rate in bits per second at which data reaches the terminal. The line editor uses this value to compensate for a slow terminal by delaying updates to the display until necessary. If the parameter is unset, or the value is zero, the compensation mechanism is turned off. The parameter is not set by default.

This parameter may be profitably set in some circumstances — e.g. for slow modems dialing into a communications server, or on a slow wide area network. Set it to the baud rate of the slowest part of the link for best performance. be nice to your slow link 🐢

cdpath <S> <Z> (CDPATH <S>) — An array (colon-separated list) of directories specifying the search path for the cd command.
COLUMNS <S> — The number of columns for this terminal session. Used for printing select lists and for the line editor.
CORRECT_IGNORE — If set, is treated as a pattern during spelling correction. Any potential correction that matches the pattern is ignored. For example, if the value is _* then completion functions (which, by convention, have names beginning with _) will never be offered as spelling corrections. The pattern does not apply to the correction of file names as applied by the CORRECT_ALL option — so with the example just given, files beginning with _ in the current directory would still be completed. sneaky ✨
CORRECT_IGNORE_FILE — If set, is treated as a pattern during spelling correction of file names. Any file name matching the pattern is never offered as a correction. For example, if the value is .* then dot file names will never be offered as spelling corrections. This is useful with the CORRECT_ALL option.
DIRSTACKSIZE — The maximum size of the directory stack; by default there is no limit. If the stack gets larger than this, it will be truncated automatically. Useful with the AUTO_PUSHD option.
ENV — If the ENV environment variable is set when zsh is invoked as sh or ksh, $ENV is sourced after the profile scripts. The value of ENV is subjected to parameter expansion, command substitution, and arithmetic expansion before being interpreted as a pathname. Note that ENV is not used unless the shell is interactive and zsh is emulating sh or ksh. two conditions, no shortcuts 💅
FCEDIT — The default editor for the fc builtin. If FCEDIT is not set, the parameter EDITOR is used; if that's not set either, a builtin default (usually vi) is used.
fignore <S> <Z> (FIGNORE <S>) — An array (colon separated list) containing the suffixes of files to be ignored during filename completion. However, if completion only generates files with suffixes in this list, then those files are completed anyway. plot twist 🎀
fpath <S> <Z> (FPATH <S>) — An array (colon separated list) of directories specifying the search path for function definitions. This path is searched when a function with the -u attribute is referenced. If an executable file is found, then it is read and executed in the current environment.
histchars <S> — Three characters used by the shell's history and lexical analysis mechanism. The first character signals the start of a history expansion (default !). The second character signals the start of a quick history substitution (default ^). The third character is the comment character (default #).

The characters must be in the ASCII character set — any attempt to set histchars to characters with a locale-dependent meaning will be rejected with an error message. no cap, it will simply not let you 🙈

HISTCHARS <S> <Z> — Same as histchars. (Deprecated.) that's it, that's the tea.
HISTFILE — The file to save the history in when an interactive shell exits. If unset, the history is not saved. save your history bestie 😭
HISTORY_IGNORE — If set, is treated as a pattern at the time history files are written. Any potential history entry that matches the pattern is skipped. For example, if the value is fc * then commands that invoke the interactive history editor are never written to the history file.

Note that HISTORY_IGNORE defines a single pattern: to specify alternatives use the (first|second|...) syntax.

Compare the HIST_NO_STORE option or the zshaddhistory hook, either of which would prevent such commands from being added to the interactive history at all. If you wish to use HISTORY_IGNORE to stop history being added in the first place, you can define the following hook:

zshaddhistory() {
  emulate -L zsh
  ## uncomment if HISTORY_IGNORE
  ## should use EXTENDED_GLOB syntax
  # setopt extendedglob
  [[ $1 != ${~HISTORY_IGNORE} ]]
}

HISTSIZE <S> — The maximum number of events stored in the internal history list. If you use the HIST_EXPIRE_DUPS_FIRST option, setting this value larger than the SAVEHIST size will give you the difference as a cushion for saving duplicated history events.

If this is made local, it is not implicitly set to 0, but may be explicitly set locally. little subtlety for you there 🌸

HOME <S> — The default argument for the cd command. This is not set automatically by the shell in sh, ksh or csh emulation, but it is typically present in the environment anyway, and if it becomes set it has its usual special behaviour.

🍓 `IFS` `<S>`

Okay so IFS is the internal field separators girlie — by default that's space, tab, newline and NUL. She's the one who decides where words split when they come out of command or parameter expansion, and also where words split when they come out of the read builtin. ✨

Any characters from the set space, tab and newline that appear in IFS are called IFS white space. One or more IFS white space characters, or one non-IFS white space character together with any adjacent IFS white space character, delimit a field. And if an IFS white space character shows up twice consecutively in IFS, it's treated as if it weren't an IFS white space character at all. 💅

If the parameter is unset, the default is used. ⚠️ Note this has a different effect from setting the parameter to an empty string.

If the MULTIBYTE option is on and IFS contains invalid characters in the current locale, it gets reset to the default. 🫶

💅 `KEYBOARD_HACK`

This variable defines a character to be removed from the end of the command line before interpreting it — interactive shells only, bestie. It's intended to fix the problem with keys placed annoyingly close to return, and it replaces the SUNKEYBOARDHACK option which did this for backquotes only. 👀

Should the chosen character be one of singlequote, doublequote or backquote, there must also be an odd number of them on the command line for the last one to be removed. Like, them's the rules. 🎀

For backward compatibility, if the SUNKEYBOARDHACK option is explicitly set, the value of KEYBOARD_HACK reverts to backquote. If the option is explicitly unset, this variable is set to empty.

⏱️ `KEYTIMEOUT`

The time the shell waits, in hundredths of seconds, for another key to be pressed when reading bound multi-character sequences. That's it, that's the whole vibe. 🦋

🌍 `LANG` `<S>`

This variable determines the locale category for any category not specifically selected via a variable starting with LC_. She's the default queen. 👑

🌈 `LC_ALL` `<S>`

This variable overrides the value of the LANG variable and the value of any of the other variables starting with LC_. Ultimate main character energy, no cap. 💖

🔤 `LC_COLLATE` `<S>`

This variable determines the locale category for character collation information within ranges in glob brackets and for sorting. Sort girlie runs the show here. 💅

🦋 `LC_CTYPE` `<S>`

This variable determines the locale category for character handling functions. If the MULTIBYTE option is in effect, this variable or LANG should contain a value that reflects the character set in use — even if it's a single-byte character set — unless only the 7-bit subset (ASCII) is used. 👀

For example, if the character set is ISO-8859-1, a suitable value might be en_US.iso88591 (certain Linux distributions) or en_US.ISO8859-1 (MacOS). Slay, locale-specific icon. ✨

💬 `LC_MESSAGES` `<S>`

This variable determines the language in which messages should be written. Note that zsh does not use message catalogs — so that's literally the tea. ☕

🔢 `LC_NUMERIC` `<S>`

This variable affects the decimal point character and thousands separator character for the formatted input/output functions and string conversion functions. Note that zsh ignores this setting when parsing floating point mathematical expressions. Numbers said "not today." 😭

🕰️ `LC_TIME` `<S>`

This variable determines the locale category for date and time formatting in prompt escape sequences. Clock girlie 🎀

📏 `LINES` `<S>`

The number of lines for this terminal session. Used for printing select lists and for the line editor. Know your vertical real estate, bestie. 🫶

📋 `LISTMAX`

In the line editor, the number of matches to list without asking first. If the value is negative, the list will be shown if it spans at most as many lines as given by the absolute value. If set to zero, the shell asks only if the top of the listing would scroll off the screen. 👑

📬 `MAIL`

If this parameter is set and mailpath is not set, the shell looks for mail in the specified file. Snail mail but make it Unix. 💌

⏲️ `MAILCHECK`

The interval in seconds between checks for new mail. Refresh rate for the mail girlies. 🎀

📮 `mailpath` `<S>` `<Z>` (`MAILPATH` `<S>`)

An array (colon-separated list) of filenames to check for new mail. Each filename can be followed by a ? and a message that will be printed. The message will undergo parameter expansion, command substitution and arithmetic expansion with the variable $_ defined as the name of the file that has changed. The default message is You have new mail. 💖

If an element is a directory instead of a file, the shell will recursively check every file in every subdirectory of the element. She's thorough, okay? 🧚

📖 `manpath` `<S>` `<Z>` (`MANPATH` `<S>` `<Z>`)

An array (colon-separated list) whose value is not used by the shell. Plot twist 😭 The manpath array can be useful, however, since setting it also sets MANPATH, and vice versa. Synergy, bestie. ✨

🧩 `match`, `mbegin`, `mend`

Arrays set by the shell when the b globbing flag is used in pattern matches. See the subsection Globbing flags in the documentation for Filename Generation in zshexpn(1). Go read the lore. 📚

🧩 `MATCH`, `MBEGIN`, `MEND`

Set by the shell when the m globbing flag is used in pattern matches. See the subsection Globbing flags in the documentation for Filename Generation in zshexpn(1). Same cinematic universe, different case. 👑

🧱 `module_path` `<S>` `<Z>` (`MODULE_PATH` `<S>`)

An array (colon-separated list) of directories that zmodload searches for dynamically loadable modules. This is initialized to a standard pathname, usually /usr/local/lib/zsh/$ZSH_VERSION. (The /usr/local/lib part varies from installation to installation.) 🔧

⚠️ For security reasons, any value set in the environment when the shell is started will be ignored. Periodt.

These parameters only exist if the installation supports dynamic module loading. No modules, no path, no drama. 💅

🐱 `NULLCMD` `<S>`

The command name to assume if a redirection is specified with no command. Defaults to cat. For sh/ksh behavior, change this to :. For csh-like behavior, unset this parameter; the shell will print an error message if null commands are entered. Choose your fighter. 🥊

🛣️ `path` `<S>` `<Z>` (`PATH` `<S>`)

An array (colon-separated list) of directories to search for commands. When this parameter is set, each directory is scanned and all files found are put in a hash table. Directories named by relative path are not scanned for hashing. Ngl, that detail matters. 👀

🖥️ `POSTEDIT` `<S>`

This string is output whenever the line editor exits. It usually contains termcap strings to reset the terminal. Bye-bye cleanup queen. 👋✨

💖 `PROMPT` `<S>` `<Z>`, `PROMPT2` `<S>` `<Z>`, `PROMPT3` `<S>` `<Z>`, `PROMPT4` `<S>` `<Z>`

Same as PS1, PS2, PS3 and PS4, respectively. Aliases, slay. 💅

💕 `prompt` `<S>` `<Z>`

Same as PS1. Lowercase icon. 🎀

🩹 `PROMPT_EOL_MARK`

When the PROMPT_CR and PROMPT_SP options are set, the PROMPT_EOL_MARK parameter can be used to customize how the end of partial lines are shown. This parameter undergoes prompt expansion, with the PROMPT_PERCENT option set. If not set, the default behavior is equivalent to the value %B%S%#%s%b. Cute but chaotic. ✨

👑 `PS1` `<S>`

The primary prompt string, printed before a command is read. It undergoes a special form of expansion before being displayed; see EXPANSION OF PROMPT SEQUENCES in zshmisc(1). The default is %m%# . Main character prompt energy, fr fr. 💖

🌸 `PS2` `<S>`

The secondary prompt, printed when the shell needs more information to complete a command. It is expanded in the same way as PS1. The default is %_> , which displays any shell constructs or quotation marks which are currently being processed. She's the "wait bestie I'm not done" prompt. 🫶

🎀 `PS3` `<S>`

Selection prompt used within a select loop. It is expanded in the same way as PS1. The default is ?# . Choose your own adventure prompt. 🦋

🔍 `PS4` `<S>`

The execution trace prompt. Default is +%N:%i> , which displays the name of the current shell structure and the line number within it. In sh or ksh emulation, the default is + . Detective mode activated. 🕵️‍♀️

🌟 `psvar` `<S>` `<Z>` (`PSVAR` `<S>`)

An array (colon-separated list) whose elements can be used in PROMPT strings. Setting psvar also sets PSVAR, and vice versa. They're besties, obviously. 💕

📥 `READNULLCMD` `<S>`

The command name to assume if a single input redirection is specified with no command. Defaults to more. Reading girlie with the backup plan. 📖

🧠 `REPORTMEMORY`

If nonnegative, commands whose maximum resident set size (roughly speaking, main memory usage) in kilobytes is greater than this value have timing statistics reported. The format used to output statistics is the value of the TIMEFMT parameter, which is the same as for the REPORTTIME variable and the time builtin; note that by default this does not output memory usage. 📊

Appending max RSS %M to the value of TIMEFMT causes it to output the value that triggered the report. If REPORTTIME is also in use, at most a single report is printed for both triggers. This feature requires the getrusage() system call, commonly supported by modern Unix-like systems. Fancy, but read the fine print. 👀

⏰ `REPORTTIME`

If nonnegative, commands whose combined user and system execution times (measured in seconds) are greater than this value have timing statistics printed for them. Output is suppressed for commands executed within the line editor, including completion; commands explicitly marked with the time keyword still cause the summary to be printed in this case. ⚠️ Important caveat, bestie — don't get caught lacking. 💅

💌 `REPLY`

This parameter is reserved by convention to pass string values between shell scripts and shell builtins in situations where a function call or redirection are impossible or undesirable. The read builtin and the select complex command may set REPLY, and filename generation both sets and examines its value when evaluating certain expressions. Some modules also employ REPLY for similar purposes. She's the messenger. 🕊️

💌 `reply`

As REPLY, but for array values rather than strings. Array-flavored messenger, periodt. ✨

➡️ `RPROMPT` `<S>`, `RPS1` `<S>`

This prompt is displayed on the right-hand side of the screen when the primary prompt is being displayed on the left. This does not work if the SINGLE_LINE_ZLE option is set. It is expanded in the same way as PS1. Right-side queen. 👑

➡️ `RPROMPT2` `<S>`, `RPS2` `<S>`

This prompt is displayed on the right-hand side of the screen when the secondary prompt is being displayed on the left. This does not work if the SINGLE_LINE_ZLE option is set. It is expanded in the same way as PS2. Same energy, secondary edition. 🎀

🗃️ `SAVEHIST`

The maximum number of history events to save in the history file. The memory keeper of your command-line past. 📜

If this is made local, it is not implicitly set to 0, but may be explicitly set locally. That's the whole note — don't let anyone tell you otherwise. 💅

`SPROMPT` <S>

okay so this is literally the prompt that shows up when zsh is like "bestie… did you mean this?" for spelling correction 💅 The sequence %R expands to the string which presumably needs spelling correction, and %r expands to the proposed correction. All other prompt escapes are also allowed, so go off with your usual prompt flair ✨

The actions available at the prompt are [nyae]:

n (no) (default) — Discard the correction and run the command. we're doing it my way 😌
y (yes) — Make the correction and run the command. slay, thank you 💖
a (abort) — Discard the entire command line without running it. nvm nvm nvm 🙈
e (edit) — Resume editing the command line. let me fix it myself, main character energy 👑

`STTY`

If this parameter is set in a command's environment, the shell runs the stty command with the value of this parameter as arguments in order to set up the terminal before executing the command. The modes apply only to the command, and are reset when it finishes or is suspended. If the command is suspended and continued later with the fg or wait builtins it will see the modes specified by STTY, as if it were not suspended. This (intentionally) does not apply if the command is continued via kill -CONT. 🫶

ngl there are rules here and you gotta respect them: STTY is ignored if the command is run in the background, or if it is in the environment of the shell but not explicitly assigned to in the input line. This avoids running stty at every external command by accidentally exporting it. Also note that STTY should not be used for window size specifications — those will not be local to the command. fr fr.

If the parameter is set and empty, all of the above applies except that stty is not run. This can be useful as a way to freeze the tty around a single command, blocking its changes to tty settings, similar to the ttyctl builtin. lowkey genius 💡

`TERM` <S>

The type of terminal in use. This is used when looking up termcap sequences. An assignment to TERM causes zsh to re-initialize the terminal, even if the value does not change (e.g., TERM=$TERM). It is necessary to make such an assignment upon any change to the terminal definition database or terminal type in order for the new settings to take effect. yes, assigning a thing to itself on purpose, it's giving self-care 🎀

`TERMINFO` <S>

A reference to your terminfo database, used by the terminfo library when the system has it; see terminfo(5). If set, this causes the shell to reinitialise the terminal, making the workaround TERM=$TERM unnecessary. iconic, no notes ✨

`TERMINFO_DIRS` <S>

A colon-separated list of terminfo databases, used by the terminfo library when the system has it; see terminfo(5). This variable is only used by certain terminal libraries, in particular ncurses; see terminfo(5) to check support on your system. If set, this causes the shell to reinitialise the terminal, making the workaround TERM=$TERM unnecessary. 💅

heads up bestie: unlike other colon-separated arrays this is not tied to a zsh array. she's independent, she's mysterious 🦋

`TIMEFMT`

The format of process time reports with the time keyword. ⏱️ The default is:

%J  %U user %S system %P cpu %*E total

It recognizes the following escape sequences, although not all may be available on all systems, and some that are available may not be useful:

%% — A %.
%U — CPU seconds spent in user mode.
%S — CPU seconds spent in kernel mode.
%E — Elapsed time in seconds.
%P — The CPU percentage, computed as 100*(%U+%S)/%E.
%W — Number of times the process was swapped.
%X — The average amount in (shared) text space used in kilobytes.
%D — The average amount in (unshared) data/stack space used in kilobytes.
%K — The total space used (%X+%D) in kilobytes.
%M — The maximum memory the process had in use at any time in kilobytes.
%F — The number of major page faults (page needed to be brought from disk).
%R — The number of minor page faults.
%I — The number of input operations.
%O — The number of output operations.
%r — The number of socket messages received.
%s — The number of socket messages sent.
%k — The number of signals received.
%w — Number of voluntary context switches (waits).
%c — Number of involuntary context switches.
%J — The name of this job.

A star may be inserted between the percent sign and flags printing time (e.g., %*E); this causes the time to be printed in hh:mm:ss.ttt format (hours and minutes are only printed if they are not zero). Alternatively, m or u may be used (e.g., %mE) to produce time output in milliseconds or microseconds, respectively. cute, precise, obsessed ⭐

`TMOUT`

If this parameter is nonzero, the shell will receive an ALRM signal if a command is not entered within the specified number of seconds after issuing a prompt. If there is a trap on SIGALRM, it will be executed and a new alarm is scheduled using the value of the TMOUT parameter after executing the trap. If no trap is set, and the idle time of the terminal is not less than the value of the TMOUT parameter, zsh terminates. Otherwise a new alarm is scheduled to TMOUT seconds after the last keypress. so basically a vibe check timer 😭

`TMPPREFIX`

A pathname prefix which the shell will use for all temporary files. Note that this should include an initial part for the file name as well as any directory names. The default is /tmp/zsh. 🗂️

`TMPSUFFIX`

A filename suffix which the shell will use for temporary files created by process substitutions (e.g., =(list)). Note that the value should include a leading dot . if intended to be interpreted as a file extension. The default is not to append any suffix, thus this parameter should be assigned only when needed and then unset again. like, use it and put it back, we don't leave mess 💅

`WORDCHARS` <S>

A list of non-alphanumeric characters considered part of a word by the line editor. tiny but powerful ✨

`ZBEEP`

If set, this gives a string of characters, which can use all the same codes as the bindkey command as described in the zsh/zle module entry in zshmodules(1), that will be output to the terminal instead of beeping. 🎀

This may have a visible instead of an audible effect; for example, the string \e[?5h\e[?5l on a vt100 or xterm will have the effect of flashing reverse video on and off (if you usually use reverse video, you should use the string \e[?5l\e[?5h instead). This takes precedence over the NOBEEP option. we said no beeping and we meant it 🔥

`ZDOTDIR`

The directory to search for shell startup files (.zshrc, etc), if not $HOME. 🏠

`zle_bracketed_paste`

Many terminal emulators have a feature that allows applications to identify when text is pasted into the terminal rather than being typed normally. For ZLE, this means that special characters such as tabs and newlines can be inserted instead of invoking editor commands. Furthermore, pasted text forms a single undo event and if the region is active, pasted text will replace the region. 🦋

This two-element array contains the terminal escape sequences for enabling and disabling the feature. These escape sequences are used to enable bracketed paste when ZLE is active and disable it at other times. Unsetting the parameter has the effect of ensuring that bracketed paste remains disabled. periodt 💜

`zle_highlight`

An array describing contexts in which ZLE should highlight the input text. See Character Highlighting in zshzle(1). 🌈

`ZLE_LINE_ABORTED`

This parameter is set by the line editor when an error occurs. It contains the line that was being edited at the point of the error. print -zr -- $ZLE_LINE_ABORTED can be used to recover the line. Only the most recent line of this kind is remembered. the "not me crying over a lost command" parameter 😭

`ZLE_REMOVE_SUFFIX_CHARS` and `ZLE_SPACE_SUFFIX_CHARS`

These parameters are used by the line editor. In certain circumstances suffixes (typically space or slash) added by the completion system will be removed automatically, either because the next editing command was not an insertable character, or because the character was marked as requiring the suffix to be removed. 💅

These variables can contain the sets of characters that will cause the suffix to be removed. If ZLE_REMOVE_SUFFIX_CHARS is set, those characters will cause the suffix to be removed; if ZLE_SPACE_SUFFIX_CHARS is set, those characters will cause the suffix to be removed and replaced by a space.

If ZLE_REMOVE_SUFFIX_CHARS is not set, the default behaviour is equivalent to:

ZLE_REMOVE_SUFFIX_CHARS=$' \t\n;&|'

If ZLE_REMOVE_SUFFIX_CHARS is set but is empty, no characters have this behaviour. ZLE_SPACE_SUFFIX_CHARS takes precedence, so that the following:

ZLE_SPACE_SUFFIX_CHARS=$'&|'

causes the characters & and | to remove the suffix but to replace it with a space.

To illustrate the difference, suppose that the option AUTO_REMOVE_SLASH is in effect and the directory DIR has just been completed, with an appended /, following which the user types &. The default result is DIR&. With ZLE_REMOVE_SUFFIX_CHARS set but without including & the result is DIR/&. With ZLE_SPACE_SUFFIX_CHARS set to include & the result is DIR &. 💀

Note that certain completions may provide their own suffix removal or replacement behaviour which overrides the values described here. See the completion system documentation in zshcompsys(1).

`ZLE_RPROMPT_INDENT` <S>

If set, used to give the indentation between the right hand side of the right prompt in the line editor as given by RPS1 or RPROMPT and the right hand side of the screen. If not set, the value 1 is used. 📏

Typically this will be used to set the value to 0 so that the prompt appears flush with the right hand side of the screen. This is not the default as many terminals do not handle this correctly, in particular when the prompt appears at the extreme bottom right of the screen. Recent virtual terminals are more likely to handle this case correctly. Some experimentation is necessary. go forth and fiddle, bestie 🧚

zsh 5.9.2                        July 12, 2026                      ZSHPARAM(1)

✨ SPECIFYING OPTIONS ✨

okay so bestie, options are mostly known by name. The names are case insensitive and underscores literally don't matter at all. Like, allexport is equivalent to A__lleXP_ort. iconic, right? 💅

You can flip an option's whole vibe by putting no in front of it, so setopt No_Beep is equivalent to unsetopt beep. But — and this is the catch — you can only do the inversion once, so nonobeep is not a synonym for beep. Same energy: tify is not a synonym for nonotify (which is the inversion of notify). no cap, it just doesn't work 🫶

Some options also have one or more single letter names. There are two sets of single letter options: one used by default, and another used to emulate sh/ksh (used when the SH_OPTION_LETTERS option is set). You can use the single letter options on the shell command line, or with the set, setopt and unsetopt builtins, as normal Unix options preceded by -.

The sense of the single letter options can be inverted by using + instead of -. Some of the single letter option names refer to an option being off, in which case the inversion of that name refers to the option being on. For example, +n is the short name of exec, and -n is the short name of its inversion, noexec. it's giving mirror universe 🪞

In strings of single letter options supplied to the shell at startup, trailing whitespace will be ignored — for example the string -f will be treated just as -f, but the string -f i is an error. This is because many systems which implement the #! mechanism for calling scripts do not strip trailing whitespace. so polite of zsh to compensate 😭

It is possible for options to be set within a function scope. See the description of the option LOCAL_OPTIONS below.

🎀 DESCRIPTION OF OPTIONS 🎀

In the following list, options set by default in all emulations are marked <D>; those set by default only in csh, ksh, sh, or zsh emulations are marked <C>, <K>, <S>, <Z> as appropriate. When listing options (by setopt, unsetopt, set -o or set +o), those turned on by default appear in the list prefixed with no. Hence (unless KSH_OPTION_PRINT is set), setopt shows all options whose settings are changed from the default. cute little filter, honestly 🌸

📁 Changing Directories 📁

AUTO_CD (-J) — if a command is issued that can't be executed as a normal command, and the command is the name of a directory, perform the cd command to that directory. This option is only applicable if the option SHIN_STDIN is set, i.e. if commands are being read from standard input. It's designed for interactive use; the recommendation is that cd be used explicitly in scripts to avoid ambiguity. just be direct, bestie 🎯
AUTO_PUSHD (-N) — make cd push the old directory onto the directory stack.
CDABLE_VARS (-T) — if the argument to a cd command (or an implied cd with the AUTO_CD option set) is not a directory, and does not begin with a slash, try to expand the expression as if it were preceded by a ~ (see the section Filename Expansion). wishful thinking but make it functional 💫
CD_SILENT — never print the working directory after a cd (whether explicit or implied with the AUTO_CD option set). cd normally prints the working directory when the argument given to it was -, a stack entry, or the name of a directory found under CDPATH. Note that this is distinct from pushd's stack-printing behaviour, which is controlled by PUSHD_SILENT. This option overrides the printing-related effects of POSIX_CD. shhh, no narration 🤫
CHASE_DOTS — when changing to a directory containing a path segment .. which would otherwise be treated as canceling the previous segment in the path (in other words, foo/.. would be removed from the path, or if .. is the first part of the path, the last part of the current working directory would be removed), instead resolve the path to the physical directory. This option is overridden by CHASE_LINKS.

For example, suppose /foo/bar is a link to the directory /alt/rod. Without this option set, cd /foo/bar/.. changes to /foo; with it set, it changes to /alt. The same applies if the current directory is /foo/bar and cd .. is used. Note that all other symbolic links in the path will also be resolved. lowkey a plot twist every time 🌀

CHASE_LINKS (-w) — resolve symbolic links to their true values when changing directory. This also has the effect of CHASE_DOTS, i.e. a .. path segment will be treated as referring to the physical parent, even if the preceding path segment is a symbolic link. 🔗
POSIX_CD <K> <S> — modifies the behaviour of cd, chdir and pushd commands to make them more compatible with the POSIX standard. The behaviour with the option unset is described in the documentation for the cd builtin in zshbuiltins(1). If the option is set, the shell does not test for directories beneath the local directory (.) until after all directories in cdpath have been tested, and the cd and chdir commands do not recognise arguments of the form {+|-}n as directory stack entries.

Also, if the option is set, the conditions under which the shell prints the new directory after changing to it are modified. It is no longer restricted to interactive shells (although printing of the directory stack with pushd is still limited to interactive shells); and any use of a component of CDPATH, including a . but excluding an empty component that is otherwise treated as ., causes the directory to be printed. compliance era 👑

PUSHD_IGNORE_DUPS — don't push multiple copies of the same directory onto the directory stack. no duplicates allowed, we're not doing reruns 💅
PUSHD_MINUS — exchanges the meanings of + and - when used with a number to specify a directory in the stack. chaos, but consensual 🎭
PUSHD_SILENT (-E) — do not print the directory stack after pushd or popd.
PUSHD_TO_HOME (-D) — have pushd with no arguments act like pushd $HOME. home sweet home 🏠

🌈 Completion 🌈

ALWAYS_LAST_PROMPT <D> — if unset, key functions that list completions try to return to the last prompt if given a numeric argument. If set these functions try to return to the last prompt if given no numeric argument. flip it, reverse it 🔄
ALWAYS_TO_END — if a completion is performed with the cursor within a word, and a full completion is inserted, the cursor is moved to the end of the word. That is, the cursor is moved to the end of the word if either a single match is inserted or menu completion is performed. cursor said bye, moving on ➡️
AUTO_LIST (-9) <D> — automatically list choices on an ambiguous completion. 📋
AUTO_MENU <D> — automatically use menu completion after the second consecutive request for completion, for example by pressing the tab key repeatedly. This option is overridden by MENU_COMPLETE. tab tab and we're vibing ✨
AUTO_NAME_DIRS — any parameter that is set to the absolute name of a directory immediately becomes a name for that directory, that will be used by the %~ and related prompt sequences, and will be available when completion is performed on a word starting with ~. (Otherwise, the parameter must be used in the form ~param first.) instant nickname privileges 🎀
AUTO_PARAM_KEYS <D> — if a parameter name was completed and a following character (normally a space) automatically inserted, and the next character typed is one of those that have to come directly after the name (like }, :, etc.), the automatically added character is deleted, so that the character typed comes immediately after the parameter name. Completion in a brace expansion is affected similarly: the added character is a ,, which will be removed if } is typed next. it reads the room, honestly 🫶
AUTO_PARAM_SLASH <D> — if a parameter is completed whose content is the name of a directory, then add a trailing slash instead of a space. slashes > spaces, we been knew ⚡
AUTO_REMOVE_SLASH <D> — when the last character resulting from a completion is a slash and the next character typed is a word delimiter, a slash, or a character that ends a command (such as a semicolon or an ampersand), remove the slash. cleanup crew 🧹
BASH_AUTO_LIST — on an ambiguous completion, automatically list choices when the completion function is called twice in succession. This takes precedence over AUTO_LIST. The setting of LIST_AMBIGUOUS is respected. If AUTO_MENU is set, the menu behaviour will then start with the third press. Note that this will not work with MENU_COMPLETE, since repeated completion calls immediately cycle through the list in that case. yes this is specific, yes it matters 💀
COMPLETE_ALIASES — prevents aliases on the command line from being internally substituted before completion is attempted. The effect is to make the alias a distinct command for completion purposes. main character alias energy 👑
COMPLETE_IN_WORD — if unset, the cursor is set to the end of the word if completion is started. Otherwise it stays there and completion is done from both ends. both ends?? we love a multitasker 🦋
GLOB_COMPLETE — when the current word has a glob pattern, do not insert all the words resulting from the expansion but generate matches as for completion and cycle through them like MENU_COMPLETE. The matches are generated as if a * was added to the end of the word, or inserted at the cursor when COMPLETE_IN_WORD is set. This actually uses pattern matching, not globbing, so it works not only for files but for any completion, such as options, user names, etc.

Note that when the pattern matcher is used, matching control (for example, case-insensitive or anchored matching) cannot be used. This limitation only applies when the current word contains a pattern; simply turning on the GLOB_COMPLETE option does not have this effect. read the fine print bestie, it's important 📖

HASH_LIST_ALL <D> — whenever a command completion or spelling correction is attempted, make sure the entire command path is hashed first. This makes the first completion slower but avoids false reports of spelling errors. slow start, flawless finish 🔥
LIST_AMBIGUOUS <D> — this one only does anything when AUTO_LIST or BASH_AUTO_LIST is also set, okurr? If there's an unambiguous prefix that can be inserted on the command line, that gets inserted and no completion list shows up — like, auto-listing only kicks in when literally nothing would be inserted. And for BASH_AUTO_LIST specifically, that means the list gets pushed back to the third call of the function. bestie, third. 💅
LIST_BEEP <D> — beep at us when a completion is ambiguous 😭. More accurately (because we're being precise here): it forces the completion widgets to return status 1 on an ambiguous completion, and that is what makes the shell beep — but only if the option BEEP is also set. And this can get modified if completion is called from a user-defined widget. so like, chain reaction vibes.
LIST_PACKED — try to make the completion list smaller, i.e. take up fewer lines, by printing the matches in columns of differing widths. cramming it in, we love efficiency ✨
LIST_ROWS_FIRST — lay the matches out in completion lists sorted horizontally. so the second match sits to the right of the first one instead of underneath it like usual. reading across, main character energy ➡️
LIST_TYPES (-X) <D> — when you're listing files that are possible completions, show the type of each file with a trailing identifying mark. lil name tags for every file 🎀
MENU_COMPLETE (-Y) — on an ambiguous completion, instead of listing the possibilities or beeping at you, it just inserts the first match immediately. then when you ask for completion again, it removes the first match and inserts the second, and so on, and so on. when there are no more matches left it loops right back to the first one. reverse-menu-complete will walk the list in the other direction if you want to go back. this option overrides AUTO_MENU. iconic behavior honestly 👑
REC_EXACT (-S) — if the string on the command line exactly matches one of the possible completions, it gets accepted — even if there's another completion that also matches (like, that same string with something else tacked onto it). exact match wins, periodt.

🌈 Expansion and Globbing 🌈

BAD_PATTERN (+2) <C> <Z> — if a pattern for filename generation is badly formed, print an error message. that's it, that's the whole thing. (and if this option is unset, the pattern just gets left unchanged, no complaints.) 🫠
BARE_GLOB_QUAL <Z> — in a glob pattern, treat a trailing set of parentheses as a qualifier list, as long as it doesn't contain any of |, ( or (when special) ~ characters. see the section Filename Generation.
BRACE_CCL — expand expressions in braces which wouldn't otherwise go through brace expansion into a lexically ordered list of all the characters. see the section Brace Expansion. 🦋
CASE_GLOB <D> — make globbing (filename generation) case-sensitive. note that other uses of patterns are always case-sensitive no matter what. if the option is unset, the presence of any character that's special to filename generation causes case-insensitive matching. for example, cvs(/) can match the directory CVS because of that globbing flag — unless the option BARE_GLOB_QUAL is unset. sneaky little exception, iykyk 💀
CASE_MATCH <D> — make regular expressions that use the zsh/regex module (including matches with =~) sensitive to case.
CASE_PATHS — if CASE_PATHS is not set (which is the default, btw), CASE_GLOB affects how every path component is interpreted, whenever a special character appears in any component. when CASE_PATHS is set, file path components that don't contain special filename generation characters are always case-sensitive — which restricts NO_CASE_GLOB to just the components that actually contain globbing characters. ngl that's a mouthful.

heads up: if the filesystem itself isn't case-sensitive, then CASE_PATHS has no effect at all. zip. nada. 🍓

CSH_NULL_GLOB <C> — if a pattern for filename generation has no matches, delete the pattern from the argument list. don't report an error unless all the patterns in a command have no matches. overrides NOMATCH.
EQUALS <Z> — perform = filename expansion. (see the section Filename Expansion.)
EXTENDED_GLOB — treat the #, ~ and ^ characters as part of patterns for filename generation, etc. (and an initial unquoted ~ always produces named directory expansion, no matter what.)
FORCE_FLOAT — constants in arithmetic evaluation get treated as floating point even if you never typed a decimal point; values of integer variables get converted to floating point when they're used in arithmetic expressions. integers in any base get converted too. floats everywhere, no escape 🔥
GLOB (+F, ksh: +f) <D> — perform filename generation (globbing). (see the section Filename Generation.)
GLOB_ASSIGN <C> — if this is set, filename generation (globbing) is performed on the right hand side of scalar parameter assignments of the form name=pattern (e.g. foo=*). if the result has more than one word, the parameter becomes an array with those words as arguments. this option is provided for backwards compatibility only — globbing is always performed on the right hand side of array assignments of the form name=(value) (e.g. foo=(*)), and that form is the recommended one for clarity. with this option set, it's genuinely not possible to predict whether the result will be an array or a scalar. 😭 so like, don't.
GLOB_DOTS (-4) — don't require a leading . in a filename to be matched explicitly. hidden files, come on out 🌸
GLOB_STAR_SHORT — when this option is set and the default zsh-style globbing is in effect, the pattern **/* can be abbreviated to **, and the pattern ***/* can be abbreviated to ***. hence **.c finds a file ending in .c in any subdirectory, and ***.c does the same thing while also following symbolic links. a / immediately after the ** or the *** forces the pattern to be treated as the unabbreviated form. typos: none, shortcuts: iconic ⭐
GLOB_SUBST <C> <K> <S> — treat any characters resulting from parameter expansion as eligible for filename expansion and filename generation, and any characters resulting from command substitution as eligible for filename generation. braces (and the commas in between them) do not become eligible for expansion.
HIST_SUBST_PATTERN — substitutions using the :s and :& history modifiers get performed with pattern matching instead of string matching. this happens wherever history modifiers are valid, including glob qualifiers and parameters. see the section Modifiers in zshexpn(1). patterns everywhere, we stay winning 💜
IGNORE_BRACES (-I) <S> — do not perform brace expansion. and for historical reasons, this also includes the effect of the IGNORE_CLOSE_BRACES option. package deal.
IGNORE_CLOSE_BRACES — when neither this option nor IGNORE_BRACES is set, a lone close brace character } is syntactically significant at any point on a command line. the effect of that is: no semicolon or newline is necessary before the brace that terminates a function or the current shell construct. when either option is set, a closing brace is syntactically significant only in command position. and unlike IGNORE_BRACES, this option does not disable brace expansion. ok so with both options unset, you can define a function like this:

args() { echo $# }

but if either option is set, that doesn't work, and you need something equivalent to this instead:

args() { echo $#; }

the semicolon is doing so much heavy lifting there, respect 🫶

KSH_GLOB <K> — in pattern matching, the interpretation of parentheses is affected by a preceding @, *, +, ? or !. see the section Filename Generation.
MAGIC_EQUAL_SUBST — all unquoted arguments of the form anything=expression that appear after the command name have filename expansion performed on expression as if it were a parameter assignment — that's the kind where expression has a leading ~ or =. the argument isn't treated specially in any other way though: it gets passed to the command as a single argument, and it is not used as an actual parameter assignment. for example, in echo foo=~/bar:~/rod, both occurrences of ~ would be replaced. note that this happens anyway with typeset and similar statements.

this option respects the setting of the KSH_TYPESET option. in other words, if both options are in effect, arguments that look like assignments will not undergo word splitting. fr fr.

MARK_DIRS (-8, ksh: -X) — append a trailing / to all directory names resulting from filename generation (globbing). lil slash, big clarity 🎀
MULTIBYTE <D> — respect multibyte characters when they're found in strings. when this option is set, strings get examined using the system library to figure out how many bytes form a character, depending on the current locale. this affects the way characters are counted in pattern matching, in parameter values, and in various delimiters. 🌈

the option is on by default if the shell was compiled with MULTIBYTE_SUPPORT; otherwise it's off by default and has no effect if you turn it on.

if the option is off, a single byte is always treated as a single character. this setting is designed purely for examining strings known to contain raw bytes or other values that may not be characters in the current locale. and honestly? it is not necessary to unset the option merely because the character set for the current locale doesn't contain multibyte characters.

the option does not affect the shell's editor, which always uses the locale to determine multibyte characters — because the character set displayed by the terminal emulator is independent of the shell settings. ik, ik, it's a lot 💀

NOMATCH (+3) <C> <Z> — if a pattern for filename generation has no matches, print an error instead of leaving it unchanged in the argument list. this also applies to file expansion of an initial ~ or =. so like, no silent failures here bestie.
NULL_GLOB (-G) — if a pattern for filename generation has no matches, delete the pattern from the argument list instead of reporting an error. overrides NOMATCH. poof, gone ✨
NUMERIC_GLOB_SORT — if numeric filenames get matched by a filename generation pattern, sort those filenames numerically rather than lexicographically. 2 before 10, the way nature intended 🔢
RC_EXPAND_PARAM (-P) — array expansions of the form foo${xx}bar, where the parameter xx is set to (a b c), get substituted with fooabar foobbar foocbar instead of the default fooa b cbar. and note that an empty array will therefore cause all arguments to be removed. bear that in mind 😭
REMATCH_PCRE — bestie, flip this on and your =~ regex matching goes full glow-up with Perl-Compatible Regular Expressions straight from the PCRE library ✨ (heads up tho: the zsh/pcre module has to be available, no module no magic). Leave it unset and regexes fall back to the extended regexp syntax the system libraries hand you.
SH_GLOB <K> <S> — this one's a vibe-killer for the fancy stuff: it disables the special meaning of (, |, ) and < when you're globbing the result of parameter and command substitutions, plus in some other spots where the shell accepts patterns. Sneaky detail: if SH_GLOB is set but KSH_GLOB isn't, the shell still allows interpretation of subshell expressions in parentheses in some cases where there's no space before the opening parenthesis — like, !(true) gets read as if there were a space after the !. Also it's set by default if zsh is invoked as sh or ksh. iykyk 💅
UNSET (+u, ksh: +u) <K> <S> <Z> — treats unset parameters as if they were empty when substituting, and as if they were zero when reading their values in arithmetic expansion and arithmetic commands 🫶. Without it? Unset params are treated as an error. dramatic but fair.
WARN_CREATE_GLOBAL — prints a warning whenever a global parameter gets created inside a function, via assignment or in math context 👀. Translation: you probably forgot to declare something local when you should've. Parameters you explicitly declare global from inside a function with typeset -g do not trigger a warning. Also noted: there's no warning when a local parameter is assigned to in a nested function — which may also indicate an error, so like, stay woke 😭
WARN_NESTED_VAR — yells at you when an existing parameter from an enclosing function scope (or a global) gets set inside a function by assignment or in math context 📢. Assignment to shell special parameters doesn't cause a warning. It's the companion to WARN_CREATE_GLOBAL, except here the warning only fires when a parameter is not created. Where possible, using typeset -g to set the parameter suppresses the error — but note you gotta do that every time the parameter is set. Want to restrict the effect to a single function scope? Use functions -W.

Take this example, which produces a warning for the assignment inside nested, since it overrides the value within toplevel:

toplevel() {
  local foo="in fn"
  nested
}
nested() {
     foo="in nested"
}
setopt warn_nested_var
toplevel

📜 History 🕰️

APPEND_HISTORY <D> — zsh sessions append their history list to the history file instead of replacing it 💖. So multiple parallel zsh sessions all get their new entries added to the history file, in the order they exit. The file still gets periodically re-written to trim it when the line count grows 20% beyond the value in $SAVEHIST (peep the HIST_SAVE_BY_COPY option too).
BANG_HIST (+K) <C> <Z> — does the textual history expansion thing, csh-style, treating the character ! as special 💥.
EXTENDED_HISTORY <C> — saves each command's beginning timestamp (in seconds since the epoch) and the duration (in seconds) to the history file. The prefixed data looks like this:

`: <beginning time>:<elapsed seconds>;<command>'.

HIST_ALLOW_CLOBBER — adds | to output redirections in the history. This lets history references clobber files even when CLOBBER is unset 😳 (yes, that's the whole point, be careful).
HIST_BEEP <D> — beeps in ZLE when a widget tries to access a history entry that isn't there 🔔.
HIST_EXPIRE_DUPS_FIRST — when the internal history needs trimming to fit the current command line, this makes the oldest duplicate event get lost before a unique event from the list does 🗑️. Pro tip: set HISTSIZE bigger than SAVEHIST so you've got room for those duplicated events — otherwise this option behaves just like HIST_IGNORE_ALL_DUPS once the history fills up with unique events. no cap.
HIST_FCNTL_LOCK — when writing out the history file, zsh normally uses ad-hoc file locking to dodge known locking problems on some operating systems. With this option, locking happens via the system's fcntl call, where that method is available ✨. On recent operating systems this may give better performance — especially avoiding history corruption when files live on NFS. iconic.
HIST_FIND_NO_DUPS — when searching history entries in the line editor, don't display duplicates of a line you already found, even if the duplicates aren't contiguous 🙅‍♀️.
HIST_IGNORE_ALL_DUPS — if a new command line being added duplicates an older one, the older command gets removed from the list (even if it's not the previous event) 🔁.
HIST_IGNORE_DUPS (-h) — don't enter command lines into the history list if they're duplicates of the previous event.
HIST_IGNORE_SPACE (-g) — removes command lines from the history list when the first character on the line is a space, or when one of the expanded aliases contains a leading space 🫥. Only normal aliases (not global or suffix aliases) have this behaviour. Note that the command lingers in the internal history until the next command is entered before it vanishes, letting you briefly reuse or edit the line. Want it gone right now without entering another command? Type a space and press return 💅
HIST_LEX_WORDS — by default, shell history read in from files is split into words on all white space, which means arguments with quoted whitespace aren't handled correctly — consequence being that references to words in history lines read from a file may be inaccurate 😬. Set this option and words read from a history file get divided up similarly to normal shell command line handling. It's more accurately delimited, but if the history file is large this can be slow. Trial and error is necessary to decide, fr fr.
HIST_NO_FUNCTIONS — removes function definitions from the history list 🚫. Same lingering note as before: the function sticks around in the internal history until the next command is entered before it vanishes, so you can briefly reuse or edit the definition.
HIST_NO_STORE — removes the history (fc -l) command from the history list when invoked. And yeah — it lingers in the internal history until the next command is entered before vanishing, so you can briefly reuse or edit the line ✨.
HIST_REDUCE_BLANKS — strips superfluous blanks from each command line being added to the history list 🧹.
HIST_SAVE_BY_COPY <D> — when the history file is re-written, zsh normally writes out a copy named $HISTFILE.new and then renames it over the old one. Unset this option and instead the old history file gets truncated and the new version written in-place 😱. If one of the history-appending options is enabled, this option only has an effect when the enlarged history file needs re-writing to trim it down to size. Only disable this if you have special needs — doing so makes it possible to lose history entries if zsh gets interrupted during the save. like, genuinely, don't.

When writing out a copy of the history file, zsh preserves the old file's permissions and group information, but will refuse to write out a new file if it would change the history file's owner 👑

HIST_SAVE_NO_DUPS — when writing out the history file, older commands that duplicate newer ones are omitted 🙌.
HIST_VERIFY — whenever you enter a line with history expansion, don't execute it directly; instead perform the history expansion and reload the line into the editing buffer 👀 (a little "are you sure bestie" for your shell).
INC_APPEND_HISTORY — works like APPEND_HISTORY except new history lines are added to $HISTFILE incrementally, as soon as they're entered, rather than waiting until the shell exits ⚡. The file still gets periodically re-written to trim it when the line count grows 20% beyond the value specified by $SAVEHIST (see also the HIST_SAVE_BY_COPY option).
INC_APPEND_HISTORY_TIME — a variant of INC_APPEND_HISTORY where, where possible, the history entry is written out to the file after the command finishes, so the time taken by the command is recorded correctly in the history file in EXTENDED_HISTORY format ⏱️. Downside: the history entry won't be available immediately from other instances of the shell using the same history file.

This option is only useful if INC_APPEND_HISTORY and SHARE_HISTORY are turned off. The three options should be considered mutually exclusive. pick one, babe 🎀

SHARE_HISTORY <K> — the social butterfly option 🦋: it both imports new commands from the history file and causes your typed commands to be appended to the history file (that latter bit is like specifying INC_APPEND_HISTORY, which should be turned off if this option is in effect). History lines are also output with timestamps ala EXTENDED_HISTORY, which makes it easier to find the spot where we left off reading the file after it gets re-written.

By default, history movement commands visit the imported lines as well as the local lines, but you can toggle this on and off with the set-local-history zle binding. It's also possible to create a zle widget that'll make some commands ignore imported commands, and some include them ✨.

If you want more control over when commands get imported, you may wish to turn SHARE_HISTORY off, turn INC_APPEND_HISTORY or INC_APPEND_HISTORY_TIME (see above) on, and then manually import commands whenever you need them using fc -RI 💅

✨ Initialisation ✨

ALL_EXPORT (-a, ksh: -a) — all parameters you define after this are automatically exported. Set it and forget it, bestie. 💖
GLOBAL_EXPORT <Z> — if this option is set, passing the -x flag to the builtins declare, float, integer, readonly and typeset (but not local) will also set the -g flag; hence parameters exported to the environment will not be made local to the enclosing function, unless they were already or the flag +g is given explicitly. If the option is unset, exported parameters will be made local in just the same way as any other parameter. This option is set by default for backward compatibility; it is not recommended that its behaviour be relied upon. Note that the builtin export always sets both the -x and -g flags, and hence its effect extends beyond the scope of the enclosing function; this is the most portable way to achieve this behaviour. 🫶
GLOBAL_RCS (+d) <D> — if this option is unset, the startup files /etc/zsh/zprofile, /etc/zsh/zshrc, /etc/zsh/zlogin and /etc/zsh/zlogout will not be run. It can be disabled and re-enabled at any time, including inside local startup files (.zshrc, etc.). ✨
RCS (+f) <D> — after /etc/zsh/zshenv is sourced on startup, source the .zshenv, /etc/zsh/zprofile, .zprofile, /etc/zsh/zshrc, .zshrc, /etc/zsh/zlogin, .zlogin, and .zlogout files, as described in the section Files. If this option is unset, the /etc/zsh/zshenv file is still sourced, but any of the others will not be; it can be set at any time to prevent the remaining startup files after the currently executing one from being sourced. 🎀

🦋 Input/Output 🦋

ALIASES <D> — expand aliases. iconic behavior, honestly. 💅
CLOBBER (+C, ksh: +C) <D> — allows > redirection to truncate existing files. Otherwise >! or >| must be used to truncate a file. If the option is not set, and the option APPEND_CREATE is also not set, >>! or >>| must be used to create a file. If either option is set, >> may be used. 🔥
CLOBBER_EMPTY — this option is only used if the option CLOBBER is not set: note that it is set by default. If this option is set, then regular files of zero length may be overwritten (clobbered). Note that it is possible another process has written to the file between this test and use of the file by the current process. This option should therefore not be used in cases where files to be clobbered may be written to asynchronously. 😭
CORRECT (-0) — try to correct the spelling of commands. Note that, when the HASH_LIST_ALL option is not set or when some directories in the path are not readable, this may falsely report spelling errors the first time some commands are used. The shell variable CORRECT_IGNORE may be set to a pattern to match words that will never be offered as corrections. ⭐
CORRECT_ALL (-O) — try to correct the spelling of all arguments in a line. The shell variable CORRECT_IGNORE_FILE may be set to a pattern to match file names that will never be offered as corrections. 🧚
DVORAK — use the Dvorak keyboard instead of the standard qwerty keyboard as a basis for examining spelling mistakes for the CORRECT and CORRECT_ALL options and the spell-word editor command. keyboard nerds, this one’s for you. ⌨️
FLOW_CONTROL <D> — if this option is unset, output flow control via start/stop characters (usually assigned to ^S/^Q) is disabled in the shell’s editor. 🫶
IGNORE_EOF (-7) — do not exit on end-of-file. Require the use of exit or logout instead. However, ten consecutive EOFs will cause the shell to exit anyway, to avoid the shell hanging if its tty goes away. Also, if this option is set and the Zsh Line Editor is used, widgets implemented by shell functions can be bound to EOF (normally Control-D) without printing the normal warning message. This works only for normal widgets, not for completion widgets. 💀
INTERACTIVE_COMMENTS (-k) <K> <S> — allow comments even in interactive shells. yes bestie, comment your heart out. 💬
HASH_CMDS <D> — note the location of each command the first time it is executed. Subsequent invocations of the same command will use the saved location, avoiding a path search. If this option is unset, no path hashing is done at all. However, when CORRECT is set, commands whose names do not appear in the functions or aliases hash tables are hashed in order to avoid reporting them as spelling errors. ⚡
HASH_DIRS <D> — whenever a command name is hashed, hash the directory containing it, as well as all directories that occur earlier in the path. Has no effect if neither HASH_CMDS nor CORRECT is set. 🗂️
HASH_EXECUTABLES_ONLY — when hashing commands because of HASH_CMDS, check that the file to be hashed is actually an executable. This option is unset by default because if the path contains a large number of commands, or consists of many remote files, the additional tests can take a long time. Trial and error is needed to show if this option is beneficial. 🔍
MAIL_WARNING (-U) — print a warning message if a mail file has been accessed since the shell last checked. old-school but valid. 📬
PATH_DIRS (-Q) — perform a path search even on command names with slashes in them. Thus if /usr/local/bin is in the user’s path, and he or she types X11/xinit, the command /usr/local/bin/X11/xinit will be executed (assuming it exists). Commands explicitly beginning with /, ./ or ../ are not subject to the path search. This also applies to the . and source builtins. Note that subdirectories of the current directory are always searched for executables specified in this form. This takes place before any search indicated by this option, and regardless of whether . or the current directory appear in the command search path. 🌈
PATH_SCRIPT <K> <S> — if this option is not set, a script passed as the first non-option argument to the shell must contain the name of the file to open. If this option is set, and the script does not specify a directory path, the script is looked for first in the current directory, then in the command path. See the section INVOCATION in zsh(1). 🧭
PRINT_EIGHT_BIT — print eight bit characters literally in completion lists, etc. This option is not necessary if your system correctly returns the printability of eight bit characters (see ctype(3)). 🖨️
PRINT_EXIT_VALUE (-1) — print the exit value of programs with non-zero exit status. This is only available at the command line in interactive shells. 👀
RC_QUOTES — allow the character sequence ''' to signify a single quote within singly quoted strings. Note this does not apply in quoted strings using the format $'...', where a backslashed single quote can be used. 💅
RM_STAR_SILENT (-H) <K> <S> — do not query the user before executing rm * or rm path/*. bold move, bestie. 💀
RM_STAR_WAIT — if querying the user before executing rm * or rm path/*, first wait ten seconds and ignore anything typed in that time. This avoids the problem of reflexively answering yes to the query when one didn’t really mean it. The wait and query can always be avoided by expanding the * in ZLE (with tab). ⏳
SHORT_LOOPS <C> <Z> — allow the short forms of for, repeat, select, if, and function constructs. ✨
SHORT_REPEAT — allow the short form repeat as SHORT_LOOPS but without enabling it for the other constructs. 🌀
SUN_KEYBOARD_HACK (-L) — if a line ends with a backquote, and there are an odd number of backquotes on the line, ignore the trailing backquote. This is useful on some keyboards where the return key is too small, and the backquote key lies annoyingly close to it. As an alternative the variable KEYBOARD_HACK lets you choose the character to be removed. ⌨️

💅 Job Control 💅

AUTO_CONTINUE — with this option set, stopped jobs that are removed from the job table with the disown builtin command are automatically sent a CONT signal to make them running. 🫶
AUTO_RESUME (-W) — treat single word simple commands without redirection as candidates for resumption of an existing job. 🔁
BG_NICE (-6) <C> <Z> — run all background jobs at a lower priority. This option is set by default. 😴
CHECK_JOBS <Z> — report the status of background and suspended jobs before exiting a shell with job control; a second attempt to exit the shell will succeed. NO_CHECK_JOBS is best used only in combination with NO_HUP, else such jobs will be killed automatically. The check is omitted if the commands run from the previous command line included a jobs command, since it is assumed the user is aware that there are background or suspended jobs. A jobs command run from one of the hook functions defined in the section SPECIAL FUNCTIONS in zshmisc(1) is not counted for this purpose. 🧐
CHECK_RUNNING_JOBS <Z> — check for both running and suspended jobs when CHECK_JOBS is enabled. When this option is disabled, zsh checks only for suspended jobs, which matches the default behavior of bash. This option has no effect unless CHECK_JOBS is set. 🏃
HUP <Z> — send the HUP signal to running jobs when the shell exits. 👋
LONG_LIST_JOBS (-R) — print job notifications in the long format by default. 📋
MONITOR (-m, ksh: -m) — allow job control. Set by default in interactive shells. 👑
NOTIFY (-5, ksh: -b) <Z> — report the status of background jobs immediately, rather than waiting until just before printing a prompt. 🔔
POSIX_JOBS <K> <S> — this option makes job control more compliant with the POSIX standard. When the option is not set, the MONITOR option is unset on entry to subshells, so that job control is no longer active. When the option is set, the MONITOR option and job control remain active in the subshell, but note that the subshell has no access to jobs in the parent shell. 🌸
PROMPT_BANG <K> — okay so when this one's on, ! gets treated specially during prompt expansion, bestie. Peep EXPANSION OF PROMPT SEQUENCES in zshmisc(1) for the deets ✨
PROMPT_CR (+V) <D> — prints a carriage return right before a prompt shows up in the line editor. It's on by default, and lowkey it has to be — multi-line editing literally only works if the editor knows where the start of the line is 🎀
PROMPT_SP <D> — the lil' rescue mission for partial lines 😭 like, a line that didn't end with a newline and would otherwise get absolutely bodied by the command prompt thanks to PROMPT_CR. How's it do it? It outputs some cursor-control characters — including a whole series of spaces — that should make your terminal wrap to the next line when a partial line is chillin' there. Quick note tho: this only slays if your terminal has automatic margins, which, tbh, is most of them 💅

When a partial line does get preserved, by default you'll see an inverse+bold character hanging at the end of it: a % for a normal user or a # for root. And if you set it, the shell parameter PROMPT_EOL_MARK lets you customize how the end of partial lines gets shown 👑

NOTE: if PROMPT_CR is not set, turning this on does literally nothing. This option is on by default though 🫶

PROMPT_PERCENT <C> <Z> — if set, % gets treated specially in prompt expansion. Again, EXPANSION OF PROMPT SEQUENCES in zshmisc(1) is your bestie here 🌸
PROMPT_SUBST <K> <S> — set it and parameter expansion, command substitution, and arithmetic expansion all happen inside your prompts 🎀 And no stress: substitutions within prompts don't affect the command status, periodt.
TRANSIENT_RPROMPT — this one yeets any right prompt from the display when you accept a command line. Could be useful if your terminal's got other cut/paste methods going on 🦋

📜 Scripts and Functions 📜

ALIAS_FUNC_DEF <S> — so by default, zsh does not let you define functions with the name () syntax if name got expanded as an alias — it throws an error at you. And honestly that's usually what you want, because an alias plus a function based on the same definition can easily cause problems, iykyk 💀

When this option is set, aliases can be used for defining functions.

For example, consider these definitions as they might show up in a startup file:

alias foo=bar
foo() {
  print This probably does not do what you expect.
}

Here, foo gets expanded as an alias to bar before the () is even encountered, so the function defined would be named bar 😭. By default this is instead an error in native mode. Note that quoting any part of the function name, or using the keyword function, avoids the problem entirely — so do that when your function name could also be an alias. Highly recommended, no cap 🫶

C_BASES — outputs hexadecimal numbers in the standard C format, like 0xFF instead of the usual 16#FF ✨ And if OCTAL_ZEROES is also set (it is not by default), octal numbers get treated similarly, so they'd show up as 077 instead of 8#77. This option has no effect on the choice of output base, nor on the output of bases other than hexadecimal and octal. Also note: these formats will be understood on input regardless of the C_BASES setting 👑
C_PRECEDENCES — this one alters the precedence of arithmetic operators to be more like C and other programming languages 💅 The section ARITHMETIC EVALUATION in zshmisc(1) has the explicit list if you wanna check it.
DEBUG_BEFORE_CMD <D> — runs the DEBUG trap before each command; otherwise it runs after each command. Setting this mimics the behaviour of ksh 93; with it unset, you get the behaviour of ksh 88 🦋
ERR_EXIT (-e, ksh: -e) — if a command has a non-zero exit status, execute the ZERR trap, if set, and exit. This is disabled while running initialization scripts 🎀

The behaviour is also disabled inside DEBUG traps. In this case the option gets handled specially: it's unset on entry to the trap. If DEBUG_BEFORE_CMD is set, as it is by default, and ERR_EXIT is found to have been set on exit, then the command for which the DEBUG trap is being executed gets skipped. The option is restored after the trap exits 👑

Non-zero status in a command list containing && or || is ignored for commands not at the end of the list. Hence

false && true

does not trigger exit. Slay.

Exiting due to ERR_EXIT has certain interactions with asynchronous jobs noted in the section JOBS in zshmisc(1) ⭐

ERR_RETURN — if a command has a non-zero exit status, return immediately from the enclosing function. The logic is similar to ERR_EXIT, except that an implicit return statement is executed instead of an exit. This will trigger an exit at the outermost level of a non-interactive script 🌸

Normally this option inherits the behaviour of ERR_EXIT that code followed by && || does not trigger a return. Hence in the following:

summit || true

no return is forced, since the combined effect always has a zero return status 💅

Note, however, that if summit in the above example is itself a function, code inside it is considered separately: it may force a return from summit (assuming the option remains set within summit), but not from the enclosing context. This behaviour is different from ERR_EXIT, which is unaffected by function scope 👀

EVAL_LINENO <Z> — if set, line numbers of expressions evaluated using the builtin eval are tracked separately from the enclosing environment ✨ This applies both to the parameter LINENO and the line number output by the prompt escape %i. If the option is set, the prompt escape %N will output the string (eval) instead of the script or function name as an indication. (The two prompt escapes are typically used in the parameter PS4 to be output when the option XTRACE is set.) If EVAL_LINENO is unset, the line number of the surrounding script or function is retained during the evaluation 🫶
EXEC (+n, ksh: +n) <D> — do execute commands. Without this option, commands are read and checked for syntax errors, but not executed 💀 This option can't be turned off in an interactive shell, except when -n is supplied to the shell at startup.
FUNCTION_ARGZERO <C> <Z> — when executing a shell function or sourcing a script, set $0 temporarily to the name of the function/script 🎀 Note that toggling FUNCTION_ARGZERO from on to off (or off to on) does not change the current value of $0. Only the state upon entry to the function or script has an effect. Compare POSIX_ARGZERO ⭐
LOCAL_LOOPS — when this option is not set, the effect of break and continue commands may propagate outside function scope, affecting loops in calling functions 😱 When the option is set in a calling function, a break or a continue that isn't caught within a called function (regardless of the setting of the option within that function) produces a warning and the effect is cancelled 💅
LOCAL_OPTIONS <K> — if this option is set at the point of return from a shell function, most options (including this one) which were in force upon entry to the function are restored; the options that are not restored are PRIVILEGED and RESTRICTED 👑 Otherwise, only this option, plus the LOCAL_LOOPS, XTRACE and PRINT_EXIT_VALUE options, get restored. Hence if this is explicitly unset by a shell function, the other options in force at the point of return will remain so. A shell function can also guarantee itself a known shell configuration with a formulation like emulate -L zsh — the -L activates LOCAL_OPTIONS ✨
LOCAL_PATTERNS — if this option is set at the point of return from a shell function, the state of pattern disables — as set with the builtin command disable -p — is restored to what it was when the function was entered 🌸 The behaviour of this option is similar to the effect of LOCAL_OPTIONS on options; hence emulate -L sh (or indeed any other emulation with the -L option) activates LOCAL_PATTERNS 🦋
LOCAL_TRAPS <K> — if this option is set when a signal trap is set inside a function, then the previous status of the trap for that signal will be restored when the function exits 🎀 Note that this option must be set prior to altering the trap behaviour in a function; unlike LOCAL_OPTIONS, the value on exit from the function is irrelevant. However, it doesn't need to be set before any global trap for that to be correctly restored by a function. For example,

unsetopt localtraps
trap - INT
fn() { setopt localtraps; trap '' INT; sleep 3; }

will restore normal handling of SIGINT after the function exits 💅

MULTI_FUNC_DEF <Z> — allows definitions of multiple functions at once in the form fn1 fn2...(); if the option isn't set, this causes a parse error 😭 Definition of multiple functions with the function keyword is always allowed though. Multiple function definitions aren't often used and can cause obscure errors, so like… be careful bestie 🙈
MULTIOS <Z> — performs implicit tees or cats when multiple redirections are attempted (see the section Redirection) 🔥
OCTAL_ZEROES <S> — okay so if u flip this on, any integer constant starting with a 0 gets read as octal, per IEEE Std 1003.2-1992 (ISO 9945-2:1993). It's not on by default bestie, because it literally breaks parsing of stuff like date and time strings that have leading zeroes 😭. Important nuance tho: sequences of digits that indicate a numeric base — like the 08 component inside 08#77 — are always interpreted as decimal, no matter how many leading zeroes they rock. 💅
PIPE_FAIL — so by default, when a pipeline exits, the exit status the shell records and hands back in $? reflects the rightmost element of the pipeline. If u set this option, it instead reflects the status of the rightmost element that was non-zero, or zero if every element exited with zero status. slay 🫶
SOURCE_TRACE — set this and zsh prints an informational message announcing the name of every file it loads ✨. Format of the output is similar to what u get with XTRACE, with the message <sourcetrace>. Files can get loaded by the shell itself at startup and shutdown (Startup/Shutdown Files) or by using the source and dot builtin commands. no cap, u get receipts for every file 👑
TYPESET_SILENT — if this is unset, running any of the typeset family of commands with no options and a list of parameters that have no values to assign but already exist will display the value of the parameter 👀. If the option is set, they'll only be shown when parameters are selected with the -m option. The -p option is available whether or not this option is set — iconic behaviour fr.
TYPESET_TO_UNSET <K> <S> — when u declare a new parameter with any of the typeset family of related commands, the parameter stays unset unless and until a value is explicitly assigned to it — either in the typeset command itself or as a later assignment statement. literally no vibes until u give it vibes 💀
VERBOSE (-v, ksh: -v) — prints shell input lines as they're read. that's it, that's the whole thing 📖
XTRACE (-x, ksh: -x) — prints commands and their arguments as they're executed. The output is preceded by the value of $PS4, formatted as described in the section EXPANSION OF PROMPT SEQUENCES in zshmisc(1). main character energy for debugging, honestly 🔥

🐚 Shell Emulation 🐚

APPEND_CREATE <K> <S> — this option only applies when NO_CLOBBER (-C) is in effect, so keep that in mind bestie.

If this option is not set, the shell reports an error when an append redirection (>>) is used on a file that does not already exist (that's the traditional zsh behaviour of NO_CLOBBER). If the option is set, no error is reported (POSIX behaviour). ✨

BASH_REMATCH — when set, matches performed with the =~ operator will set the BASH_REMATCH array variable instead of the default MATCH and match variables. The first element of BASH_REMATCH holds the entire matched text, and subsequent elements hold the extracted substrings 🎀. This option makes way more sense when KSH_ARRAYS is also set, so the entire matched portion is stored at index 0 and the first substring is at index 1. Without this option, MATCH contains the entire matched text and the match array variable contains the substrings. iykyk 💅
BSD_ECHO <S> — makes the echo builtin compatible with the BSD echo(1) command. This disables backslashed escape sequences in echo strings unless the -e option is specified. cute, consistent, done 🦋
CONTINUE_ON_ERROR — if a fatal error is encountered (see the section ERRORS in zshmisc(1)) and the code is running in a script, the shell will resume execution at the next statement in the script at the top level — in other words, outside all functions or shell constructs like loops and conditions. This mimics the behaviour of interactive shells, where the shell returns to the line editor to read a new command; it was the normal behaviour in versions of zsh before 5.0.1. she picks herself up and keeps going 😭💪
CSH_JUNKIE_HISTORY <C> — a history reference without an event specifier will always refer to the previous command. Without this option, such a history reference refers to the same event as the previous history reference on the current command line, defaulting to the previous command. 🕰️
CSH_JUNKIE_LOOPS <C> — lets loop bodies take the form list; end instead of do list; done. retro but make it fashion 💅
CSH_JUNKIE_QUOTES <C> — changes the rules for single- and double-quoted text to match csh. These require embedded newlines to be preceded by a backslash; unescaped newlines will cause an error message 🚨. In double-quoted strings, it becomes impossible to escape $, ` or " (and \ itself no longer needs escaping). Command substitutions are only expanded once and cannot be nested. chaotic csh energy, handle with care 😌
CSH_NULLCMD <C> — do not use the values of NULLCMD and READNULLCMD when running redirections with no command. This makes such redirections fail (see the section Redirection). ngl, it's giving strict 🍓
KSH_ARRAYS <K> <S> — emulates ksh array handling as closely as possible. If set: array elements are numbered from zero, an array parameter without a subscript refers to the first element instead of the whole array, and braces are required to delimit a subscript (${path[2]} rather than just $path[2]) or to apply modifiers to any parameter (${PWD:h} rather than $PWD:h). brace urself bestie, literally 🎀
KSH_AUTOLOAD <K> <S> — emulates ksh function autoloading. This means when a function is autoloaded, the corresponding file is merely executed and must define the function itself. (By default, the function is defined to the contents of the file. However, the most common ksh-style case — of the file containing only a simple definition of the function — is always handled in the ksh-compatible manner.) so ur usual setup just works, periodt ✨
KSH_OPTION_PRINT <K> — alters the way option settings are printed: instead of separate lists of set and unset options, all options are shown, marked on if they're in the non-default state, off otherwise. full transparency, we love to see it 👑
KSH_TYPESET — this option is now obsolete: a better approximation to the behaviour of other shells is obtained with the reserved word interface to declare, export, float, integer, local, readonly and typeset. Note the option is only applied when the reserved word interface is not in use.

It alters the way arguments to the typeset family of commands — including declare, export, float, integer, local and readonly — are processed. Without this option, zsh performs normal word splitting after command and parameter expansion in arguments of an assignment; with it, word splitting does not take place in those cases. 💀 (also, yes, the original really does spell it "appropximation" — we're keeping it, character for character 🫶)

KSH_ZERO_SUBSCRIPT — treats use of a subscript of value zero in array or string expressions as a reference to the first element, i.e. the element that usually has the subscript 1. Ignored if KSH_ARRAYS is also set.

If neither this option nor KSH_ARRAYS is set, accesses to an element of an array or string with subscript zero return an empty element or string, while attempts to set element zero of an array or string are treated as an error 😤. However, attempts to set an otherwise valid subscript range that includes zero will succeed. For example, if KSH_ZERO_SUBSCRIPT is not set,

array[0]=(element)

is an error, while

array[0,1]=(element)

is not, and will replace the first element of the array. This option is for compatibility with older versions of the shell and is not recommended in new code. bestie, don't do it 🙈

POSIX_ALIASES <K> <S> — when set, reserved words are not candidates for alias expansion: it's still possible to declare any of them as an alias, but the alias will never be expanded. Reserved words are described in the section RESERVED WORDS in zshmisc(1).

Alias expansion takes place while text is being read; hence when this option is set it doesn't take effect until the end of any function or other piece of shell code parsed as one unit ⏳. Note this may cause differences from other shells even when the option is in effect. For example, when running a command with zsh -c, or even zsh -o posixaliases -c, the entire command argument is parsed as one unit, so aliases defined within the argument are not available even in later lines. If in doubt, avoid use of aliases in non-interactive code. fr fr, just don't 🍓

POSIX_ARGZERO — this option may be used to temporarily disable FUNCTION_ARGZERO and thereby restore the value of $0 to the name used to invoke the shell (or as set by the -c command line option). For compatibility with previous versions of the shell, emulations use NO_FUNCTION_ARGZERO instead of POSIX_ARGZERO, which may result in unexpected scoping of $0 if the emulation mode is changed inside a function or script 😭. To avoid this, explicitly enable POSIX_ARGZERO in the emulate command:

emulate sh -o POSIX_ARGZERO

Note that NO_POSIX_ARGZERO has no effect unless FUNCTION_ARGZERO was already enabled upon entry to the function or script. sneaky, but that's the tea ☕

POSIX_BUILTINS <K> <S> — when this option is set, the command builtin can be used to execute shell builtin commands. Parameter assignments specified before shell functions and special builtins are kept after the command completes unless the special builtin is prefixed with the command builtin. Special builtins are ., :, break, continue, declare, eval, exit, export, integer, local, readonly, return, set, shift, source, times, trap and unset. 👑

In addition, various error conditions associated with the above builtins or exec cause a non-interactive shell to exit and an interactive shell to return to its top-level processing.

Furthermore, functions and shell builtins are not executed after an exec prefix; the command to be executed must be an external command found in the path.

Furthermore, the getopts builtin behaves in a POSIX-compatible fashion in that the associated variable OPTIND is not made local to functions, and its value is calculated differently to match other shells.

Moreover, the warning and special exit code from [[ -o non_existent_option ]] are suppressed. quiet, discreet, POSIX 😌✨

POSIX_IDENTIFIERS <K> <S> — bestie, when this one's on, the only characters allowed in identifiers (aka names of shell parameters and modules) are the ASCII a to z, A to Z, 0 to 9 and _. That's it, that's the guest list.
- It also nerfs parameter substitution with no braces, so $# gets treated as the parameter $# even if a valid parameter name is sitting right next to it. When it's unset, zsh lets expressions like $#name mean "the length of $name" — even for special variables, like in expressions such as $#- and $#*.
- Another lil difference: with the option set, assignment to an unset variable in arithmetic context creates that variable as a scalar instead of a numeric type. So after unset t; (( t = 3 )), without POSIX_IDENTIFIERS set t has integer type, and with it set it has scalar type. Iconic behavior switch, honestly.
- When the option is unset and multibyte character support is enabled (i.e. it's compiled in and the MULTIBYTE option is set), then you also get any alphanumeric characters in the local character set usable in identifiers. ⚠️ Note that scripts and functions written with this feature are not portable, and also that both options must be set before the script or function is parsed — setting them during execution is not sufficient because the syntax variable=value has already been parsed as a command rather than an assignment. The parser does not wait for your vibe check.
- If multibyte character support is not compiled into the shell, this option is ignored; all octets with the top bit set may be used in identifiers. Non-standard? Yes. But it's the traditional zsh behaviour, so 💅.
POSIX_STRINGS <K> <S> — this one's all about processing quoted strings. Currently it only affects the behaviour of null characters, i.e. character 0 in the portable character set corresponding to US ASCII.
- When this option is not set, null characters embedded within strings of the form $'...' are treated as ordinary characters. The entire string is maintained within the shell and output to files where necessary, although — owing to restrictions of the library interface — the string is truncated at the null character in file names, environment variables, or in arguments to external programs.
- When this option is set, the $'...' expression is truncated at the null character. Note that the remaining parts of the same string beyond the termination of the quotes are not truncated. Sneaky little detail, don't sleep on it.
- Example time: the command line argument a$'b\0c'd is treated with the option off as the characters a, b, null, c, d, and with the option on as the characters a, b, d. Two totally different little guys. 🫶
POSIX_TRAPS <K> <S> — when this is set, the usual zsh behaviour of executing traps for EXIT on exit from shell functions gets suppressed. In that case, manipulating EXIT traps always alters the global trap for exiting the shell; the LOCAL_TRAPS option is ignored for the EXIT trap.
- Also, a return statement executed in a trap with no argument passes back from the function the value from the surrounding context, not from code executed within the trap.
- Furthermore, if a trap is set to be ignored, this state persists when a subshell is entered. Without the option, the trap would be reset to its default state at this point. So like, choose your fighter.
SH_FILE_EXPANSION <K> <S> — does filename expansion (e.g. ~ expansion) before parameter expansion, command substitution, arithmetic expansion and brace expansion. If this option is unset, it's performed after brace expansion, so things like ~$USERNAME and ~{pfalstad,rc} will work. Certifiably slaying.
SH_NULLCMD <K> <S> — do not use the values of NULLCMD and READNULLCMD when doing redirections, use : instead (see the section Redirection).
SH_OPTION_LETTERS <K> <S> — if this is set, the shell tries to interpret single letter options (which are used with set and setopt) like ksh does. This also affects the value of the - special parameter.
SH_WORD_SPLIT (-y) <K> <S> — causes field splitting to be performed on unquoted parameter expansions. Note that this option has nothing to do with word splitting. (See zshexpn(1).) Yes, the naming is giving confusing on purpose, we cope. 😭
TRAPS_ASYNC — while waiting for a program to exit, handle signals and run traps immediately. Otherwise the trap is run after a child process has exited. Note this does not affect the point at which traps are run for any case other than when the shell is waiting for a child process. Very specific queen.

Shell State 👑

INTERACTIVE (-i, ksh: -i) — this is an interactive shell. This option is set upon initialisation if the standard input is a tty and commands are being read from standard input. (See the discussion of SHIN_STDIN.) This heuristic may be overridden by specifying a state for this option on the command line. The value of this option can only be changed via flags supplied at invocation of the shell — it cannot be changed once zsh is running. Sorry bestie, no take-backs.
LOGIN (-l, ksh: -l) — this is a login shell. If this option is not explicitly set, the shell becomes a login shell if the first character of the argv[0] passed to the shell is a -. One (1) hyphen and suddenly you're a login shell, that's the whole personality test.
PRIVILEGED (-p, ksh: -p) — turns on privileged mode. Typically this is used when a script is to be run with elevated privileges. This should be done as follows, directly with the -p option to zsh, so that it takes effect during startup:

#!/bin/zsh -p

The option is enabled automatically on startup if the effective user (group) ID is not equal to the real user (group) ID. In this case, turning the option off causes the effective user and group IDs to be set to the real user and group IDs. ⚠️ Be aware that if that fails the shell may be running with different IDs than was intended, so a script should check for failure and act accordingly, for example:

unsetopt privileged || exit

The PRIVILEGED option disables sourcing user startup files. If zsh is invoked as sh or ksh with this option set, /etc/suid_profile is sourced (after /etc/profile on interactive shells). Sourcing ~/.profile is disabled and the contents of the ENV variable is ignored. This option cannot be changed using the -m option of setopt and unsetopt, and changing it inside a function always changes it globally regardless of the LOCAL_OPTIONS option. No sneaky local overrides, it's giving absolute authority.

RESTRICTED (-r) — enables restricted mode. This option cannot be changed using unsetopt, and setting it inside a function always changes it globally regardless of the LOCAL_OPTIONS option. See the section Restricted Shell. Once you're in, you're in. 🔥
SHIN_STDIN (-s, ksh: -s) — commands are being read from the standard input. Commands are read from standard input if no command is specified with -c and no file of commands is specified. If SHIN_STDIN is set explicitly on the command line, any argument that would otherwise have been taken as a file to run will instead be treated as a normal positional parameter. Note that setting or unsetting this option on the command line does not necessarily affect the state the option will have while the shell is running — that is purely an indicator of whether or not commands are actually being read from standard input. The value of this option can only be changed via flags supplied at invocation of the shell. It cannot be changed once zsh is running.
SINGLE_COMMAND (-t, ksh: -t) — if the shell is reading from standard input, it exits after a single command has been executed. This also makes the shell non-interactive, unless the INTERACTIVE option is explicitly set on the command line. The value of this option can only be changed via flags supplied at invocation of the shell. It cannot be changed once zsh is running. One command and dip. 👋

Zle 🎀

BEEP (+B) <D> — beep on error in ZLE. Tiny little sound effect, honestly kind of a jumpscare.
COMBINING_CHARS — assume that the terminal displays combining characters correctly. Specifically, if a base alphanumeric character is followed by one or more zero-width punctuation characters, assume that the zero-width characters will be displayed as modifications to the base character within the same width. Not all terminals handle this, so heads up. If this option is not set, zero-width characters are displayed separately with special mark-up.
- If this option is set, the pattern test [[:WORD:]] matches a zero-width punctuation character on the assumption that it will be used as part of a word in combination with a word character. Otherwise the base shell does not handle combining characters specially.
EMACS — if ZLE is loaded, turning on this option has the equivalent effect of bindkey -e. In addition, the VI option is unset. Turning it off has no effect. The option setting is not guaranteed to reflect the current keymap. This option is provided for compatibility; bindkey is the recommended interface. Ngl, it's the legacy option, use the icon instead.
OVERSTRIKE — start up the line editor in overstrike mode.
SINGLE_LINE_ZLE (-M) <K> — use single-line command line editing instead of multi-line.
- Note that although this is on by default in ksh emulation, it only provides superficial compatibility with the ksh line editor and reduces the effectiveness of the zsh line editor. As it has no effect on shell syntax, many users may wish to disable this option when using ksh emulation interactively. Respectfully, it's giving downgrade. 💀
VI — if ZLE is loaded, turning on this option has the equivalent effect of bindkey -v. In addition, the EMACS option is unset. Turning it off has no effect. The option setting is not guaranteed to reflect the current keymap. This option is provided for compatibility; bindkey is the recommended interface.
ZLE (-Z) — use the zsh line editor. Set by default in interactive shells connected to a terminal. Main character energy, as it should be. ✨
BRACE_EXPAND → NO_IGNORE_BRACES (ksh and bash compatibility, bestie 💅)
DOT_GLOB → GLOB_DOTS (bash compatibility)
HASH_ALL → HASH_CMDS (bash compatibility)
HIST_APPEND → APPEND_HISTORY (bash compatibility)
HIST_EXPAND → BANG_HIST (bash compatibility)
LOG → NO_HIST_NO_FUNCTIONS (ksh compatibility)
MAIL_WARN → MAIL_WARNING (bash compatibility)
ONE_CMD → SINGLE_COMMAND (bash compatibility)
PHYSICAL → CHASE_LINKS (ksh and bash compatibility)
PROMPT_VARS → PROMPT_SUBST (bash compatibility)
STDIN → SHIN_STDIN (ksh compatibility)
TRACK_ALL → HASH_CMDS (ksh compatibility)

Like, literally never expect to see these alias names in output — the shell only ever prints the real ones. But when you're setting options you can totally use them like any normal option name. iykyk 🫶

✨ SINGLE LETTER OPTIONS ✨

Ok so this is the cheat sheet, the little one-letter glow-ups. Buckle up, there's a lot of them 💖

🎀 Default set

-0 — CORRECT
-1 — PRINT_EXIT_VALUE
-2 — NO_BAD_PATTERN
-3 — NO_NOMATCH
-4 — GLOB_DOTS
-5 — NOTIFY
-6 — BG_NICE
-7 — IGNORE_EOF
-8 — MARK_DIRS
-9 — AUTO_LIST
-B — NO_BEEP
-C — NO_CLOBBER
-D — PUSHD_TO_HOME
-E — PUSHD_SILENT
-F — NO_GLOB
-G — NULL_GLOB
-H — RM_STAR_SILENT
-I — IGNORE_BRACES
-J — AUTO_CD
-K — NO_BANG_HIST
-L — SUN_KEYBOARD_HACK
-M — SINGLE_LINE_ZLE
-N — AUTO_PUSHD
-O — CORRECT_ALL
-P — RC_EXPAND_PARAM
-Q — PATH_DIRS
-R — LONG_LIST_JOBS
-S — REC_EXACT
-T — CDABLE_VARS
-U — MAIL_WARNING
-V — NO_PROMPT_CR
-W — AUTO_RESUME
-X — LIST_TYPES
-Y — MENU_COMPLETE
-Z — ZLE
-a — ALL_EXPORT
-e — ERR_EXIT
-f — NO_RCS
-g — HIST_IGNORE_SPACE
-h — HIST_IGNORE_DUPS
-i — INTERACTIVE
-k — INTERACTIVE_COMMENTS
-l — LOGIN
-m — MONITOR
-n — NO_EXEC
-p — PRIVILEGED
-r — RESTRICTED
-s — SHIN_STDIN
-t — SINGLE_COMMAND
-u — NO_UNSET
-v — VERBOSE
-w — CHASE_LINKS
-x — XTRACE
-y — SH_WORD_SPLIT

🦋 sh/ksh emulation set

-C — NO_CLOBBER
-T — TRAPS_ASYNC
-X — MARK_DIRS
-a — ALL_EXPORT
-b — NOTIFY
-e — ERR_EXIT
-f — NO_GLOB
-i — INTERACTIVE
-l — LOGIN
-m — MONITOR
-n — NO_EXEC
-p — PRIVILEGED
-r — RESTRICTED
-s — SHIN_STDIN
-t — SINGLE_COMMAND
-u — NO_UNSET
-v — VERBOSE
-x — XTRACE

🌸 Also note

-A — used by set for setting arrays
-b — used on the command line to specify end of option processing
-c — used on the command line to specify a single command
-m — used by setopt for pattern-matching option setting
-o — used in all places to allow use of long option names
-s — used by set to sort positional parameters

zsh 5.9.2                        July 12, 2026                    ZSHOPTIONS(1)

ZSHBUILTINS(1)              General Commands Manual              ZSHBUILTINS(1)

👑 NAME 👑

zshbuiltins — zsh built-in commands ✨

🧰 SHELL BUILTIN COMMANDS 🧰

Okay so bestie, some shell builtin commands take options, and those get described in their own individual entries. Down below in the list we usually call them flags — that's on purpose, to stop you from confusing them with shell options, which are a whole different thing that can also change how a builtin behaves. Context is everything 💅

And just to be clear: in this intro section specifically, the word option always means an option to a command, like, the kind of thing any regular command-line user already knows and loves.

Typically, options are single letters with a hyphen (-) in front. Options that want an argument will take it either right up against the option letter, or after some white space — both of these are literally equivalent:

print -C3 {1..9}
print -C 3 {1..9}

And no, arguments to options are not the same thing as arguments to the command — the docs always tell you which is which. Options that don't take an argument can be mashed together into one word, so these two are equivalent as well:

print -rca -- *
print -r -c -a -- *

Some shell builtin commands also take options that start with + instead of -. Which commands? The list below makes it obvious, don't worry 🫶

Options (plus their own little arguments, if they have any) have to show up as a group before any non-option arguments. The moment the first non-option argument appears, option processing is done, over, periodt.

Every builtin command except echo and the precommand modifiers — even the ones with zero options — can be handed -- to end option processing. It tells the shell "everything after this is a non-option argument", and other than that it's just ignored. Super useful when your arguments might start with -. For historical reasons, most builtin commands (including echo) also recognize a single - sitting in its own word for the same purpose — but ngl that's less standard, so use --, ok? 👑

`-` — simple command

Check the section Precommand Modifiers in zshmisc(1) for the tea 🍓

`.` `file [ arg ... ]`

Reads commands from file and runs them in the current shell environment. No subshell, no escape 😌

If file doesn't contain a slash, or if PATH_DIRS is set, the shell digs through the components of $path to find the directory holding file. Files sitting in the current directory are not read unless . shows up somewhere in $path. And if there's a file named file.zwc that's newer than file and is the compiled form of file (made with the zcompile builtin), then the commands get read from that one instead. Efficiency slay ✨

If you pass any arguments arg, they become the positional parameters — and the old positional parameters get restored once the file finishes executing. But if you give no arguments at all, the positional parameters just stay whatever they were in the calling context, and nothing gets restored. Careful with that one 🫶

Return statuses, because we love specifics:

file not found → return status 127
file found but has a syntax error → return status 126
otherwise → the exit status of the last command executed

`:` `[ arg ... ]`

This command does absolutely nothing, though normal argument expansion still happens, which can totally have side effects on shell parameters. A zero exit status is returned. Main character energy: present, doing nothing, flawless 💅

`alias` `[ {+|-}gmrsL ] [ name[=value] ... ]`

For each name that has a matching value, define an alias with that value. If value ends in a trailing space, the next word gets checked for alias expansion too. Neat.

If the -g flag is present, you're defining a global alias — and global aliases get expanded even when they're not in command position, which is kind of iconic:

% perldoc --help 2>&1 | grep 'built-in functions'
    -f   Search Perl built-in functions
% alias -g HG='--help 2>&1 | grep'
% perldoc HG 'built-in functions'
    -f   Search Perl built-in functions

If the -s flag is present, you're defining a suffix alias: when the command word on a command line looks like text.name, where text is any non-empty string, it gets replaced by value text.name. Heads up: name is treated as a literal string here, not a pattern. And a trailing space in value is not special in this case. For example:

alias -s ps='gv --'

…will make the command *.ps expand to gv -- *.ps. And since alias expansion happens earlier than globbing, that *.ps then gets globbed. Suffix aliases live in their own separate name space from other aliases — so in the example above you can still make a regular alias for the command ps — and the two sets are never listed together. Never ever 🙈

For each name with no value, print the value of name, if there is one. With no arguments at all, print every currently defined alias except suffix aliases. If the -m flag is given, the arguments are treated as patterns (quote them so the shell doesn't glob them first!), and every alias matching those patterns gets printed. When printing aliases, if one of -g, -r or -s is present, the printing is restricted to global, regular or suffix aliases respectively — and a regular alias is one that's neither global nor suffix. Oh, and using + instead of -, or ending the option list with a single +, stops the alias values from being printed at all.

If the -L flag is present, each alias is printed in a form suitable for dropping straight into a startup script. The exit status is nonzero if you ask for a name (with no value) that has no alias defined. Rude, but fair 💀

Want more on aliases, including the classic common problems? See the section ALIASING in zshmisc(1).

`autoload` `[ {+|-}RTUXdkmrtWz ] [ -w ] [ name ... ]`

Full details live in the section Autoloading Functions in zshmisc(1) — go read it, it's giving essential lore. The fpath parameter gets searched to find the function definition when the function is first referenced.

If name is an absolute path, the function is defined to load from the file you gave (searching as usual for dump files in that location). The function's name is the basename — the non-directory part — of the file. Normally it's an error if the function isn't found in that location; however, if the -d option is given, the search for the function defaults to $fpath. And if a function is loaded by absolute path, any functions loaded from it that are marked for autoload without an absolute path get the load path of the parent function temporarily prepended to $fpath. Family tree realness 🌈

If the -r or -R option is given, the function is searched for immediately and the location is recorded internally for use when the function actually executes; a relative path gets expanded using the value of $PWD. This protects you against $fpath changing after the autoload call — genuinely such a glow-up for reliability. With -r, if the function isn't found, it's silently left unresolved until execution (mystery, suspense, drama). With -R, an error message is printed and command processing is aborted immediately when the search fails — meaning at the autoload command itself, not later at function execution. Blunt, but honest 🔥

okay so we're still deep in autoload and the flags are lowkey doing the most, so let's go one by one bestie 💅

-X — bestie, this one is only allowed inside a shell function, no exceptions. What it does: it marks the calling function for autoloading and then immediately loads and executes it, with the current array of positional parameters as arguments. It replaces the previous definition of the function, like a glow-up that erases the before photo 😭. If no function definition is found, an error is printed and the function stays undefined and marked for autoloading. If an argument is given, it's used as a directory (i.e. it does not include the name of the function) in which the function is to be found; this may be combined with the -d option to allow the function search to default to $fpath if it's not in the given location. iykyk 🫶
+X — this one attempts to load each name as an autoloaded function, but it does not execute it. Exit status is zero (success) if the function was not previously defined and a definition for it was found. It does not replace any existing definition of the function. Exit status is nonzero (failure) if the function was already defined or when no definition was found. In that latter case the function remains undefined and marked for autoloading. If ksh-style autoloading is enabled, the function created will contain the contents of the file plus a call to the function itself appended to it, thus giving normal ksh autoloading behaviour on the first call to the function. And if the -m flag is also given, each name is treated as a pattern and all functions already marked for autoload that match the pattern are loaded ✨
-t — turns on execution tracing. periodt.
-T — turns on execution tracing only for the current function, turning it off on entry to any called functions that don't also have tracing enabled. main character energy only 🔥
-U — alias expansion is suppressed when the function is loaded ✨
-w — the names are taken as names of files compiled with the zcompile builtin, and all functions defined in them are marked for autoloading 🎀
-z — marks the function to be autoloaded using the zsh style, as if the option KSH_AUTOLOAD were unset.
-k — marks the function to be autoloaded using the ksh style, as if the option KSH_AUTOLOAD were set. The flags override the setting of the option at the time the function is loaded 👑

ngl bestie, autoload makes no attempt to ensure the shell options set during the loading or execution of the file have any particular value. That's simply not its job 🙈

If you actually want that, the emulate command is your girl:

emulate zsh -c 'autoload -Uz func'

this arranges that when func is loaded the shell is in native zsh emulation, and this emulation is also applied when func is run 💜

Also, some of the functions of autoload are also provided by functions -u or functions -U, but autoload is a more comprehensive interface. she ate and left no crumbs 🍓

🌈 bg

bg [ job ... ]
job ... &

Put each specified job in the background, or the current job if none is specified. classic move, no cap 💅

🌸 bindkey

See the section Zle Builtins' in zshzle(1)`. (she's over there living her best life)

✨ break

break [ n ]

Exit from an enclosing for, while, until, select or repeat loop. If an arithmetic expression n is specified, then break n levels instead of just one. escape the loop bestie, you don't have to stay 😭

🦋 builtin

builtin name [ args ... ]

Executes the builtin name, with the given args. no external nonsense, all in-house 💅

🎀 bye

Same as exit. that's it. that's the whole vibe. bye 👋

🧚 cap

See the section The zsh/cap Module' in zshmodules(1)`.

👑 cd

cd [ -qsLP ] [ arg ]
cd [ -qsLP ] old new
cd [ -qsLP ] {+|-}n

Change the current directory. In the first form, change the current directory to arg, or to the value of $HOME if arg is not specified. If arg is -, change to the previous directory. (the "take me back" of it all 💖)

Otherwise, if arg begins with a slash, attempt to change to the directory given by arg.

If arg does not begin with a slash, the behaviour depends on whether the current directory . occurs in the list of directories contained in the shell parameter cdpath. If it does not, first attempt to change to the directory arg under the current directory, and if that fails but cdpath is set and contains at least one element, attempt to change to the directory arg under each component of cdpath in turn until successful. If . occurs in cdpath, then cdpath is searched strictly in order so that . is only tried at the appropriate point. it's giving careful, ordered, iconic 🔥

The order of testing cdpath is modified if the option POSIX_CD is set, as described in the documentation for the option 👀

If no directory is found, the option CDABLE_VARS is set, and a parameter named arg exists whose value begins with a slash, treat its value as the directory. In that case, the parameter is added to the named directory hash table ✨

The second form of cd substitutes the string new for the string old in the name of the current directory, and tries to change to this new directory. find-and-replace but make it directory 🫶

The third form of cd extracts an entry from the directory stack, and changes to that directory. An argument of the form +n identifies a stack entry by counting from the left of the list shown by the dirs command, starting with zero. An argument of the form -n counts from the right. If the PUSHD_MINUS option is set, the meanings of + and - in this context are swapped 🙈. If the POSIX_CD option is set, this form of cd is not recognised and will be interpreted as the first form.

-q (quiet) — if specified, the hook function chpwd and the functions in the array chpwd_functions are not called. This is useful for calls to cd that do not change the environment seen by an interactive user. stealth mode 💅
-s — cd refuses to change the current directory if the given pathname contains symlinks.
-P — symbolic links are resolved to their true values. (also happens if the CHASE_LINKS option is set)
-L — symbolic links are retained in the directory (and not resolved) regardless of the state of the CHASE_LINKS option. do not perceive me 💀

⭐ chdir

Same as cd. literally the same. twin behaviour 💖

🌈 clone

See the section The zsh/clone Module' in zshmodules(1)`.

💜 command

command [ -pvV ] simple command

The simple command argument is taken as an external command instead of a function or builtin and is executed. If the POSIX_BUILTINS option is set, builtins will also be executed but certain special properties of them are suppressed. The -p flag causes a default path to be searched instead of that in $path. With the -v flag, command is similar to whence and with -V, it is equivalent to whence -v. slay 💅

See also the section Precommand Modifiers' in zshmisc(1)`.