re(3)

📦

version 29.0.6 · 3 versions

🫶 re(3) — Erlang Module Definition 🫶

💖 NAME 💖

re — bestie, this is the module: it's the one holding all the regular expression matching functions for strings and binaries. That's the whole job description. Slay. 🦋

🌸 DESCRIPTION 🌸

Okay so first thing: the syntax and semantics of these regular expressions resemble Perl's. Very that, very iconic. 🎀

The matching algorithms inside this library are based on the PCRE2 library — but not all of PCRE2 is interfaced, and some parts of this library go beyond what PCRE2 offers. Right now, PCRE2 version 10.45 (release date 2024-11-27) is what's running underneath. The sections of the PCRE2 documentation that are relevant to this module are included right here, so there's no need to go hunting elsewhere. 📚

NOTE: The Erlang literal syntax for strings uses the \ (backslash) character as an escape code. You need to escape backslashes in literal strings — both in your code and in the shell — with an extra backslash, that is, "\\" or <<"\\">>. Since Erlang/OTP 27 you can use verbatim sigils to write literal strings. The example above would be written as ~S"\\" or ~B"\\". No cap, this one trips up everybody, so please respect the backslashes. 💅

🦋 Perl-Like Regular Expression Syntax 🦋

The following sections contain the reference material for the regular expressions used by this module. All of it is based on the PCRE2 documentation, with changes noted wherever this module behaves differently from the PCRE2 library. 🫶

CHANGE: Starting with Erlang/OTP 28, the underlying library for handling regular expressions via the re module changes from PCRE to PCRE2. This upgrade brings Erlang's regular expression capabilities more in line with modern standards, particularly Perl 🐚, but it also introduces several breaking changes and differences in behavior compared to PCRE — see PCRE incompatabilities. Read that part, for real. 🚨

🌈 PCRE2 Regular Expression Details 🌈

The syntax and semantics of the regular expressions supported by PCRE2 are described in detail below. Perl's regular expressions are described in its own documentation, and regular expressions in general are covered in a number of books, some of which have copious examples. Jeffrey Friedl's "Mastering Regular Expressions", published by O'Reilly, covers regular expressions in great detail. This description of PCRE2's regular expressions is intended as reference material. 📖

The reference material is divided into the following sections:

Special Start-of-Pattern Items
Characters and Metacharacters
Backslash
Circumflex and Dollar
Full Stop (Period, Dot) and \N
Matching a Single Code Unit
Square Brackets and Character Classes
Perl Extended Character Classes
Posix Character Classes
Vertical Bar
Internal Option Setting
Groups
Duplicate Group Numbers
Named Capture Groups
Repetition
Atomic Grouping and Possessive Quantifiers
Backreferences
Assertions
Non-Atomic Assertions
Scan Substring Assertions
Script Runs
Conditional Groups
Comments
Recursive Patterns
Groups as Subroutines
Recursion and subroutines with returned capture groups
Oniguruma Subroutine Syntax
Backtracking Control

⭐ Special Start-of-Pattern Items ⭐

A bunch of options that can be passed to compile/2 can also be set by special items right at the start of a pattern. These are not Perl-compatible — they exist so that pattern writers who aren't able to change the program processing the pattern can still reach these options. You can have any number of these items, but ⚠️ they must all be together, right at the start of the pattern string, and the letters must be in upper case. Periodt. 💅

UTF support

Unicode support is basically UTF-8 based. To use Unicode characters, you either call compile/2 or run/3 with option unicode, or the pattern must start with one of these special sequences: (*UTF), which is equivalent to setting unicode. Heads up though — with these instructions, the automatic conversion of lists to UTF-8 is not performed by the re functions. Because of that, using these sequences is not recommended. Add option unicode when running compile/2 instead. ✨

Some applications that allow their users to supply patterns may wish to restrict them to non-UTF data for security reasons. If the never_utf option is passed to compile/2, (*UTF) is not allowed, and its appearance in a pattern causes an error. 🚫

Unicode Property Support

Another special sequence that may appear at the start of a pattern is (*UCP). Same effect as setting the ucp option: it causes sequences such as \d and \w to use Unicode properties to determine character types, instead of recognizing only characters with codes less than 256 via a lookup table. It also causes upper/lower casing operations to use Unicode properties for characters with code points greater than 127, even when UTF is not set. These behaviours can be changed within the pattern; see Internal Option Setting. 🌈

Some applications that allow their users to supply patterns may wish to restrict them for security reasons. If the never_ucp option is passed to compile/2, (*UCP) is not allowed, and its appearance in a pattern causes an error. 🚫

Locking Out Empty String Matching

Starting a pattern with (*NOTEMPTY) or (*NOTEMPTY_ATSTART) has the same effect as passing the notempty or notempty_atstart option to whichever matching function is subsequently called to match the pattern. These options lock out the matching of empty strings — either entirely, or only at the start of the subject. 🙅‍♀️

Disabling Start-Up Optimizations

If a pattern starts with (*NO_START_OPT), it has the same effect as setting the no_start_optimize option. This disables several optimizations for quickly reaching "no match" results. 🛑

Disabling Automatic Anchoring

If a pattern starts with (*NO_DOTSTAR_ANCHOR). This disables optimizations that apply to patterns whose top-level branches all start with .* (match any number of arbitrary characters).

Setting Match Resource Limits

The caller of run/3 can set a limit on this counter, which therefore limits the amount of computing resource used for a match. The maximum depth of nested backtracking can also be limited; this indirectly restricts the amount of heap memory that is used, but there is also an explicit memory limit that can be set. 🧠

These facilities are provided to catch runaway matches that are provoked by patterns with huge matching trees. A common example is a pattern with nested unlimited repeats applied to a long string that does not match. When one of these limits is reached, run/3 gives an error return. The limits can also be set by items at the start of the pattern of the form

(*LIMIT_HEAP=d)
(*LIMIT_MATCH=d)
(*LIMIT_DEPTH=d)

where d is any number of decimal digits. However, the value of the setting must be less than the value set as default for it to have any effect. In other words: the pattern writer can lower the limits set by the programmer, but not raise them. If there is more than one setting of one of these limits, the lower value is used. The heap limit is specified in kibibytes (units of 1024 bytes). 📏

The default value for LIMIT_MATCH and LIMIT_DEPTH is 10,000,000 in the Erlang VM. Notice that the recursion limit does not affect the stack depth of the VM, as PCRE for Erlang is compiled in such a way that the match function never does recursion on the C stack. 💀

Prior to release 10.30, LIMIT_DEPTH was called LIMIT_RECURSION. This name is still recognized for backwards compatibility. 🫶

Newline Conventions

PCRE2 supports six different conventions for indicating line breaks in strings: a single CR (carriage return) character, a single LF (linefeed) character, the two-character sequence CRLF, any of the three preceding, any Unicode newline sequence, or the NUL character (binary zero).

It is also possible to specify a newline convention by starting a pattern string with one of the following sequences:

(*CR) — carriage return
(*LF) — linefeed
(*CRLF) — carriage return, followed by linefeed
(*ANYCRLF) — any of the three above
(*ANY) — all Unicode newline sequences
(*NUL) — the NUL character (binary zero)

These override the default and the options given to the compiling function. For example, on a Unix system where LF is the default newline sequence, the pattern

(*CR)a.b

changes the convention to CR. That pattern matches "a\nb" because LF is no longer a newline. If more than one of these settings is present, the last one is used. 👑

The newline convention affects where the circumflex and dollar assertions are true. It also affects the interpretation of the dot metacharacter when dotall is not set, and the behaviour of \N when not followed by an opening brace. However, it does not affect what the \R escape sequence matches. By default, this is any Unicode newline sequence, for Perl compatibility. However, this can be changed; see the next section and the description of \R in Newline Sequences. A change of \R setting can be combined with a change of newline convention. 💫

Specifying What \R Matches

It is possible to restrict \R to match only CR, LF, or CRLF (instead of the complete set of Unicode line endings) by setting the option bsr_anycrlf at compile time. This effect can also be achieved by starting a pattern with (*BSR_ANYCRLF). For completeness, (*BSR_UNICODE) is also recognized, corresponding to bsr_anycrlf. 🌈

🎀 Characters and Metacharacters 🎀

A regular expression is a pattern that is matched against a subject string from left to right. Most characters stand for themselves in a pattern, and match the corresponding characters in the subject. As a trivial example, the pattern

The quick brown fox

matches a portion of a subject string that is identical to itself. 🦊

When caseless matching is specified (the caseless option or (?i) within the pattern), letters are matched independently of case. Note that there are two ASCII characters, K and S, that, in addition to their lower case ASCII equivalents, are case-equivalent with Unicode U+212A (Kelvin sign) and U+017F (long S) respectively when either unicode or ucp is set. 🔤

The power of regular expressions comes from the ability to include wild cards, character classes, alternatives, and repetitions in the pattern. These are encoded in the pattern by the use of metacharacters, which do not stand for themselves but instead are interpreted in some special way. ✨

There are two different sets of metacharacters: those that are recognized anywhere in the pattern except within square brackets, and those that are recognized within square brackets. Outside square brackets, the metacharacters are as follows:

\ — general escape character with several uses
^ — assert start of string (or line, in multiline mode)
$ — asserts end of string (or end of line, if we're in multiline mode) ✨
. — matches any character except newline (by default, bestie, keep reading)
[ — starts a character class definition
| — starts an alternative branch
( — starts a group or a control verb
) — ends a group or a control verb
* — the 0 or more quantifier
+ — the 1 or more quantifier; also the "possessive quantifier"
? — the 0 or 1 quantifier; also the quantifier minimizer
{ — potential start of a min/max quantifier

Brace characters { and } are also doing double duty, enclosing data for constructions like \g{2} or \k{name}. In almost every use of braces, space and/or horizontal tab characters that follow { or precede } are allowed and are ignored. And for quantifiers specifically, they may also show up before or after the comma. okay so braces are just casually flexible like that 💅

Part of a pattern that lives inside square brackets is called a "character class". And inside a character class, the only metacharacters are:

\ — the general escape character
^ — negate the class, but only if it's the first character
- — indicates a character range
[ — POSIX character class (if followed by POSIX syntax)
] — terminates the character class

If a pattern is compiled with the extended option, most white space in the pattern is ignored — other than in a character class, within a \Q...\E sequence, or between a # outside a character class and the next newline, inclusive. And an escaping backslash can be used to include a white space or a # character as part of the pattern. no cap.

NOTE: Only these two characters are ignored — not the full set of pattern white space characters that are ignored outside a character class. Option settings can be changed within a pattern; see Internal Option Setting.

The following sections describe the use of each of the metacharacters.

🦋 Backslash 🦋

The backslash character has several uses, and honestly it's giving multitalented. First up: if it's followed by a character that is not a digit or a letter, it takes away any special meaning that character may have. This use of backslash as an escape character applies both inside and outside character classes. 💅

For example, if you want to match a * character, you must write \* in the pattern. This escaping action applies whether or not the following character would otherwise be interpreted as a metacharacter, so it is always safe to precede a non-alphanumeric with backslash to specify that it stands for itself. In particular, if you want to match a backslash, you write \\.

Only ASCII digits and letters have any special meaning after a backslash. All other characters (in particular, those whose code points are greater than 127) are treated as literals. periodt.

If you want to treat all characters in a sequence as literals, you can do so by putting them between \Q and \E. Note that this includes white space even when the extended option is set so that most other white space is ignored. The behaviour is different from Perl in that $ and @ are handled as literals in \Q...\E sequences in PCRE2, whereas in Perl, $ and @ cause variable interpolation. Also, Perl does "double-quotish backslash interpolation" on any backslashes between \Q and \E which, its documentation says, "may lead to confusing results". PCRE2 treats a backslash between \Q and \E just like any other character. Note the following examples:

  Pattern            PCRE2 matches     Perl matches

  \Qabc$xyz\E        abc$xyz           abc followed by the contents of $xyz
  \Qabc\$xyz\E       abc\$xyz          abc$xyz
  \Qabc\E\$\Qxyz\E   abc$xyz           abc$xyz
  \QA\0\E            A\0               A\0
  \Q.\.\E            .\.               .\.

The \Q...\E sequence is recognized both inside and outside character classes. An isolated \E that is not preceded by \Q is ignored. If \Q is not followed by \E later in the pattern, the literal interpretation continues to the end of the pattern (that is, \E is assumed at the end). If the isolated \Q is inside a character class, this causes an error, because the character class is then not terminated by a closing square bracket. 💀 so watch out.

Another difference from Perl is that any appearance of \Q or \E inside what might otherwise be a quantifier causes PCRE2 not to recognize the sequence as a quantifier. Perl recognizes a quantifier if (redundantly) either of the numbers is inside \Q...\E, but not if the separating comma is. When not recognized as a quantifier a sequence such as {\Q1\E,2} is treated as the literal string "{1,2}". iconic tbh.

🌸 Non-Printing Characters 🌸

A second use of backslash provides a way of encoding non-printing characters in patterns in a visible manner. There is no restriction on the appearance of non-printing characters in a pattern, but when a pattern is being prepared by text editing, it is often easier to use one of the following escape sequences instead of the binary character it represents. In an ASCII or Unicode environment, these escapes are as follows:

\a — alarm, that is, the BEL character (hex 07)
\cx — "control-x", where x is a non-control ASCII character
\e — escape (hex 1B)
\f — form feed (hex 0C)
\n — linefeed (hex 0A)
\r — carriage return (hex 0D) (but see below 👀)
\t — tab (hex 09)
\0dd — character with octal code 0dd
\ddd — character with octal code ddd, or back reference
\o{ddd..} — character with octal code ddd..
\xhh — character with hex code hh
\x{hhh..} — character with hex code hhh..
\N{U+hhh..} — character with Unicode hex code point hhh..

A description of how back references work is given later, following the discussion of parenthesized groups. (stay tuned bestie 🫶)

By default, after \x that is not followed by {, one or two hexadecimal digits are read (letters can be in upper or lower case). If the character that follows \x is neither { nor a hexadecimal digit, an error occurs. This is different from Perl's default behaviour, which generates a NUL character, but is in line with the behaviour of Perl's 'strict' mode in re. so PCRE2 is the responsible friend here.

Any number of hexadecimal digits may appear between \x{ and }. If a character other than a hexadecimal digit appears between \x{ and }, or if there is no terminating }, an error occurs.

Characters whose code points are less than 256 can be defined by either of the two syntaxes for \x or by an octal sequence. There is no difference in the way they are handled. For example, \xdc is exactly the same as \x{dc} or \334. However, using the braced versions does make such sequences easier to read. 💅 readability slay.

The \N{U+hhh..} escape sequence is recognized only when PCRE2 is operating in UTF mode. Perl also uses \N{name} to specify characters by Unicode name; PCRE2 does not support this. Note that when \N is not followed by an opening brace (curly bracket) it has an entirely different meaning, matching any character that is not a newline. sneaky little double life fr fr.

There are some legacy applications where the escape sequence \r is expected to match a newline. 🙈

An error occurs if \c is not followed by a character whose ASCII code point is in the range 32 to 126. The precise effect of \cx is as follows: if x is a lower case letter, it is converted to upper case. Then bit 6 of the character (hex 40) is inverted. Thus \cA to \cZ become hex 01 to hex 1A (A is 41, Z is 5A), but \c{ becomes hex 3B ({ is 7B), and \c; becomes hex 7B (; is 3B). If the code unit following \c has a code point less than 32 or greater than 126, a compile-time error occurs. math is mathing ✨

🎀 Octal Escapes and Back References 🎀

The escape \o must be followed by a sequence of octal digits, enclosed in braces. An error occurs if this is not the case. This escape provides a way of specifying character code points as octal numbers greater than 0777, and it also allows octal numbers and backreferences to be unambiguously distinguished. obsessed.

If braces are not used, after \0 up to two further octal digits are read.

Inside a character class, when a backslash is followed by any octal digit, up to three octal digits are read to generate a code point. Any subsequent digits stand for themselves. The sequences \8 and \9 are treated as the literal characters "8" and "9".

Outside a character class, Perl's handling of a backslash followed by a digit other than 0 is complicated by ambiguity, and Perl has changed over time, causing PCRE2 also to change. 😭 the drama never ends.

For greater clarity and unambiguity, it is best to avoid following \e by a digit greater than zero. Instead, use \o{...} or \x{...} to specify numerical character code points, and \g{...} to specify backreferences.

⭐ Perl Rules for Non-Class Backslash 1-9 ⭐

All the digits that follow the backslash are read as a decimal number. If the number is less than 10, begins with the digit 8 or 9, or if there are at least that many previous capture groups in the expression, the entire sequence is taken as a back reference. Otherwise, up to three octal digits are read to form a character code. For example:

\40 is another way of writing an ASCII space
\040 is the same, provided there are fewer than 40 previous capture groups
\7 is always a backreference
\11 might be a backreference, or another way of writing a tab
\011 is always a tab
\0113 is a tab followed by the character "3"
\113 might be a backreference, otherwise the character with octal code 113
\377 might be a backreference, otherwise the value 255 (decimal)
\81 is always a backreference

Note that octal values of 100 or greater that are specified using this syntax must not be introduced by a leading zero, because no more than three octal digits are ever read. 🔥

💜 Python Rules for Non-Class Backslash 1-9 💜

If there are at least three octal digits after the backslash, exactly three are read as an octal code point number, but the value must be no greater than \377, even in modes where higher code point values are supported. Any subsequent digits stand for themselves. If there are fewer than three octal digits, the sequence is taken as a decimal back reference. Thus, for example, \12 is always a back reference, independent of how many captures there are in the pattern. An error is generated for a reference to a non-existent capturing group. rude but valid, no notes.

🌈 Constraints on Character Values 🌈

Characters that are specified using octal or hexadecimal numbers are limited to certain values, as follows:

8-bit non-UTF mode — no greater than 0xff
8-bit UTF-8 mode — no greater than 0x10ffff and a valid code point

Invalid Unicode code points are all those in the range 0xd800 to 0xdfff (the so-called "surrogate" code points).

🫶 Escape Sequences in Character Classes 🫶

All the sequences that define a single character value can be used both inside and outside character classes. In addition, inside a character class, \b is interpreted as the backspace character (hex 08). 💅

So when \N isn't followed by an opening brace, it's not allowed in a character class, periodt. 🙅‍♀️ And \B, \R, and \X? Bestie, they're not special inside a character class either. Just like every other unrecognized alphabetic escape sequence, they cause an error. 💀 But outside a character class, these sequences have totally different meanings, iykyk. ✨

🚫 Unsupported Escape Sequences 🚫

In Perl, the sequences \F, \l, \L, \u, and \U are recognized by its string handler and used to modify the case of following characters. By default, PCRE2 does not support these escape sequences in patterns. Sorry bestie, no glow-up for your letters here. 💅

🔢 Absolute and Relative Backreferences 🔢

The sequence \g followed by a signed or unsigned number, optionally enclosed in braces, is an absolute or relative backreference. A named backreference can be coded as \g{name}.

Backreferences get discussed later, right after the whole parenthesized-groups conversation. 🫶

📞 Absolute and Relative Subroutine Calls 📞

For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or a number enclosed either in angle brackets or single quotes, is an alternative syntax for referencing a capture group as a subroutine. Details are discussed later.

⚠️ Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not synonymous. The former is a backreference; the latter is a subroutine call. Not me mixing those up, they are literally different things. 😭

🎀 Generic Character Types 🎀

Another use of backslash is for specifying generic character types:

\d — any decimal digit
\D — any character that is not a decimal digit
\h — any horizontal white space character
\H — any character that is not a horizontal white space character
\N — any character that is not a newline
\s — any white space character
\S — any character that is not a white space character
\v — any vertical white space character
\V — any character that is not a vertical white space character
\w — any "word" character
\W — any "non-word" character

The \N escape sequence has the same meaning as the . metacharacter when dotall is not set, but setting dotall does not change the meaning of \N. Note that when \N is followed by an opening brace it has a different meaning — go check Non-Printing Characters. 💖 Perl also uses \N{name} to specify characters by Unicode name; PCRE2 does not support this.

Each pair of lower and upper case escape sequences partitions the complete set of characters into two disjoint sets. Any given character matches one, and only one, of each pair. The sequences can appear both inside and outside character classes. They each match one character of the appropriate type. If the current matching point is at the end of the subject string, all of them fail, because there is no character to match. 🫠

The default \s characters are HT (9), LF (10), VT (11), FF (12), CR (13), and space (32), which are defined as white space in the "C" locale. This list may vary if locale-specific matching is taking place. For example, in some locales the "non-breaking space" character (\xA0) is recognized as white space, and in others the VT character is not. ✨

A "word" character is an underscore or any character that is a letter or digit. By default, the definition of letters and digits is controlled by PCRE2's low-valued character tables, and may vary if locale-specific matching is taking place. For example, in a French locale such as "fr_FR" in Unix-like systems, or "french" in Windows, some character codes greater than 127 are used for accented letters, and these are then matched by \w.

🌈 The use of locales with Unicode is discouraged. Just don't, bestie.

By default, characters whose code points are greater than 127 never match \d, \s, or \w, and always match \D, \S, and \W, although this may be different for characters in the range 128-255 when locale-specific matching is happening. These escape sequences retain their original meanings from before Unicode support was available, mainly for efficiency reasons. 🙌

If the ucp option is set, the behaviour is changed so that Unicode properties are used to determine character types, as follows:

\d — Any character that matches \p{Nd} (decimal digit)
\s — Any character that matches \p{Z} or \h or \v
\w — Any character that matches \p{L}, \p{N}, \p{Mn}, or \p{Pc}

The addition of \p{Mn} (non-spacing mark) and the replacement of an explicit test for underscore with a test for \p{Pc} (connector punctuation) happened in PCRE2 release 10.43. This brings PCRE2 into line with Perl. 👑

The upper case escapes match the inverse sets of characters. Note that \d matches only decimal digits, whereas \w matches any Unicode digit, as well as other character categories. Note also that ucp affects \b, and \B because they are defined in terms of \w and \W. Matching these sequences is noticeably slower when ucp is set. 🐌

The effect of ucp on any one of these escape sequences can be negated by the options within a pattern by means of an Internal Option Setting.

The sequences \h, \H, \v, and \V, in contrast to the other sequences, which match only ASCII characters by default, always match a specific list of code points, whether or not ucp is set.

The horizontal space characters are:

U+0009 — Horizontal tab (HT)
U+0020 — Space
U+00A0 — Non-break space
U+1680 — Ogham space mark
U+180E — Mongolian vowel separator
U+2000 — En quad
U+2001 — Em quad
U+2002 — En space
U+2003 — Em space
U+2004 — Three-per-em space
U+2005 — Four-per-em space
U+2006 — Six-per-em space
U+2007 — Figure space
U+2008 — Punctuation space
U+2009 — Thin space
U+200A — Hair space
U+202F — Narrow no-break space
U+205F — Medium mathematical space
U+3000 — Ideographic space

The vertical space characters are:

U+000A — Linefeed (LF)
U+000B — Vertical tab (VT)
U+000C — Form feed (FF)
U+000D — Carriage return (CR)
U+0085 — Next line (NEL)
U+2028 — Line separator
U+2029 — Paragraph separator

In 8-bit, non-UTF-8 mode, only the characters with code points less than 256 are relevant. 🧚

🌙 Newline Sequences 🌙

Outside a character class, by default, the escape sequence \R matches any Unicode newline sequence.

In 8-bit non-UTF-8 mode \R is equivalent to the following:

(?>\r\n|\n|\x0b|\f|\r|\x85)

This is an example of an "atomic group", details of which are given below. This particular group matches either the two-character sequence CR followed by LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab, U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next line, U+0085). Because this is an atomic group, the two-character sequence is treated as a single unit that cannot be split. 🔥

In other modes, two additional characters whose code points are greater than 255 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029). Unicode support is not needed for these characters to be recognized.

It is possible to restrict \R to match only CR, LF, or CRLF (instead of the complete set of Unicode line endings) by setting the option bsr_anycrlf at compile time. (BSR is an abbreviation for "backslash R".) This can be made the default when PCRE2 is built; if this is the case, the other behaviour can be requested via the bsr_anycrlf option. 😌

It is also possible to specify these settings by starting a pattern string with one of the following sequences:

(*BSR_ANYCRLF) — CR, LF, or CRLF only
(*BSR_UNICODE) — any Unicode newline sequence

These override the default and the options given to the compiling function.

⚠️ Note that these special settings, which are not Perl-compatible, are recognized only at the very start of a pattern, and that they must be in upper case. If more than one of them is present, the last one is used.

They can be combined with a change of newline convention; for example, a pattern can start with:

(*ANY)(*BSR_ANYCRLF)

They can also be combined with the (*UTF) or (*UCP) special sequences.

💀 Inside a character class, \R is treated as an unrecognized escape sequence, and causes an error.

🌸 Unicode Character Properties 🌸

When PCRE2 is built with Unicode support (the default — bestie, it's the norm), three additional escape sequences that match characters with specific properties are available. They can be used in any mode, though in 8-bit non-UTF modes these sequences are of course limited to testing characters whose code points are less than U+0100.

Matching characters by Unicode property is not fast, because PCRE2 has to do a multistage table lookup in order to find a character's property. That is why the traditional escape sequences such as \d and \w do not use Unicode properties in PCRE2 by default, though you can make them do so by setting the ucp option or by starting the pattern with (*UCP). 🫡

The extra escape sequences that provide property support are:

\p{_xx_} — a character with the _xx_ property
\P{_xx_} — a character without the _xx_ property
\X — a Unicode extended grapheme cluster

For compatibility with Perl, negation can be specified by including a circumflex between the opening brace and the property. For example, \p{^Lu} is the same as \P{Lu}. ✨

In accordance with Unicode's "loose matching" rules, ASCII white space characters, hyphens, and underscores are ignored in the properties represented by _xx_ above. As well as the space character, ASCII white space can be tab, linefeed, vertical tab, formfeed, or carriage return. 🌈

Some properties are specified as a name only; others as a name and a value, separated by a colon or an equals sign. The names and values consist of ASCII letters and digits (with one Perl-specific exception, see below). They are not case sensitive.

🙈 Note, however, that the escapes themselves, \p and \P, _are_ case sensitive.

There are abbreviations for many names. The following examples are all equivalent:

\p{bidiclass=al}
\p{BC=al}
\p{ Bidi_Class : AL }
\p{ Bi-di class = Al }
\P{ ^ Bi-di class = Al }

There is support for Unicode script names, Unicode general category properties, "Any", which matches any character (including newline), Bidi_Class, a number of binary (yes/no) properties, and some special PCRE2 properties (described below). Certain other Perl properties such as "InMusicalSymbols" are not supported by PCRE2.

💀 Note that \P{Any} does not match any characters, so always causes a match failure.

✨ Script Properties for `\p` and `\P` ✨

Okay so there are three different syntax forms for matching a script, and you're gonna want to know all of them. Every Unicode character has a basic script and, optionally, a list of other scripts ("Script Extensions") it commonly hangs out with. Take Adlam as the example: \p{sc:Adlam} matches characters whose basic script is Adlam, whereas \p{scx:Adlam} matches those plus the characters that have Adlam in their extensions list. The full property-type names script and script extensions are recognized too, and — same as for every property spec — you can use an equals sign instead of the colon. And if you give a script name without a property type, like \p{Adlam}, it gets treated as \p{scx:Adlam}. Perl switched to this interpretation at release 5.26, and PCRE2 switched at release 10.40. ✨

Unassigned characters get assigned the Unknown script. Everyone else who isn't part of an identified script gets lumped together as Common. 🫶

✨ The General Category Property for `\p` and `\P` ✨

Every single character has exactly one Unicode general category property, and it's specified by a two-letter abbreviation. If you only give one letter with \p or \P, it includes all the general category properties that start with that letter. And in that case — as long as there's no negation — the curly brackets in the escape sequence are optional, bestie. These two examples do the exact same thing:

\p{L}
\pL

The following general category property codes are supported:

C — Other
Cc — Control
Cf — Format
Cn — Unassigned
Co — Private use
Cs — Surrogate
L — Letter
Lc — Cased letter
Ll — Lower case letter
Lm — Modifier letter
Lo — Other letter
Lt — Title case letter
Lu — Upper case letter
M — Mark
Mc — Spacing mark
Me — Enclosing mark
Mn — Non-spacing mark
N — Number
Nd — Decimal number
Nl — Letter number
No — Other number
P — Punctuation
Pc — Connector punctuation
Pd — Dash punctuation
Pe — Close punctuation
Pf — Final punctuation
Pi — Initial punctuation
Po — Other punctuation
Ps — Open punctuation
S — Symbol
Sc — Currency symbol
Sk — Modifier symbol
Sm — Mathematical symbol
So — Other symbol
Z — Separator
Zl — Line separator
Zp — Paragraph separator
Zs — Space separator

Storytime 💅: Perl originally used the name L& for the Lc property. Perl still supports it, but it's discouraged, and PCRE2 still supports it too. This property matches any character with the Lu, Ll, or Lt property — in other words, any letter that isn't classified as a modifier or as "other". From release 10.45 of PCRE2, Lu, Ll, and Lt are all treated as Lc when case-independent matching gets set by the caseless option or by (?i) inside the pattern. None of the other properties are affected by caseless matching.

The Cs (Surrogate) property only applies to characters whose code points live in the range U+D800 to U+DFFF. But they're not valid in Unicode strings, so PCRE2 can't test for them in UTF mode. 😭

The long synonyms for property names that Perl supports (like \p{Letter}) are not supported by PCRE2, and you're also not allowed to prefix any of these properties with Is.

And no character that's actually in the Unicode table has the Cn (unassigned) property. Instead, that property is just assumed for any code point that isn't in the Unicode table.

🌈 Binary (Yes/No) Properties for `\p` and `\P` 🌈

Unicode defines a bunch of binary properties — meaning properties whose only possible values are true or false. You can get a list of the ones that are recognized by \p and \P, along with their abbreviations, by running this command:

pcre2test -LP

🦋 The Bidi_Class Property for `\p` and `\P` 🦋

{Bidi_Class:<class>}

matches a character with the given class

{BC:<class>}

matches a character with the given class

The recognized classes are:

AL — Arabic letter
AN — Arabic number
B — paragraph separator
BN — boundary neutral
CS — common separator
EN — European number
ES — European separator
ET — European terminator
FSI — first strong isolate
L — left-to-right
LRE — left-to-right embedding
LRI — left-to-right isolate
LRO — left-to-right override
NSM — non-spacing mark
ON — other neutral
PDF — pop directional format
PDI — pop directional isolate
R — right-to-left
RLE — right-to-left embedding
RLI — right-to-left isolate
RLO — right-to-left override
S — segment separator
WS — white space

Just like with all property specifications, you can use an equals sign instead of a colon, and the class names are case-insensitive. Only the short names listed above are recognized, though — PCRE2 does not at present support any long alternatives. 🙈

🎀 Extended Grapheme Clusters 🎀

The \X escape matches any number of Unicode characters that form an "extended grapheme cluster", and it treats the sequence as an atomic group (more on that below). Unicode supports various kinds of composite character by giving each character a grapheme breaking property, and then having rules that use these properties to define the boundaries of extended grapheme clusters. Those rules are defined in Unicode Standard Annex 29, "Unicode Text Segmentation". Unicode 11.0.0 straight-up abandoned the use of some previous properties that had been used for emojis, and instead introduced various emoji-specific properties. PCRE2 uses only the Extended Pictographic property.

\X always matches at least one character. Then it decides whether to pull in additional characters according to these rules for ending a cluster:

End at the end of the subject string.
Do not end between CR and LF; otherwise end after any control character.
Do not break Hangul (a Korean script) syllable sequences. Hangul characters come in five types: L, V, T, LV, and LVT. An L character may be followed by an L, V, LV, or LVT character; an LV or V character may be followed by a V or T character; an LVT or T character may be followed only by a T character.
Do not end before extending characters or spacing marks or the zero-width joiner (ZWJ) character. Characters with the "mark" property always have the "extend" grapheme breaking property.
Do not end after prepend characters.
Do not end within emoji modifier sequences or emoji ZWJ (zero-width joiner) sequences. An emoji ZWJ sequence consists of a character with the Extended_Pictographic property, optionally followed by one or more characters with the Extend property, followed by the ZWJ character, followed by another Extended_Pictographic character.
Do not break within emoji flag sequences. That is, don't break between regional indicator (RI) characters if there are an odd number of RI characters before the break point.
Otherwise, end the cluster.

💜 PCRE2's Additional Properties 💜

On top of all the standard Unicode properties we just talked about, PCRE2 supports four more that let you convert traditional escape sequences like \w and \s into Unicode-property versions. PCRE2 uses these non-standard, non-Perl properties internally whenever ucp is set. But you can also just use them explicitly. Here they are:

Xan — Any alphanumeric character
Xps — Any POSIX space character
Xsp — Any Perl space character
Xwd — Any Perl "word" character

Xan matches characters that have either the L (letter) or the N (number) property. Xps matches the characters tab, linefeed, vertical tab, form feed, or carriage return, plus any other character that has the Z (separator) property (and yes, this includes the space character). Xsp is the same as Xps; back in PCRE1 it used to exclude vertical tab, for Perl compatibility, but Perl changed. Xwd matches the same characters as Xan, plus the ones that match Mn (non-spacing mark) or Pc (connector punctuation, which includes underscore).

There's another non-standard property, Xuc, which matches any character that can be represented by a Universal Character Name in C++ and other programming languages. Those are the characters $, @, ` (grave accent), and every character with a Unicode code point greater than or equal to U+00A0, except for the surrogates U+D800 to U+DFFF. Heads up: most base (ASCII) characters are excluded. (Universal Character Names are of the form \uHHHH or \UHHHHHHHH where H is a hexadecimal digit. Also note that the Xuc property doesn't match these sequences — it matches the characters that they represent.) 🔥

⭐ Resetting the Match Start ⭐

In normal use, the escape sequence \K causes any previously matched characters to not be included in the final matched sequence that gets returned. For example, the pattern:

foo\Kbar

matches "foobar", but reports that it matched "bar". And \K does not interact with anchoring in any way. So this pattern:

^foo\Kbar

matches only when the subject begins with "foobar" (in single line mode), though it again reports the matched string as "bar". This feature is similar to a lookbehind assertion (described below), but the part of the pattern before \K is not constrained to match a limited number of characters, which is a requirement for a lookbehind assertion. Using \K also doesn't interfere with the setting of captured substrings. For example, when the pattern

(foo)\Kbar

matches "foobar", the first substring is still set to "foo". 👑

From version 5.32.0, Perl forbids the use of \K in lookaround assertions. From release 10.38, PCRE2 also forbids this by default.

🌸 Simple Assertions 🌸

Last use of backslash, and it's for certain simple assertions. An assertion specifies a condition that has to be met at a particular point in a match, without consuming any characters from the subject string. Using groups for the more complicated assertions is described below. The backslashed assertions are:

\b — matches at a word boundary
\B — matches when not at a word boundary
\A — matches at the start of the subject
\Z — matches at the end of the subject; also matches before a newline at the end of the subject
\z — matches only at the end of the subject
\G — matches at the first matching position in the subject

Inside a character class, \b is doing a whole different thing, bestie — it matches the backspace character. And if any other of these assertions shows up inside a character class, an "invalid escape sequence" error is generated. Periodt.

A word boundary is a position in the subject string where the current character and the previous character do not both match \w or \W — i.e. one matches \w and the other matches \W — or the start or end of the string if the first or last character matches \w, respectively. When PCRE2 is built with Unicode support, the meanings of \w and \W can be changed by setting the ucp option. When this is done, it also affects \b and \B. Neither PCRE2 nor Perl has a separate "start of word" or "end of word" metasequence. However, whatever follows \b normally determines which it is. For example, the fragment \ba matches "a" at the start of a word.

The \A, \Z, and \z assertions differ from the traditional circumflex and dollar (described in the next section) in that they only ever match at the very start and end of the subject string, whatever options are set. Thus, they are independent of multiline mode. These three assertions are not affected by the notbol or noteol options, which affect only the behaviour of the circumflex and dollar metacharacters. However, if the startoffset argument of run/3 is non-zero, indicating that matching is to start at a point other than the beginning of the subject, \A can never match. The difference between \Z and \z is that \Z matches before a newline at the end of the string as well as at the very end, whereas \z matches only at the end.

The \G assertion is true only when the current matching position is at the start point of the matching process, as specified by the startoffset argument of run/3. It differs from \A when the value of startoffset is non-zero. By calling run/3 multiple times with appropriate arguments, you can mimic Perl's /g option, and it is in this kind of implementation where \G can be useful.

Note, however, that PCRE2's implementation of \G, being true at the starting character of the matching process, is subtly different from Perl's, which defines it as true at the end of the previous match. In Perl, these can be different when the previously matched string was empty. Because PCRE2 does just one match at a time, it cannot reproduce this behaviour.

If all the alternatives of a pattern begin with \G, the expression is anchored to the starting match position, and the "anchored" flag is set in the compiled regular expression.

💫 Circumflex and Dollar 💫

The circumflex and dollar metacharacters are zero-width assertions. That is, they test for a particular condition being true without consuming any characters from the subject string. These two metacharacters are concerned with matching the starts and ends of lines. If the newline convention is set so that only the two-character sequence CRLF is recognized as a newline, isolated CR and LF characters are treated as ordinary data characters, and are not recognized as newlines.

Outside a character class, in the default matching mode, the circumflex character is an assertion that is true only if the current matching point is at the start of the subject string. If the startoffset argument of run/3 is non-zero, or if notbol is set, circumflex can never match if the multiline option is unset. Inside a character class, circumflex has an entirely different meaning (see below).

Circumflex need not be the first character of the pattern if a number of alternatives are involved, but it should be the first thing in each alternative in which it appears if the pattern is ever to match that branch. If all possible alternatives start with a circumflex, that is, if the pattern is constrained to match only at the start of the subject, it is said to be an "anchored" pattern. (There are also other constructs that can cause a pattern to be anchored.)

The dollar character is an assertion that is true only if the current matching point is at the end of the subject string, or immediately before a newline at the end of the string (by default), unless noteol is set. Note, however, that it does not actually match the newline. Dollar need not be the last character of the pattern if a number of alternatives are involved, but it should be the last item in any branch in which it appears. Dollar has no special meaning in a character class.

The meaning of dollar can be changed so that it matches only at the very end of the string, by setting the dollar_endonly option at compile time. This does not affect the \Z assertion.

The meanings of the circumflex and dollar metacharacters are changed if the multiline option is set. When this is the case, a dollar character matches before any newlines in the string, as well as at the very end, and a circumflex matches immediately after internal newlines as well as at the start of the subject string. It does not match after a newline that ends the string, for compatibility with Perl.

For example, the pattern /^abc$/ matches the subject string "def\nabc" (where \n represents a newline) in multiline mode, but not otherwise. Consequently, patterns that are anchored in single line mode because all branches start with ^ are not anchored in multiline mode, and a match for circumflex is possible when the startoffset argument of run/3 is non-zero. The dollar_endonly option is ignored if multiline is set.

When the Newline Conventions recognizes the two-character sequence CRLF as a newline, this is preferred, even if the single characters CR and LF are also recognized as newlines. For example, if the newline convention is "any", a multiline mode circumflex matches before "xyz" in the string "abc\r\nxyz" rather than after CR, even though CR on its own is a valid newline. (It also matches at the very start of the string, of course.)

Note that the sequences \A, \Z, and \z can be used to match the start and end of the subject in both modes, and if all branches of a pattern start with \A it is always anchored, whether or not multiline is set.

🌼 Full Stop (Period, Dot) and . 🌼

Outside a character class, a dot in the pattern matches any one character in the subject string except (by default) a character that signifies the end of a line. One or more characters may be specified as line terminators (see Newline Conventions).

Dot never matches a single line-ending character. When the two-character sequence CRLF is the only line ending, dot does not match CR if it is immediately followed by LF, but otherwise it matches all characters (including isolated CRs and LFs). When ANYCRLF is selected for line endings, no occurrences of CR or LF match dot. When all Unicode line endings are being recognized, dot does not match CR or LF or any of the other line ending characters.

The behaviour of dot with regard to newlines can be changed. If the dotall option is set, a dot matches any one character, without exception. If the two-character sequence CRLF is present in the subject string, it takes two dots to match it.

The handling of dot is entirely independent of the handling of circumflex and dollar, the only relationship being that they both involve newlines. Dot has no special meaning in a character class.

The escape sequence \N when not followed by an opening brace behaves like a dot, except that it is not affected by the dotall option. In other words, it matches any character except one that signifies the end of a line.

When \N is followed by an opening brace it has a different meaning. See Non-Printing Characters. Perl also uses \N{name} to specify characters by Unicode name; PCRE2 does not support this.

🌈 Matching a Single Code Unit 🌈

Outside a character class, the escape sequence \C matches any one code unit, whether or not a UTF mode is set. In the 8-bit library, one code unit is one byte. Unlike a dot, \C always matches line-ending characters. The feature is provided in Perl in order to match individual bytes in UTF-8 mode, but it is unclear how it can usefully be used.

Because \C breaks up characters into individual code units, matching one unit with \C in UTF-8 mode means that the rest of the string may start with a malformed UTF character. This has undefined results, because PCRE2 assumes that it is matching character by character in a valid UTF string (by default it checks the subject string's validity at the start of processing).

PCRE2 does not allow \C to appear in lookbehind assertions (described below) in UTF-8 mode, because this would make it impossible to calculate the length of the lookbehind.

In general, the \C escape sequence is best avoided. However, one way of using it that avoids the problem of malformed UTF-8 characters is to use a lookahead to check the length of the next character, as in this pattern, which could be used with a UTF-8 string (ignore white space and line breaks):

       (?| (?=[\x00-\x7f])(\C) |
             (?=[\x80-\x{7ff}])(\C)(\C) |
             (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) |
             (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C))

In this example, a group that starts with (?| resets the capturing parentheses numbers in each alternative (see Duplicate Group Numbers). The assertions at the start of each branch check the next UTF-8 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The character's individual bytes are then captured by the appropriate number of \C groups.

💅 Square Brackets and Character Classes 💅

An opening square bracket introduces a character class, terminated by a closing square bracket. A closing square bracket on its own is not special by default. If a closing square bracket is required as a member of the class, it should be the first data character in the class (after an initial circumflex, if present) or escaped with a backslash. This means that, by default, an empty class cannot be defined.

A character class matches a single character in the subject. A matched character must be in the set of characters defined by the class, unless the first character in the class definition is a circumflex, in which case the subject character must not be in the set defined by the class. If a circumflex is actually required as a member of the class, ensure it is not the first character, or escape it with a backslash.

For example, the character class [aeiou] matches any lower case English vowel, whereas [^aeiou] matches all other characters. Note that a circumflex is just a convenient notation for specifying the characters that are in the class by enumerating those that are not. A class that starts with a circumflex is not an assertion; it still consumes a character from the subject string, and therefore it fails to match if the current pointer is at the end of the string.

okay so bestie, characters in a class can be specified by their code points using \o, \x, or \N{U+hh..} in the usual way. ✨ When caseless matching is set, any letters in a class represent both their upper case and lower case versions, so for example, a caseless [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a caseful version would. 💅 Note: there are two ASCII characters, K and S, that, in addition to their lower case ASCII equivalents, are case-equivalent with Unicode U+212A (Kelvin sign) and U+017F (long S) respectively when either unicode or ucp is set. 👑

Characters that might indicate line breaks are never treated in any special way when matching character classes, whatever line-ending sequence is in use, and whatever setting of the dotall and multiline options is used. A class such as [^a] always matches one of these characters. 💀

The generic character type escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v, \V, \w, and \W may appear in a character class, and add the characters that they match to the class. For example, [\dABCDEF] matches any hexadecimal digit. In UTF modes, the ucp option affects the meanings of \d, \s, \w and their upper case partners, just as it does when they appear outside a character class, as described in Generic character types. The escape sequence \b has a different meaning inside a character class; it matches the backspace character. The sequences \B, \R, and \X are not special inside a character class. Like any other unrecognized escape sequences, they cause an error. The same is true for \N when not followed by an opening brace. 🫶

The minus (hyphen) character can be used to specify a range of characters in a character class. For example, [d-m] matches any letter between d and m, inclusive. If a minus character is required in a class, it must be escaped with a backslash or appear in a position where it cannot be interpreted as indicating a range, typically as the first or last character in the class, or immediately after a range. For example, [b-d-z] matches letters in the range b to d, a hyphen character, or z. ✨

Perl treats a hyphen as a literal if it appears before or after a POSIX class (see below) or before or after a character type escape such as \d or \H. However, unless the hyphen is the last character in the class, Perl outputs a warning in its warning mode, as this is most likely a user error. As PCRE2 has no facility for warning, an error is given in these cases. 😭

It is not possible to have the literal character "]" as the end character of a range. A pattern such as [W-]46] is interpreted as a class of two characters ("W" and "-") followed by a literal string "46]", so it would match "W46]" or "-46]". However, if the "]" is escaped with a backslash it is interpreted as the end of a range, so [W-\]46] is interpreted as a class containing a range and two other characters. The octal or hexadecimal representation of "]" can also be used to end a range. 💅

Ranges normally include all code points between the start and end characters, inclusive. They can also be used for code points specified numerically, for example [\000-\037]. Ranges can include any characters that are valid for the current mode. In any UTF mode, the so-called "surrogate" characters (those whose code points lie between 0xd800 and 0xdfff inclusive) may not be specified explicitly by default. However, ranges such as [\x{d7ff}-\x{e000}], which include the surrogates, are always permitted. 🌈

If a range that includes letters is used when caseless matching is set, it matches the letters in either case. For example, [W-c] is equivalent to

tables for a French locale are in use, [\xc8-\xcb] matches accented E characters in both cases.

A circumflex can conveniently be used with the upper case character types to specify a more restricted set of characters than the matching lower case type. For example, the class [^\W_] matches any letter or digit, but not underscore, whereas [\w] includes underscore. A positive character class should be read as "something OR something OR ..." and a negative class as "NOT something AND NOT something AND NOT ...". 💖

The metacharacters that are recognized in character classes are backslash, hyphen (when it can be interpreted as specifying a range), circumflex (only at the start), and the terminating closing square bracket. An opening square bracket is also special when it can be interpreted as introducing a POSIX class (see "Posix character classes" below), or a special compatibility feature (see Compatibility feature for word boundaries" below). Escaping any non-alphanumeric character in a class turns it into a literal, whether or not it would otherwise be a metacharacter. ✨

[](){: #sect7X}

✨ Perl Extended Character Classes ✨

From release 10.45 PCRE2 supports Perl's (?[...]) extended character class syntax. This can be used to perform set operations such as intersection on character classes. 💅

The syntax permitted within (?[...]) is quite different to ordinary character classes. Inside the extended class, there is an expression syntax consisting of "atoms", operators, and ordinary parentheses "()" used for grouping. Such classes always have the Perl /xx modifier (PCRE2 option PCRE2_EXTENDED_MORE) turned on within them. This means that literal space and tab characters are ignored everywhere in the class. 🎀

The allowed atoms are individual characters specified by escape sequences such as \n or \x{123}, character types such as \d, POSIX classes such as [:alpha:], and nested ordinary (non-extended) character classes. For example, in (?[\d & [...]]) the nested class [...] follows the usual rules for ordinary character classes, in which parentheses are not metacharacters, and character literals and ranges are permitted. ✨

Character literals and ranges may not appear outside a nested ordinary character class because they are not atoms in the extended syntax. The extended syntax does not introduce any additional escape sequences, so (?[\y]) is an unknown escape, as it would be in [\y]. 💀

In the extended syntax, ^ does not negate a class (except within an ordinary class nested inside an extended class); it is instead a binary operator. 👑

The binary operators are "&" (intersection), "|" or "+" (union), "-" (subtraction) and "^" (symmetric difference). These are left-associative and "&" has higher (tighter) precedence, while the others have equal lower precedence. The one prefix unary operator is "!" (complement), with highest precedence. 🔥

[](){: #sect8 }

🌸 Posix Character Classes 🌸

Perl supports the POSIX notation for character classes. This uses names enclosed by [: and :] within the enclosing square brackets. PCRE2 also supports this notation. in both ordinary and extended classes. For example,

[01[:alpha:]%]

matches "0", "1", any alphabetic character, or "%". The supported class names are:

alnum — letters and digits
alpha — letters
ascii — character codes 0-127
blank — space or tab only
cntrl — control characters
digit — decimal digits (same as \d)
graph — printing characters, excluding space
lower — lower case letters
print — printing characters, including space
punct — printing characters, excluding letters and digits and space
space — white space (the same as \s from PCRE2 8.34)
upper — upper case letters
word — "word" characters (same as \w)
xdigit — hexadecimal digits

The default "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and space (32). If locale-specific matching is taking place, the list of space characters may be different; there may be fewer or more of them. "Space" and \s match the same set of characters, as do "word" and \w. 💖

The name "word" is a Perl extension, and "blank" is a GNU extension from Perl 5.8. Another Perl extension is negation, which is indicated by a ^ character after the colon. For example,

[12[:^digit:]]

matches "1", "2", or any non-digit. PCRE2 (and Perl) also recognize the POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not supported, and an error is given if they are encountered. 🚫

By default, characters with values greater than 127 do not match any of the POSIX character classes, although this may be different for characters in the range 128-255 when locale-specific matching is happening. However, in UCP mode, unless certain options are set (see below), some of the classes are changed so that Unicode character properties are used. This is achieved by replacing POSIX classes with other sequences, as follows:

[:alnum:] — Becomes \p{Xan}
[:alpha:] — Becomes \p{L}
[:blank:] — Becomes \h
[:cntrl:] — Becomes {Cc}
[:digit:] — Becomes \p{Nd}
[:lower:] — Becomes \p{Ll}
[:space:] — Becomes \p{Xps}
[:upper:] — Becomes \p{Lu}
[:word:] — Becomes \p{Xwd}

Negated versions, such as [:^alpha:], use \P instead of \p. Four other POSIX classes are handled specially in UCP mode:

[:graph:] — This matches characters that have glyphs that mark the page when printed. In Unicode property terms, it matches all characters with the L, M, N, P, S, or Cf properties, except for:
- U+061C — Arabic Letter Mark
- U+180E — Mongolian Vowel Separator
- U+2066 - U+2069 — Various "isolate"s
[:print:] — This matches the same characters as [:graph:] plus space characters that are not controls, that is, characters with the Zs property.
[:punct:] — This matches all characters that have the Unicode P (punctuation) property, plus those characters with code points less than 256 that have the S (Symbol) property.
[:xdigit:] — In addition to the ASCII hexadecimal digits, this also matches the "fullwidth" versions of those characters, whose Unicode code points start at U+FF10. This is a change that was made in PCRE2 release 10.43 for Perl compatibility.

The other POSIX classes are unchanged by ucp, and match only characters with code points less than 256. 💅

okay bestie, so there are two options that can be used to restrict the POSIX classes to ASCII characters when ucp is set. The option PCRE2_EXTRA_ASCII_DIGIT affects just [:digit:] and [:xdigit:]. Within a pattern, this can be set and unset by (?aT) and (?-aT). The PCRE2_EXTRA_ASCII_POSIX option disables UCP processing for all POSIX classes, including [:digit:] and [:xdigit:]. Within a pattern, (?aP) and (?-aP) set and unset both these options for consistency. 💖

✨ Compatibility Feature for Word Boundaries ✨

In the POSIX.2 compliant library that was included in 4.4BSD Unix, the ugly syntax [[:<:]] and [[:>:]] is used for matching "start of word" and "end of word". PCRE2 treats these items as follows:

[[:<:]]  is converted to  \b(?=\w)
[[:>:]]  is converted to  \b(?<=\w)

Only these exact character sequences are recognized. A sequence such as [a[:<:]b] provokes error for an unrecognized POSIX class name. This support is not compatible with Perl. It is provided to help migrations from other environments, and is best not used in any new patterns. Note that \b matches at the start and the end of a word (see "Simple assertions" above), and in a Perl-style pattern the preceding or following character normally shows which is wanted, without the need for the assertions that are used above in order to give exactly the POSIX behaviour. Note also that the ucp option changes the meaning of \w (and therefore \b) by default, so it also affects these POSIX sequences. 🫶

🌈 Vertical Bar 🌈

Vertical bar characters are used to separate alternative patterns. For example, the pattern

gilbert|sullivan

matches either "gilbert" or "sullivan". Any number of alternatives may appear, and an empty alternative is permitted (matching the empty string). The matching process tries each alternative in turn, from left to right, and the first one that succeeds is used. If the alternatives are within a group (defined below), "succeeds" means matching the rest of the main pattern as well as the alternative in the group. 💅

💅 Internal Option Setting 💅

The settings of several options can be changed within a pattern by a sequence of letters enclosed between (? and ). The option letters are:

i for caseless
m for multiline
n for PCRE2_NO_AUTO_CAPTURE
s for dotall
x for extended
xx for PCRE2_EXTENDED_MORE

For example, (?im) sets caseless, multiline matching. It is also possible to unset these options by preceding the relevant letters with a hyphen, for example (?-im). The two "extended" options are not independent; unsetting either one cancels the effects of both of them.

A combined setting and unsetting such as (?im-sx), which sets caseless and multiline while unsetting dotall and extended, is also permitted. Only one hyphen may appear in the options string. If a letter appears both before and after the hyphen, the option is unset. An empty options setting (?) is allowed. Needless to say, it has no effect. 🙈

If the first character following (? is a circumflex, it causes all of the above options to be unset. Letters may follow the circumflex to cause some options to be re-instated, but a hyphen may not appear.

Some PCRE2-specific options can be changed by the same mechanism using these pairs or individual letters:

aD for PCRE2_EXTRA_ASCII_BSD
aS for PCRE2_EXTRA_ASCII_BSS
aW for PCRE2_EXTRA_ASCII_BSW
aP for PCRE2_EXTRA_ASCII_POSIX and PCRE2_EXTRA_ASCII_DIGIT
aT for PCRE2_EXTRA_ASCII_DIGIT
r for PCRE2_EXTRA_CASELESS_RESTRICT
J for dupnames
U for ungreedy

However, except for r, these are not unset by (?^), which is equivalent to (?-imnrsx). If a is not followed by any of the upper case letters shown above, it sets (or unsets) all the ASCII options. 👑

PCRE2_EXTRA_ASCII_DIGIT has no additional effect when PCRE2_EXTRA_ASCII_POSIX is set, but including it in (?aP) means that (?-aP) suppresses all ASCII restrictions for POSIX classes.

When one of these option changes occurs at top level (that is, not inside group parentheses), the change applies until a subsequent change, or the end of the pattern. An option change within a groups affects only that part of the group that follows it. At the end of the group these options are reset to the state they were before the group. For example,

(a(?i)b)c

matches abc and aBc and no other strings (assuming caseless is not set externally). Any changes made in one alternative do carry on into subsequent branches within the same group. For example,

(a(?i)b|c)

matches "ab", "aB", "c", and "C", even though when matching "C" the first branch is abandoned before the option setting. This is because the effects of option settings happen at compile time. There would be some very weird behaviour otherwise. 😭

As a convenient shorthand, if any option settings are required at the start of a non-capturing group (see the next section), the option letters may appear between the ? and the :. Thus the two patterns

(?i:saturday|sunday)
(?:(?i)saturday|sunday)

match exactly the same set of strings.

NOTE: There are other PCRE2-specific options, applying to the whole pattern, which can be set by the application when the compiling function is called. In addition, the pattern can contain special leading sequences such as (*CRLF) to override what the application has set or what has been defaulted. Details are given in Newline Sequences. There are also the (*UTF) and (*UCP) leading sequences that can be used to set UTF and Unicode property modes; they are equivalent to setting the unicode and ucp options, respectively. However, the application can set the never_utf or never_ucp options, which lock out the use of the (*UTF) and (*UCP) sequences. 🔥

🎀 Groups 🎀

Groups are delimited by parentheses (round brackets), which can be nested. Turning part of a pattern into a group does two things:

It localizes a set of alternatives. For example, the pattern

cat(aract|erpillar|)

matches "cataract", "caterpillar", or "cat". Without the parentheses, it would match "cataract", "erpillar" or an empty string.

It creates a "capture group". This means that, when the whole pattern matches, the portion of the subject string that matched the group is passed back to the caller, separately from the portion that matched the whole pattern.

Opening parentheses are counted from left to right (starting from 1) to obtain numbers for capture groups. For example, if the string "the red king" is matched against the pattern

the ((red|white) (king|queen))

the captured substrings are "red king", "red", and "king", and are numbered 1, 2, and 3, respectively. 💖

The fact that plain parentheses fulfil two functions is not always helpful. There are often times when grouping is required without capturing. If an opening parenthesis is followed by a question mark and a colon, the group does not do any capturing, and is not counted when computing the number of any subsequent capture groups. For example, if the string "the white queen" is matched against the pattern the ((?:red|white) (king|queen)) the captured substrings are "white queen" and "queen", and are numbered 1 and 2. The maximum number of capture groups is 65535. 🌸

As a convenient shorthand, if any option settings are required at the start of a non-capturing group, the option letters may appear between the ? and the :. Thus the two patterns

(?i:saturday|sunday)
(?:(?i)saturday|sunday)

match exactly the same set of strings. Because alternative branches are tried from left to right, and options are not reset until the end of the group is reached, an option setting in one branch does affect subsequent branches, so the above patterns match "SUNDAY" as well as "Saturday". 🦋

🦋 Duplicate Group Numbers 🦋

Perl 5.10 introduced a feature whereby each alternative in a group uses the same numbers for its capturing parentheses. Such a group starts with (?| and is itself a non-capturing group. For example, consider this pattern:

(?|(Sat)ur|(Sun))day

Because the two alternatives are inside a (?| group, both sets of capturing parentheses are numbered one. Thus, when the pattern matches, you can look at captured substring number one, whichever alternative matched. This construct is useful when you want to capture part, but not all, of one of a number of alternatives. Inside a (?| group, parentheses are numbered as usual, but the number is reset at the start of each branch. The numbers of any capturing parentheses that follow the whole group start after the highest number used in any branch. The following example is taken from the Perl documentation. The numbers underneath show in which buffer the captured content will be stored.

# before  ---------------branch-reset----------- after
/ ( a )  (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
# 1            2         2  3        2     3     4

A backreference to a capture group uses the most recent value that is set for the group. The following pattern matches "abcabc" or "defdef":

/(?|(abc)|(def))\1/

In contrast, a subroutine call to a capture group always refers to the first one in the pattern with the given number. The following pattern matches "abcabc" or "defabc":

/(?|(abc)|(def))(?1)/

A relative reference such as (?-1) is no different: it is just a convenient way of computing an absolute group number. 💜

If a condition test for a group's having matched refers to a non-unique number, the test is true if any group with that number has matched.

An alternative approach to using this "branch reset" feature is to use duplicate named groups, as described in the next section. ⭐

👑 Named Capture Groups 👑

Identifying capture groups by number is simple, but it can be very hard to keep track of the numbers in complicated patterns. Furthermore, if an expression is modified, the numbers may change. To help with this difficulty, PCRE2 supports the naming of capture groups. This feature was not added to Perl until release 5.10. Python had the feature earlier, and PCRE1 introduced it at release 4.0, using the Python syntax. PCRE2 supports both the Perl and the Python syntax. 🎀

In PCRE2, a capture group can be named in one of three ways: (?<name>...) or (?'name'...) as in Perl, or (?P<name>...) as in Python. Names may be up to 128 code units long. When unicode is not set, they may contain only ASCII alphanumeric characters and underscores, but must start with a non-digit. When unicode is set, the syntax of group names is extended to allow any Unicode letter or Unicode decimal digit. In other words, group names must match one of these patterns:

^_A-Za-z*.RS 2 when unicode is not set

when unicode is set

okayyy so that was the last of the name rules 💅 and now onto the good part — references to capture groups from other places in the pattern, like backreferences, recursion, and conditions? You can make all of those by name as well as by number. Number or name, both slay. 🎀

And here's the tea ☕: named capture groups get allocated numbers too, exactly as if the names weren't even there. In both PCRE2 and Perl, capture groups are primarily identified by their numbers — any names are just lil aliases for those numbers, no cap. The PCRE2 API provides function calls for extracting the complete name-to-number translation table out of a compiled pattern, plus convenience functions for extracting captured substrings by name. Obsessed. ✨

⚠️ WARNING ⚠️ When more than one capture group has the same number — like we covered in the previous section — a name given to one of them applies to all of them. And here's where it gets spicy 🌶️: Perl allows identically numbered groups to have different names. Consider this pattern, where there are two capture groups, both numbered 1: (?|(?<AA>aa)|(?<BB>bb)) Perl allows this, with both names AA and BB as aliases of group 1. So after a successful match, both names yield the same value — either "aa" or "bb". In an attempt to reduce confusion, PCRE2 does not allow the same group number to be associated with more than one name. That example above? Compile-time error, bestie. It's giving 💀. But ngl, there's still scope for confusion. Consider this pattern: (?|(?<AA>aa)|(bb)) Although the second group number 1 is not explicitly named, the name AA is still an alias for any group 1. So whether the pattern matches "aa" or "bb", a reference by name to group AA yields the matched string. 🙈

By default, a name must be unique within a pattern — except that duplicate names are permitted for groups with the same number, for example:

(?|(?<AA>aa)|(?<AA>bb))

Want to turn that constraint off? Set the dupnames option at compile time, or use (?J) within the pattern — see Internal Option Setting for the whole situation. 🌸

Duplicate names can genuinely be useful for patterns where only one instance of the named capture group can match. Suppose you wanna match the name of a weekday, either as a 3-letter abbreviation or as the full name, and in both cases you want to extract the abbreviation. This pattern (ignoring the line breaks, they're just for vibes) does the job:

(?J)
(?<DN>Mon|Fri|Sun)(?:day)?|
(?<DN>Tue)(?:sday)?|
(?<DN>Wed)(?:nesday)?|
(?<DN>Thu)(?:rsday)?|
(?<DN>Sat)(?:urday)?

There are five capture groups, but only one is ever set after a match. The convenience functions for extracting the data by name return the substring for the first — and in this example, the only — group of that name that matched. Which saves you the searching to find which numbered group it was. Queen behaviour 👑 (and psst, an alternative way of solving this problem is to use a "branch reset" group, as described in the previous section.)

If you make a backreference to a non-unique named group from elsewhere in the pattern, the groups to which the name refers get checked in the order in which they appear in the overall pattern. The first one that is set is used for the reference. For example, this pattern matches both "foofoo" and "barbar" but not "foobar" or "barfoo":

(?J)(?:(?<n>foo)|(?<n>bar))\k<n>

If you make a subroutine call to a non-unique named group, the one that corresponds to the first occurrence of the name is used. In the absence of duplicate numbers, that's the one with the lowest number. 💜

If you use a named reference in a condition test (peep the section about conditions below), either to check whether a capture group has matched, or to check for recursion, then all groups with the same name are tested. If the condition is true for any one of them, the overall condition is true — same behaviour as testing by number. 🫶

✨ Repetition ✨

Repetition is specified by quantifiers, and a quantifier may follow any one of these items:

a literal data character
the dot metacharacter
the \C escape sequence
the \R escape sequence
the \X escape sequence
any escape sequence that matches a single character
a character class
a backreference
a parenthesized group (including lookaround assertions)
a subroutine call (recursive or otherwise)

If a quantifier doesn't follow a repeatable item, an error occurs. Periodt. 💅

The general repetition quantifier specifies a minimum and a maximum number of permitted matches by giving two numbers in curly brackets (braces), separated by a comma. The numbers must be less than 65536, and the first must be less than or equal to the second. For example,

z{2,4}

matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special character. If the second number is omitted but the comma is present, there's no upper limit; and if the second number and the comma are both omitted, the quantifier specifies an exact number of required matches. So

[aeiou]{3,}

matches at least 3 successive vowels, but may match many more, whereas

\d{8}

matches exactly 8 digits. If the first number is omitted, the lower limit is taken as zero — and in this case the upper limit must be present.

X{,4} is interpreted as X{0,4}

This is a change in behaviour that happened in Perl 5.34.0 and PCRE2 10.43. In earlier versions, such a sequence was not interpreted as a quantifier. And other regular expression engines may behave either way, so vibe check your engine before you fight about it. 🙈

If the characters that follow an opening brace don't match the syntax of a quantifier, the brace is taken as a literal character. In particular, that means {,} is a literal string of three characters.

Note that not every opening brace is potentially the start of a quantifier, because braces are used in other items too, such as \N{U+345} or \k{name}. 🌈

In UTF modes, quantifiers apply to characters rather than to individual code units. So, for example, \x{100}{2} matches two characters, each of which is represented by a two-byte sequence in a UTF-8 string. Similarly, \X{3} matches three Unicode extended grapheme clusters, each of which may be several code units long (and they may be of different lengths). No cap, lengths can differ. ✨

The quantifier {0} is permitted, causing the expression to behave as if the previous item and the quantifier were not present. This may be useful for capture groups that are referenced as subroutines from elsewhere in the pattern (see Defining capture groups for use by reference only). Except for parenthesized groups, items that have a {0} quantifier are omitted from the compiled pattern.

For convenience, the three most common quantifiers have single-character abbreviations:

* — is equivalent to {0,}
+ — is equivalent to {1,}
? — is equivalent to {0,1}

It is possible to construct infinite loops by following a group that can match no characters with a quantifier that has no upper limit, for example:

(a?)*

Earlier versions of Perl and PCRE1 used to give an error at compile time for such patterns. But because there are cases where this can be useful, such patterns are now accepted — however, whenever an iteration of such a group matches no characters, matching moves on to the next item in the pattern instead of repeatedly matching an empty string. And this does not prevent backtracking into any of the iterations if a subsequent item fails to match. 🫶

By default, quantifiers are "greedy", that is, they match as much as possible (up to the maximum number of permitted repetitions), without causing the rest of the pattern to fail. The classic example of where this gives problems is in trying to match comments in C programs. These appear between /* and */ and within the comment, individual * and / characters may appear. An attempt to match C comments by applying the pattern

/\*.*\*/

to the string

/* first comment */  not comment  /* second comment */

fails, because it matches the entire string owing to the greediness of the .* item. It's giving 💀, honestly. However, if a quantifier is followed by a question mark, it ceases to be greedy, and instead matches the minimum number of times possible, so the pattern

/\*.*?\*/

does the right thing with C comments. The meaning of the various quantifiers is not otherwise changed, just the preferred number of matches. And do not confuse this use of question mark with its use as a quantifier in its own right — because it has two uses, it can sometimes appear doubled, as in

\d??\d

which matches one digit by preference, but can match two if that is the only way the rest of the pattern matches.

If the ungreedy option is set (an option that is not available in Perl), the quantifiers are not greedy by default — but individual ones can be made greedy again by following them with a question mark. In other words, it inverts the default behaviour. 🌸

When a parenthesized group is quantified with a minimum repeat count that is greater than 1, or with a limited maximum, more memory is required for the compiled pattern, in proportion to the size of the minimum or maximum.

If a pattern starts with .* or .{0,} and the dotall option (equivalent to Perl's /s) is set — thus allowing the dot to match newlines — the pattern is implicitly anchored, because whatever follows will be tried against every character position in the subject string, so there's no point in retrying the overall match at any position after the first. PCRE2 normally treats such a pattern as though it were preceded by \A. iconic ✨

In cases where it is known that the subject string contains no newlines, it is worth setting dotall in order to obtain this optimization — or alternatively, using ^ to indicate anchoring explicitly. 👑

However, there are some cases where the optimization cannot be used. When .* is inside capturing parentheses that are the subject of a backreference elsewhere in the pattern, a match at the start may fail where a later one succeeds. Consider, for example:

(.*)abc\1

If the subject is "xyz123abc123" the match point is the fourth character. For this reason, such a pattern is not implicitly anchored.

Another case where implicit anchoring is not applied is when the leading .* is inside an atomic group. Once again, a match at the start may fail where a later one succeeds. Consider this pattern:

(?>.*?a)b

It matches "ab" in the subject "aab". The use of the backtracking control verbs (*PRUNE) and (*SKIP) also disables this optimization. 🙈

When a capture group is repeated, the value captured is the substring that matched the final iteration. For example, after

(tweedle[dume]{3}\s*)+

…has matched "tweedledum tweedledee", the value of the captured substring is "tweedledee". Cute! Obsessed! 💅

But bestie, heads up: if there are nested capture groups, the corresponding captured values may have been set in previous iterations and just… lingered there. For example, after

(a|(b))+

matches "aba", the value of the second captured substring is "b". Yeah, it's stale. It's giving leftovers in the fridge energy 🙈. Not a bug, just how the girl works.

🔥 Atomic Grouping and Possessive Quantifiers 🔥

Okay so with both maximizing ("greedy") and minimizing ("ungreedy" or "lazy") repetition, when whatever comes after fails, the normal move is to re-evaluate the repeated item to see if a different number of repeats lets the rest of the pattern match. Makes sense, right? But sometimes you wanna prevent that — either because you're trying to change the nature of the match, or because you want it to straight-up fail earlier than it otherwise might, specifically when whoever wrote the pattern already knows there's no point in carrying on. Periodt. 🎀

Take the pattern \d+foo applied to the subject line

123456bar

After matching all 6 digits and then failing to match "foo", the normal action of the matcher is to try again with only 5 digits matching the \d+ item, then with 4, and so on, before ultimately failing. Exhausting. 😭 So: atomic grouping (a term taken from Jeffrey Friedl's book) is the thing that lets you say "once this group has matched, we are NOT re-evaluating it this way."

If we use atomic grouping for that previous example, the matcher gives up immediately on failing to match "foo" the first time. The notation is a kind of special parenthesis, starting with (?> as in this example:

(?>\d+)foo

Perl 5.28 introduced an experimental alphabetic form starting with (* which may be easier to remember:

(*atomic:\d+)foo

This kind of parenthesized group "locks up" the part of the pattern it contains once it has matched, and a failure further into the pattern is prevented from backtracking into it. Backtracking past it to previous items, however, works as normal. Bestie stays locked in, but you can still walk back out of the room. 🚪✨

An alternative description: a group of this type matches exactly the string of characters that an identical standalone pattern would match, if anchored at the current point in the subject string. Slay.

Atomic groups are not capture groups, btw. Simple cases such as the above example can be thought of as a maximizing repeat that must swallow everything it can. So while both \d+ and \d+? are prepared to adjust the number of digits they match in order to make the rest of the pattern match, (?>\d+) can only match an entire sequence of digits. It's giving "my way or the highway" 🫶.

Atomic groups in general can of course contain arbitrarily complicated expressions, and can be nested. However, when the contents of an atomic group is just a single repeated item, as in the example above, a simpler notation called a "possessive quantifier" can be used. This consists of an additional + character following a quantifier. Using this notation, the previous example can be rewritten as

\d++foo

Note that a possessive quantifier can be used with an entire group, for example:

(abc|xyz){2,3}+

Possessive quantifiers are always greedy — the setting of the ungreedy option is ignored. Rude of them, honestly. They are a convenient notation for the simpler forms of atomic group. However, there is no difference in the meaning of a possessive quantifier and the equivalent atomic group, though there may be a performance difference; possessive quantifiers should be slightly faster. 🏃‍♀️💨

The possessive quantifier syntax is an extension to the Perl 5.8 syntax. Jeffrey Friedl originated the idea (and the name) in the first edition of his book. Mike McCloskey liked it, so implemented it when he built Sun's Java package, and PCRE1 copied it from there. It found its way into Perl at release 5.10. History lesson, but make it fashion 👑

PCRE2 has an optimization that automatically "possessifies" certain simple pattern constructs. For example, the sequence A+B is treated as A++B because there is no point in backtracking into a sequence of A's when B must follow. Like, why would you 😭. This feature can be disabled by the PCRE2_NO_AUTO_POSSESS option, by starting the pattern with (*NO_AUTO_POSSESS).

When a pattern contains an unlimited repeat inside a group that can itself be repeated an unlimited number of times, the use of an atomic group is the only way to avoid some failing matches taking a very long time indeed. Not exaggerating. The pattern

(\D+|<\d+>)*[!?]

matches an unlimited number of substrings that either consist of non-digits, or digits enclosed in <>, followed by either ! or ?. When it matches, it runs quickly. However, if it is applied to

aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

it takes a long time before reporting failure. This is because the string can be divided between the internal \D+ repeat and the external * repeat in a large number of ways, and all have to be tried. 💀 (The example uses [!?] rather than a single character at the end, because both PCRE2 and Perl have an optimization that allows for fast failure when a single character is used. They remember the last single character that is required for a match, and fail early if it is not present in the string.) If the pattern is changed so that it uses an atomic group, like this:

((?>\D+)|<\d+>)*[!?]

sequences of non-digits cannot be broken, and failure happens quickly. Fixed bestie! ✨

💖 Backreferences 💖

Outside a character class, a backslash followed by a digit greater than 0 (and possibly further digits) is a backreference to a capture group earlier — that is, to its left — in the pattern, provided there have been that many previous capture groups.

However! If the decimal number following the backslash is less than 8, it is always taken as a backreference, and causes an error only if there are not that many capture groups in the entire pattern. In other words, the group that is referenced need not be to the left of the reference for numbers less than 8. A "forward backreference" of this type can make sense when a repetition is involved and the group to the right has participated in an earlier iteration. Sneaky but legal 😏

It is not possible to have a numerical "forward backreference" to a group whose number is 8 or more using this syntax, because a sequence such as \50 is interpreted as a character defined in octal. See Non-Printing Characters for further details of the handling of digits following a backslash. Other forms of backreferencing do not suffer from this restriction. In particular, there is no problem when named capture groups are used. 🧚

Another way of avoiding the ambiguity inherent in the use of digits following a backslash is to use the \g escape sequence. This escape must be followed by a signed or unsigned number, optionally enclosed in braces. These examples are all identical:

(ring), \1
(ring), \g1
(ring), \g{1}

An unsigned number specifies an absolute reference without the ambiguity that is present in the older syntax. It is also useful when literal digits follow the reference. A signed number is a relative reference. Consider this example:

(abc(def)ghi)\g{-1}

The sequence \g{-1} is a reference to the capture group whose number is one less than the number of the next group to be started, so in this example (where the next group would be numbered 3) it is equivalent to \2, and \g{-2} would be equivalent to \1. Note that if this construct is inside a capture group, that group is included in the count, so in this example \g{-2} also refers to group 1:

(A)(\g{-2}B)

The use of relative references can be helpful in long patterns, and also in patterns that are created by joining together fragments that contain references within themselves. Building with blocks 🧱💜

The sequence \g{+1} is a reference to the next capture group that is started after this item, and \g{+2} refers to the one after that, and so on. This kind of forward reference can be useful in patterns that repeat. Perl does not support the use of + in this way. Noted, bestie ✍️

A backreference matches whatever actually most recently matched the capture group in the current subject string, rather than anything at all that matches the group (see Groups as Subroutines below for a way of doing that). So the pattern

(sens|respons)e and \1ibility

matches "sense and sensibility" and "response and responsibility", but not "sense and responsibility". If caseful matching is in force at the time of the backreference, the case of letters is relevant. For example,

((?i)rah)\s+\1

matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original capture group is matched caselessly. It's giving consistency era 💅

There are several different ways of writing backreferences to named capture groups. The .NET syntax is \k{name}, the Python syntax is (?=name), and the original Perl syntax is \k<name> or \k'name'. All of these are now supported by both Perl and PCRE2. Perl 5.10's unified backreference syntax, in which \g can be used for both numeric and named references, is also supported by PCRE2. We could rewrite the above example in any of the following ways:

(?<p1>(?i)rah)\s+\k<p1>
(?'p1'(?i)rah)\s+\k{p1}
(?P<p1>(?i)rah)\s+(?P=p1)
(?<p1>(?i)rah)\s+\g{p1}

A capture group that is referenced by name may appear in the pattern before or after the reference. No spoilers, no problem 🫶

There may be more than one backreference to the same group. If a group has not actually been used in a particular match, backreferences to it always fail by default. For example, the pattern

(a|(bc))\2

always fails if it starts to match "a" rather than "bc". Nothing there to grab onto, so it simply cannot. 🙈

Because there may be many capture groups in a pattern, all digits following a backslash are taken as part of a potential backreference number. If the pattern continues with a digit character, some delimiter must be used to terminate the backreference. If the extended or PCRE2_EXTENDED_MORE option is set, this can be white space. Otherwise, the \g{} syntax or an empty comment (see Comments below) can be used. Cute lil escape hatch 🌸

🌀 Recursive Backreferences 🌀

A backreference that occurs inside the group to which it refers fails when the group is first used, so, for example, (a\1) never matches. It just can't, bestie, the group hasn't matched anything yet 😭 However, such references can be useful inside repeated groups. For example, the pattern

(a|b\1)+

matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of the group, the backreference matches the character string corresponding to the previous iteration. Glow-up per iteration, honestly ✨ In order for this to work, the pattern must be such that the first iteration does not need to match the backreference. This can be done using alternation, as in the example above, or by a quantifier with a minimum of zero.

For versions of PCRE2 less than 10.25, backreferences of this type used to cause the group that they reference to be treated as an atomic group. This restriction no longer applies, and backtracking into such groups can occur as normal. Ancient history, we've healed 🕊️

An assertion is a lil test that doesn't eat any characters — it just looks, judges, and keeps it moving. But bestie, it has to pass or the whole match is done. 💅 The simple ones you already met up top: \b, \B, \A, \G, \Z, \z, ^ and $.

The fancier assertions are written as parenthesized groups. If that group matches, matching carries on right after it — but the matching position in the subject string gets reset to whatever it was before the assertion even got processed. No character theft, only vibes. ✨

One special flavor is the "scan substring" assertion, which matches a subpattern against a substring you captured earlier. That one lives in Scan Substring Assertions. Heads up though: it's a PCRE2 extension, so it does not play nice with Perl. 🙈

The other group-based assertions come in two eras of energy: some look ahead from where you are, some look behind you. And in both cases an assertion can be positive (it's true only if it matches) or negative (it's true only if it does not match). Iconic duality. 🎀

The Perl-compatible lookaround assertions are atomic. Meaning: even if the assertion is true and then something later fails to match, there is no backtracking into the assertion. Periodt. That said, there are some cases where non-atomic assertions hit different — PCRE2 has some support for those, covered in Non-atomic assertions, but they are not Perl-compatible.

A lookaround assertion can also show up as the condition in a conditional group — in that case, whether the assertion matches decides which branch of the condition you take. Main character energy, honestly. 👑

Assertion groups are not capture groups. If an assertion has capture groups inside it, those still get counted when numbering the capture groups of the whole pattern. And inside each branch of an assertion, locally captured substrings can be referenced the usual way. For example, a sequence like (.)\g{-1} can be used to check that two adjacent characters are the same. Cute, right?

When a branch inside an assertion fails to match, any substrings it captured get discarded (same as any pattern branch that fails — bye 💀). A negative assertion is only true when all its branches fail to match, which means: no captured substrings are ever retained after a successful negative assertion. And when an assertion contains a matching branch? What happens next depends on the type of assertion.

For a positive assertion, internally captured substrings in the successful branch are retained, and matching continues with the next pattern item after the assertion.
For a negative assertion, a matching branch means the assertion is not true. If that assertion is being used as a condition in a conditional group, captured substrings are retained, because matching continues with the "no" branch of the condition. For any other failing negative assertion, control passes to the previous backtracking point, so any captured strings inside the assertion get discarded.

Most assertion groups may be repeated. Ngl, asserting the same thing several times makes zero sense, but the side effect of capturing in positive assertions may occasionally be useful, so. 🙈 However — an assertion that forms the condition for a conditional group may not be quantified. Also, PCRE2 used to restrict the repetition of assertions, but from release 10.35 the only restriction left is that an unlimited maximum repetition gets changed to be one more than the minimum. Like, {3,} is treated as {3,4}. Little glow-up. ✨

Alphabetic Assertion Names 🌸

Traditionally, symbolic sequences like (?= and (?<= have been how we specify lookaround assertions. Perl 5.28 rolled out some experimental alphabetic alternatives that might be easier to remember. They all start with (* instead of (? and must be written in lower case letters. PCRE2 supports these synonyms:

(*positive_lookahead:  or (*pla: is the same as (?=
(*negative_lookahead:  or (*nla: is the same as (?!
(*positive_lookbehind: or (*plb: is the same as (?<=
(*negative_lookbehind: or (*nlb: is the same as (?<!

So for example, (*pla:foo) is the same assertion as (?=foo). In the sections below, all the assertions are described using the original symbolic forms. Okay? Okay. 💖

Lookahead Assertions 👀

Lookahead assertions start with (?= for positive assertions and (?! for negative ones. For example:

\w+(?=;)

matches a word followed by a semicolon, but it does not include the semicolon in the match. And this:

foo(?!bar)

matches any occurrence of "foo" that is not followed by "bar". Now watch out, because the apparently similar pattern:

(?!foo)bar

does not find an occurrence of "bar" that is preceded by something other than "foo" — it finds any occurrence of "bar" whatsoever, because the assertion (?!foo) is always true when the next three characters are "bar". If you want that other effect, you need a lookbehind assertion. Snuck up on you, didn't it. 🦋

If you want to force a matching failure at some point in a pattern, the most convenient way to do it is with (?!) — because an empty string always matches, so an assertion that requires there not to be an empty string must always fail. Flawless logic, no notes. The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).

Lookbehind Assertions 🔙

Lookbehind assertions start with (?<= for positive assertions and (?<! for negative ones. Example:

(?<!foo)bar

this one does find an occurrence of "bar" that is not preceded by "foo". But the contents of a lookbehind assertion are restricted: there must be a known maximum to the lengths of all the strings it matches. Two cases, bestie:

If every top-level alternative matches a fixed length, for example

(?<=colour|color)

there is a limit of 65535 characters to the lengths — and they don't have to be the same, as that example demonstrates. This is the only kind of lookbehind supported by PCRE2 versions earlier than 10.43.

In PCRE2 10.43 and later, run/3 supports lookbehind assertions in which one or more top-level alternatives can match more than one string length, for example

(?<=colou?r)

The maximum matching length for any branch of the lookbehind is limited to a value set by the calling program (default 255 characters). Unlimited repetition (for example \d*) is not supported. In some cases, the escape sequence \K (see above) can be used instead of a lookbehind assertion at the start of a pattern to get round the length limit restriction. Workaround queen. 💅

In UTF-8, PCRE2 does not allow the \C escape (which matches a single code unit even in a UTF mode) to appear in lookbehind assertions, because it makes it impossible to calculate the length of the lookbehind. And the \X and \R escapes, which can match different numbers of code units, are never permitted in lookbehinds. Not up for debate. 🚫

"Subroutine" calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long as the called capture group matches a limited-length string. However, recursion — that is, a "subroutine" call into a group that is already active — is not supported.

PCRE2 supports backreferences in lookbehinds, but only if certain conditions are met:

There must be no use of (?| in the pattern (it creates duplicate group numbers).
If the backreference is by name, the name must be unique.
And obviously, the referenced group must itself match a limited length substring.

This pattern matches words containing at least two characters that begin and end with the same character:

\b(\w)\w++(?<=\1)

Possessive quantifiers can be used together with lookbehind assertions to specify efficient matching at the end of subject strings. Consider a simple pattern like:

abcd$

applied to a long string that does not match. Because matching proceeds left to right, PCRE2 will look for each "a" in the subject and then check whether what follows matches the rest of the pattern. If the pattern is instead written as:

^.*abcd$

the initial .* matches the entire string at first, but when that fails (because there's no following "a"), it backtracks to match all but the last character, then all but the last two characters, and so on. And once again the search for "a" covers the entire string, from right to left, so we are no better off. 🙃 However, if the pattern is written as:

^.*+(?<=abcd)

there can be no backtracking for the .*+ item because of the possessive quantifier — it can match only the entire string. The subsequent lookbehind assertion does a single test on the last four characters. If it fails, the match fails immediately. For long strings, this approach makes a significant difference to the processing time. Efficiency slay fr fr. 🔥

Using Multiple Assertions 🫶

Several assertions (of any sort) may occur in succession. For example:

(?<=`\d{3}`)(?<!999)foo

matches "foo" preceded by three digits that are not "999". Notice that each assertion is applied independently at the same point in the subject string. First there's a check that the previous three characters are all digits, and then there's a check that the same three characters are not "999". Important: this pattern does not match "foo" preceded by six characters, the first of which are digits and the last three of which are not "999". For example, it doesn't match "123abcfoo". A pattern that does do that is:

(?<=\d{3}...)(?<!999)foo

This time the first assertion looks at the preceding six characters, checking that the first three are digits, and then the second assertion checks that the preceding three characters are not "999". Teamwork. ✨

Assertions can be nested in any combination. For example:

(?<=(?<!foo)bar)baz

matches an occurrence of "baz" that is preceded by "bar" which in turn is not preceded by "foo". Meanwhile:

(?<=\d{3}(?!999)...)foo

is another pattern that matches "foo" preceded by three digits and any three characters that are not "999".

Non-Atomic Assertions 💫

Traditional lookaround assertions are atomic. That is, if an assertion is true but there's a subsequent matching failure, there is no backtracking into the assertion. However, there are some cases where non-atomic positive assertions can be useful. PCRE2 provides these using the following syntax:

(*non_atomic_positive_lookahead:  or (*napla: or (?*
(*non_atomic_positive_lookbehind: or (*naplb: or (?<*

Consider the problem of finding the right-most word in a string that also appears earlier in the string — that is, it must appear at least twice in total. This pattern returns the required result as captured substring 1:

^(?x)(*napla: .* \b(\w++)) (?> .*? \b\1\b ){2}

Like, say our subject is "word1 word2 word3 word2 word3 word4" — the result is "word3". But how, bestie? 💭 At the very start, ^(?x) anchors the pattern and turns on the "x" option, which is what makes all the white space (thrown in there purely for readability) get ignored. Inside the assertion, the greedy .* at first eats the entire string, but then it has to backtrack until the rest of the assertion can match a word, which gets captured by group 1. Translation: when the assertion first succeeds, it's holding the right-most word in the string. 👑

Then the current matching point gets reset back to the start of the subject, and the rest of the pattern match checks for two occurrences of that captured word, using an ungreedy .*? to scan from the left. If that works, we're done, periodt. But if the last word in the string doesn't show up twice, this part of the pattern fails. And here's the tea: if a traditional atomic lookahead ((?= or (*pla:) had been used, the assertion could not be re-entered, and the whole match would fail. The pattern would only succeed if the very last word in the subject was found twice.

Using a non-atomic lookahead, though, means that when the last word doesn't occur twice in the string, the lookahead can backtrack and find the second-last word, and so on, until either the match succeeds or all the words have been tested. 🦋

Two conditions have to be met for a non-atomic assertion to actually be useful:

the contents of one or more capturing groups must change after a backtrack into the assertion, and
there must be a backreference to a changed group later in the pattern.

If that's not the case, the rest of the pattern match fails exactly as before, because nothing has changed — so using a non-atomic assertion is just wasting resources, lowkey pointless. 💀

There's one exception to backtracking into a non-atomic assertion. If an (*ACCEPT) control verb is triggered, the assertion succeeds atomically. That is, a subsequent match failure cannot backtrack into the assertion.

Note that assertions that appear as conditions for conditional groups must be atomic. No discussion. 💅

✨ Scan Substring Assertions ✨

A special kind of assertion, not compatible with Perl, makes it possible to check the contents of a captured substring by matching it with a subpattern. Because this involves capturing.

A scan substring assertion starts with the sequence (*scan_substring: or (*scs: which is followed by a list of substring numbers (absolute or relative) and/or substring names enclosed in single quotes or angle brackets, all within parentheses. The rest of the item is the subpattern that is applied to the substring, as shown in these examples:

(*scan_substring:(1)...)
(*scs:(-2)...)
(*scs:('AB')...)
(*scs:(1,'AB',-2)...)

The list of groups is checked in the order they are given, and it's the contents of the first one that is found to be set that get scanned. When dupnames is set and there are ambiguous group names, all groups with the same name are checked in numerical order. A scan substring assertion fails if none of the groups it references have been set. 🚫

The pattern match on the substring is always anchored — like, it must match from the start of the substring. There is no "bumpalong" if it doesn't match at the start. The end of the subject is temporarily reset to be the end of the substring, so \Z, \z, and $ will match there. However, the start of the subject is not reset. This means ^ matches only if the substring is actually at the start of the main subject, but it also means that lookbehind assertions into what precedes the substring are possible. 👀

Here's a super simple example: find a word that contains the rare (in English) sequence of letters "rh" not at the start:

\b(\w++)(*scs:(1).+rh)

The first group captures a word which is then scanned by the second group. This example doesn't actually need this heavyweight feature; the same match can be achieved with:

\b\w+?rh\w*\eb

When things get more complicated, though, scanning a captured substring can be a genuinely useful way to describe the required match. For example, there's a rather complicated pattern in the PCRE2 test data that checks an entire subject string for a palindrome — that is, the sequence of letters is the same in both directions. Suppose you want to search for individual words of two or more characters such as "level" that are palindromes:

(\b\w{2,}+\b)(*scs:(1)...palindrome-matching-pattern...)

Within a substring scanning subpattern, references to other groups work as normal. Capturing groups may appear, and will retain their values during ongoing matching if the assertion succeeds. 🎀

🌈 Script Runs 🌈

In concept, a script run is a sequence of characters that are all from the same Unicode script, such as Latin or Greek. But because some scripts are commonly used together, and because some diacritical and other marks are used with multiple scripts, it's not that simple. There's a full description of the rules that PCRE2 uses in the section entitled "Script Runs" in the pcre2unicode documentation. 📖

If part of a pattern is enclosed between (*script_run: or (*sr: and a closing parenthesis, it fails if the sequence of characters that it matches are not a script run. After a failure, normal backtracking occurs. Script runs can be used to detect spoofing attacks using characters that look the same but are from different scripts. The string "paypal.com" is an infamous example, where the letters could be a mixture of Latin and Cyrillic. 😭 This pattern ensures that the matched characters in a sequence of non-spaces that follow white space are a script run:

\s+(*sr:\S+)

To be sure that they're all from the Latin script (for example), a lookahead can be used:

\s+(?=\p{Latin})(*sr:\S+)

This works as long as the first character is expected to be a character in that script, and not (for example) punctuation, which is allowed with any script. If that's not the case, a more creative lookahead is needed. For example, if digits, underscore, and dots are permitted at the start:

\s+(?=[0-9_.]*\p{Latin})(*sr:\S+)

In many cases, backtracking into a script run pattern fragment is not desirable. The script run can employ an atomic group to prevent this. Because this is a common requirement, a shorthand notation is provided by (*atomic_script_run: or (*asr::

(*asr:...) is the same as (*sr:(?>...))

Note that the atomic group is inside the script run. Putting it outside would not prevent backtracking into the script run pattern. 🫶

Support for script runs is not available if PCRE2 is compiled without Unicode support. A compile-time error is given if any of the above constructs is encountered.

WARNING: The (*ACCEPT) control verb (see below) should not be used within a script run group, because it causes an immediate exit from the group, bypassing the script run checking. Do not do it. ⚠️

💜 Conditional Groups 💜

It's possible to cause the matching process to obey a pattern fragment conditionally, or to choose between two alternative fragments, depending on the result of an assertion, or whether a specific capture group has already been matched. The two possible forms of conditional group are:

(?(condition)yes-pattern)
(?(condition)yes-pattern|no-pattern)

If the condition is satisfied, the yes-pattern is used; otherwise the no-pattern (if present) is used. An absent no-pattern is equivalent to an empty string (it always matches). If there are more than two alternatives in the group, a compile-time error occurs. Each of the two alternatives may itself contain nested groups of any form, including conditional groups; the restriction to two alternatives applies only at the level of the condition itself. This pattern fragment is an example where the alternatives are complex:

(?(1) (A|B|C) |  (D |  (?(2)E|F) |  E) )

There are five kinds of condition: references to capture groups, references to recursion, two pseudo-conditions called DEFINE and VERSION, and assertions. ✨

Checking for a Used Capture Group by Number

If the text between the parentheses consists of a sequence of digits, the condition is true if a capture group of that number has previously matched. If there is more than one capture group with the same number (see Duplicate Group Numbers), the condition is true if any of them have matched. An alternative notation, which is a PCRE2 extension not supported by Perl, is to precede the digits with a plus or minus sign. In this case, the group number is relative rather than absolute. The most recently opened capture group (which could be enclosing this condition) can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside loops it can also make sense to refer to subsequent groups. The next capture group to be opened can be referenced as (?(+1), and so on. The value zero in any of these forms is not used; it provokes a compile-time error. 💀

Consider the following pattern, which contains non-significant white space to make it more readable (assume the extended option) and to divide it into three parts for ease of discussion:

( ( )?
  [^()]+ (?(1) ) )

The first part matches an optional opening parenthesis, and if that character is present, sets it as the first captured substring. The second part matches one or more characters that are not parentheses. The third part is a conditional group that tests whether or not the first capture group matched. If it did — that is, if the subject started with an opening parenthesis — the condition is true, and so the yes-pattern is executed and a closing parenthesis is required. Otherwise, since no-pattern is not present, the conditional group matches nothing. In other words, this pattern matches a sequence of non-parentheses, optionally enclosed in parentheses. 👑

If you were embedding this pattern in a larger one, you could use a relative reference:

...other stuff... ( ( )?
  [^()]+    (?(-1) ) )  ...

This makes the fragment independent of the parentheses in the larger pattern. Slay.

Checking for a Used Capture Group by Name

Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used capture group by name. For compatibility with earlier versions of PCRE1, which had this facility before Perl, the syntax (?(name)...) is also recognized. Note, however, that undelimited names consisting of the letter R followed by digits are ambiguous (see the following section). Rewriting the above example to use a named group gives this:

(?<OPEN> ( )?
  [^()]+ (?(<OPEN>) ) )

If the name used in a condition of this kind is a duplicate, the test is applied to all groups of the same name, and is true if any one of them has matched. 🦋

Checking for Pattern Recursion

Defining Capture Groups for Use by Reference Only 🎀

Okay so before that, real quick, because the word "recursion" is doing a LOT of heavy lifting here: in this sense it just means any subroutine-ish call from one part of the pattern to another — doesn't matter whether it's actually recursive or not. Want the full lore on that? Go read Recursive Patterns and Groups as Subroutines. 💅

Now. The conditions:

If your condition is the string (R), and there is no capture group named R, the condition is true when matching is currently inside a recursion or subroutine call to the whole pattern, or to any capture group. 🌀
If digits follow that letter R, and there's no group with that name, the condition is true if the most recent call went into a group with the given number — and that group has to exist somewhere in the overall pattern.

Here's a little contrived example, which is exactly equivalent to a+b:

((?(R1)a+|(?1)b))

But — and this is the plot twist — in both of those cases, if there does happen to be a capture group with a matching name, the condition tests whether it is set, the way the section above describes, instead of testing for recursion at all. So like, if you create a group named R1 by adding (?<R1>) to that pattern up there? Completely different meaning. Not the same pattern anymore, bestie. 😭

If a name with an ampersand in front follows the letter R, like this:

(?(R&name)...)

then the condition is true if the most recent recursion went into a group of that name — which, yes, must exist within the pattern.

Heads up: this condition does not check the entire recursion stack. It tests only the current level. And if the name you're using in a condition like this is a duplicate, the test gets applied to all groups bearing the same name, and it's true if any one of them is the most recent recursion. 🫶

Also: at "top level", all these recursion test conditions are false. Simple as that.

If the condition is the string (DEFINE), it is always false — even if there's a group actually named DEFINE. (Iconic of it, honestly.) In that case there can only be one alternative in the rest of the conditional group, and it's always skipped whenever control reaches this point in the pattern. The whole idea of DEFINE is that you use it to define subroutines that get referenced from elsewhere. (More on subroutines below, don't worry. 👑)

So for example, a pattern to match an IPv4 address like "192.168.23.245" could be written like this — ignore the white space and line breaks:

(?(DEFINE) (?<byte> 2[0-4]\d |  25[0-5] |  1\d\d |  [1-9]?\d) )
\b (?&byte) (\.(?&byte)){3} \b

The first chunk of the pattern is a DEFINE group, and inside it there's another group named "byte" being defined. That one matches an individual component of an IPv4 address (i.e. a number less than 256). When matching actually happens, this part gets skipped, because DEFINE behaves like a false condition. Then the rest of the pattern uses references to that named group to match the four dot-separated components of an IPv4 address, insisting on a word boundary at each end. Slay. ✨

Checking the PCRE2 Version 🔍

Programs that link with a PCRE2 library can check the version by calling pcre2_config() with appropriate arguments. But if you're just a user of an application and you don't have access to the underlying code? You can't do that. 🙈

Enter: a special "condition" called VERSION, which exists purely so those users can find out which version of PCRE2 they're dealing with — you use this condition to match a string such as "yesno". VERSION must be followed by either = or >=, and then a version number. For example:

(?(VERSION>=10.4)yes|no)

This pattern matches "yes" if the PCRE2 version is greater than or equal to 10.4, or "no" otherwise. And the fractional part of the version number can be omitted entirely. 💖

Assertion Conditions 🌸

If the condition isn't in any of the formats above, it has to be a parenthesized assertion. That can be a positive or negative lookahead, or a lookbehind. However — it has to be a traditional atomic assertion. Non-atomic assertions do not qualify here, periodt. 💅

Check out this pattern, which again has non-significant white space, and puts its two alternatives on the second line:

(?(?=[^a-z]*[a-z])
\d{2}-[a-z]{3}-\d{2}  |  \d{2}-\d{2}-\d{2} )

The condition is a positive lookahead assertion that matches an optional sequence of non-letters followed by a letter. In other words, it's testing whether the subject contains at least one letter. If a letter is found, the subject gets matched against the first alternative; otherwise it's matched against the second. So this pattern matches strings in one of two forms: dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits. 🌈

And here's the detail people always miss: when an assertion that's being used as a condition contains capture groups, any capturing that happens in a matching branch is retained afterwards — for both positive and negative assertions — because matching always continues after the assertion, whether it succeeded or failed. (Compare that with non-conditional assertions, where captures are retained only for positive assertions that succeed.) 🔥

Comments 💬

There are two ways of including comments in patterns processed by PCRE2. In both cases, the start of the comment must not be inside a character class, and it must not be in the middle of any other sequence of related characters — things like (?:, or a group name or number, or a Unicode property name. And the characters that make up a comment play no part in the pattern matching whatsoever.

The sequence (?# marks the start of a comment that runs up to the next closing parenthesis. Nested parentheses are not permitted. 🙅
If the extended or PCRE2_EXTENDED_MORE option is set, an unescaped # character also introduces a comment — and in this case it continues to immediately after the next newline character or character sequence in the pattern. Which characters count as newlines is controlled either by an option passed to the compiling function, or by a special sequence at the start of the pattern, as described in Newline Conventions. Note that the end of this type of comment is a literal newline sequence in the pattern; escape sequences that just happen to represent a newline do not count.

So like, consider this pattern when extended is set, and the default newline convention (a single linefeed character) is in force:

abc #comment \n still comment

On encountering the # character, compile/2 skips along, looking for a newline in the pattern. At this stage the sequence \n is still literal, so it does not terminate the comment. Only an actual character with the code value 0x0a (the default newline) terminates it. Sneaky! 👀

Recursive Patterns 🌀

Okay bestie, imagine the problem: matching a string in parentheses, allowing for unlimited nested parentheses. Without recursion, the best you can do is a pattern that matches up to some fixed depth of nesting. Handling arbitrary nesting depth is just not possible. That's the whole point of this section.

For some time, Perl has provided a facility that lets regular expressions recurse (amongst other things). It does this by interpolating Perl code in the expression at run time, and that code can refer to the expression itself. A Perl pattern using code interpolation to solve the parentheses problem can be created like this:

$re =  qr{\( (?: (?>[^()]+) |  (?p{$re}) )* \)}x;

The (?p{...}) item interpolates Perl code at run time, and in this case it refers recursively to the pattern in which it appears.

Obviously, PCRE2 cannot support the interpolation of Perl code. Instead, it supports special syntax for recursion of the entire pattern, and also for individual capture group recursion. After being introduced in PCRE1 and Python, this kind of recursion was subsequently introduced into Perl at release 5.10. 📜

A special item consisting of (?, followed by a number greater than zero, followed by a closing parenthesis, is a recursive subroutine call of the capture group of the given number — provided that it occurs inside that group. (If it's not inside that group, it's a non-recursive subroutine call instead, which gets described in the next section.) And the special item (?R) or (?0) is a recursive call of the entire regular expression. 💫

This PCRE2 pattern solves the nested parentheses problem — assume the extended option is set so that white space is ignored:

\( ( [^()]++ | (?R) )* \)

First it matches an opening parenthesis. Then it matches any number of substrings, which can either be a sequence of non-parentheses, or a recursive match of the pattern itself — that is, a correctly parenthesized substring. Finally, there's a closing parenthesis. Note the use of a possessive quantifier to avoid backtracking into sequences of non-parentheses. 🫶

If this were part of a larger pattern, you wouldn't want to recurse the entire pattern, so instead you could use this:

( \( ( [^()]++ | (?1) )* \) )

We've put the pattern into parentheses, and made the recursion refer to those instead of the whole pattern. Chef's kiss. 👑

In a larger pattern, keeping track of parenthesis numbers can be tricky. That's made easier by relative references — instead of (?1) in the pattern above, you can write (?-2) to refer to the second most recently opened parentheses preceding the recursion. Put another way: a negative number counts capturing parentheses leftwards, from the point at which it's encountered.

Be aware though: if duplicate capture group numbers are in use, relative references refer to the earliest group with the appropriate number. 🚨

Consider, for example:

(?|(a)|(b)) (c) (?-2)

The first two capture groups (a) and (b) are both numbered 1, and group (c) is number 2. When the reference (?-2) is encountered, the second most recently opened parentheses has the number 1 — but it's the first such group, the (a) group, that the recursion refers to. And it'd be exactly the same if an absolute reference (?1) were used. In other words: relative references are just a shorthand for computing a group number. Nothing magical, sorry. 💀

It's also possible to refer to subsequent capture groups, by writing references such as (?+2). However, those cannot be recursive, because the reference isn't inside the parentheses being referenced. They're always non-recursive subroutine calls — again, described in the next section.

An alternative approach is to use named parentheses. The Perl syntax for this is (?&name); PCRE1's earlier syntax (?P>name) is also supported. So we could rewrite the example above like this:

(?<pn> \( ( [^()]++ | (?&pn) )* \) )

If there's more than one group with the same name, the earliest one is used. ✨

Now — the example pattern we've been staring at contains nested unlimited repeats, and so the use of a possessive quantifier for matching strings of non-parentheses is genuinely important when you apply the pattern to strings that don't match. For example, when this pattern is applied to

(aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()

it yields "no match" quickly. But if a possessive quantifier is not used, the match runs for a very long time indeed — because there are so many different ways the + and * repeats can carve up the subject, and all of them have to be tested before failure can be reported. 😭 Not a vibe. Use the possessive quantifier, bestie. 💅

Okay so real talk about what happens to your capture groups at the end of a match: the values of capturing parentheses are those from the outermost level. That's the rule. If you want to obtain intermediate values, a callout function can be used (see below and the pcre2callout documentation) — that's your girl for the receipts. 💅

So if the pattern above is matched against

(ab(cd)ef)

the value for the inner capturing parentheses (numbered 2) is "ef", which is the last value taken on at the top level. And if a capture group is not matched at the top level, its final captured value is unset, even if it was (temporarily) set at a deeper level during the matching process. It doesn't get to keep the souvenir, bestie. 😭

Do not confuse the (?R) item with the condition (R), which tests for recursion. Different jobs, similar letters, do not mix them up. 🫠 Consider this pattern, which matches text in angle brackets, allowing for arbitrary nesting. Only digits are allowed in nested brackets (that is, when recursing), whereas any characters are permitted at the outer level.

< (?: (?(R) \d++  |  [^<>]*+) |  (?R)) *  >

In this pattern, (?(R) is the start of a conditional group, with two different alternatives for the recursive and non-recursive cases. The (?R) item is the actual recursive call. See? Iconic teamwork. ✨

🌈 Differences in Recursion Processing Between PCRE2 and Perl 🌈

Some former differences between PCRE2 and Perl no longer exist. Character development arc, honestly. 🦋

Before release 10.30, recursion processing in PCRE2 differed from Perl in that a recursive subroutine call was always treated as an atomic group. That is, once it had matched some of the subject string, it was never re-entered, even if it contained untried alternatives and there was a subsequent matching failure. (Historical note, and this is lowkey iconic: PCRE implemented recursion before Perl did. 👑)

Starting with release 10.30, recursive subroutine calls are no longer treated as atomic. That is, they can be re-entered to try unused alternatives if there is a matching failure later in the pattern. This is now compatible with the way Perl works. And if you want a subroutine call to be atomic, you must explicitly enclose it in an atomic group — it's not gonna do it for you anymore. 💅

Supporting backtracking into recursions simplifies certain types of recursive pattern. For example, this pattern matches palindromic strings:

^((.)(?1)\2|.?)$

The second branch in the group matches a single central character in the palindrome when there are an odd number of characters, or nothing when there are an even number of characters, but in order to work it has to be able to try the second case when the rest of the pattern match fails. Like, it needs that option or the whole thing falls apart. 🙈

If you want to match typical palindromic phrases, the pattern has to ignore all non-word characters, which can be done like this:

^\W*+((.)\W*+(?1)\W*+\2|\W*+.?)\W*+$

If run with the caseless option, this pattern matches phrases such as "A man, a plan, a canal: Panama!". Note the use of the possessive quantifier *+ to avoid backtracking into sequences of non-word characters — this is the whole trick, do not skip it. Without this, PCRE2 takes a great deal longer (ten times or more) to match typical phrases, and Perl takes so long that you think it has gone into a loop. 😭 fr fr.

Another way in which PCRE2 and Perl used to differ in their recursion processing is in the handling of captured values. Formerly in Perl, when a group was called recursively or as a subroutine (see Groups as Subroutines), it had no access to any values that were captured outside the recursion, whereas in PCRE2 these values can be referenced. Consider this pattern:

^(.)(\1|a(?2))

This pattern matches "bab". Here's the play-by-play: the first capturing parentheses match "b", then in the second group, when the backreference \1 fails to match "b", the second alternative matches "a" and then recurses. In the recursion, \1 does now match "b" and so the whole match succeeds. This match used to fail in Perl, but in later versions (I tried 5.024) it now works. Glow up. ✨

🎀 Groups as Subroutines 🎀

If the syntax for a recursive group call (either by number or by name) is used outside the parentheses to which it refers, it operates a bit like a subroutine in a programming language. More accurately, PCRE2 treats the referenced group as an independent subpattern which it tries to match at the current matching position. The called group may be defined before or after the reference. A numbered reference can be absolute or relative, as in these examples:

(...(absolute)...)...(?2)...
(...(relative)...)...(?-1)...
(...(?+1)...(relative)...

An earlier example pointed out that the pattern

(sens|respons)e and \1ibility

matches "sense and sensibility" and "response and responsibility", but not "sense and responsibility". If instead the pattern

(sens|respons)e and (?1)ibility

is used, it does match "sense and responsibility" as well as the other two strings. Another example is given in the discussion of DEFINE above.

Like recursions, subroutine calls used to be treated as atomic, but this changed at PCRE2 release 10.30, so backtracking into subroutine calls can now occur. However, any capturing parentheses that are set during the subroutine call revert to their previous values afterwards. So the subroutine doesn't leave its mess behind. 💅

Processing options such as case-independence are fixed when a group is defined, so if it is used as a subroutine, such options cannot be changed for different calls. For example, consider this pattern:

(abc)(?i:(?-1))

It matches "abcabc". It does not match "abcABC" because the change of processing option does not affect the called group. Periodt. 👑

The behaviour of backtracking control verbs in groups when called as subroutines is described in Backtracking verbs in subroutines.

🦋 Recursion and subroutines with returned capture groups 🦋

Since PCRE2 10.46, recursion and subroutine calls may also specify a list of capture groups to return. This is a PCRE2 syntax extension not supported by Perl. The pattern matching recurses into the referenced expression as described above, however, when the recursion returns to the calling expression the subgroups captured during the recursion can be retained when the calling expression's context is restored.

When used as a subroutine, this allows the subroutine's capture groups to be used as return values. Obsessed, actually. 🫶

Only the specific capture groups listed by the caller will be retained, using the following syntax:

(?R(grouplist))       recurse whole pattern, returning capture groups
(?n(grouplist))       )
(?+n(grouplist))      )
(?-n(grouplist))      ) call subroutine, returning capture groups
(?&name(grouplist))   )
(?P>name(grouplist))  )

The list of capture groups "grouplist" is a comma-separated list of (absolute or relative) group numbers, and group names enclosed in single quotes or angle brackets.

Here is an example which first uses the DEFINE condition to create a re-usable routine for matching a weekday, then calls that subroutine and retains the groups it captures for use later:

(?x: # ignore whitespace for clarity
  # Define the routine "weekendday" which matches Saturday or
  # Sunday, and returns the Sat/Sun prefix as \k<short>.
  (?(DEFINE) (?<weekendday>
      (?|(?<short>Sat)urday|(?<short>Sun)day) ) )
  # Call the routine. Matches "Saturday,Sat" or "Sunday,Sun".
  (?&weekendday(<short>)),\k<short> )

This feature is not available using the Oniguruma syntax \g<...> or \g'...' below. Sorry babes, that's just how it is. 🙈

🌸 Oniguruma Subroutine Syntax 🌸

For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or a number enclosed either in angle brackets or single quotes, is an alternative syntax for calling a group as a subroutine, possibly recursively. Here are two of the examples used above, rewritten using this syntax:

(?<pn> \( (  (?>[^()]+) |  \g<pn> )* \) )
(sens|respons)e and \g'1'ibility

PCRE2 supports an extension to Oniguruma: if a number is preceded by a plus or a minus sign it is taken as a relative reference. For example:

(abc)(?i:\g<-1>)

Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not synonymous. The former is a backreference; the latter is a subroutine call. That's the plot twist, don't miss it. 🎀

🔥 Backtracking Control 🔥

There are a number of special "Backtracking Control Verbs" (to use Perl's terminology) that modify the behaviour of backtracking during matching. They are generally of the form (*VERB) or (*VERB:NAME). Some verbs take either form, and may behave differently depending on whether or not a name argument is present. The names are not required to be unique within the pattern.

By default, for compatibility with Perl, a name is any sequence of characters that does not include a closing parenthesis. The name is not processed in any way, and it is not possible to include a closing parenthesis in the name.

A closing parenthesis can be included in a name either as \) or between \Q and \E. In addition to backslash processing, if the extended or PCRE2_EXTENDED_MORE option is also set, unescaped white space in verb names is skipped, and #-comments are recognized, exactly as in the rest of the pattern. extended and PCRE2_EXTENDED_MORE do not affect verb names.

The maximum length of a name is 255 in the 8-bit library. If the name is empty, that is, if the closing parenthesis immediately follows the colon, the effect is as if the colon were not there. Any number of these verbs may occur in a pattern. Except for (*ACCEPT), they may not be quantified.

Since these verbs are specifically related to backtracking, most of them can be used only when the pattern is to be matched using the traditional matching function, because it uses backtracking algorithms. With the exception of (*FAIL), which behaves like a failing negative assertion.

The behaviour of these verbs in repeated groups, assertions, and in capture groups called as subroutines (whether or not recursively) is documented below.

⭐ Optimizations that affect backtracking verbs ⭐

PCRE2 contains some optimizations that are used to speed up matching by running some checks at the start of each match attempt. For example, it may know the minimum length of matching subject, or that a particular character must be present. When one of these optimizations bypasses the running of a match, any included backtracking verbs will not, of course, be processed. You can suppress the start-of-match optimizations by setting the no_start_optimize option when calling compile/2 or by starting the pattern with (*NO_START_OPT).

Experiments with Perl suggest that it too has similar optimizations, and like PCRE2, turning them off can change the result of a match. Vibe check: sometimes the fast path is a different path. 🌀

👑 Verbs that act immediately 👑

The following verbs act as soon as they are encountered.

(*ACCEPT) or (*ACCEPT:NAME)

This verb makes the match end successfully, just straight-up skipping the rest of the pattern. Obsessed. But bestie listen — when it's inside a capture group that gets called as a subroutine, it only ends that group successfully, and matching then continues at the outer level. And if (*ACCEPT) is triggered inside a positive assertion, the assertion succeeds; inside a negative assertion, the assertion fails. 💅

If (*ACCEPT) is inside capturing parentheses, the data so far gets captured. Like, for example:

A((?:A|B(*ACCEPT)|C)D)

This matches "AB", "AAD", or "ACD"; and when it matches "AB", the "B" is captured by the outer parentheses. slay.

(*ACCEPT) is the only backtracking verb that's allowed to be quantified, because an ungreedy quantification with a minimum of zero acts only when a backtrack happens. Consider, for example:

(A(*ACCEPT)??B)C

where A, B, and C may be complex expressions. After matching "A", the matcher processes "BC"; if that fails, causing a backtrack, (*ACCEPT) is triggered and the match succeeds. In both cases, all but C is captured. And whereas (*COMMIT) (see below) means "fail on backtrack", a repeated (*ACCEPT) of this type means "succeed on backtrack". periodt.

WARNING: (*ACCEPT) should not be used within a script run group, because it causes an immediate exit from the group, bypassing the script run checking. no cap, don't do it 🙈

✨ `(FAIL)` or `(FAIL:NAME)` ✨

This verb causes a matching failure, forcing backtracking to occur. It may be abbreviated to (*F). It's equivalent to (?!) but easier to read. The Perl documentation notes that it's probably useful only when combined with (?{}) or (??{}). Those are, of course, Perl features that are not present in PCRE2. The nearest equivalent is the callout feature, as for example in this pattern:

a+(?C)(*FAIL)

A match with the string "aaaa" always fails, but the callout is taken before each backtrack happens (in this example, 10 times).

(*ACCEPT:NAME) and (*FAIL:NAME) behave the same as (*MARK:NAME)(*ACCEPT) and (*MARK:NAME)(*FAIL), respectively — that is, a (*MARK) is recorded just before the verb acts. 🎀

🌸 Recording which path was taken 🌸

There's one verb whose main purpose is to track how a match was arrived at, though it also has a secondary use in conjunction with advancing the match starting point (see (*SKIP) below).

✨ `(MARK:NAME)` or `(:NAME)` ✨

A name is always required with this verb. For all the other backtracking control verbs, a NAME argument is optional.

When a match succeeds, the name of the last-encountered mark name on the matching path is passed back to the caller as described in the section entitled "Other information about the match" in the pcre2api documentation. This applies to all instances of (*MARK) and other verbs, including those inside assertions and atomic groups. However, there are differences in those cases when (*MARK) is used in conjunction with (*SKIP) as described below.

The mark name that was last encountered on the matching path is passed back. A verb without a NAME argument is ignored for this purpose. Here's an example of pcre2test output, where the "mark" modifier requests the retrieval and outputting of (*MARK) data:

    re> /X(*MARK:A)Y|X(*MARK:B)Z/mark
data> XY
   0: XY
MK: A
XZ
   0: XZ
MK: B

The (*MARK) name is tagged with "MK:" in this output, and in this example it indicates which of the two alternatives matched. This is a more efficient way of obtaining this information than putting each alternative in its own capturing parentheses. 💖

If a verb with a name is encountered in a positive assertion that is true, the name is recorded and passed back if it's the last-encountered. This does not happen for negative assertions or failing positive assertions.

After a partial match or a failed match, the last encountered name in the entire match process is returned. For example:

    re> /X(*MARK:A)Y|X(*MARK:B)Z/mark
data> XP
No match, mark `=`  B

Note that in this unanchored example the mark is retained from the match attempt that started at the letter "X" in the subject. Subsequent match attempts starting at "P" and then with an empty string do not get as far as the (*MARK) item, but nevertheless do not reset it.

If you're interested in (*MARK) values after failed matches, you should probably either set the no_start_optimize option to ensure that the match is always attempted. 👑

👑 Verbs that act after backtracking 👑

The following verbs do nothing when they're encountered. Matching continues with what follows, but if there's a subsequent match failure, causing a backtrack to the verb, a failure is forced. That is, backtracking cannot pass to the left of the verb. However, when one of these verbs appears inside an atomic group or in an atomic lookaround assertion that is true, its effect is confined to that group, because once the group has been matched, there's never any backtracking into it. Backtracking from beyond an atomic assertion or group ignores the entire group, and seeks a preceding backtracking point.

These verbs differ in exactly what kind of failure occurs when backtracking reaches them. The behaviour described below is what happens when the verb is not in a subroutine or an assertion. Subsequent sections cover these special cases.

✨ `(COMMIT)` or `(COMMIT:NAME)` ✨

This verb causes the whole match to fail outright if there's a later matching failure that causes backtracking to reach it. Even if the pattern is unanchored, no further attempts to find a match by advancing the starting point take place. If (*COMMIT) is the only backtracking verb that's encountered, once it's been passed run/3 is committed to finding a match at the current starting point, or not at all. For example:

a+(*COMMIT)b

This matches "xxaab" but not "aacaab". It can be thought of as a kind of dynamic anchor, or "I've started, so I must finish." 💪

The behaviour of (*COMMIT:NAME) is not the same as (*MARK:NAME)(*COMMIT). It's like (*MARK:NAME) in that the name is remembered for passing back to the caller. However, (*SKIP:NAME) searches only for names that are set with (*MARK), ignoring those set by any of the other backtracking verbs.

If there's more than one backtracking verb in a pattern, a different one that follows (*COMMIT) may be triggered first, so merely passing (*COMMIT) during a match does not always guarantee that a match must be at this starting point.

Note that (*COMMIT) at the start of a pattern is not the same as an anchor, unless PCRE2's start-of-match optimizations are turned off, as shown in this output from pcre2test:

    re> /(*COMMIT)abc/
data> xyzabc
   0: abc
data>
re> /(*COMMIT)abc/no_start_optimize
data> xyzabc
No match

For the first pattern, PCRE2 knows that any match must start with "a", so the optimization skips along the subject to "a" before applying the pattern to the first set of data. The match attempt then succeeds. The second pattern disables the optimization that skips along to the first character. The pattern is now applied starting at "x", and so the (*COMMIT) causes the match to fail without trying any other starting points. 😭

✨ `(PRUNE)` or `(PRUNE:NAME)` ✨

This verb causes the match to fail at the current starting position in the subject if there's a later matching failure that causes backtracking to reach it. If the pattern is unanchored, the normal "bumpalong" advance to the next starting character then happens. Backtracking can occur as usual to the left of (*PRUNE), before it's reached, or when matching to the right of (*PRUNE), but if there's no match to the right, backtracking cannot cross (*PRUNE). In simple cases, the use of (*PRUNE) is just an alternative to an atomic group or possessive quantifier, but there are some uses of (*PRUNE) that cannot be expressed in any other way. In an anchored pattern (*PRUNE) has the same effect as (*COMMIT).

The behaviour of (*PRUNE:NAME) is not the same as (*MARK:NAME)(*PRUNE). It's like (*MARK:NAME) in that the name is remembered for passing back to the caller. However, (*SKIP:NAME) searches only for names set with (*MARK), ignoring those set by other backtracking verbs.

✨ `(*SKIP)` ✨

This verb, when given without a name, is like (*PRUNE), except that if the pattern is unanchored, the "bumpalong" advance is not to the next character, but to the position in the subject where (*SKIP) was encountered. (*SKIP) signifies that whatever text was matched leading up to it cannot be part of a successful match if there's a later mismatch. Consider:

a+(*SKIP)b

If the subject is "aaaac...", after the first match attempt fails (starting at the first character in the string), the starting point skips on to start the next attempt at "c". Note that a possessive quantifier does not have the same effect as this example; although it would suppress backtracking during the first match attempt, the second attempt would start at the second character instead of skipping on to "c". 🦋

If (*SKIP) is used to specify a new starting position that is the same as the starting position of the current match, or (by being inside a lookbehind) earlier, the position specified by (*SKIP) is ignored, and instead the normal "bumpalong" occurs.

✨ `(*SKIP:NAME)` ✨

When (*SKIP) has an associated name, its behaviour is modified. When such a (*SKIP) is triggered, the previous path through the pattern is searched for the most recent (*MARK) that has the same name. If one is found, the "bumpalong" advance is to the subject position that corresponds to that (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a matching name is found, the (*SKIP) is ignored.

The search for a (*MARK) name uses the normal backtracking mechanism, which means that it does not see (*MARK) settings that are inside atomic groups or assertions, because they're never re-entered by backtracking. Compare the following pcre2test examples:

    re> /a(?>(*MARK:X))(*SKIP:X)(*F)|(.)/
data: abc
   0: a
   1: a
data:
    re> /a(?:(*MARK:X))(*SKIP:X)(*F)|(.)/
data: abc
   0: b
   1: b

In the first example, the (*MARK) setting is in an atomic group, so it doesn't get seen when (*SKIP:X) triggers, which means (*SKIP) is just… ignored, bestie. 💀 That lets the second branch of the pattern be tried at the first character position. In the second example, the (*MARK) setting is not in an atomic group, so (*SKIP:X) can find the (*MARK) when it backtracks, and that makes a whole new matching attempt start at the second character. This time the (*MARK) never gets seen, because "a" doesn't match "b", so the matcher immediately jumps to the second branch of the pattern. slay. ✨

Heads up though: (*SKIP:NAME) searches only for names set by (*MARK:NAME). It ignores names that are set by other backtracking verbs. no cap. 🙅‍♀️

🌸 (THEN) or (THEN:NAME) 🌸

This verb causes a skip to the next innermost alternative when backtracking reaches it. Basically it cancels any further backtracking within the current alternative. Its name comes from the observation that it can be used for a pattern-based if-then-else block:

( COND1 (*THEN) FOO |  COND2 (*THEN) BAR |  COND3 (*THEN) BAZ )  ...

If the COND1 pattern matches, FOO is tried (and possibly further items after the end of the group if FOO succeeds); on failure, the matcher skips to the second alternative and tries COND2, without backtracking into COND1. If that succeeds and BAR fails, COND3 is tried. If subsequently BAZ fails, there are no more alternatives, so there's a backtrack to whatever came before the entire group. If (*THEN) is not inside an alternation, it acts like (*PRUNE).

The behaviour of (*THEN:NAME) is not the same as (*MARK:NAME)(*THEN). It's like (*MARK:NAME) in that the name is remembered for passing back to the caller. However, (*SKIP:NAME) searches only for names set with (*MARK), ignoring those set by other backtracking verbs.

A group that does not contain a | character is just a part of the enclosing alternative; it is not a nested alternation with only one alternative. The effect of (*THEN) extends beyond such a group to the enclosing alternative. Consider this pattern, where A, B, etc. are complex pattern fragments that do not contain any | characters at this level:

A (B(*THEN)C) |  D

If A and B are matched, but there's a failure in C, matching does not backtrack into A; instead it moves to the next alternative, that is, D. However, if the group containing (*THEN) is given an alternative, it behaves differently:

A (B(*THEN)C |  (*FAIL)) |  D

The effect of (*THEN) is now confined to the inner group. After a failure in C, matching moves to (*FAIL), which causes the whole group to fail because there are no more alternatives to try. In this case, matching does backtrack into A.

Note that a conditional group is not considered as having two alternatives, because only one is ever used. In other words, the | character in a conditional group has a different meaning. Ignoring white space, consider:

^.*? (?(?=a) a |  b(*THEN)c )

If the subject is "ba", this pattern does not match. Because .*? is ungreedy, it initially matches zero characters. The condition (?=a) then fails, the character "b" is matched, but "c" is not. At this point, matching does not backtrack to .*? as might perhaps be expected from the presence of the | character. The conditional group is part of the single alternative that comprises the whole pattern, and so the match fails. (If there was a backtrack into .*?, allowing it to match "b", the match would succeed.)

The verbs just described provide four different "strengths" of control when subsequent matching fails. (*THEN) is the weakest, carrying on the match at the next alternative. (*PRUNE) comes next, failing the match at the current starting position, but allowing an advance to the next character (for an unanchored pattern). (*SKIP) is similar, except that the advance may be more than one character. (*COMMIT) is the strongest, causing the entire match to fail. it's giving power scale. 👑

💜 More than one backtracking verb 💜

If more than one backtracking verb is present in a pattern, the one that is backtracked onto first acts. For example, consider this pattern, where A, B, etc. are complex pattern fragments:

(A(*COMMIT)B(*THEN)C|ABD)

If A matches but B fails, the backtrack to (*COMMIT) causes the entire match to fail. However, if A and B match, but C fails, the backtrack to (*THEN) causes the next alternative (ABD) to be tried. This behaviour is consistent, but is not always the same as Perl's. It means that if two or more backtracking verbs appear in succession, all but the last of them has no effect. Consider this example:

If there is a matching failure to the right, backtracking onto (*PRUNE) causes it to be triggered, and its action is taken. There can never be a backtrack onto (*COMMIT).

🎀 Backtracking verbs in repeated groups 🎀

PCRE2 sometimes differs from Perl in its handling of backtracking verbs in repeated groups. For example, consider:

/(a(*COMMIT)b)+ac/

If the subject is "abac", Perl matches unless its optimizations are disabled, but PCRE2 always fails because the (*COMMIT) in the second repeat of the group acts. 💅

🦋 Backtracking verbs in assertions 🦋

(*FAIL) in any assertion has its normal effect: it forces an immediate backtrack. The behaviour of the other backtracking verbs depends on whether or not the assertion is standalone or acting as the condition in a conditional group.

(*ACCEPT) in a standalone positive assertion causes the assertion to succeed without any further processing; captured strings and a mark name (if set) are retained. In a standalone negative assertion, (*ACCEPT) causes the assertion to fail without any further processing; captured substrings and any mark name are discarded.

If the assertion is a condition, (*ACCEPT) causes the condition to be true for a positive assertion and false for a negative one; captured substrings are retained in both cases.

The remaining verbs act only when a later failure causes a backtrack to reach them. This means that, for the Perl-compatible assertions, their effect is confined to the assertion, because Perl lookaround assertions are atomic. A backtrack that occurs after such an assertion is complete does not jump back into the assertion. Note in particular that a (*MARK) name that is set in an assertion is not "seen" by an instance of (*SKIP:NAME) later in the pattern.

PCRE2 now supports non-atomic positive assertions and also "scan substring" assertions, as described in the sections entitled "Non-atomic assertions" and "Scan substring assertions" above. These assertions must be standalone (not used as conditions). They are not Perl-compatible. For these assertions, a later backtrack does jump back into the assertion, and therefore verbs such as (*COMMIT) can be triggered by backtracks from later in the pattern.

The effect of (*THEN) is not allowed to escape beyond an assertion. If there are no more branches to try, (*THEN) causes a positive assertion to be false, and a negative assertion to be true. This behaviour differs from Perl when the assertion has only one branch.

The other backtracking verbs are not treated specially if they appear in a standalone positive assertion. In a conditional positive assertion, backtracking (from within the assertion) into (*COMMIT), (*SKIP), or (*PRUNE) causes the condition to be false. However, for both standalone and conditional negative assertions, backtracking into (*COMMIT), (*SKIP), or (*PRUNE) causes the assertion to be true, without considering any further alternative branches.

🌈 Backtracking verbs in subroutines 🌈

These behaviours occur whether or not the group is called recursively.

(*ACCEPT) in a group called as a subroutine causes the subroutine match to succeed without any further processing. Matching then continues after the subroutine call. Perl documents this behaviour. Perl's treatment of the other verbs in subroutines is different in some cases.

(*FAIL) in a group called as a subroutine has its normal effect: it forces an immediate backtrack.

(*COMMIT), (*SKIP), and (*PRUNE) cause the subroutine match to fail when triggered by being backtracked to in a group called as a subroutine. There is then a backtrack at the outer level.

(*THEN), when triggered, skips to the next alternative in the innermost enclosing group that has alternatives (its normal behaviour). However, if there is no such group within the subroutine's group, the subroutine match fails and there is a backtrack at the outer level.

okay bestie that's literally the whole page 😭💖 you made it to the end!! give yourself main character energy because you now understand backtracking verbs better than most people fr fr. this iconic wisdom was brought to you by Ericsson AB, stdlib 8.0.4, page re(3) — go forth and regex responsibly, queen. 👑✨🫶

📦 where this page comes from

ubuntu stonking: erlang-doc 1:29.0.6+dfsg-1 · Apache-2.0
💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on 2026-11-03: a language model's rewrite of the original, under its terms