pandoc(1)

💖 pandoc(1) — Pandoc User's Guide 💖

okay bestie, sit down, hydrate, because we are about to meet the single most iconic little command-line guy in the whole markup universe: pandoc. this is the manual page, but make it make sense. let's go 🫶

---

✨ NAME ✨

pandoc — general markup converter 💅

✨ SYNOPSIS ✨

pandoc [options] [input-file]...

✨ DESCRIPTION ✨

So. Pandoc is two things at once (main character energy fr): it's a Haskell library for converting from one markup format to another, and it's also a command-line tool that uses that library. both. at the same time. no cap.

Pandoc can convert between numerous markup and word processing formats — including, but absolutely not limited to, various flavors of Markdown, HTML, LaTeX and Word docx. for the full lists of input and output formats, go look at the --from and --to options below. Pandoc can also produce PDF output: see creating a PDF, below. 👑

Pandoc's enhanced version of Markdown includes syntax for tables, definition lists, metadata blocks, footnotes, citations, math, and much more. see below under Pandoc's Markdown. she's got range.

Pandoc has a modular design: it consists of

• a set of readers</>, which parse text in a given format and produce a native representation of the document (an abstract syntax tree or AST), and
• a set of writers</>, which convert this native representation into a target format.

so adding an input or output format requires only adding a reader or writer. that's it. that's the whole trick. ✨ Users can also run custom pandoc filters to modify the intermediate AST.

okay but here's the honest part, and we need to talk about it: because pandoc's intermediate representation of a document is less expressive than many of the formats it converts between, one should not expect perfect conversions between every format and every other. Pandoc attempts to preserve the structural elements of a document, but not formatting details such as margin size. And some document elements, such as complex tables, may not fit into pandoc's simple document model. While conversions from pandoc's Markdown to all formats aspire to be perfect, conversions from formats more expressive than pandoc's Markdown can be expected to be lossy. it's giving "we tried our best, bestie." 😭

🌸 Using pandoc 🌸

If no input-files are specified, input is read from stdin. Output goes to stdout by default. For output to a file, use the -o/--output option:

pandoc -o output.html input.txt

By default, pandoc produces a document fragment. To produce a standalone document (e.g. a valid HTML file including <head> and <body>), use the -s or --standalone flag:

pandoc -s -o output.html input.txt

For more information on how standalone documents are produced, see Templates below.

If multiple input files are given, pandoc will concatenate them all (with blank lines between them) before parsing. (Use --file-scope to parse files individually.) 🧚

🌈 Specifying formats 🌈

The format of the input and output can be specified explicitly using command-line options. The input format can be specified using the -f/--from option, the output format using the -t/--to option. Thus, to convert hello.txt from Markdown to LaTeX, you could type:

pandoc -f markdown -t latex hello.txt

To convert hello.html from HTML to Markdown:

pandoc -f html -t markdown hello.html

Supported input and output formats are listed below under Options (see -f for input formats and -t for output formats). You can also use pandoc --list-input-formats and pandoc --list-output-formats to print lists of supported formats. (obsessed with a tool that just... tells you its own options. ✨)

If the input or output format is not specified explicitly, pandoc will attempt to guess it from the extensions of the filenames. Thus, for example,

pandoc -o hello.tex hello.txt

will convert hello.txt from Markdown to LaTeX. If no output file is specified (so that output goes to stdout), or if the output file's extension is unknown, the output format will default to HTML. If no input file is specified (so that input comes from stdin), or if the input files' extensions are unknown, the input format will be assumed to be Markdown.

🦋 Character encoding 🦋

Pandoc uses the UTF-8 character encoding for both input and output. If your local character encoding is not UTF-8, you should pipe input and output through iconv:

iconv -t utf-8 input.txt | pandoc | iconv -f utf-8

Note that in some output formats (such as HTML, LaTeX, ConTeXt, RTF, OPML, DocBook, and Texinfo), information about the character encoding is included in the document header, which will only be included if you use the -s/--standalone option. so don't forget her 💅

🎀 Creating a PDF 🎀

To produce a PDF, specify an output file with a .pdf extension:

pandoc test.txt -o test.pdf

By default, pandoc will use LaTeX to create the PDF, which requires that a LaTeX engine be installed (see --pdf-engine below). Alternatively, pandoc can use ConTeXt, roff ms, or HTML as an intermediate format. To do this, specify an output file with a .pdf extension, as before, but add the --pdf-engine option or -t context, -t html, or -t ms to the command line. The tool used to generate the PDF from the intermediate format may be specified using --pdf-engine.

You can control the PDF style using variables, depending on the intermediate format used: see variables for LaTeX, variables for ConTeXt, variables for wkhtmltopdf, variables for ms. When HTML is used as an intermediate format, the output can be styled using --css.

To debug the PDF creation, it can be useful to look at the intermediate representation: instead of -o test.pdf, use for example -s -o test.tex to output the generated LaTeX. You can then test it with pdflatex test.tex.

When using LaTeX, the following packages need to be available (they are included with all recent versions of TeX Live): amsfonts, amsmath, lm, unicode-math, iftex, listings (if the --listings option is used), fancyvrb, longtable, caption, booktabs, multirow (if the document contains a table with cells that cross multiple rows), graphicx (if the document contains images), bookmark, xcolor, soul, geometry (with the geometry variable set), setspace (with linestretch), and babel (with lang). If CJKmainfont is set, xeCJK is needed if xelatex is used, else luatexja is needed if lualatex is used. framed is required if code is highlighted in a scheme that use a colored background. The use of xelatex or lualatex as the PDF engine requires fontspec. lualatex uses selnolig and lua-ul. xelatex uses bidi (with the dir variable set). If the mathspec variable is set, xelatex will use mathspec instead of unicode-math. The csquotes package will be used for typography if the csquotes variable or metadata field is set to a true value. The natbib, biblatex, bibtex, and biber packages can optionally be used for citation rendering. If math with \cancel, \bcancel, or \xcancel is used, the cancel package is needed. The following packages will be used to improve output quality if present, but pandoc does not require them to be present: upquote (for straight quotes in verbatim environments), microtype (for better spacing adjustments), parskip (for better inter-paragraph spaces), xurl (for better line breaks in URLs), and footnotehyper or footnote (to allow footnotes in tables).

okay that was a LOT of package names. bestie I know. but you asked for the full manual and we keep every single one. 💀

🌐 Reading from the Web 🌐

Instead of an input file, an absolute URI may be given. In this case pandoc will fetch the content using HTTP:

pandoc -f html -t markdown https://www.fsf.org

It is possible to supply a custom User-Agent string or other header when requesting a document from a URL:

pandoc -f html -t markdown --request-header User-Agent:"Mozilla/5.0" \
  https://www.fsf.org

---

✨ OPTIONS ✨

💅 General options 💅

• -f FORMAT, -r FORMAT, --from=FORMAT, --read=FORMAT — Specify input format. FORMAT can be:
  • asciidoc (AsciiDoc markup)
  • bibtex (BibTeX bibliography)
  • biblatex (BibLaTeX bibliography)
  • bits (BITS XML, alias for jats)
  • commonmark (CommonMark Markdown)
  • commonmark_x (CommonMark Markdown with extensions)
  • creole (Creole 1.0)
  • csljson (CSL JSON bibliography)
  • csv (CSV table)
  • tsv (TSV table)
  • djot (Djot markup)
  • docbook (DocBook)
  • docx (Word docx)
  • dokuwiki (DokuWiki markup)
  • endnotexml (EndNote XML bibliography)
  • epub (EPUB)
  • fb2 (FictionBook2 e-book)
  • gfm (GitHub-Flavored Markdown), or the deprecated and less accurate markdown_github; use markdown_github only if you need extensions not supported in gfm.
  • haddock (Haddock markup)
  • html (HTML)
  • ipynb (Jupyter notebook)
  • jats (JATS XML)
  • jira (Jira/Confluence wiki markup)
  • json (JSON version of native AST)
  • latex (LaTeX)
  • markdown (Pandoc's Markdown)
  • markdown_mmd (MultiMarkdown)
  • markdown_phpextra (PHP Markdown Extra)
  • markdown_strict (original unextended Markdown)
  • mediawiki (MediaWiki markup)
  • man (roff man)
  • mdoc (mdoc manual page markup)
  • muse (Muse)
  • native (native Haskell)
  • odt (OpenDocument text document)
  • opml (OPML)
  • org (Emacs Org mode)
  • pod (Perl's Plain Old Documentation)
  • pptx (PowerPoint)
  • ris (RIS bibliography)
  • rtf (Rich Text Format)
  • rst (reStructuredText)
  • t2t (txt2tags)
  • textile (Textile)
  • tikiwiki (TikiWiki markup)
  • twiki (TWiki markup)
  • typst (typst)
  • vimwiki (Vimwiki)
  • xlsx (Excel spreadsheet)
  • xml (XML version of native AST)
  • the path of a custom Lua reader, see Custom readers and writers below

Extensions can be individually enabled or disabled by appending +EXTENSION or -EXTENSION to the format name. See Extensions below, for a list of extensions and their names. See --list-input-formats and --list-extensions, below.

• -t FORMAT, -w FORMAT, --to=FORMAT, --write=FORMAT — Specify output format. FORMAT can be:
  • ansi (text with ANSI escape codes, for terminal viewing)
  • asciidoc (modern AsciiDoc as interpreted by Asciidoctor)
  • asciidoc_legacy (AsciiDoc as interpreted by asciidoc-py).
  • asciidoctor (deprecated synonym for asciidoc)
  • bbcode BBCode
  • bbcode_fluxbb BBCode (FluxBB)
  • bbcode_phpbb BBCode (phpBB)
  • bbcode_steam BBCode (Steam)
  • bbcode_hubzilla BBCode (Hubzilla)
  • bbcode_xenforo BBCode (xenForo)
  • beamer (LaTeX beamer slide show)
  • bibtex (BibTeX bibliography)
  • biblatex (BibLaTeX bibliography)
  • chunkedhtml (zip archive of multiple linked HTML files)
  • commonmark (CommonMark Markdown)
• commonmark_x — CommonMark Markdown, but with extensions bestie ✨
• context — ConTeXt 💅
• csljson — CSL JSON bibliography, for the citation girlies 🎀
• djot — Djot markup 🌸
• docbook or docbook4 — DocBook 4
• docbook5 — DocBook 5
• docx — Word docx, the classic 📄
• dokuwiki — DokuWiki markup
• epub or epub3 — EPUB v3 book, e-reader slay 📚
• epub2 — EPUB v2
• fb2 — FictionBook2 e-book
• gfm — GitHub-Flavored Markdown 🌈 — or the deprecated and less accurate markdown_github; use markdown_github only if you need extensions that gfm doesn't support, iykyk 💀
• haddock — Haddock markup
• html or html5 — HTML, aka HTML5/XHTML polyglot markup, the polyglot queen 👑
• html4 — XHTML 1.0 Transitional
• icml — InDesign ICML
• ipynb — Jupyter notebook 🧪
• jats_archiving — JATS XML, Archiving and Interchange Tag Set
• jats_articleauthoring — JATS XML, Article Authoring Tag Set
• jats_publishing — JATS XML, Journal Publishing Tag Set
• jats — alias for jats_archiving
• jira — Jira/Confluence wiki markup
• json — JSON version of native AST
• latex — LaTeX, academic baddie energy 📐
• man — roff man, we're so meta rn
• markdown — Pandoc's Markdown
• markdown_mmd — MultiMarkdown
• markdown_phpextra — PHP Markdown Extra
• markdown_strict — original unextended Markdown, no extras, no notes
• markua — Markua
• mediawiki — MediaWiki markup
• ms — roff ms
• muse — Muse
• native — native Haskell 🦋
• odt — OpenDocument text document
• opml — OPML
• opendocument — OpenDocument XML
• org — Emacs Org mode
• pdf — PDF, the final boss 💅
• plain — plain text, no vibes just words
• pptx — PowerPoint slide show
• rst — reStructuredText
• rtf — Rich Text Format
• texinfo — GNU Texinfo
• textile — Textile
• t2t — txt2tags
• slideous — Slideous HTML and JavaScript slide show
• slidy — Slidy HTML and JavaScript slide show
• dzslides — DZSlides HTML5 + JavaScript slide show
• revealjs — reveal.js HTML5 + JavaScript slide show
• s5 — S5 HTML and JavaScript slide show
• tei — TEI Simple
• typst — typst ✨
• vimdoc — Vimdoc
• xml — XML version of native AST
• xwiki — XWiki markup
• zimwiki — ZimWiki markup
• the path of a custom Lua writer — see Custom readers and writers below bestie, it's giving DIY 🔥

Heads up: odt, docx, epub, and pdf output will not go to stdout unless you force it with -o -. No cap.

Extensions can be individually turned on or off by appending +EXTENSION or -EXTENSION to the format name. See Extensions below for the full list of extensions and their names, and see --list-output-formats and --list-extensions below too. 🫶

---

• -o FILE, --output=FILE — write the output to FILE instead of stdout. If FILE is -, output goes to stdout even if you picked a non-textual format (docx, odt, epub2, epub3), so yes you can pipe a docx, we love a flexible icon 💖. If the output format is chunkedhtml and FILE has no extension, then instead of making a .zip file, pandoc will create a directory named FILE and unpack the zip archive in there — unless FILE already exists, in which case it just raises an error, so like, don't do that 🙈
• --data-dir=DIRECTORY — tell pandoc which user data directory to search for its data files. Skip this option and it uses the default user data directory. On *nix and macOS that's the pandoc subdirectory of the XDG data directory (by default $HOME/.local/share, and you can override that by setting the XDG_DATA_HOME environment variable). If that directory doesn't exist but $HOME/.pandoc does, it'll use that one instead — backwards compatibility core 💅. On Windows the default user data directory is %APPDATA%\pandoc. Wanna know your system's default? Just look at the output of pandoc --version. Data files you drop in there (like reference.odt, reference.docx, epub.css, templates) will override pandoc's normal defaults, main character behaviour honestly 👑. (Note: the user data directory is not created by pandoc, so you have to make it yourself if you want to use it. Do it. It's worth it.)
• -d FILE, --defaults=FILE — load a set of default option settings. FILE is a YAML or JSON file whose fields map to command-line option settings. Literally all options for document conversion, including input and output files, can be set with a defaults file 🌈. Pandoc searches for the file first in the working directory, then in the defaults subdirectory of the user data directory (see --data-dir). If FILE has no extension, pandoc will also try adding the .yaml extension if FILE isn't found. Check the Defaults files section for the tea on the file format ☕. Settings from the defaults file can be overridden or extended by options that come later on the command line, so order matters bestie.
• --completion=SHELL — generate a shell completion script for the given shell, one of bash, zsh, or fish. To actually enable completion with pandoc, evaluate the output of this command in your shell's startup file. For example, for bash, add this to your .bashrc:

eval "$(pandoc --completion=bash)"

For zsh, you'll need to create a directory to store zsh completions (e.g., ~/.local/share/zsh/site-functions) or use one of the existing directories listed by zsh --pathdirs. Create the pandoc completion script in that directory:

pandoc --completion=zsh > ~/.local/share/zsh/site-functions/_pandoc

Then add these lines to your .zshrc:

fpath=(~/.local/share/zsh $fpath)
autoload -Uz compinit && compinit

• --bash-completion — Deprecated. Use --completion=bash instead bestie, we moved on 💀
• --sandbox[=true|false] — run pandoc in a sandbox, which limits IO operations in readers and writers to reading the files you named on the command line. Important caveat: this option does not limit IO operations by filters or in the production of PDF documents 😭. But it does give you security against stuff like, say, disclosure of files through the use of include directives. If you're running pandoc on untrusted user input, you should use this option, period:

Note: some readers and writers (e.g., docx) need access to data files. If those live on the file system, pandoc won't be able to find them when running in --sandbox mode and will raise an error. For those situations, we recommend a pandoc binary compiled with the embed_data_files option, which bakes the data files into the binary instead of storing them on the file system. Iconic ✨

• --verbose — give verbose debugging output, spill the tea 💬
• --quiet — suppress warning messages. Silence is golden 🎀
• --fail-if-warnings[=true|false] — exit with error status if there are any warnings at all. Zero tolerance era 🚨
• --log=FILE — write log messages to FILE in machine-readable JSON format. All messages above DEBUG level get written, regardless of your verbosity settings (--verbose, --quiet).
• --list-input-formats — list supported input formats, one per line. Window shopping ✨
• --list-output-formats — list supported output formats, one per line.
• --list-extensions[=FORMAT] — list supported extensions for FORMAT, one per line, each preceded by a + or - telling you whether it's enabled by default in FORMAT. If you don't specify FORMAT, you get the defaults for pandoc's Markdown. Chef's kiss 💅
• --list-highlight-languages — list supported languages for syntax highlighting, one per line.
• --list-highlight-styles — list supported styles for syntax highlighting, one per line. See --syntax-highlighting 🌈
• -v, --version — print version. Iconic numbers only.
• -h, --help — show usage message. We love a helpful queen 🫶

---

🦋 Reader options 🦋

• --shift-heading-level-by=NUMBER — shift heading levels by a positive or negative integer. For example, with --shift-heading-level-by=-1, level 2 headings become level 1 headings and level 3 headings become level 2 headings. Headings can't have a level less than 1, so any heading that would get shifted below level 1 just becomes a regular paragraph 😌. Exception cutie: with a shift of -N, a level-N heading at the beginning of the document replaces the metadata title. --shift-heading-level-by=-1 is a great choice when you're converting HTML or Markdown documents that use an initial level-1 heading for the doc title and level-2+ headings for sections. And --shift-heading-level-by=1 may be a great choice for converting Markdown documents that use level-1 headings for sections into HTML, since pandoc uses a level-1 heading to render the document title. Context is everything bestie 💖
• --base-header-level=NUMBER — Deprecated. Use --shift-heading-level-by=X instead, where X = NUMBER - 1. Specify the base level for headings (defaults to 1).
• --indented-code-classes=CLASSES — specify classes to use for indented code blocks — for example, perl,numberLines or haskell. Multiple classes can be separated by spaces or commas 🎀
• --default-image-extension=EXTENSION — specify a default extension to use when image paths/URLs have no extension. This lets you reuse the same source for formats that need different kinds of images, so you're not copy-pasting like a fool 🌸. Currently this option only affects the Markdown and LaTeX readers.
• --file-scope[=true|false] — parse each file individually before combining everything for multifile documents. This lets footnotes in different files with the same identifiers work the way you'd expect 🙌. But if you set this option, footnotes and links will not work across files. Also, reading binary files (docx, odt, epub) implies --file-scope.

If two or more files are processed using --file-scope, prefixes based on the filenames get added to identifiers so they can be disambiguated, and internal links get adjusted accordingly. For example, a header with identifier foo in subdir/file1.txt will have its identifier changed to subdir__file1.txt__foo. Very demure, very mindful ✨

• -F PROGRAM, --filter=PROGRAM — bestie, this is how you plug an executable in as a filter that transforms the pandoc AST after the input gets parsed but before the output gets written. The executable should read JSON from stdin and write JSON to stdout. That JSON has to be formatted exactly like pandoc’s own JSON input and output. The name of the output format gets passed to the filter as the first argument. So this:

pandoc --filter ./caps.py -t latex

is equivalent to this:

pandoc -t json | ./caps.py latex | pandoc -f json -t latex

The latter form may be useful for debugging filters, no cap. ✨

Filters may be written in any language. Text.Pandoc.JSON exports toJSONFilter to make writing filters in Haskell easier. If you’d rather write filters in python, you can use the pandocfilters module, installable from PyPI. There are also pandoc filter libraries in PHP, perl, and JavaScript/node.js. iykyk 💅

In order of preference, pandoc will look for filters in:

1. a specified full or relative path (executable or non-executable),
2. $DATADIR/filters (executable or non-executable) where $DATADIR is the user data directory (see --data-dir, above),
3. $PATH (executable only).

Filters, Lua-filters, and citeproc processing are applied in the order specified on the command line. periodt 🔥

• -L SCRIPT, --lua-filter=SCRIPT — transform the document in a similar fashion as JSON filters (see --filter), but use pandoc’s built-in Lua filtering system. The given Lua script is expected to return a list of Lua filters which will be applied in order. Each Lua filter must contain element-transforming functions indexed by the name of the AST element on which the filter function should be applied.

The pandoc Lua module provides helper functions for element creation. It is always loaded into the script’s Lua environment.

See the Lua filters documentation for further details.

In order of preference, pandoc will look for Lua filters in:

1. a specified full or relative path,
2. $DATADIR/filters where $DATADIR is the user data directory (see --data-dir, above).

Filters, Lua filters, and citeproc processing are applied in the order specified on the command line. ate 💖

• -M KEY[=VAL], --metadata=KEY[:VAL] — set the metadata field KEY to the value VAL. A value specified on the command line overrides a value specified in the document using YAML metadata blocks. Values will be parsed as YAML boolean or string values. If no value is specified, the value will be treated as Boolean true. Like --variable, --metadata causes template variables to be set. But unlike --variable, --metadata affects the metadata of the underlying document (which is accessible from filters and may be printed in some output formats) and metadata values will be escaped when inserted into the template. the way that works 😭
• --metadata-file=FILE — read metadata from the supplied YAML (or JSON) file. This option can be used with every input format, but string scalars in the metadata file will always be parsed as Markdown. (If the input format is Markdown or a Markdown variant, then the same variant will be used to parse the metadata file; if it is a non-Markdown format, pandoc’s default Markdown extensions will be used.) This option can be used repeatedly to include multiple metadata files; values in files specified later on the command line will be preferred over those specified in earlier files. Metadata values specified inside the document, or by using -M, overwrite values specified with this option. The file will be searched for first in the working directory, and then in the metadata subdirectory of the user data directory (see --data-dir). fr fr 🍓
• -p, --preserve-tabs[=true|false] — preserve tabs instead of converting them to spaces. (By default, pandoc converts tabs to spaces before parsing its input.) Note that this will only affect tabs in literal code spans and code blocks. Tabs in regular text are always treated as spaces. ngl this one is niche but iconic 🎀
• --tab-stop=NUMBER — specify the number of spaces per tab (default is 4). simple as 💙
• --track-changes=accept|reject|all — specifies what to do with insertions, deletions, and comments produced by the MS Word “Track Changes” feature. accept (the default) processes all the insertions and deletions. reject ignores them. Both accept and reject ignore comments. all includes all insertions, deletions, and comments, wrapped in spans with insertion, deletion, comment-start, and comment-end classes, respectively. The author and time of change is included. all is useful for scripting: only accepting changes from a certain reviewer, say, or before a certain date. If a paragraph is inserted or deleted, track-changes=all produces a span with the class paragraph-insertion/paragraph-deletion before the affected paragraph break. This option only affects the docx reader. main character energy for real 👑
• --extract-media=DIR|FILE.zip — extract images and other media contained in or linked from the source document to the path DIR, creating it if necessary, and adjust the images references in the document so they point to the extracted files. Media are downloaded, read from the file system, or extracted from a binary container (e.g. docx), as needed. The original file paths are used if they are relative paths not containing ... Otherwise filenames are constructed from the SHA1 hash of the contents.

If the path given ends in .zip, then instead of creating a directory, pandoc will create a zip archive containing the media files. obsessed 💅

• --abbreviations=FILE — specifies a custom abbreviations file, with abbreviations one to a line. If this option is not specified, pandoc will read the data file abbreviations from the user data directory or fall back on a system default. To see the system default, use pandoc --print-default-data-file=abbreviations. The only use pandoc makes of this list is in the Markdown reader. Strings found in this list will be followed by a nonbreaking space, and the period will not produce sentence-ending space in formats like LaTeX. The strings may not contain spaces. lowkey useful 🌸
• --typst-input=KEY[=VAL] — set a parameter value that will be made available to the typst parser in sys.inputs, like --input in the typst CLI. Either : or = may be used to separate KEY from VAL. Values containing spaces must be quoted. vibe check passed ✅
• --trace[=true|false] — print diagnostic output tracing parser progress to stderr. This option is intended for use by developers in diagnosing performance issues. for the devs 🧚

General writer options 💖

• -s, --standalone — produce output with an appropriate header and footer (e.g. a standalone HTML, LaTeX, TEI, or RTF file, not a fragment). This option is set automatically for pdf, epub, epub3, fb2, docx, and odt output. For native output, this option causes metadata to be included; otherwise, metadata is suppressed. iconic behavior 🌈
• --template=FILE|URL — use the specified file as a custom template for the generated document. Implies --standalone. See Templates, below, for a description of template syntax. If the template is not found, pandoc will search for it in the templates subdirectory of the user data directory (see --data-dir). If no extension is specified and an extensionless template is not found, pandoc will look for a template with an extension corresponding to the writer, so that --template=special looks for special.html for HTML output. If this option is not used, a default template appropriate for the output format will be used (see -D/--print-default-template). not me loving templates 🙈
• -V KEY[=VAL], --variable=KEY[=VAL] — set the template variable KEY to the string value VAL when rendering the document in standalone mode. Either : or = may be used to separate KEY from VAL. If no VAL is specified, the key will be given the value true. Structured values (lists, maps) cannot be assigned using this option, but they can be assigned in the variables section of a defaults file or using the --variable-json option. If the variable already has a list value, the value will be added to the list. If it already has another kind of value, it will be made into a list containing the previous and the new value. For example, -V author=Joe -V author=Sue makes author contain a list of strings: Joe and Sue. the way that stacks 💜
• --variable-json=KEY[=JSON] — set the template variable KEY to the value specified by a JSON string (this may be a boolean, a string, a list, or a mapping; a number will be treated as a string). For example, --variable-json foo=false will give foo the boolean false value, while --variable-json foo='"false"' will give it the string value "false". Either : or = may be used to separate KEY from VAL. If the variable already has a value, this value will be replaced. slay ✨
• -D FORMAT, --print-default-template=FORMAT — print the system default template for an output FORMAT. (See -t for a list of possible FORMATs.) Templates in the user data directory are ignored. This option may be used with -o/--output to redirect output to a file, but -o/--output must come before --print-default-template on the command line.

Note that some of the default templates use partials, for example styles.html. To print the partials, use --print-default-data-file: for example, --print-default-data-file=templates/styles.html. order matters, bestie 💙

• --print-default-data-file=FILE — print a system default data file. Files in the user data directory are ignored. This option may be used with -o/--output to redirect output to a file, but -o/--output must come before --print-default-data-file on the command line. same energy 🍃
• --eol=crlf|lf|native — manually specify line endings: crlf (Windows), lf (macOS/Linux/UNIX), or native (line endings appropriate to the OS on which pandoc is being run). The default is native. periodt 💖
• --dpi=NUMBER — sets the default dpi (dots per inch) value used when converting pixels into inches/centimeters and back the other way. (Okay so technically the correct term would be ppi: pixels per inch. We all know. We're moving on. 💅) The default is 96dpi. BUT — when an image has its own internal dpi info baked in, that encoded value gets used instead of whatever you set here. So this is a fallback, not a dictatorship. 👑
• --wrap=auto|none|preserve — decides how text gets wrapped in the output (like, the source code of the output, not the rendered pretty version, bestie). With auto — the default — pandoc tries to wrap lines to the column width from --columns (default 72). With none, pandoc does not wrap anything at all, no thoughts head empty. With preserve, pandoc tries to keep the wrapping from the source document — meaning where there are nonsemantic newlines in the source, there will be nonsemantic newlines in the output too. In ipynb output, this option controls wrapping of the contents of Markdown cells. 🫶
• --columns=NUMBER — sets how long lines are, in characters. This affects text wrapping in the generated source code (see --wrap above, bestie, they're a package deal). It also affects the calculation of column widths for plain text tables (see Tables below). ✨
• --toc[=true|false], --table-of-contents[=true|false] — drops an automatically generated table of contents into the output document (or, for latex, context, docx, odt, opendocument, rst, or ms, an instruction to create one). Heads up though: this does nothing unless -s/--standalone is used, and it also does nothing at all for man, docbook4, docbook5, or jats output.

Also note: if you're making a PDF via ms and using (the default) groff as your --pdf-engine, the table of contents will show up at the end of the document. Kinda cursed ngl. If you'd rather have it at the beginning, before the title, use --pdf-engine=pdfroff. 🎀

And because the entries in a table of contents need heading identifiers to link to, this option forces the auto_identifiers extension on for the input format — even when it's off by default (as it is for typst). It's giving main character, it takes what it needs.

• --toc-depth=NUMBER — how many section levels get included in the table of contents. Default is 3, which means level-1, level-2, and level-3 headings all get listed. 💖
• --lof[=true|false], --list-of-figures[=true|false] — drops in an automatically generated list of figures (or, in some formats, an instruction to create one) in the output document. Has no effect unless -s/--standalone is used, and it only does anything for latex, context, and docx output. 🌸
• --lot[=true|false], --list-of-tables[=true|false] — same energy, but for tables: an automatically generated list of tables (or, in some formats, an instruction to create one) in the output document. Also no effect unless -s/--standalone is used, and it only works on latex, context, and docx output. 🦋
• --strip-comments[=true|false] — strips out HTML comments from the Markdown or Textile source, instead of passing them along to Markdown, Textile, or HTML output as raw HTML. 🩷 This does not apply to HTML comments sitting inside raw HTML blocks when the markdown_in_html_blocks extension is not set.
• --syntax-highlighting=default|none|idiomatic|STYLE|FILE — which method to use for code syntax highlighting. Setting a specific STYLE makes highlighting run through the internal highlighting engine, using KDE syntax definitions and styles. The idiomatic method uses a format-specific highlighter if one exists, or falls back to the default style if the target format has no idiomatic highlighting method. Setting this to none disables all syntax highlighting, full stop. The default method uses a format-specific default.

The default for HTML, EPUB, Docx, Ms, Man, and LaTeX output is to use the internal highlighter with the default style; for Typst, it's to use Typst's own syntax highlighting system. 👑

Style options are pygments (the default), kate, monochrome, breezeDark, espresso, zenburn, haddock, and tango. For more info on syntax highlighting in pandoc, see Syntax highlighting, below. See also --list-highlight-styles. ✨

Instead of a STYLE name, you can hand over a JSON file with a .theme extension. That gets parsed as a KDE syntax highlighting theme and (if it's valid) used as the highlighting style. 🎀

To generate the JSON version of an existing style, use --print-highlight-style.

• --no-highlight — ⚠️ Deprecated, use --syntax-highlighting=none instead.

Disables syntax highlighting for code blocks and inlines, even when a language attribute is given. 😭

• --highlight-style=STYLE|FILE — ⚠️ Deprecated, use --syntax-highlighting=STYLE|FILE instead.

Specifies the coloring style used in highlighted source code.

• --print-highlight-style=STYLE|FILE — prints out a JSON version of a highlighting style, which you can then modify, save with a .theme extension, and use with --syntax-highlighting. 💅 This option can be used with -o/--output to redirect output to a file — but -o/--output must come before --print-highlight-style on the command line. Order matters, bestie, no cap.
• --syntax-definition=FILE — tells pandoc to load a KDE XML syntax definition file, which then gets used for syntax highlighting of appropriately marked code blocks. Use this to add support for brand-new languages or to swap in altered syntax definitions for languages you already have. This option may be repeated to add multiple syntax definitions. 🔥
• -H FILE, --include-in-header=FILE|URL — includes the contents of FILE, verbatim, at the end of the header. Use it for things like slipping special CSS or JavaScript into HTML documents. This option can be used repeatedly to include multiple files, and they'll be included in the order you specified. Implies --standalone. 🌈
• -B FILE, --include-before-body=FILE|URL — includes the contents of FILE, verbatim, at the beginning of the document body (think: after the <body> tag in HTML, or the \begin{document} command in LaTeX). Perfect for navigation bars or banners in HTML documents. This option can be used repeatedly to include multiple files, and they'll be included in the order specified. Implies --standalone. 💜 Note that if the output format is odt, this file must be in OpenDocument XML format suitable for insertion into the body of the document, and if the output is docx, this file must be in appropriate OpenXML format.
• -A FILE, --include-after-body=FILE|URL — includes the contents of FILE, verbatim, at the end of the document body (before the </body> tag in HTML, or the \end{document} command in LaTeX). This option can be used repeatedly to include multiple files, included in the order specified. Implies --standalone. 💖 Note that if the output format is odt, this file must be in OpenDocument XML format suitable for insertion into the body of the document, and if the output is docx, this file must be in appropriate OpenXML format.
• --resource-path=SEARCHPATH — the list of paths to search for images and other resources. 🌸 Paths get separated by : on Linux, UNIX, and macOS systems, and by ; on Windows. If --resource-path isn't specified, the default resource path is just the working directory. ⚠️ Big note: if --resource-path is specified, the working directory must be explicitly listed or it will not be searched. Sneaky, I know. For example:

--resource-path=.:test

will search the working directory and the test subdirectory, in that order. This option can be used repeatedly. Search path components that come later on the command line get searched before those that come earlier, so

--resource-path foo:bar --resource-path baz:bim

is equivalent to

--resource-path baz:bim:foo:bar

Note that this option only has an effect when pandoc itself needs to find an image (e.g., in producing a PDF or docx, or when --embed-resources is used). It will not cause image paths to be rewritten in other cases (e.g., when pandoc is generating LaTeX or HTML). 🫶

• --request-header=NAME:VAL — sets the request header NAME to the value VAL whenever pandoc makes HTTP requests (for instance, when a URL is given on the command line, or when resources used in a document have to be downloaded). 🦋 If you're behind a proxy, you also need to set the environment variable http_proxy to http://....
• --no-check-certificate[=true|false] — disables certificate verification so you can access unsecure HTTP resources (for example when the certificate is no longer valid, or it's self signed). 🙈 Live a little, dangerously.

---

✨ Options affecting specific writers ✨

• --self-contained[=true|false] — ⚠️ deprecated synonym for --embed-resources --standalone. It's the old name, it still vibes, but like… use the new one.
• --embed-resources[=true|false] — produces a standalone HTML file with zero external dependencies, using data: URIs to fold in the contents of linked scripts, stylesheets, images, and videos. The resulting file should be "self-contained," in the sense that it needs no external files and no net access to display properly in a browser. Iconic. 👑

This option works only with HTML output formats, including html4, html5, html+lhs, html5+lhs, s5, slidy, slideous, dzslides, and revealjs. Scripts, images, and stylesheets at absolute URLs will be downloaded; those at relative URLs will be sought relative to the working directory (if the first source file is local) or relative to the base URL (if the first source file is remote). Elements with the attribute data-external="1" will be left alone — the documents they link to will not be incorporated into the document. 🚫

⚠️ Limitation, bestie: resources loaded dynamically through JavaScript cannot be incorporated. As a result, fonts may be missing when --math-method=mathjax is used, and some advanced features (e.g. zoom or speaker notes) may not work in an offline "self-contained" reveal.js slide show. 😭 It's not a bug, it's just the rules of the game.

For SVG images, img tags with data: URIs are used, unless the image has the class inline-svg, in which case an inline SVG element is inserted. This approach is highkey recommended when there are many occurrences of the same SVG in a document, because <use> elements get used to cut down on duplication. 💅

• --link-images[=true|false] — Instead of embedding images in ODT, just link to them. (This one currently only affects ODT output, bestie, so don't go expecting it everywhere.)
• --html-q-tags[=true|false] — Use <q> tags for quotes in HTML. 🎀 Heads up tho: this only does anything if the smart extension is enabled for the input format you're using.
• --ascii[=true|false] — ASCII only, no fancy Unicode, periodt. ✨ Currently supported for XML and HTML formats (which use entities instead of UTF-8 when you pick this), CommonMark, gfm, and Markdown (entities again), roff man and ms (hexadecimal escapes, slay), and, to a limited degree, LaTeX (which uses standard commands for accented characters when it can).
• --reference-links[=true|false] — Use reference-style links instead of inline links when writing Markdown or reStructuredText. By default it's inline links. Where those references land is up to --reference-location. 🦋
• --reference-location=block|section|document — Pick where footnotes (and references, if reference-links is set) go: end of the current (top-level) block, the current section, or the document. Default is document. Currently this only affects the markdown, muse, html, epub, slidy, s5, slideous, dzslides, and revealjs writers. In slide formats, --reference-location=section renders notes at the bottom of a slide. 👑
• --figure-caption-position=above|below — Do figure captions sit above or below the figure? Default is below. Only affects HTML, LaTeX, Docx, ODT, and Typst output.
• --table-caption-position=above|below — Same energy, but for tables. Default is above. Again, only HTML, LaTeX, Docx, ODT, and Typst output care. 🌸
• --markdown-headings=setext|atx — ATX-style (#-prefixed) or Setext-style (underlined) headings for level 1 and 2 headings in Markdown output? Default is atx. Levels 3+ are always ATX. This also affects Markdown cells in ipynb output. 💖
• --list-tables[=true|false] — Render tables as list tables in RST output.
• --top-level-division=default|section|chapter|part — Treat top-level headings as this division type in LaTeX, ConTeXt, DocBook, and TEI output. Hierarchy order goes part → chapter → section, and all headings get shifted so the top-level heading becomes whatever type you picked. Default behaviour is to figure out the best division type via heuristics: unless other conditions apply, section is chosen. When the documentclass variable is set to report, book, or memoir (unless the article option is specified), chapter is implied. And if beamer is the output format, specifying chapter or part makes top-level headings become \part{..}, while second-level headings stay their default type.

In Docx output, this option adds section breaks before first-level headings if chapter is selected, and before first- and second-level headings if part is selected. Footnote numbers restart with each section break unless the reference doc modifies that. 💀

• -N, --number-sections=[true|false] — Number those section headings in LaTeX, ConTeXt, HTML, Docx, ms, or EPUB output. By default sections are not numbered. Sections with class unnumbered will never be numbered, even if --number-sections is specified. No exceptions, fr fr.
• --number-offset=NUMBER[,NUMBER,...] — Offsets for section heading numbers. First number gets added to the level-1 section number, second to level-2, and so on. So like, if you want your first level-1 heading to be numbered "6" instead of "1", say --number-offset=5. If your doc starts with a level-2 heading you want numbered "1.5", say --number-offset=1,4. --number-offset only directly affects the number of the first section heading in a document — after that, numbers just increment normally. Implies --number-sections. Currently this feature only affects HTML and Docx output. 🌈
• --listings[=true|false] — Deprecated, use --syntax-highlighting=idiomatic or --syntax-highlighting=default instead.

Use the listings package for LaTeX code blocks. The package does not support multi-byte encoding for source code. To handle UTF-8 you'd need a custom template. This issue is fully documented here: Encoding issue with the listings package.

• -i, --incremental[=true|false] — Make list items in slide shows show up incrementally, one by one, like a little reveal moment. ✨ Default is all at once.
• --slide-level=NUMBER — Says that headings at this level create slides (for beamer, revealjs, pptx, s5, slidy, slideous, dzslides). Headings above this level split the slide show into sections; headings below it create subheads inside a slide. Valid values are 0-6. If you set slide level 0, slides won't be split automatically on headings at all, and you have to use horizontal rules to mark slide boundaries. If you don't specify a slide level explicitly, it gets set automatically based on the document contents — see Structuring the slide show. 🎀
• --section-divs[=true|false] — Wrap sections in <section> tags (or <div> tags for html4), and attach identifiers to the enclosing <section> (or <div>) instead of the heading itself (see Heading identifiers, below). This only affects HTML output (and does not affect HTML slide formats).
• --email-obfuscation=none|javascript|references — How do we obfuscate mailto: links in HTML documents? none leaves them exactly as they are. javascript obfuscates them with JavaScript. references obfuscates them by printing their letters as decimal or hexadecimal character references. Default is none. 🫶
• --id-prefix=STRING — A prefix added to all identifiers and internal links in HTML and DocBook output, and to footnote numbers in Markdown and Haddock output. Super useful for dodging duplicate identifiers when you're generating fragments to drop into other pages. 💜
• -T STRING, --title-prefix=STRING — Use STRING as a prefix at the beginning of the title that shows up in the HTML header (but not in the title as it appears at the beginning of the HTML body). Implies --standalone.
• -c URL, --css=URL — Link to a CSS stylesheet. Can be used repeatedly to pull in multiple files, and they get included in the order you specified. Only affects HTML (including HTML slide shows) and EPUB output. Use it together with -s/--standalone, because the stylesheet link goes in the document header. 🌸

A stylesheet is required for generating EPUB. If none is given via this option (or the css / stylesheet metadata fields), pandoc looks for a file epub.css in the user data directory (see --data-dir). If it's not found there either, sensible defaults will be used.

• --reference-doc=FILE|URL — Use the specified file as a style reference when producing a docx or ODT file. 👑

Docx — For best results, the reference docx should be a modified version of a docx produced using pandoc. The contents of the reference docx get ignored, but its stylesheets and document properties (margins, page size, header, footer — all of it) are used in the new docx. If no reference docx is specified on the command line, pandoc looks for a file reference.docx in the user data directory (see --data-dir). If that's not found either, sensible defaults get used.

To produce a custom reference.docx, first grab a copy of the default one:

pandoc -o custom-reference.docx --print-default-data-file reference.docx

Then open custom-reference.docx in Word or LibreOffice, tweak the styles however you want, and save it. For best results, don't change anything in this file other than the styles pandoc uses: 💅

Paragraph styles:

• Normal
• Body Text
• First Paragraph
• Compact
• Title
• Subtitle
• Author
• Date
• Abstract
• AbstractTitle
• Bibliography
• Heading 1
• Heading 2
• Heading 3
• Heading 4
• Heading 5
• Heading 6
• Heading 7
• Heading 8
• Heading 9
• Block Text [for block quotes]
• Footnote Block Text [for block quotes in footnotes]
• Source Code
• Footnote Text
• Definition Term
• Definition
• Caption
• Table Caption
• Image Caption
• Figure
• Captioned Figure
• TOC Heading

Character styles:

• Default Paragraph Font
• Verbatim Char
• Footnote Reference
• Hyperlink
• Section Number

Table style:

• Table

• ODT 💖 — for the bestest results, your reference ODT should lowkey be a modified version of an ODT that was produced using pandoc. The contents of the reference ODT get ignored, but its stylesheets? Those are what get used in the new ODT. If you don't specify a reference ODT on the command line, pandoc will go looking for a file called reference.odt in the user data directory (see --data-dir). And if that's not found either, sensible defaults get used, no stress.

To produce a custom reference.odt, first grab a copy of the default one:

pandoc -o custom-reference.odt --print-default-data-file reference.odt

Then open custom-reference.odt in LibreOffice, modify the styles however you want, and save the file. Periodt. 💅

• PowerPoint ✨ — templates included with Microsoft PowerPoint 2013 (either with .pptx or .potx extension) are known to work, and so are most templates derived from those.

The specific requirement is that the template should contain layouts with the following names (as seen within PowerPoint):

• Title Slide
• Title and Content
• Section Header
• Two Content
• Comparison
• Content with Caption
• Blank

For each name, the first layout found with that name will be used. If no layout is found with one of the names, pandoc will output a warning and use the layout with that name from the default reference doc instead. (How these layouts are used is described in PowerPoint layout choice.)

All templates included with a recent version of MS PowerPoint will fit these criteria. (You can click on Layout under the Home menu to check, bestie 👀.)

You can also modify the default reference.pptx: first run

pandoc -o custom-reference.pptx --print-default-data-file reference.pptx

and then modify custom-reference.pptx in MS PowerPoint (pandoc will use the layouts with the names listed above). 🎀

---

• --split-level=NUMBER — specify the heading level at which to split an EPUB or chunked HTML document into separate files. The default is to split into chapters at level-1 headings. In the case of EPUB, this option only affects the internal composition of the EPUB, not the way chapters and sections are displayed to users. Some readers may be slow if the chapter files are too large, so for large documents with few level-1 headings, one might want to use a chapter level of 2 or 3. For chunked HTML, this option determines how much content goes in each "chunk." 🦋
• --chunk-template=PATHTEMPLATE — specify a template for the filenames in a chunkedhtml document. In the template, %n will be replaced by the chunk number (padded with leading 0s to 3 digits), %s with the section number of the chunk, %h with the heading text (with formatting removed), and %i with the section identifier. For example, section-%s-%i.html might be resolved to section-1.1-introduction.html. The characters / and \ are not allowed in chunk templates and will be ignored. The default is %s-%i.html. 🌸
• --epub-chapter-level=NUMBER — deprecated synonym for --split-level. (Old news, still works, but like… use --split-level okay? 💀)
• --epub-cover-image=FILE — use the specified image as the EPUB cover. It is recommended that the image be less than 1000px in width and height. Note that in a Markdown source document you can also specify cover-image in a YAML metadata block (see EPUB Metadata, below). 🖼️💖
• --epub-title-page=true|false — determines whether a the title page is included in the EPUB (default is true).
• --epub-metadata=FILE — look in the specified XML file for metadata for the EPUB. The file should contain a series of Dublin Core elements. For example:

<dc:rights>Creative Commons</dc:rights>
<dc:language>es-AR</dc:language>

By default, pandoc will include the following metadata elements: <dc:title> (from the document title), <dc:creator> (from the document authors), <dc:date> (from the document date, which should be in ISO 8601 format), <dc:language> (from the lang variable, or, if is not set, the locale), and <dc:identifier id="BookId"> (a randomly generated UUID). Any of these may be overridden by elements in the metadata file.

Note: if the source document is Markdown, a YAML metadata block in the document can be used instead. See below under EPUB Metadata.

• --epub-embed-font=FILE — embed the specified font in the EPUB. This option can be repeated to embed multiple fonts. Wildcards can also be used: for example, DejaVuSans-*.ttf. However, if you use wildcards on the command line, be sure to escape them or put the whole filename in single quotes, to prevent them from being interpreted by the shell. 🐚✨ To use the embedded fonts, you will need to add declarations like the following to your CSS (see --css):

@font-face {
   font-family: DejaVuSans;
   font-style: normal;
   font-weight: normal;
   src:url("../fonts/DejaVuSans-Regular.ttf");
}
@font-face {
   font-family: DejaVuSans;
   font-style: normal;
   font-weight: bold;
   src:url("../fonts/DejaVuSans-Bold.ttf");
}
@font-face {
   font-family: DejaVuSans;
   font-style: italic;
   font-weight: normal;
   src:url("../fonts/DejaVuSans-Oblique.ttf");
}
@font-face {
   font-family: DejaVuSans;
   font-style: italic;
   font-weight: bold;
   src:url("../fonts/DejaVuSans-BoldOblique.ttf");
}
body { font-family: "DejaVuSans"; }

• --epub-subdirectory=DIRNAME — specify the subdirectory in the OCF container that is to hold the EPUB-specific contents. The default is EPUB. To put the EPUB contents in the top level, use an empty string.
• --ipynb-output=all|none|best — determines how ipynb output cells are treated. all means that all of the data formats included in the original are preserved. none means that the contents of data cells are omitted. best causes pandoc to try to pick the richest data block in each output cell that is compatible with the output format. The default is best. 🧚
• --pdf-engine=PROGRAM — use the specified engine when producing PDF output. Valid values are pdflatex, lualatex, xelatex, latexmk, tectonic, wkhtmltopdf, weasyprint, pagedjs-cli, prince, context, groff, pdfroff, and typst. If the engine is not in your PATH, the full path of the engine may be specified here. If this option is not specified, pandoc uses the following defaults depending on the output format specified using -t/--to:
  • -t latex or none: pdflatex (other options: xelatex, lualatex, tectonic, latexmk)
  • -t context: context
  • -t html: weasyprint (other options: prince, wkhtmltopdf, pagedjs-cli; see print-css.rocks for a good introduction to PDF generation from HTML/CSS)
  • -t ms: groff
  • -t typst: typst

This option is normally intended to be used when a PDF file is specified as -o/--output. However, it may still have an effect when other output formats are requested. For example, ms output will include .pdfhref macros only if a --pdf-engine is selected, and the macros will be differently encoded depending on whether groff or pdfroff is specified. 👑

• --pdf-engine-opt=STRING — use the given string as a command-line argument to the pdf-engine. For example, to use a persistent directory foo for latexmk's auxiliary files, use --pdf-engine-opt=-outdir=foo. Note that no check for duplicate options is done.

---

🌈 Citation rendering 🌈

• -C, --citeproc — process the citations in the file, replacing them with rendered citations and adding a bibliography. Citation processing will not take place unless bibliographic data is supplied, either through an external file specified using the --bibliography option or the bibliography field in metadata, or via a references section in metadata containing a list of citations in CSL YAML format with Markdown formatting. The style is controlled by a CSL stylesheet specified using the --csl option or the csl field in metadata. (If no stylesheet is specified, the chicago-author-date style will be used by default.) The citation processing transformation may be applied before or after filters or Lua filters (see --filter, --lua-filter): these transformations are applied in the order they appear on the command line. For more information, see the section on Citations. 📚💖

Note: if this option is specified, the citations extension will be disabled automatically in the writer, to ensure that the citeproc-generated citations will be rendered instead of the format's own citation syntax.

• --bibliography=FILE — set the bibliography field in the document's metadata to FILE, overriding any value set in the metadata. If you supply this argument multiple times, each FILE will be added to bibliography. If FILE is a URL, it will be fetched via HTTP. If FILE is not found relative to the working directory, it will be sought in the resource path (see --resource-path). 🔥
• --csl=FILE — set the csl field in the document's metadata to FILE, overriding any value set in the metadata. (This is equivalent to --metadata csl=FILE.) If FILE is a URL, it will be fetched via HTTP. If FILE is not found relative to the working directory, it will be sought in the resource path (see --resource-path) and finally in the csl subdirectory of the pandoc user data directory. 🫶
• --citation-abbreviations=FILE — bestie, this sets the citation-abbreviations field in the doc's metadata to FILE, and it straight-up overrides whatever was already chilling in the metadata. It's literally the same thing as --metadata citation-abbreviations=FILE, no cap. If FILE is a URL? It gets fetched via HTTP, iconic. And if FILE isn't found relative to the working directory, pandoc goes looking in the resource path (peep --resource-path) and then finally in the csl subdirectory of the pandoc user data directory. 🫶
• --natbib — use natbib for citations in LaTeX output. ⚠️ Do not use this with the --citeproc option or with PDF output, periodt. It's meant for making a LaTeX file that you can then process with bibtex.
• --biblatex — use biblatex for citations in LaTeX output. Same warning, bestie: not for use with --citeproc or with PDF output. It's for producing a LaTeX file that can be processed with bibtex or biber. 💅

---

✨ Math rendering ✨

• --math-method=METHOD[:URL] — this picks the method used to display TeX math. Here are all the possible methods, some of which take an optional URL separated from the method name by a colon. Oh and the default value is mathml. 🌸
  • plain — renders math using Unicode characters, to the extent that's possible. When it's not possible, it falls back to plain TeX. Simple, lowkey elegant.
  • mathjax[:URL] — renders math using MathJaX. TeX math gets put between \(...\) (for inline math) or \[...\] (for display math) and wrapped in <span> tags with class math. Then the MathJax JavaScript renders it. 🔥 The URL should point to the MathJax.js load script. If no URL is provided, a link to the Cloudflare CDN gets inserted. This method works in HTML and HTML-derived formats.
  • mathml — converts TeX math to MathML. MathML is supported natively by the main web browsers and most e-book readers. Works in HTML and XML formats. 👑
  • webtex[:URL] — converts TeX formulas to <img> tags that link to an external script which converts formulas to images. The formula gets URL-encoded and concatenated with the URL you provided. For SVG images you can, for example, use --math-method=webtex:https://latex.codecogs.com/svg.latex?. If no URL is specified, the CodeCogs URL generating PNGs gets used (https://latex.codecogs.com/png.latex?). Note: this method works not just in HTML but in Markdown too, which is super useful if you're targeting a version of Markdown without native math support. 🎀
  • katex[:URL] — uses KaTeX to display embedded TeX math in HTML output. The URL is the base URL for the KaTeX library, and that directory should contain a katex.min.js and a katex.min.css file. If a URL is not provided, a link to the KaTeX CDN gets inserted. Works in HTML and HTML-derived formats. ✨
  • gladtex — encloses TeX math in <eq> tags in HTML output. The resulting HTML can then be processed by GladTeX to produce SVG images of the typeset formulas and an HTML file with these images embedded.

pandoc -s --math-method=gladtex input.md -o myfile.htex
gladtex -d image_dir myfile.htex
# produces myfile.html and images in image_dir

• --mathjax[=URL] — deprecated, bestie 💀 use --math-method=mathjax[:URL] instead.
• --mathml — deprecated 💀 use --math-method=mathml instead.
• --webtex[=URL] — deprecated 💀 use --math-method=webtex[:URL] instead.
• --katex[=URL] — deprecated 💀 use --math-method=katex[:URL] instead.
• --gladtex — deprecated 💀 use --math-method=gladtex instead.

---

🧚 Options for wrapper scripts 🧚

• --dump-args[=true|false] — prints info about the command-line arguments to stdout, then exits. This one is mainly for use in wrapper scripts. The first line of output contains the name of the output file specified with the -o option, or - (for stdout) if no output file was specified. The remaining lines contain the command-line arguments, one per line, in the order they appear. These do not include regular pandoc options and their arguments, but they do include any options appearing after a -- separator at the end of the line. 🙈
• --ignore-args[=true|false] — ignore command-line arguments (for use in wrapper scripts). Regular pandoc options are not ignored. So like, for example,

pandoc --ignore-args -o foo.html -s foo.txt -- -e latin1

is equivalent to

pandoc -o foo.html -s

---

💀 EXIT CODES 💀

If pandoc completes successfully, it returns exit code 0. Nonzero exit codes have the following meanings, bestie — memorize these, they're giving cheat sheet:

• 1 — PandocIOError
• 3 — PandocFailOnWarningError
• 4 — PandocAppError
• 5 — PandocTemplateError
• 6 — PandocOptionError
• 21 — PandocUnknownReaderError
• 22 — PandocUnknownWriterError
• 23 — PandocUnsupportedExtensionError
• 24 — PandocCiteprocError
• 25 — PandocBibliographyError
• 31 — PandocEpubSubdirectoryError
• 43 — PandocPDFError
• 44 — PandocXMLError
• 47 — PandocPDFProgramNotFoundError
• 61 — PandocHttpError
• 62 — PandocShouldNeverHappenError
• 63 — PandocSomeError
• 64 — PandocParseError
• 66 — PandocMakePDFError
• 67 — PandocSyntaxMapError
• 83 — PandocFilterError
• 84 — PandocLuaError
• 89 — PandocNoScriptingEngine
• 91 — PandocMacroLoop
• 92 — PandocUTF8DecodingError
• 93 — PandocIpynbDecodingError
• 94 — PandocUnsupportedCharsetError
• 95 — PandocInputNotTextError
• 97 — PandocCouldNotFindDataFileError
• 98 — PandocCouldNotFindMetadataFileError
• 99 — PandocResourceNotFound

---

🌈 DEFAULTS FILES 🌈

Okay so the --defaults option lets you specify a whole package of options, in the form of a YAML or JSON file. All the examples in this section are gonna be in YAML, but the equivalent forms in JSON work too. 💜

Fields that you omit just get their regular default values. So a defaults file can literally be as simple as one line:

verbosity: INFO

or in JSON:

{ "verbosity": "INFO" }

In fields that expect a file path (or a list of file paths), you can use the following syntax to interpolate environment variables:

csl:  ${HOME}/mycsldir/special.csl

${USERDATA} may also be used; this always resolves to the user data directory that is current when the defaults file is parsed, regardless of the setting of the environment variable USERDATA. Iconic behaviour tbh. ✨

${.} resolves to the directory containing the defaults file itself. This lets you refer to resources contained in that directory:

epub-cover-image: ${.}/cover.jpg
epub-metadata: ${.}/meta.xml
resource-path:
- .             # the working directory from which pandoc is run
- ${.}/images   # the images subdirectory of the directory
                # containing this defaults file

This environment variable interpolation syntax only works in fields that expect file paths. No cap.

Defaults files can be placed in the defaults subdirectory of the user data directory and used from any directory. So like, you could create a file specifying defaults for writing letters, save it as letter.yaml in the defaults subdirectory of the user data directory, and then invoke these defaults from any directory using pandoc --defaults letter or pandoc -dletter. Obsessed. 🎀

When multiple defaults are used, their contents get combined.

Note that, where command-line arguments may be repeated (--metadata-file, --css, --include-in-header, --include-before-body, --include-after-body, --variable, --metadata, --syntax-definition), the values specified on the command line will combine with values specified in the defaults file, rather than replacing them. ✨

The following tables show the mapping between the command line and defaults file entries. 🫶

• foo.md maps to input-file: foo.md
• foo.md bar.md maps to:

input-files:
  - foo.md
  - bar.md

The value of input-files may be left empty to indicate input from stdin, and it can be an empty sequence [] for no input.

---

🌸 General options 🌸

• --from markdown+emoji maps to either from: markdown+emoji or reader: markdown+emoji
• --to markdown+hard_line_breaks maps to either to: markdown+hard_line_breaks or writer: markdown+hard_line_breaks
• --output foo.pdf maps to output-file: foo.pdf
• --output - maps to:

output-file:

• --data-dir dir maps to data-dir: dir
• --defaults file maps to:

defaults:
- file

• --verbose maps to verbosity: INFO
• --quiet maps to verbosity: ERROR
• --fail-if-warnings maps to fail-if-warnings: true
• --sandbox maps to sandbox: true
• --log=FILE maps to log-file: FILE

Options specified in a defaults file itself always have priority over those in another file included with a defaults: entry. Periodt. 👑

verbosity can have the values ERROR, WARNING, or INFO.

---

🦋 Reader options 🦋

• --shift-heading-level-by -1 maps to shift-heading-level-by: -1
• --indented-code-classes python maps to:

indented-code-classes:
  - python

• --default-image-extension ".jpg" maps to default-image-extension: '.jpg'
• --file-scope maps to file-scope: true
• this whole command line:

--citeproc \
 --lua-filter count-words.lua \
 --filter special.lua

maps to:

filters:
  - citeproc
  - count-words.lua
  - type: json
    path: special.lua

• this whole command line:

--metadata key=value \
 --metadata key2

maps to:

metadata:
  key: value
  key2: true

• --metadata-file meta.yaml maps to either:

metadata-files:
  - meta.yaml

or:

metadata-file: meta.yaml

• --preserve-tabs maps to preserve-tabs: true
• --tab-stop 8 maps to tab-stop: 8
• --track-changes accept maps to track-changes: accept
• --extract-media dir → extract-media: dir
• --abbreviations abbrevs.txt → abbreviations: abbrevs.txt
• --typst-input foo=bar →

typst-inputs:
  foo: bar

• --trace → trace: true

heads up bestie 💖 the values you drop into a defaults file get parsed as plain literal string text, not as Markdown. no markdown glow-ups in there, okay?

Also, filters get sniffed out automatically: if they end in .lua they're assumed to be Lua filters, otherwise they're JSON filters. But you can also say the type out loud if you want, like the example above does. And they run in the order you listed them, no shuffling. Want the built-in citeproc filter in the mix? Just write citeproc or {type: citeproc}. periodt 💅

---

✨ General writer options ✨

• --standalone → standalone: true
• --template letter → template: letter
• --variable key=val (and a second --variable key2) →

variables:
  key: val
  key2: true

• --eol nl → eol: nl
• --dpi 300 → dpi: 300
• --wrap preserve → wrap: "preserve"
• --columns 72 → columns: 72
• --table-of-contents → table-of-contents: true
• --toc → toc: true
• --toc-depth 3 → toc-depth: 3
• --strip-comments → strip-comments: true
• --no-highlight → syntax-highlighting: 'none'
• --syntax-highlighting kate → syntax-highlighting: kate
• --syntax-definition mylang.xml →

syntax-definitions:
  - mylang.xml

or, the singular way: syntax-definition: mylang.xml

• --include-in-header inc.tex →

include-in-header:
  - inc.tex

• --include-before-body inc.tex →

include-before-body:
  - inc.tex

• --include-after-body inc.tex →

include-after-body:
  - inc.tex

• --resource-path .:foo → resource-path: ['.','foo']
• --request-header foo:bar →

request-headers:
  - ["User-Agent", "Mozilla/5.0"]

• --no-check-certificate → no-check-certificate: true

---

💫 Options affecting specific writers 💫

• --self-contained → self-contained: true
• --link-images → link-images: true
• --html-q-tags → html-q-tags: true
• --ascii → ascii: true
• --reference-links → reference-links: true
• --reference-location block → reference-location: block
• --figure-caption-position=above → figure-caption-position: above
• --table-caption-position=below → table-caption-position: below
• --markdown-headings atx → markdown-headings: atx
• --list-tables → list-tables: true
• --top-level-division chapter → top-level-division: chapter
• --number-sections → number-sections: true
• --number-offset=1,4 → number-offset: \[1,4\]
• --listings → listings: true
• --list-of-figures → list-of-figures: true
• --lof → lof: true
• --list-of-tables → list-of-tables: true
• --lot → lot: true
• --incremental → incremental: true
• --slide-level 2 → slide-level: 2
• --section-divs → section-divs: true
• --email-obfuscation references → email-obfuscation: references
• --id-prefix ch1 → identifier-prefix: ch1
• --title-prefix MySite → title-prefix: MySite
• --css styles/screen.css plus --css styles/special.css →

css:
  - styles/screen.css
  - styles/special.css

• --reference-doc my.docx → reference-doc: my.docx
• --epub-cover-image cover.jpg → epub-cover-image: cover.jpg
• --epub-title-page=false → epub-title-page: false
• --epub-metadata meta.xml → epub-metadata: meta.xml
• --epub-embed-font special.otf plus --epub-embed-font headline.otf →

epub-fonts:
  - special.otf
  - headline.otf

• --split-level 2 → split-level: 2
• --chunk-template="%i.html" → chunk-template: "%i.html"
• --epub-subdirectory="" → epub-subdirectory: ''
• --ipynb-output best → ipynb-output: best
• --pdf-engine xelatex → pdf-engine: xelatex
• --pdf-engine-opt=--shell-escape →

pdf-engine-opts:
  - '-shell-escape'

or the single one: pdf-engine-opt: '-shell-escape'

---

🎀 Citation rendering 🎀

• --citeproc → citeproc: true
• --bibliography logic.bib → bibliography: logic.bib
• --csl ieee.csl → csl: ieee.csl
• --citation-abbreviations ab.json → citation-abbreviations: ab.json
• --natbib → cite-method: natbib
• --biblatex → cite-method: biblatex

okay so listen — cite-method can be citeproc, natbib, or biblatex. this only affects LaTeX output, mmkay? and if your whole goal is to have citeproc format your citations, you should also set citeproc: true.

And if you need to control when the citeproc processing happens compared to your other filters? then don't do it this way — instead put citeproc in your list of filters (peep the Reader options section for that one). 🫶

---

🌈 Math rendering in HTML 🌈

• --math-method=mathjax →

math-method:
  method: mathjax

• --math-method=webtex:URL →

math-method:
  method: webtex
  url: mathjax

• --math-method=mathml →

math-method:
  method: mathml

If the command line option takes a URL argument, you can add an url: field to math-method:. easy ✨

---

🦋 Options for wrapper scripts 🦋

• --dump-args → dump-args: true
• --ignore-args → ignore-args: true

---

👑 TEMPLATES 👑

So when you use the -s/--standalone option, pandoc pulls up a template to slap on the header and footer material — the stuff you need for a document that can stand on its own two feet. 💅

Wanna see the default template it's using? Just type:

pandoc -D *FORMAT*

…where FORMAT is the name of the output format. 💖

You can bring your own custom template with the --template option. You can also override the system default templates for a given output format FORMAT by dropping a file templates/default.*FORMAT* into the user data directory (that's --data-dir, covered up above). 🎀

Exceptions though, bestie, because of course there are exceptions:

• For odt output, customize the default.opendocument template.
• For docx output, customize the default.openxml template.
• For pdf output, customize the default.latex template (or the default.context template if you use -t context, or the default.ms template if you use -t ms, or the default.html template if you use -t html).
• pptx has no template. 💀

Also note that docx, odt, and pptx output can also be customized with --reference-doc. Here's the vibe split: use a reference doc to adjust the styles inside your document; use a template to handle variable interpolation and customize the presentation of metadata, where the table of contents sits, boilerplate text, etc. 🌸

Templates contain variables, which let you inject arbitrary info at any point in the file. Set them at the command line with the -V/--variable option. If a variable isn't set, pandoc goes looking for that key in the document's metadata — which you can set with either YAML metadata blocks or the -M/--metadata option. On top of that, pandoc gives some variables default values. Peep the Variables section below for the full list of variables used in pandoc's default templates. 👑

If you're rolling with custom templates, you might need to revise them as pandoc changes. Our recommendation: track the changes in the default templates and modify your custom ones to match. Easiest way is to fork the pandoc-templates repository and merge in changes after each pandoc release. fr fr, do this. 🫶

---

🧚 Template syntax 🧚

💜 Comments

Anything sitting between the sequence $-- and the end of the line gets treated as a comment and left out of the output. gone, not sorry. 💅

💜 Delimiters

To mark variables and control structures in your template, you can use either $...$ or ${...} as delimiters. You can mix the styles in the same template, no problem — but the opening and closing delimiter must match in each case. 🔥

The opening delimiter can be followed by one or more spaces or tabs, which get ignored. The closing delimiter can be preceded by one or more spaces or tabs, also ignored. so chill about your spacing.

Wanna put a literal $ in the document? Use $$.

💜 Interpolated variables

A slot for an interpolated variable is a variable name wrapped in matched delimiters. Variable names must begin with a letter and can contain letters, numbers, _, -, and ..

Important: the keywords it, if, else, endif, for, sep, and endfor can't be used as variable names. 🙈

Examples, bestie:

$foo$
$foo.bar.baz$
$foo_bar.baz-bim$
$ foo $
${foo}
${foo.bar.baz}
${foo_bar.baz-bim}
${ foo }

Variable names with periods are how you get at structured variable values. So for example, employee.salary returns the value of the salary field of the object that is the value of the employee field. iykyk ✨

• If the value of the variable is a simple value, it gets rendered verbatim. (Note that no escaping is done here — the assumption is that the calling program will escape the strings appropriately for the output format.)
• If the value is a list, the values get concatenated.
• If the value is a map, the string true will be rendered.
• Every other value gets rendered as the empty string.

💜 Conditionals

A conditional begins with if(variable) (wrapped in matched delimiters) and ends with endif (wrapped in matched delimiters). It can optionally contain an else (also wrapped in matched delimiters). The if section is what gets used when variable has a true value — otherwise the else section is used (if it's there).

These values count as true, no cap:

• any map
• any array containing at least one true value
• any nonempty string
• boolean True

okay so quick lil thing you NEED to know before we go any further, bestie: in YAML metadata (and any metadata you pass on the command line with -M/--metadata), an unquoted true or false gets read as a Boolean value. But a variable you set with -V/--variable on the command line will always get a string value. So a conditional like if(foo) fires if you write -V foo=false, but it does NOT fire if you write -M foo=false. subtle but it will absolutely get you, so note it 💅

Examples, in all their glory:

$if(foo)$bar$endif$

$if(foo)$
  $foo$
$endif$

$if(foo)$
part one
$else$
part two
$endif$

${if(foo)}bar${endif}

${if(foo)}
  ${foo}
${endif}

${if(foo)}
${ foo.bar }
${else}
no foo!
${endif}

And the keyword elseif exists so you can de-spaghetti your deeply nested conditionals 🍝✨

$if(foo)$
XXX
$elseif(bar)$
YYY
$else$
ZZZ
$endif$

---

✨ For loops ✨

A for loop starts with for(variable) (wrapped in matched delimiters) and ends with endfor (also wrapped in matched delimiters). Here's the vibe check on what happens inside:

• If variable is an array, the stuff inside the loop gets evaluated over and over, with variable set to each value of the array in turn, and then concatenated. 🎀
• If variable is a map, the material inside will be set to the map.
• If the value of the associated variable is neither an array nor a map, a single iteration is performed on its value.

Examples, serve:

$for(foo)$$foo$$sep$, $endfor$

$for(foo)$
  - $foo.last$, $foo.first$
$endfor$

${ for(foo.bar) }
  - ${ foo.bar.last }, ${ foo.bar.first }
${ endfor }

$for(mymap)$
$it.name$: $it.office$
$endfor$

You can optionally drop a separator between consecutive values with sep (wrapped in matched delimiters, obviously). Everything between sep and the endfor is the separator. iconic.

${ for(foo) }${ foo }${ sep }, ${ endfor }

And instead of naming variable inside the loop, you can use the special anaphoric keyword it:

${ for(foo.bar) }
  - ${ it.last }, ${ it.first }
${ endfor }

---

🦋 Partials 🦋

Partials are sub-templates stored in different files, and you pull one in by writing its name followed by (), like:

${ styles() }

Partials get looked for in the directory holding the main template. If the file name has no extension, it's assumed to have the same extension as the main template. And when you call the partial, you can totally use the full name with the file extension included:

${ styles.html() }

(If a partial isn't found in the directory of the template, and the template path was given as a relative path, it'll also be sought in the templates subdirectory of the user data directory.) sneaky but helpful fr fr

Partials can optionally be applied to variables using a colon:

${ date:fancy() }

${ articles:bibentry() }

If articles is an array, this iterates over its values, applying the partial bibentry() to each one. So that second example up there is equivalent to:

${ for(articles) }
${ it:bibentry() }
${ endfor }

Heads up 💖 the anaphoric keyword it must be used when you're iterating over partials. In the examples above, the bibentry partial should contain it.title (and so on) rather than articles.title.

Final newlines are omitted from included partials. And yes — partials may include other partials. it's giving nesting dolls 🪆

A separator between values of an array can be specified in square brackets, immediately after the variable name or partial:

${months[, ]}

${articles:bibentry()[; ]}

The separator here is literal, and (unlike with sep in an explicit for loop) it cannot contain interpolated variables or other template directives. no cap.

---

🌸 Nesting 🌸

To make sure content is "nested," meaning subsequent lines get indented, use the ^ directive:

$item.number$  $^$$item.description$ ($item.price$)

In this example, if item.description has multiple lines, they'll all be indented to line up with the first line:

00123  A fine bottle of 18-year old
      Oban whiskey. ($148)

To nest multiple lines to the same level, align them with the ^ directive in the template. For example:

$item.number$  $^$$item.description$ ($item.price$)
               (Available til $item.sellby$.)

will produce:

00123  A fine bottle of 18-year old
      Oban whiskey. ($148)
      (Available til March 30, 2020.)

And if a variable occurs by itself on a line — preceded by whitespace, and not followed by any further text or directives on that same line — and that variable's value contains multiple lines, it'll be nested automatically. the template just knows ✨

---

🎀 Breakable spaces 🎀

Normally, spaces in the template itself (not the values of the interpolated variables) are not breakable. But you can make them breakable in part of the template with the ~ keyword (which is ended with another ~). look at her go:

$~$This long line may break if the document is rendered
with a short line length.$~$

---

💅 Pipes 💅

A pipe transforms the value of a variable or a partial. Pipes are specified using a slash (/) between the variable name (or partial) and the pipe name. Example time:

$for(name)$
$name/uppercase$
$endfor$

$for(metadata/pairs)$
- $it.key$: $it.value$
$endfor$

$employee:name()/uppercase$

And pipes may be chained, which is so fun actually:

$for(employees/pairs)$
$it.key/alpha/uppercase$. $it.name$
$endfor$

Some pipes take parameters too:

|----------------------|------------|
$for(employee)$
$it.name.first/uppercase/left 20 "| "$$it.name.salary/right 10 " | " " |"$
$endfor$
|----------------------|------------|

Currently the following pipes are predefined (collect them all, bestie):

• pairs: Converts a map or array to an array of maps, each with key and value fields. If the original value was an array, the key will be the array index, starting with 1.
• uppercase: Converts text to uppercase.
• lowercase: Converts text to lowercase.
• length: Returns the length of the value: number of characters for a textual value, number of elements for a map or array.
• reverse: Reverses a textual value or array, and has no effect on other values.
• first: Returns the first value of an array, if applied to a non-empty array; otherwise returns the original value.
• last: Returns the last value of an array, if applied to a non-empty array; otherwise returns the original value.
• rest: Returns all but the first value of an array, if applied to a non-empty array; otherwise returns the original value.
• allbutlast: Returns all but the last value of an array, if applied to a non-empty array; otherwise returns the original value.
• chomp: Removes trailing newlines (and breakable space).
• nowrap: Disables line wrapping on breakable spaces.
• alpha: Converts textual values that can be read as an integer into lowercase alphabetic characters a..z (mod 26). This can be used to get lettered enumeration from array indices. To get uppercase letters, chain with uppercase.
• roman: Converts textual values that can be read as an integer into lowercase roman numerals. This can be used to get lettered enumeration from array indices. To get uppercase roman, chain with uppercase.
• left n "leftborder" "rightborder": Renders a textual value in a block of width n, aligned to the left, with an optional left and right border. Has no effect on other values. This can be used to align material in tables. Widths are positive integers indicating the number of characters. Borders are strings inside double quotes; literal " and \ characters must be backslash-escaped.
• right n "leftborder" "rightborder": Renders a textual value in a block of width n, aligned to the right, and has no effect on other values.
• center n "leftborder" "rightborder": Renders a textual value in a block of width n, aligned to the center, and has no effect on other values.

---

👑 Variables 👑

🌈 Metadata variables 🌈

• title, author, date — allow identification of basic aspects of the document. Included in PDF metadata through LaTeX and ConTeXt. These can be set through a pandoc title block, which allows for multiple authors, or through a YAML metadata block:

---
author:
- Aristotle
- Peter Abelard
...

Note that if you just want to set PDF or HTML metadata, without including a title block in the document itself, you can set the title-meta, author-meta, and date-meta variables. (By default these are set automatically, based on title, author, and date.) The page title in HTML is set by pagetitle, which is equal to title by default.

• subtitle — document subtitle, included in HTML, EPUB, LaTeX, ConTeXt, and docx documents
• abstract — document summary, included in HTML, LaTeX, ConTeXt, AsciiDoc, and docx documents
• abstract-title — title of abstract, currently used only in HTML, EPUB, docx, and Typst. This will be set automatically to a localized value, depending on lang, but can be manually overridden.
• keywords — list of keywords to be included in HTML, PDF, ODT, pptx, docx and AsciiDoc metadata; repeat as for author, above
• subject — document subject, included in ODT, PDF, docx, EPUB, and pptx metadata
• description — document description, included in ODT, docx and pptx metadata. Some applications show this as Comments metadata.
• category — document category, included in docx and pptx metadata

Additionally, any root-level string metadata that isn't included in ODT, docx or pptx metadata gets added as a custom property. This YAML metadata block for instance:

---
title:  'This is the title'
subtitle: "This is the subtitle"
author:
- Author One
- Author Two
description: |
    This is a long
    description.

    It consists of two paragraphs
...

will include title, author and description as standard document properties and subtitle as a custom property when converting to docx, ODT or pptx. ✨

💬 Language variables

• lang — this one tells pandoc what the main language of your doc is, using IETF language tags (so we're following the BCP 47 standard, bestie, we stay ✨compliant✨). Stuff like en or en-GB. The Language subtag lookup tool can look up or verify these tags for you. It affects most formats, and it's what controls hyphenation in PDF output when you're using LaTeX (via babel and polyglossia) or ConTeXt. And the cute part: you can switch languages mid-doc by using native pandoc Divs and Spans with the lang attribute:

---
lang: en-GB
...

Text in the main document language (British English).

::: {lang=fr-CA}
> Cette citation est écrite en français canadien.
:::

More text in English. ['Zitat auf Deutsch.']{lang=de}

• dir — the base script direction, either rtl (right-to-left) or ltr (left-to-right). For bidirectional documents, native pandoc spans and divs with the dir attribute (value rtl or ltr) can be used to override the base direction in some output formats. This may not always be necessary if the final renderer (e.g. the browser, when generating HTML) supports the Unicode Bidirectional Algorithm. When using LaTeX for bidirectional documents, only the xelatex engine is fully supported (use --pdf-engine=xelatex). 👑

---

✨ Variables for HTML

• document-css — turns on inclusion of most of the CSS in the styles.html partial (go peek at it with pandoc --print-default-data-file=templates/styles.html). Unless you use --css, this variable is set to true by default. You can disable it with e.g. pandoc -M document-css=false. 🎀
• mainfont — sets the CSS font-family property on the html element.
• fontsize — sets the base CSS font-size, which you'd usually set to e.g. 20px, but it also accepts pt (12pt = 16px in most browsers).
• fontcolor — sets the CSS color property on the html element.
• linkcolor — sets the CSS color property on all links.
• monofont — sets the CSS font-family property on code elements.
• monobackgroundcolor — sets the CSS background-color property on code elements and adds extra padding.
• linestretch — sets the CSS line-height property on the html element, which is preferred to be unitless.
• maxwidth — sets the CSS max-width property (default is 36em).
• backgroundcolor — sets the CSS background-color property on the html element.
• margin-left, margin-right, margin-top, margin-bottom — sets the corresponding CSS padding properties on the body element.

To override or extend some CSS for just one document, include for example:

---
header-includes: |
  <style>
  blockquote {
    font-style: italic;
  }
  tr.even {
    background-color: #f0f0f0;
  }
  td, th {
    padding: 0.5em 2em 0.5em 0.5em;
  }
  tbody {
    border-bottom: none;
  }
  </style>
---

---

🧮 Variables for HTML math

• classoption — when you're using --math-method=katex, you can render display math equations flush left using YAML metadata or with -M classoption=fleqn. ship it 💖

---

🌈 Variables for HTML slides

These affect HTML output when producing slide shows with pandoc.

• institute — author affiliations: can be a list when there are multiple authors
• revealjs-url — base URL for reveal.js documents (defaults to https://unpkg.com/reveal.js@^5)
• s5-url — base URL for S5 documents (defaults to s5/default)
• slidy-url — base URL for Slidy documents (defaults to https://www.w3.org/Talks/Tools/Slidy2)
• slideous-url — base URL for Slideous documents (defaults to slideous)
• title-slide-attributes — additional attributes for the title slide of reveal.js slide shows. See background in reveal.js, beamer, and pptx for an example.
• highlightjs-theme — highlight.js theme for code highlighting when using --syntax-highlighting=idiomatic with reveal.js (defaults to monokai). See the highlight.js demo page for available themes.

All reveal.js configuration options are available as variables. To turn off boolean flags that default to true in reveal.js, use 0. 🦋

---

🎬 Variables for Beamer slides

These variables change the appearance of PDF slides using beamer.

• aspectratio — slide aspect ratio (43 for 4:3 [default], 169 for 16:9, 1610 for 16:10, 149 for 14:9, 141 for 1.41:1, 54 for 5:4, 32 for 3:2)
• beameroption — add extra beamer option with \setbeameroption{}
• institute — author affiliations: can be a list when there are multiple authors
• logo — logo image for slides
• logooptions — options for logo image (e.g., width, height)
• navigation — controls navigation symbols (default is empty for no navigation symbols; other valid values are frame, vertical, and horizontal)
• section-titles — enables "title pages" for new sections (default is true)
• theme, colortheme, fonttheme, innertheme, outertheme — beamer themes
• themeoptions, colorthemeoptions, fontthemeoptions, innerthemeoptions, outerthemeoptions — options for LaTeX beamer themes (lists)
• titlegraphic — image for title slide: can be a list
• titlegraphicoptions — options for title slide image (e.g., width, height)
• shorttitle, shortsubtitle, shortauthor, shortinstitute, shortdate — some beamer themes use short versions of the title, subtitle, author, institute, date

---

📊 Variables for PowerPoint

These variables control the visual aspects of a slide show that are not easily controlled via templates.

• monofont — font to use for code.

---

📚 Variables for LaTeX

Pandoc uses these variables when creating a PDF with a LaTeX engine.

---

🧱 Layout

• block-headings — make \paragraph and \subparagraph (fourth- and fifth-level headings, or fifth- and sixth-level with book classes) free-standing rather than run-in; it requires further formatting to distinguish from \subsubsection (third- or fourth-level headings). And hey, instead of using this option, KOMA-Script can adjust headings more extensively:

---
documentclass: scrartcl
header-includes: |
  \RedeclareSectionCommand[
    beforeskip=-10pt plus -2pt minus -1pt,
    afterskip=1sp plus -1sp minus 1sp,
    font=\normalfont\itshape]{paragraph}
  \RedeclareSectionCommand[
    beforeskip=-10pt plus -2pt minus -1pt,
    afterskip=1sp plus -1sp minus 1sp,
    font=\normalfont\scshape,
    indent=0pt]{subparagraph}
...

• classoption — option for document class, e.g. oneside; repeat for multiple options:

---
classoption:
- twocolumn
- landscape
...

• documentclass — document class: usually one of the standard classes, article, book, and report; the KOMA-Script equivalents, scrartcl, scrbook, and scrreprt, which default to smaller margins; or memoir
• geometry — option for geometry package, e.g. margin=1in; repeat for multiple options:

---
geometry:
- top=30mm
- left=20mm
- heightrounded
...

• shorthands — Enable language-specific shorthands when loading babel. (By default, pandoc includes shorthands=off when loading babel, disabling language-specific shorthands.)
• hyperrefoptions — option for hyperref package, e.g. linktoc=all; repeat for multiple options:

---
hyperrefoptions:
- linktoc=all
- pdfwindowui
- pdfpagemode=FullScreen
...

• indent — if true, pandoc will use document class settings for indentation (the default LaTeX template otherwise removes indentation and adds space between paragraphs)
• linestretch — adjusts line spacing using the setspace package, e.g. 1.25, 1.5
• margin-left, margin-right, margin-top, margin-bottom — sets margins if geometry is not used (otherwise geometry overrides these)
• pagestyle — control \pagestyle{}: the default article class supports plain (default), empty (no running heads or page numbers), and headings (section titles in running heads)
• papersize — paper size, e.g. letter, a4
• secnumdepth — numbering depth for sections (with --number-sections option or numbersections variable)
• beamerarticle — produce an article from Beamer slides. Note: if you set this variable, you must specify the beamer writer but use the default LaTeX template: for example, pandoc -Vbeamerarticle -t beamer --template default.latex.
• handout — produce a handout version of Beamer slides (with overlays condensed into single slides)
• csquotes — load csquotes package and use \enquote or \enquote* for quoted text.
• csquotesoptions — options to use for csquotes package (repeat for multiple options).
• babeloptions — options to pass to the babel package (may be repeated for multiple options). This defaults to provide=* if the main language isn't a European language written with Latin or Cyrillic script or Vietnamese. Most users will not need to adjust the default setting. 🫶

---

🔤 Fonts

• fontenc — allows font encoding to be specified through fontenc package (with pdflatex); default is T1 (see LaTeX font encodings guide)
• fontfamily — font package for use with pdflatex: TeX Live includes many options, documented in the LaTeX Font Catalogue. The default is Latin Modern.
• fontfamilyoptions — options for package used as fontfamily; repeat for multiple options. For example, to use the Libertine font with proportional lowercase (old-style) figures through the libertinus package:

---
fontfamily: libertinus
fontfamilyoptions:
- osf
- p
...

• fontsize — the size of your body text, bestie. The standard classes are like, only here for 10pt, 11pt, and 12pt. Wanna go rogue with another size? Set documentclass to one of the KOMA-Script classes, like scrartcl or scrbook. Slay. 💅
• mainfont, sansfont, monofont, mathfont, CJKmainfont, CJKsansfont, CJKmonofont — font families for when you're rolling with xelatex or lualatex: just drop the name of any system font and it goes through the fontspec package. And CJKmainfont? It pulls in the xecjk package if you're using xelatex, or the luatexja package if you're using lualatex. ✨
• mainfontoptions, sansfontoptions, monofontoptions, mathfontoptions, CJKoptions, luatexjapresetoptions — the options that ride along with mainfont, sansfont, monofont, mathfont, CJKmainfont in xelatex and lualatex. You can use literally anything fontspec offers, and repeat them for multiple options. For example, to use the TeX Gyre version of Palatino with lowercase figures:

---
mainfont: TeX Gyre Pagella
mainfontoptions:
- Numbers=Lowercase
- Numbers=Proportional
...

• mainfontfallback, sansfontfallback, monofontfallback — the fonts to try when a glyph just isn't found in mainfont, sansfont, or monofont respectively. These are lists, okay? The font name has to be followed by a colon and optionally a set of options, like this:

---
mainfontfallback:
  - "FreeSans:"
  - "NotoColorEmoji:mode=harf"
...
Font fallbacks currently only work with lualatex.

• babelfonts — a map of Babel language names (think chinese) to the font you want used with that language:

---
babelfonts:
  chinese-hant: "Noto Serif CJK TC"
  russian: "Noto Serif"
...

• microtypeoptions — options to pass to the microtype package. easy.

✨ Links ✨

• colorlinks — adds color to your link text. It's automatically enabled if any of linkcolor, filecolor, citecolor, urlcolor, or toccolor are set, so it kinda turns itself on. 💖
• boxlinks — puts a visible box around links. Highkey does nothing if colorlinks is set.
• linkcolor, filecolor, citecolor, urlcolor, toccolor — the color for internal links, external links, citation links, linked URLs, and links in the table of contents, respectively. It uses the options xcolor allows, including the dvipsnames, svgnames, and x11names lists. 🌈
• links-as-notes — prints links as footnotes instead. it's giving academic.
• urlstyle — the style for URLs (e.g., tt, rm, sf, and, the default, same).

🎀 Front matter 🎀

• lof, lot — include list of figures, list of tables (can also be set using --lof/--list-of-figures, --lot/--list-of-tables).
• thanks — contents of acknowledgments footnote after document title.
• toc — include table of contents (can also be set using --toc/--table-of-contents).
• toc-depth — level of section to include in table of contents.

📚 BibLaTeX Bibliographies 📚

These variables function when using BibLaTeX for citation rendering.

• biblatexoptions — list of options for biblatex.
• biblio-style — bibliography style, when used with --natbib and --biblatex.
• biblio-title — bibliography title, when used with --natbib and --biblatex.
• bibliography — bibliography to use for resolving references.
• natbiboptions — list of options for natbib.

🌈 Other 🌈

• pdf-trailer-id — the PDF trailer ID; it must be two PDF byte strings if you set it, conventionally with 16 bytes each. E.g., <00112233445566778899aabbccddeeff> <00112233445566778899aabbccddeeff>. See the section on reproducible builds.
• pdfstandard — PDF standard(s) for the document, e.g. ua-2, a-4f. Supports PDF/A, PDF/X, and PDF/UA variants. Requires LuaLaTeX and LaTeX 2023+. Repeat for multiple standards:

---
pdfstandard:
- ua-2
- a-4f
...

---

🌸 Variables for ConTeXt 🌸

Pandoc uses these variables when creating a PDF with ConTeXt.

• fontsize — font size for body text (e.g. 10pt, 12pt).
• headertext, footertext — text to be placed in running header or footer (see ConTeXt Headers and Footers); repeat up to four times for different placement.
• indenting — controls indentation of paragraphs, e.g. yes,small,next (see ConTeXt Indentation); repeat for multiple options.
• interlinespace — adjusts line spacing, e.g. 4ex (using setupinterlinespace); repeat for multiple options.
• layout — options for page margins and text arrangement (see ConTeXt Layout); repeat for multiple options.
• linkcolor, contrastcolor — color for links outside and inside a page, e.g. red, blue (see ConTeXt Color).
• linkstyle — typeface style for links, e.g. normal, bold, slanted, boldslanted, type, cap, small.
• lof, lot — include list of figures, list of tables.
• mainfont, sansfont, monofont, mathfont — font families: take the name of any system font (see ConTeXt Font Switching).
• mainfontfallback, sansfontfallback, monofontfallback — list of fonts to try, in order, if a glyph is not found in the main font. Use \definefallbackfamily-compatible font name syntax. Emoji fonts are unsupported. 😭
• margin-left, margin-right, margin-top, margin-bottom — sets margins, if layout is not used (otherwise layout overrides these).
• pagenumbering — page number style and location (using setuppagenumbering); repeat for multiple options.
• papersize — paper size, e.g. letter, A4, landscape (see ConTeXt Paper Setup); repeat for multiple options.
• pdfa — adds to the preamble the setup necessary to generate PDF/A of the type specified, e.g. 1a:2005, 2a. If no type is specified (i.e. the value is set to True, by e.g. --metadata=pdfa or pdfa: true in a YAML metadata block), 1b:2005 will be used as default, for reasons of backwards compatibility. Using --variable=pdfa without specified value is not supported. To successfully generate PDF/A the required ICC color profiles have to be available and the content and all included files (such as images) have to be standard-conforming. The ICC profiles and output intent may be specified using the variables pdfaiccprofile and pdfaintent. See also ConTeXt PDFA for more details.
• pdfaiccprofile — when used in conjunction with pdfa, specifies the ICC profile to use in the PDF, e.g. default.cmyk. If left unspecified, sRGB.icc is used as default. May be repeated to include multiple profiles. Note that the profiles have to be available on the system. They can be obtained from ConTeXt ICC Profiles.
• pdfaintent — when used in conjunction with pdfa, specifies the output intent for the colors, e.g. ISO coated v2 300% (ECI). If left unspecified, sRGB IEC61966-2.1 is used as default.
• toc — include table of contents (can also be set using --toc/--table-of-contents).
• urlstyle — typeface style for links without link text, e.g. normal, bold, slanted, boldslanted, type, cap, small.
• whitespace — spacing between paragraphs, e.g. none, small (using setupwhitespace).
• includesource — include all source documents as file attachments in the PDF file.

🦋 Variables for wkhtmltopdf 🦋

Pandoc uses these variables when creating a PDF with wkhtmltopdf. The --css option also affects the output. iykyk.

• footer-html, header-html — add information to the header and footer.
• margin-left, margin-right, margin-top, margin-bottom — set the page margins.
• papersize — sets the PDF paper size.

💜 Variables for man pages 💜

• adjusting — adjusts text to left (l), right (r), center (c), or both (b) margins.
• footer — footer in man pages.
• header — header in man pages.
• section — section number in man pages.

🧚 Variables for Texinfo 🧚

• version — version of software (used in title and title page).
• filename — name of info file to be generated (defaults to a name based on the texi filename).

⭐ Variables for Typst ⭐

• template — Typst template to use (relative path only).
• margin — A dictionary with the fields defined in the Typst documentation: x, y, top, bottom, left, right.
• papersize — Paper size: a4, us-letter, etc.
• mainfont — Name of system font to use for the main font.
• fontsize — Font size (e.g., 12pt).
• section-numbering — Schema to use for numbering sections, e.g. 1.A.1.
• page-numbering — Schema to use for numbering pages, e.g. 1 or i, or an empty string to omit page numbering.
• columns — Number of columns for body text.
• thanks — contents of acknowledgments footnote after document title. (Note: as of typst 0.15, this does not work properly, due to an upstream issue; see jgm/pandoc#11807.) 💀
• mathfont, codefont — Name of system font to use for math and code, respectively.
• linestretch — adjusts line spacing, e.g. 1.25, 1.5.
• linkcolor, filecolor, citecolor — color for external links, internal links, and citation links, respectively: expects a hexadecimal color code.

🍓 Variables for ms 🍓

• fontfamily — A (Avant Garde), B (Bookman), C (Helvetica), HN (Helvetica Narrow), P (Palatino), or T (Times New Roman). This setting does not affect source code, which is always displayed using monospace Courier. These built-in fonts are limited in their coverage of characters. Additional fonts may be installed using the script install-font.sh provided by Peter Schaffter and documented in detail on his web site.
• indent — paragraph indent (e.g. 2m).
• lineheight — line height (e.g. 12p).
• pointsize — point size (e.g. 10p).

🫶 Variables for RTF 🫶

• fontsize — font size for body text, e.g. 12pt or 12.5pt.

👑 Variables set automatically 👑

Pandoc sets these variables automatically in response to options or document contents; users can also modify them. These vary depending on the output format, and include the following:

• body — body of document.
• date-meta — the date variable but glowified into ISO 8601 YYYY-MM-DD format. It gets dropped into all the HTML-based formats (dzslides, epub, html, html4, html5, revealjs, s5, slideous, slidy). And the date formats it actually recognizes are: mm/dd/yyyy, mm/dd/yy, yyyy-mm-dd (ISO 8601, the iconique one), dd MM yyyy (so like either 02 Apr 2018 or 02 April 2018), MM dd, yyyy (like Apr. 02, 2018 or April 02, 2018), and yyyy[mm[dd]] (so 20180402, 201804 or just 2018). 💅
• header-includes — whatever you pass via -H/--include-in-header (can hold multiple values, no cap ✨)
• include-before — whatever you pass via -B/--include-before-body (also can have multiple values)
• include-after — whatever you pass via -A/--include-after-body (multiple values again, bestie)
• meta-json — a straight-up JSON representation of the whole document's metadata. Field values get transformed into whatever output format you picked.
• numbersections — non-null value if you specified -N/--number-sections
• sourcefile, outputfile — the source and destination filenames, exactly as given on the command line. sourcefile can also be a list (if your input came from multiple files), or empty (if input is coming from stdin, the drama of it all). You can use this lil snippet in your template to tell the difference:

$if(sourcefile)$
$for(sourcefile)$
$sourcefile$
$endfor$
$else$
(stdin)
$endif$

Same energy for outputfile — it can be - if output goes to the terminal. And if you need absolute paths? Use e.g. $curdir$/$sourcefile$. slay.

• pdf-engine — the name of the PDF engine, if you gave one with --pdf-engine, or the default engine for the format when PDF output is requested.
• curdir — the working directory you ran pandoc from.
• pandoc-version — the pandoc version, simple as that.
• toc — non-null value if you specified --toc/--table-of-contents
• toc-title — the title of the table of contents (this only works with EPUB, HTML, revealjs, opendocument, odt, docx, pptx, beamer, LaTeX). ⚠️ Heads up: in docx and pptx a custom toc-title will be picked up from metadata, but it cannot be set as a variable. iykyk.

---

✨ EXTENSIONS ✨

Okay bestie, some of the readers and writers let you tweak their whole behavior by turning various extensions on or off. It's giving customizable. 🎀

You turn an extension on by adding +EXTENSION to the format name, and off by adding -EXTENSION. So for example, --from markdown_strict+footnotes is strict Markdown with footnotes enabled, while --from markdown-footnotes-pipe_tables is pandoc's Markdown without footnotes or pipe tables. See how the + and - do the thing? 💅

The Markdown reader and writer make by far the most use of extensions — like, not even close. So the ones only they use get covered in the Pandoc's Markdown section below (peep Markdown variants for commonmark and gfm). Down here in the following, we're covering the extensions that also work for other formats.

Also, quick note: Markdown extensions added to the ipynb format affect Markdown cells in Jupyter notebooks (as do command-line options like --markdown-headings). 🦋

Typography

Extension: smart

Interprets straight quotes as curly quotes, --- as em-dashes, -- as en-dashes, and ... as ellipses. Nonbreaking spaces get inserted after certain abbreviations, such as "Mr." ✨

This extension can be enabled/disabled for the following formats:

• input formats — markdown, commonmark, latex, mediawiki, org, rst, twiki, html
• output formats — markdown, latex, context, org, rst
• enabled by default in — markdown, latex, context (both input and output)

📌 Note: If you're writing Markdown, then the smart extension has the reverse effect — what would have been curly quotes comes out straight. Ironic, iconic.

In LaTeX, smart means to use the standard TeX ligatures for quotation marks ( and `''` for double quotes, `` and' for single quotes) and dashes (--for en-dash and---for em-dash). Ifsmartis disabled, then when reading LaTeX, pandoc will parse these characters literally. And when writing LaTeX, enablingsmarttells pandoc to use the ligatures when possible; ifsmart` is disabled, pandoc uses unicode quotation mark and dash characters instead. 🤓

Headings and sections

Extension: auto_identifiers

A heading without an explicitly specified identifier gets automatically assigned a unique identifier based on the heading text. So your sections literally name themselves. main character energy fr fr. 👑

This extension can be enabled/disabled for the following formats:

• input formats — markdown, latex, rst, mediawiki, textile, man, typst
• output formats — markdown, muse
• enabled by default in — markdown, muse, man

The default algorithm used to derive the identifier from the heading text is:

1. Remove all formatting, links, etc.
2. Remove all footnotes.
3. Remove all non-alphanumeric characters, except underscores, hyphens, and periods.
4. Replace all spaces and newlines with hyphens.
5. Convert all alphabetic characters to lowercase.
6. Remove everything up to the first letter (identifiers may not begin with a number or punctuation mark).
7. If nothing is left after this, use the identifier section.

So like, for example:

• Heading: Heading identifiers in HTML → Identifier: heading-identifiers-in-html
• Heading: Maître d'hôtel → Identifier: maître-dhôtel
• Heading: *Dogs*?--in *my* house? → Identifier: dogs--in-my-house
• Heading: [HTML], [S5], or [RTF]? → Identifier: html-s5-or-rtf
• Heading: 3. Applications → Identifier: applications
• Heading: 33 → Identifier: section

These rules should, in most cases, let you figure out the identifier from the heading text all on your own. The exception is when several headings have the same text — in that case, the first gets an identifier as described above; the second gets the same identifier with -1 appended; the third with -2; and so on. So they never actually collide. 💖

(However, a different algorithm is used if gfm_auto_identifiers is enabled; see below.)

These identifiers are used to provide link targets in the table of contents generated by the --toc|--table-of-contents option. They also make it super easy to provide links from one section of a document to another. A link to this section, for example, might look like this:

See the section on
[heading identifiers](#heading-identifiers-in-html-latex-and-context).

Note, however, that this method of providing links to sections works only in HTML, LaTeX, and ConTeXt formats. 🙈

If the --section-divs option is specified, then each section gets wrapped in a section (or a div, if html4 was specified), and the identifier gets attached to the enclosing <section> (or <div>) tag rather than the heading itself. This lets entire sections be manipulated using JavaScript or treated differently in CSS. ✨

Extension: ascii_identifiers

Causes the identifiers produced by auto_identifiers to be pure ASCII. Accents get stripped off of accented Latin letters, and non-Latin letters are omitted. 🦋

Extension: gfm_auto_identifiers

Changes the algorithm used by auto_identifiers to conform to GitHub's method. Spaces are converted to dashes (-), uppercase characters to lowercase characters, and punctuation characters other than - and _ are removed. Emojis are replaced by their names. 😭 so GitHub-coded.

Math Input

The extensions tex_math_dollars, tex_math_gfm, tex_math_single_backslash, and tex_math_double_backslash are described in the section about Pandoc's Markdown.

However, they can also be used with HTML input. This is handy for reading web pages formatted using MathJax, for example. 🤓✨

Raw HTML/TeX

The following extensions are described in more detail in their respective sections of Pandoc's Markdown:

• raw_html allows HTML elements which are not representable in pandoc's AST to be parsed as raw HTML. By default, this is disabled for HTML input.
• raw_tex allows raw LaTeX, TeX, and ConTeXt to be included in a document. This extension can be enabled/disabled for the following formats (in addition to markdown):
  • input formats — latex, textile, html (environments, \ref, and \eqref only), ipynb
  • output formats — textile, commonmark

Note: as applied to ipynb, raw_html and raw_tex affect not only raw TeX in Markdown cells, but data with mime type text/html in output cells. Since the ipynb reader attempts to preserve the richest possible outputs when several options are given, you'll get best results if you disable raw_html and raw_tex when converting to formats like docx which don't allow raw html or tex. 💅

• native_divs causes HTML div elements to be parsed as native pandoc Div blocks. If you want them to be parsed as raw HTML, use -f html-native_divs+raw_html.
• native_spans causes HTML span elements to be parsed as native pandoc Span inlines. If you want them to be parsed as raw HTML, use -f html-native_spans+raw_html. And if you want to drop all divs and spans when converting HTML to Markdown, you can use pandoc -f html-native_divs-native_spans -t markdown. 🎀

Literate Haskell support

Extension: literate_haskell

Treat the document as literate Haskell source. 🧚

This extension can be enabled/disabled for the following formats:

• input formats — markdown, rst, latex
• output formats — markdown, rst, latex, html

If you append +lhs (or +literate_haskell) to one of the formats above, pandoc will treat the document as literate Haskell source. This means that:

• In Markdown input, "bird track" sections will be parsed as Haskell code rather than block quotations. Text between \begin{code} and \end{code} will also be treated as Haskell code. For ATX-style headings the character = will be used instead of #.
• In Markdown output, code blocks with classes haskell and literate will be rendered using bird tracks, and block quotations will be indented one space, so they will not be treated as Haskell code. In addition, headings will be rendered setext-style (with underlines) rather than ATX-style (with # characters). (This is because ghc treats # characters in column 1 as introducing line numbers.) 💀
• In restructured text input, "bird track" sections will be parsed as Haskell code.
• In restructured text output, code blocks with class haskell will be rendered using bird tracks.
• In LaTeX input, text in code environments will be parsed as Haskell code.
• In LaTeX output, code blocks with class haskell will be rendered inside code environments.
• In HTML output, code blocks with class haskell will be rendered with class literatehaskell and bird tracks.

Examples:

pandoc -f markdown+lhs -t html

okay so this corner of the extensions universe is the literate Haskell one 🧚 — it reads literate Haskell source formatted with Markdown conventions and writes out plain, ordinary HTML (no bird tracks in sight 💖):

pandoc -f markdown+lhs -t html+lhs

and this one writes HTML with the Haskell code chilling in bird tracks, so it can be copied and pasted as literate Haskell source 🐦✨

heads up bestie: GHC expects the bird tracks in the first column, so indented literate code blocks (like, say, inside an itemized environment) will not be picked up by the Haskell compiler. it's giving "invisible to the compiler" 😭

---

🌈 Other extensions 🌈

Extension: empty_paragraphs

Lets empty paragraphs exist. By default empty paragraphs are just… omitted, gone, not invited 💀

This extension can be enabled/disabled for the following formats:

• input formats: docx, html
• output formats: docx, odt, opendocument, html, latex

Extension: native_numbering

Turns on numbering of figures and tables using counter fields that are specific to whatever the target format is. This extension can be enabled/disabled for the following formats:

• output formats: odt, opendocument, docx

If this extension is not enabled, tables and figures in these formats will not be numbered. periodt. 👑

Extension: xrefs_name

Links to headings, figures and tables inside the document get swapped out for cross-references that use the name or caption of the thing being referenced. The original link text is replaced once the generated document is refreshed. This extension can be combined with xrefs_number, in which case numbers will appear before the name. 🔢➡️🏷️

note: the text in cross-references is only made consistent with the referenced item once the document has been refreshed.

This extension can be enabled/disabled for the following formats:

• output formats: odt, opendocument

Extension: xrefs_number

Links to headings, figures and tables inside the document get substituted with cross-references that use the number of the referenced item. The original link text is discarded — bye 👋. This extension can be combined with xrefs_name, in which case the name or caption numbers will appear after the number.

For xrefs_number to actually be useful, heading numbers must be enabled in the generated document, and table and figure captions must be enabled too, using for example the native_numbering extension.

note: numbers in cross-references are only visible in the final document once it has been refreshed.

This extension can be enabled/disabled for the following formats:

• output formats: odt, opendocument

Extension: styles

When converting from docx, this adds custom-styles attributes for all docx styles, no matter whether pandoc actually understands what those styles mean. Because attributes can't be added directly to paragraphs or text in the pandoc AST, paragraph styles will cause Divs to be created, and character styles will cause Spans to be created, to hold those attributes. (Table styles get added to the Table elements directly.) This extension can be used with docx custom styles. 💅

• input formats: docx

Extension: amuse

In the muse input format, this enables Text::Amuse extensions to Emacs Muse markup.

Extension: raw_markdown

In the ipynb input format, this causes Markdown cells to be included as raw Markdown blocks (allowing lossless round-tripping 🎀) rather than being parsed. Use this only when you are targeting ipynb or a Markdown-based output format.

Extension: citations (typst)

When the citations extension is enabled in typst (as it is by default), typst citations will be parsed as native pandoc citations, and native pandoc citations will be rendered as typst citations. 🔄

Extension: citations (org)

When the citations extension is enabled in org, org-cite and org-ref style citations will be parsed as native pandoc citations, and org-cite citations will be used to render native pandoc citations.

Extension: citations (docx)

When citations is enabled in docx, citations inserted by Zotero or Mendeley or EndNote plugins will be parsed as native pandoc citations. (Otherwise, the formatted citations generated by the bibliographic software will be parsed as regular text — mid, honestly.)

Extension: fancy_lists (org)

Some aspects of Pandoc's Markdown fancy lists are also accepted in org input, mimicking the option org-list-allow-alphabetical in Emacs. As in Org Mode, enabling this extension allows lowercase and uppercase alphabetical markers for ordered lists to be parsed in addition to arabic ones. Note that for Org, this does not include roman numerals or the # placeholder that are enabled by the extension in Pandoc's Markdown. 📝

Extension: element_citations

In the jats output formats, this causes reference items to be replaced with <element-citation> elements. These elements are not influenced by CSL styles, but all information on the item is included in tags. 🏷️

Extension: ntb

In the context output format this enables the use of Natural Tables (TABLE) instead of the default Extreme Tables (xtables). Natural tables allow more fine-grained global customization but come at a performance penalty compared to extreme tables. 💀 slow but fancy.

Extension: smart_quotes (org)

Interpret straight quotes as curly quotes during parsing. When writing Org, then the smart_quotes extension has the reverse effect: what would have been curly quotes comes out straight. ✨

This extension is implied if smart is enabled.

Extension: special_strings (org)

Interpret --- as em-dashes, -- as en-dashes, \- as shy hyphen, and ... as ellipses. —–‐…

This extension is implied if smart is enabled.

Extension: tagging

Enabling this extension with context output will produce markup suitable for the production of tagged PDFs. This includes additional markers for paragraphs and alternative markup for emphasized text. The emphasis-command template variable is set if the extension is enabled. 🏷️📄

---

📜 PANDOC'S MARKDOWN 📜

Pandoc understands an extended and slightly revised version of John Gruber's Markdown syntax. This document explains the syntax, and it points out the differences from original Markdown. Except where noted, these differences can be suppressed by using the markdown_strict format instead of markdown. Extensions can be enabled or disabled to specify the behavior more granularly — they're all described below 👇. See also Extensions above, for extensions that work on other formats too.

🦋 Philosophy 🦋

Markdown is designed to be easy to write and, even more importantly, easy to read:

A Markdown-formatted document should be publishable as-is, as plain text, without looking like it's been marked up with tags or formatting instructions. – John Gruber

This principle has guided pandoc's decisions in finding syntax for tables, footnotes, and other extensions. 💖

There is, however, one respect in which pandoc's aims are different from the original aims of Markdown. Whereas Markdown was originally designed with HTML generation in mind, pandoc is designed for multiple output formats. So, while pandoc allows the embedding of raw HTML, it discourages it, and it provides other, non-HTMLish ways of representing important document elements like definition lists, tables, mathematics, and footnotes. 🌈

Paragraphs

A paragraph is one or more lines of text followed by one or more blank lines. Newlines are treated as spaces, so you can reflow your paragraphs however you like. If you need a hard line break, put two or more spaces at the end of a line. ✌️

Extension: escaped_line_breaks

A backslash followed by a newline is also a hard line break. 📏

note: in multiline and grid table cells, this is the only way to create a hard line break, since trailing spaces in the cells are ignored.

Headings

There are two kinds of headings: Setext and ATX. both iconic, both valid 💅

Setext-style headings

A setext-style heading is a line of text "underlined" with a row of = signs (for a level-one heading) or - signs (for a level-two heading):

A level-one heading
===================

A level-two heading
-------------------

The heading text can contain inline formatting, such as emphasis (see Inline formatting, below).

ATX-style headings

An ATX-style heading consists of one to six # signs and a line of text, optionally followed by any number of # signs. The number of # signs at the beginning of the line is the heading level:

## A level-two heading

### A level-three heading ###

As with setext-style headings, the heading text can contain formatting:

# A level-one heading with a [link](/url) and *emphasis*

Extension: blank_before_header

Original Markdown syntax does not require a blank line before a heading. Pandoc does require this (except, of course, at the beginning of the document 🙃). The reason for the requirement is that it is all too easy for a # to end up at the beginning of a line by accident — maybe through line wrapping. Consider, for example:

I like several of their flavors of ice cream:
#22, for example, and #5.

Extension: space_in_atx_header

Many Markdown implementations do not require a space between the opening #s of an ATX heading and the heading text, so that #5 bolt and #hashtag count as headings. With this extension, pandoc does require the space. ␣

Heading identifiers

See also the auto_identifiers extension above. 👀

Extension: header_attributes

Headings can be assigned attributes using this syntax at the end of the line containing the heading text:

{#identifier .class .class key=value key=value}

Thus, for example, the following headings will all be assigned the identifier foo:

# My heading {#foo}

## My heading ##    {#foo}

My other heading   {#foo}
---------------

(This syntax is compatible with PHP Markdown Extra. 🫶)

Note that although this syntax allows assignment of classes and key/value attributes, writers generally don't use all of this information. Identifiers, classes, and key/value attributes are used in HTML and HTML-based formats such as EPUB and slidy. Identifiers are used for labels and link anchors in the LaTeX, ConTeXt, Textile, Jira markup, and AsciiDoc writers.

Headings with the class unnumbered will not be numbered, even if --number-sections is specified. A single hyphen (-) in an attribute context is equivalent to .unnumbered, and it's preferable in non-English documents. So,

# My heading {-}

is just the same as

# My heading {.unnumbered}

If the unlisted class is present in addition to unnumbered, the heading will not be included in a table of contents. (Currently this feature is only implemented for certain formats: those based on LaTeX and HTML, PowerPoint, and RTF.)

Extension: implicit_header_references

Pandoc behaves as if reference links have been defined for each heading. So, to link to a heading

# Heading identifiers in HTML

you can simply write

or

[Heading identifiers in HTML][]

or

[the section on heading identifiers][heading identifiers in HTML]

instead of spelling the identifier out by hand like

[Heading identifiers in HTML](#heading-identifiers-in-html)

okay so here's the tea: if you've got multiple headings with the exact same text, the ref is only gonna link to the first one, bestie. so if you wanna link to the others you literally have to write out those explicit links like we just talked about, no shortcuts. 🙈

and just like regular reference links, these implicit ones don't care about case. case-insensitive, periodt.

big important energy: explicit link reference definitions ALWAYS win over implicit heading references. so check this out — in the example below, the link points to bar, NOT to #foo:

# Foo

[foo]: bar

See [foo]

💬 Block quotations 💬

so markdown steals email's vibe for quoting blocks of text. a block quotation is one or more paragraphs or other block elements (think lists or headings, the whole squad) where every line starts with a > character and an optional space. and lowkey the > does not have to start at the left margin, but bestie it should not be indented more than three spaces. ✨

> This is a block quote. This
> paragraph has two lines.
>
> 1. This is a list inside a block quote.
> 2. Second item.

there's also a “lazy” form that only needs the > character on the first line of each block, and she's absolutely allowed:

> This is a block quote. This
paragraph has two lines.

> 1. This is a list inside a block quote.
2. Second item.

and guess what — among the block elements that can live inside a block quote are other block quotes. yep, block quotes can nest inside each other, it's giving inception:

> This is a block quote.
>
> > A block quote within a block quote.

here's a lil detail that matters: if the > character is followed by an optional space, that space counts as part of the block quote marker, NOT as part of the indentation of the contents. so if you wanna put an indented code block inside a block quote, you need five spaces after the >:

>     code

🌸 Extension: blank_before_blockquote 🌸

original markdown syntax doesn't require a blank line before a block quote, but pandoc does require this (except obviously at the very beginning of the document, duh). why tho? because it's all too easy for a > to accidentally end up at the start of a line (line wrapping, we've all been there bestie 😭). so unless you're using the markdown_strict format, this next thing does not make a nested block quote in pandoc:

> This is a block quote.
>> Not nested, since `blank_before_blockquote` is enabled by default

💻 Verbatim (code) blocks 💻

🌈 Indented code blocks 🌈

a block of text indented four spaces (or one tab) is treated as verbatim text: meaning special characters do NOT trigger special formatting, and every space and line break is preserved fr fr. like so:

    if (a > 3) {
      moveShip(5 * gravity, DOWN);
    }

that initial (four space or one tab) indentation is not part of the verbatim text and gets removed in the output, slay.

note bestie: blank lines inside the verbatim text don't need to begin with four spaces. 💅

🎀 Fenced code blocks 🎀

✨ Extension: fenced_code_blocks ✨

on top of the standard indented code blocks, pandoc also supports fenced code blocks. these start with a row of three or more tildes (~) and end with a row of tildes that has to be at least as long as the starting row. everything between those lines = code. no indentation required, we love convenience:

~~~~~~~
if (a > 3) {
  moveShip(5 * gravity, DOWN);
}
~~~~~~~

like regular code blocks, fenced ones have to be separated from surrounding text by blank lines. no cap.

and if the code itself contains a row of tildes or backticks, just use a longer row of tildes or backticks at the start and end:

~~~~~~~~~~~~~~~~
~~~~~~~~~~
code including tildes
~~~~~~~~~~
~~~~~~~~~~~~~~~~

💜 Extension: backtick_code_blocks 💜

same as fenced_code_blocks, but it uses backticks (`) instead of tildes (~). iconic switch-up.

🌟 Extension: fenced_code_attributes 🌟

optionally, you can attach attributes to a fenced or backtick code block using this syntax:

~~~~ {#mycode .haskell .numberLines startFrom="100"}
qsort []     = []
qsort (x:xs) = qsort (filter (< x) xs) ++ [x] ++
               qsort (filter (>= x) xs)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

so here mycode is an identifier, haskell and numberLines are classes, and startFrom is an attribute with value 100. some output formats can use this info to do syntax highlighting. currently the only output formats that use this are HTML, LaTeX, Docx, Ms, and PowerPoint. if highlighting is supported for your output format and language, then that code block above appears highlighted, with numbered lines. (to see which languages are supported, type pandoc --list-highlight-languages.) otherwise, the code block above shows up like this:

<pre id="mycode" class="haskell numberLines" startFrom="100">
  <code>
  ...
  </code>
</pre>

the numberLines (or number-lines) class makes the lines of the code block get numbered, starting with 1 or the value of the startFrom attribute. the lineAnchors (or line-anchors) class makes the lines clickable anchors in HTML output. 🫶

a shortcut form also exists for specifying the language of the code block:

qsort [] = []

which is equivalent to:

qsort [] = []

and this shortcut form can be combined with attributes:

qsort [] = []

which is equivalent to:

qsort [] = []

if the fenced_code_attributes extension is disabled but the input contains class attribute(s) for the code block, the first class attribute gets printed after the opening fence as a bare word.

to prevent all highlighting, use the --syntax-highlighting=none option. to set the highlighting style or method, use --syntax-highlighting. for more info on highlighting, see Syntax highlighting, below. 💅

📝 Line blocks 📝

🦋 Extension: line_blocks 🦋

a line block is a sequence of lines starting with a vertical bar (|) followed by a space. the division into lines is preserved in the output, and so is any leading spaces; otherwise, the lines get formatted as markdown. this is super useful for verse and addresses:

| The limerick packs laughs anatomical
| In space that is quite economical.
|    But the good ones I've seen
|    So seldom are clean
| And the clean ones so seldom are comical

| 200 Main St.
| Berkeley, CA 94718

the lines can be hard-wrapped if needed, but the continuation line has to begin with a space.

| The Right Honorable Most Venerable and Righteous Samuel L.
  Constable, Jr.
| 200 Main St.
| Berkeley, CA 94718

inline formatting (like emphasis) is allowed in the content (though it can't cross line boundaries). block-level formatting (like block quotes or lists) is not recognized, sorry bestie.

this syntax is borrowed from reStructuredText. 💜

📋 Lists 📋

🔥 Bullet lists 🔥

a bullet list is a list of bulleted list items. a bulleted list item begins with a bullet (*, +, or -). here's a simple example:

* one
* two
* three

this produces a “compact” list. if you want a “loose”... wait no — a “loose” list, where each item is formatted as a paragraph, put spaces between the items:

* one

* two

* three

the bullets don't have to be flush with the left margin; they can be indented one, two, or three spaces. the bullet must be followed by whitespace.

list items look best when subsequent lines are flush with the first line (after the bullet):

* here is my first
  list item.
* and my second.

but markdown also allows a “lazy” format:

* here is my first
list item.
* and my second.

💖 Block content in list items 💖

a list item can contain multiple paragraphs and other block-level content. however, subsequent paragraphs must be preceded by a blank line and indented to line up with the first non-space content after the list marker.

  * First paragraph.

    Continued.

  * Second paragraph. With a code block, which must be indented
    eight spaces:

        { code }

exception: if the list marker is followed by an indented code block, which must begin 5 spaces after the list marker, then subsequent paragraphs must begin two columns after the last character of the list marker:

*     code

  continuation paragraph

list items can include other lists. in that case the preceding blank line is optional. the nested list must be indented to line up with the first non-space character after the list marker of the containing list item.

* fruits
  + apples
    - macintosh
    - red delicious
  + pears
  + peaches
* vegetables
  + broccoli
  + chard

as noted above, markdown lets you write list items “lazily,” instead of indenting continuation lines. but if there are multiple paragraphs or other blocks in a list item, the first line of each must be indented.

+ A lazy, lazy, list
item.

+ Another one; this looks
bad but is legal.

    Second paragraph of second
list item.

👑 Ordered lists 👑

ordered lists work just like bulleted lists, except the items begin with enumerators instead of bullets.

in original markdown, enumerators are decimal numbers followed by a period and a space. the numbers themselves are ignored, so there's literally no difference between this list:

1.  one
2.  two
3.  three

and this one:

5.  one
7.  two
1.  three

⭐ Extension: fancy_lists ⭐

unlike original markdown, pandoc lets ordered list items be marked with uppercase and lowercase letters and roman numerals, on top of Arabic numerals. list markers can be enclosed in parentheses or followed by a single right-parenthesis or period. they must be separated from the text that follows by at least one space, and — if the list marker is a capital letter with a period — by at least two spaces. 💅

The fancy_lists extension also lets you use # as an ordered list marker instead of an actual numeral:

#. one
#. two

Heads up bestie: the # ordered list marker does not work with commonmark. 💀

And same energy as commonmark: ordered sublists have to start with 1 or the equivalent (like i or a). This helps you avoid accidental lists caused by stuff like parenthesized dates. (iykyk — nobody wants (2024) turning into a list 👑)

✨ Extension: startnum

Pandoc also pays attention to which type of list marker you used, and what number it starts at, and it preserves both of those wherever it can in the output format. So the following gives you a list with numbers followed by a single parenthesis, starting at 9, plus a sublist in lowercase roman numerals:

 9)  Ninth
10)  Tenth
11)  Eleventh
      i. subone
     ii. subtwo
    iii. subthree

The way pandoc works: every time a different type of list marker shows up, that's a brand new list. So this next one creates three lists:

(2) Two
(5) Three
1.  Four
*   Five

Want the default list markers back? Just use #. 💅

#.  one
#.  two
#.  three

✨ Extension: task_lists

Pandoc supports task lists, using the syntax of GitHub-Flavored Markdown. Slay.

- [ ] an unchecked task list item
- [x] checked item

🌸 Definition lists

✨ Extension: definition_lists

Pandoc supports definition lists, using the syntax of PHP Markdown Extra with some extensions. Obsessed.

Term 1

:   Definition 1

Term 2 with *inline markup*

:   Definition 2

        { some code, part of Definition 2 }

    Third paragraph of definition 2.

Each term has to fit on one line, which may optionally be followed by a blank line, and must be followed by one or more definitions. A definition begins with a colon : or a tilde ~, which may be indented one or two spaces.

A term can have multiple definitions, and each definition can consist of one or more indented block elements (paragraph, code block, list, etc. — the whole menu 🎀). The blocks in the definition should be indented to the column of the first non-space content after the : or ~ marker, or (if the four_space_rule extension is enabled) four spaces or one tab stop. And just like other Markdown lists, you can “lazily” skip the indentation on paragraph continuation lines:

Term 1

: Definition
with lazy continuation.

  Second paragraph of the definition.

If you leave a space before the definition (like in the example up there 👆), the text of the definition gets treated as a paragraph. In some output formats that means bigger spacing between term/definition pairs. Want a more compact definition list? Omit the space before the definition, periodt:

Term 1
  ~ Definition 1

Term 2
  ~ Definition 2a
  ~ Definition 2b

Note that space between items in a definition list is required. no cap.

💜 Numbered example lists

✨ Extension: example_lists

The special list marker @ can be used for sequentially numbered examples. The first list item with a @ marker gets numbered 1, the next 2, and so on, throughout the whole document. These numbered examples don't have to live in a single list — each new list using @ picks up right where the last one stopped. So, for example:

(@)  My first example will be numbered (1).
(@)  My second example will be numbered (2).

Explanation of examples.

(@)  My third example will be numbered (3).

Numbered examples can also be labeled and referred to elsewhere in the document:

(@good)  This is a good example.

As (@good) illustrates, ...

The label can be any string of alphanumeric characters, underscores, or hyphens. 🫶

Continuation paragraphs in example lists must always be indented four spaces, no matter how long the list marker is. Which is to say, example lists always behave as if the four_space_rule extension is set. Why? Because example labels tend to be long, and indenting content to the first non-space character after the label would be awkward. 😭

You can repeat an earlier numbered example by re-using its label:

(@foo) Sample sentence.

Intervening text...

This theory can explain the case we saw earlier (repeated):

(@foo) Sample sentence.

But ngl, this only works reliably if the repeated item is in a list all by itself, because each numbered example list gets numbered continuously from its own starting number.

In a longer (book-length) work, one might need to reset the example list counter at the beginning of each chapter. You can do that with this syntax:

(1@foo) This item will be numbered 1, and subsequent example
    list items will be numbered 2, 3, etc.

Any number may be used instead of 1. Note that if there are several example list items in a row, the number only has an effect if it's placed on the first of them. 👑

✨ Ending a list

Okay so what if you want to put an indented code block after a list? 👀

-   item one
-   item two

    { my code block }

Trouble! Here pandoc (like other Markdown implementations) will treat { my code block } as the second paragraph of item two, and not as a code block. 💀

To “cut off” the list after item two, you can insert some non-indented content, like an HTML comment, which won't produce any visible output in any format:

-   item one
-   item two

<!-- end of list -->

    { my code block }

Same trick works if you want two consecutive lists instead of one big list, bestie:

1.  one
2.  two
3.  three

<!-- -->

1.  uno
2.  dos
3.  tres

✨ Horizontal rules

A line containing a row of three or more *, -, or _ characters (optionally separated by spaces) produces a horizontal rule:

*  *  *  *

---------------

We strongly recommend that horizontal rules be separated from surrounding text by blank lines. If a horizontal rule isn't followed by a blank line, pandoc may try to interpret the lines that follow as a YAML metadata block or a table. 😭 don't do it.

✨ Tables

Four kinds of tables may be used. The first three kinds (simple, multiline, and grid) require lining up columns and thus presuppose the use of a fixed-width font, such as Courier ⭐. The fourth kind (pipe tables) can be used with proportionally spaced fonts, since it does not require lining up columns. main character energy.

✨ Extension: table_captions

A caption may optionally be provided with all 4 kinds of tables (as illustrated in the examples below). A caption is a paragraph beginning with the string Table: (or table: or just :), which will be stripped off. It may appear either before or after the table. 🎀

✨ Extension: simple_tables

Simple tables look like this:

  Right     Left     Center     Default
-------     ------ ----------   -------
     12     12        12            12
    123     123       123          123
      1     1          1             1

Table:  Demonstration of simple table syntax.

The header and table rows must each fit on one line. Column alignments are determined by the position of the header text relative to the dashed line below it:

• If the dashed line is flush with the header text on the right side but extends beyond it on the left, the column is right-aligned.
• If the dashed line is flush with the header text on the left side but extends beyond it on the right, the column is left-aligned.
• If the dashed line extends beyond the header text on both sides, the column is centered.
• If the dashed line is flush with the header text on both sides, the default alignment is used (in most cases, this will be left).

The table must end with a blank line, or a line of dashes followed by a blank line.

The column header row may be omitted, provided a dashed line is used to end the table. For example:

-------     ------ ----------   -------
     12     12        12             12
    123     123       123           123
      1     1          1              1
-------     ------ ----------   -------

When the header row is omitted, column alignments are determined on the basis of the first line of the table body. So, in the tables above, the columns would be right, left, center, and right aligned, respectively. 🦋

✨ Extension: multiline_tables

Multiline tables allow header and table rows to span multiple lines of text (but cells that span multiple columns or rows of the table are not supported 🙈). Here is an example:

-------------------------------------------------------------
 Centered   Default           Right Left
  Header    Aligned         Aligned Aligned
----------- ------- --------------- -------------------------
   First    row                12.0 Example of a row that
                                    spans multiple lines.

  Second    row                 5.0 Here's another one. Note
                                    the blank line between
                                    rows.
-------------------------------------------------------------

Table: Here's the caption. It, too, may span
multiple lines.

These work like simple tables, but with the following differences:

• They must begin with a row of dashes, before the header text (unless the header row is omitted).
• They must end with a row of dashes, then a blank line.
• The rows must be separated by blank lines.

In multiline tables, the table parser pays attention to the widths of the columns, and the writers try to reproduce these relative widths in the output. So if you find that one of the columns is too narrow in the output, try widening it in the Markdown source. 💅

The header may be omitted in multiline tables as well as simple tables:

----------- ------- --------------- -------------------------
   First    row                12.0 Example of a row that
                                    spans multiple lines.

  Second    row                 5.0 Here's another one. Note
                                    the blank line between
                                    rows.
----------- ------- --------------- -------------------------

: Here's a multiline table without a header.

It is possible for a multiline table to have just one row, but the row should be followed by a blank line (and then the row of dashes that ends the table), or the table may be interpreted as a simple table. ⭐

🫶 Extension: grid_tables

okay bestie, grid tables are the ✨fancy✨ ones and they look exactly like this:

: Sample grid table.

+---------------+---------------+--------------------+
| Fruit         | Price         | Advantages         |
+===============+===============+====================+
| Bananas       | $1.34         | - built-in wrapper |
|               |               | - bright color     |
+---------------+---------------+--------------------+
| Oranges       | $2.10         | - cures scurvy     |
|               |               | - tasty            |
+---------------+---------------+--------------------+

that row of =s is doing the absolute most — it's the divider between the header and the table body, and you can just... skip it if you want a headerless table. no notes. 💅

and the cells?? they can hold arbitrary block elements — multiple paragraphs, code blocks, lists, whatever. literally anything.

cells can also span multiple columns or rows, like this:

+---------------------+----------+
| Property            | Earth    |
+=============+=======+==========+
|             | min   | -89.2 °C |
| Temperature +-------+----------+
| 1961-1990   | mean  | 14 °C    |
|             +-------+----------+
|             | max   | 56.7 °C  |
+-------------+-------+----------+

a table header can have more than one row too, iconic behavior:

+---------------------+-----------------------+
| Location            | Temperature 1961-1990 |
|                     | in degree Celsius     |
|                     +-------+-------+-------+
|                     | min   | mean  | max   |
+=====================+=======+=======+=======+
| Antarctica          | -89.2 | N/A   | 19.8  |
+---------------------+-------+-------+-------+
| Earth               | -89.2 | 14    | 56.7  |
+---------------------+-------+-------+-------+

alignments? same deal as pipe tables — drop colons at the boundaries of the separator line after the header:

+---------------+---------------+--------------------+
| Right         | Left          | Centered           |
+==============:+:==============+:==================:+
| Bananas       | $1.34         | built-in wrapper   |
+---------------+---------------+--------------------+

and for headerless tables the colons just move up to the top line instead:

+--------------:+:--------------+:------------------:+
| Right         | Left          | Centered           |
+---------------+---------------+--------------------+

want a table foot? enclose it with separator lines that use = instead of -:

+---------------+---------------+
| Fruit         | Price         |
+===============+===============+
| Bananas       | $1.34         |
+---------------+---------------+
| Oranges       | $2.10         |
+===============+===============+
| Sum           | $3.44         |
+===============+===============+

the foot must always be placed at the very bottom of the table. no wiggle room, no exceptions, periodt. 👑

and if you're an Emacs girlie, grid tables can be created easily using Emacs' table-mode (M-x table-insert). slay.

---

🌈 Extension: pipe_tables

pipe tables are the lowkey ones, they look like this:

| Right | Left | Default | Center |
|------:|:-----|---------|:------:|
|   12  |  12  |    12   |    12  |
|  123  |  123 |   123   |   123  |
|    1  |    1 |     1   |     1  |

  : Demonstration of pipe table syntax.

the syntax is identical to PHP Markdown Extra tables, fr fr. the beginning and ending pipe characters are optional, but pipes are required between all columns. the colons indicate column alignment, as shown. the header cannot be omitted. to simulate a headerless table, include a header with blank cells. that's the hack, bestie.

and because the pipes are what mark column boundaries, columns don't need to be vertically aligned like they are in the example above. so this is a perfectly legal (though ugly) pipe table:

fruit| price
-----|-----:
apple|2.05
pear|1.37
orange|3.09

the cells of pipe tables cannot contain block elements like paragraphs and lists, and cannot span multiple lines. if any line of the Markdown source is longer than the column width (see --columns), then the table will take up the full text width and the cell contents will wrap, with the relative cell widths determined by the number of dashes in the line separating the table header from the table body. (for example ---|- would make the first column 3/4 and the second column 1/4 of the full text width.) on the other hand, if no lines are wider than column width, then cell contents will not be wrapped, and the cells will be sized to their contents. 🤷‍♀️

Note: pandoc also recognizes pipe tables of the following form, as can be produced by Emacs' orgtbl-mode:

| One | Two   |
|-----+-------|
| my  | table |
| is  | nice  |

the difference is that + is used instead of |. other orgtbl features are not supported. in particular, to get non-default column alignment, you'll need to add colons as above. no shortcuts, sorry babe. 💅

---

🎀 Extension: table_attributes

attributes may be attached to tables by including them at the end of the caption. (for the syntax, see header_attributes.)

: Here's the caption. {#ident .class key="value"}

---

💜 Table placement in LaTeX

by default, pandoc will render tables in LaTeX using the longtable environment. these tables will not "float" and they can span page boundaries. honestly kind of main character energy.

but if you prefer standard "floating" tables (which are limited to what can fit on one page), add the float class to the table's attributes (which can be specified after the caption). optionally, the latex-placement attribute may be used to pass parameters affecting the float placement:

: The caption of the table. {#ident .float latex-placement="ht"}

Note: longtable does not work with two-column layouts, so if you are producing a two-column document using LaTeX, you will need to specify the float class on every table, or use this small Lua filter to add it automatically:

function Table(el)
  el.classes:insert("float")
  return el
end

---

🌸 Metadata blocks

⭐ Extension: pandoc_title_block

if the file begins with a title block

% title
% author(s) (separated by semicolons)
% date

it will be parsed as bibliographic information, not regular text. (it will be used, for example, in the title of standalone LaTeX or HTML output.) the block may contain just a title, a date and an author, or all three elements. if you want to include an author but no title, or a title and a date but no author, you need a blank line:

%
% Author

% My title
%
% June 15, 2006

the title may occupy multiple lines, but continuation lines must begin with leading space, thusly:

% My title
  on multiple lines

if a document has multiple authors, the authors may be put on separate lines with leading space, or separated by semicolons, or both. so all of the following are equivalent:

% Author One
  Author Two

% Author One; Author Two

% Author One;
  Author Two

the date must fit on one line. that's it. no multi-line dates, bestie. 😭

all three metadata fields may contain standard inline formatting (italics, links, footnotes, etc.). go off.

title blocks will always be parsed, but they will affect the output only when the --standalone (-s) option is chosen. in HTML output, titles will appear twice: once in the document head — this is the title that will appear at the top of the window in a browser — and once at the beginning of the document body. the title in the document head can have an optional prefix attached (--title-prefix or -T option). the title in the body appears as an H1 element with class "title", so it can be suppressed or reformatted with CSS. if a title prefix is specified with -T and no title block appears in the document, the title prefix will be used by itself as the HTML title.

the man page writer extracts a title, man page section number, and other header and footer information from the title line. the title is assumed to be the first word on the title line, which may optionally end with a (single-digit) section number in parentheses. (there should be no space between the title and the parentheses.) anything after this is assumed to be additional footer and header text. a single pipe character (|) should be used to separate the footer text from the header text. thus,

% PANDOC(1)

will yield a man page with the title PANDOC and section 1.

% PANDOC(1) Pandoc User Manuals

will also have "Pandoc User Manuals" in the footer.

% PANDOC(1) Pandoc User Manuals | Version 4.0

will also have "Version 4.0" in the header. 🦋

---

🍓 Extension: yaml_metadata_block

a YAML metadata block is a valid YAML object, delimited by a line of three hyphens (---) at the top and a line of three hyphens (---) or three dots (...) at the bottom. the initial line --- must not be followed by a blank line. a YAML metadata block may occur anywhere in the document, but if it is not at the beginning, it must be preceded by a blank line. (note that JSON may be used as well, because JSON is a subset of YAML. iykyk.)

okay so here's a cute lil trick: because of the way pandoc concatenates input files when several are provided, you may also keep the metadata in a separate YAML file and pass it to pandoc as an argument, along with your Markdown files:

pandoc chap1.md chap2.md chap3.md metadata.yaml -s -o book.html

just be sure that the YAML file begins with --- and ends with --- or .... alternatively, you can use the --metadata-file option. using that approach however, you cannot reference content (like footnotes) from the main Markdown input document. 💀

metadata will be taken from the fields of the YAML object and added to any existing document metadata. metadata can contain lists and objects (nested arbitrarily), but all string scalars will be interpreted as Markdown. fields with names ending in an underscore will be ignored by pandoc. (they may be given a role by external processors.) field names must not be interpretable as YAML numbers or boolean values (so, for example, yes, True, and 15 cannot be used as field names).

a document may contain multiple metadata blocks. if two metadata blocks attempt to set the same field, the value from the second block will be taken. 👑

Each metadata block gets handled internally as its own independent YAML document, bestie. Which means… any YAML anchors you define in one block? Can't be referenced from another block. Nope. They're strangers. 💀

When pandoc is used with -t markdown to make a Markdown document, a YAML metadata block only shows up if you used the -s/--standalone option. And all of the metadata will appear in a single block at the very beginning of the document. One block, top of the file, that's the vibe.

Okay now — YAML escaping rules have to be followed, fr fr. Like, if a title contains a colon, it must be quoted. And if it contains a backslash escape, you gotta make sure it isn't treated as a YAML escape sequence. The pipe character (|) can start an indented block that gets interpreted literally, no escaping needed. And this form is necessary when the field contains blank lines or block-level formatting:

---
title:  'This is the title: it contains a colon'
author:
- Author One
- Author Two
keywords: [nothing, nothingness]
abstract: |
  This is the abstract.

  It consists of two paragraphs.
...

The literal block after the | must be indented relative to the line containing the |. If it isn't? The YAML is invalid and pandoc will not interpret it as metadata. Periodt. If you want the full tea on the complex rules governing YAML, go read the Wikipedia entry on YAML syntax.

Template variables get set automatically from the metadata. So like, in writing HTML, the variable abstract will be set to the HTML equivalent of the Markdown in the abstract field:

<p>This is the abstract.</p>
<p>It consists of two paragraphs.</p>

Variables can hold arbitrary YAML structures, but your template has to match that structure. The author variable in the default templates expects a simple list or string, but you can change it to support more complicated structures. This combo, for example, would add an affiliation to the author if one is given:

---
title: The document title
author:
- name: Author One
  affiliation: University of Somewhere
- name: Author Two
  affiliation: University of Nowhere
...

And to actually use those structured authors? You'd need a custom template, bestie:

$for(author)$
$if(author.name)$
$author.name$$if(author.affiliation)$ ($author.affiliation$)$endif$
$else$
$author$
$endif$
$endfor$

Raw content to include in the document's header can be specified with header-includes. But — and this part is important, don't skip it 🫶 — you have to mark up this content as raw code for a particular output format, using the raw_attribute extension. Otherwise it gets interpreted as Markdown. Like this:

header-includes:
- |

\let\oldsection\section \renewcommand{\section}[1]{\clearpage\oldsection{#1}}

Note: the yaml_metadata_block extension works not just with markdown but with commonmark (and it's enabled by default in gfm and commonmark_x). However — in commonmark, gfm, and commonmark_x, these restrictions apply:

• The YAML metadata block must occur at the beginning of the document (and there can be only one. one single block, no more). If multiple files are given as arguments to pandoc, only the first can be a YAML metadata block.
• The leaf nodes of the YAML structure are parsed in isolation from each other and from the rest of the document. So, for example, you can't use a reference link in these contexts if the link definition lives somewhere else in the document. 🙈

---

🌈 Backslash escapes 🌈

Extension: all_symbols_escapable

Except inside a code block or inline code, any punctuation or space character preceded by a backslash will be treated literally — even if it would normally indicate formatting. So if one writes:

*\*hello\**

one will get:

<em>*hello*</em>

instead of:

<strong>hello</strong>

This rule is way easier to remember than original Markdown's rule, which only lets you backslash-escape these characters:

\`*_{}[]()>#+-.!

(However, if the markdown_strict format is used, the original Markdown rule will be used instead. Just so you know. 💅)

A backslash-escaped space is parsed as a nonbreaking space. In TeX output, it shows up as ~. In HTML and XML output, it shows up as a literal unicode nonbreaking space character — note that it will actually look "invisible" in the generated HTML source! (You can still use the --ascii command-line option to make it appear as an explicit entity.)

A backslash-escaped newline (i.e. a backslash chillin' at the end of a line) is parsed as a hard line break. It'll appear in TeX output as \\ and in HTML as <br />. Honestly this is a nice alternative to Markdown's "invisible" way of indicating hard line breaks using two trailing spaces on a line. 👑

Backslash escapes do not work in verbatim contexts. Sorry bestie, that's the rule.

---

💖 Inline formatting 💖

Emphasis ✨

To emphasize some text, surround it with *s or _s, like this:

This text is _emphasized with underscores_, and this
is *emphasized with asterisks*.

Double * or _ produces strong emphasis:

This is **strong emphasis** and __with underscores__.

A * or _ character surrounded by spaces, or backslash-escaped, will not trigger emphasis:

This is * not emphasized *, and \*neither is this\*.

Extension: intraword_underscores

Because _ is sometimes used inside words and identifiers, pandoc does not interpret a _ surrounded by alphanumeric characters as an emphasis marker. So if you want to emphasize just part of a word, use *:

feas*ible*, not feas*able*.

Strikeout 🎀

Extension: strikeout

To strike out a section of text with a horizontal line, begin and end it with ~~. Thus, for example,

This ~~is deleted text.~~

Superscripts and subscripts 🧚

Extension: superscript, subscript

Superscripts may be written by surrounding the superscripted text by ^ characters; subscripts may be written by surrounding the subscripted text by ~ characters. Thus, for example,

H~2~O is a liquid.  2^10^ is 1024.

The text between ^...^ or ~...~ may not contain spaces or newlines. If the superscripted or subscripted text contains spaces, those spaces must be escaped with backslashes. (This is to prevent accidental superscripting and subscripting through the ordinary use of ~ and ^, and also bad interactions with footnotes.) So if you want the letter P with `a cat' in subscripts, use P~a\ cat~, not P~a cat~. 💜

Verbatim 💻

To make a short span of text verbatim, put it inside backticks:

What is the difference between `>>=` and `>>`?

If the verbatim text includes a backtick, use double backticks:

Here is a literal backtick `` ` ``.

(The spaces after the opening backticks and before the closing backticks will be ignored.)

The general rule is that a verbatim span starts with a string of consecutive backticks (optionally followed by a space) and ends with a string of the same number of backticks (optionally preceded by a space).

Note that backslash-escapes (and other Markdown constructs) do not work in verbatim contexts:

This is a backslash followed by an asterisk: `\*`.

Extension: inline_code_attributes

Attributes can be attached to verbatim text, just as with fenced code blocks:

`<$>`{.haskell}

Underline 🖍️

To underline text, use the underline class:

[Underline]{.underline}

Or, without the bracketed_spans extension (but with native_spans):

<span class="underline">Underline</span>

This will work in all output formats that support underline.

Small caps 🐜

To write small caps, use the smallcaps class:

[Small caps]{.smallcaps}

Or, without the bracketed_spans extension:

<span class="smallcaps">Small caps</span>

For compatibility with other Markdown flavors, CSS is also supported:

<span style="font-variant:small-caps;">Small caps</span>

This will work in all output formats that support small caps.

Highlighting 🔥

To highlight text, use the mark class:

[Mark]{.mark}

Or, without the bracketed_spans extension (but with native_spans):

<span class="mark">Mark</span>

This will work in all output formats that support highlighting.

---

🦋 Math 🦋

Extension: tex_math_dollars

Anything between two $ characters will be treated as TeX math. The opening $ must have a non-space character immediately to its right, while the closing $ must have a non-space character immediately to its left, and must not be followed immediately by a digit. So $20,000 and $30,000 won't parse as math — sorry to those two, they're not math today. If for some reason you need to enclose text in literal $ characters, backslash-escape them and they won't be treated as math delimiters.

For display math, use $$ delimiters. (In this case, the delimiters may be separated from the formula by whitespace. However, there can be no blank lines between the opening and closing $$ delimiters. Don't do it. 🚫)

TeX math will be printed in all output formats. How it's rendered depends on the output format:

• LaTeX — It will appear verbatim surrounded by \(...\) (for inline math) or \[...\] (for display math).
• Markdown, Emacs Org mode, ConTeXt, Txt2Tags, ZimWiki — It will appear verbatim surrounded by $...$ (for inline math) or $$...$$ (for display math).
• XWiki — It will appear verbatim surrounded by {{formula}}..{{/formula}}.
• reStructuredText — It will be rendered using an interpreted text role :math:.
• AsciiDoc — For AsciiDoc output math will appear verbatim surrounded by latexmath:[...]. For asciidoc_legacy the bracketed material will also include inline or display math delimiters.
• Texinfo — It will be rendered inside a @math command.
• roff man, Jira markup — It will be rendered verbatim without $'s.
• MediaWiki, DokuWiki — It will be rendered inside <math> tags.
• Textile — It will be rendered inside <span class="math"> tags.
• RTF, OpenDocument — It will be rendered, if possible, using Unicode characters, and will otherwise appear verbatim.
• ODT — It will be rendered, if possible, using MathML.
• DocBook — If --math-method=mathml is used, it will be rendered using MathML in an inlineequation or informalequation tag. Otherwise it will be rendered, if possible, using Unicode characters.
• Docx and PowerPoint — It will be rendered using OMML math markup.

FictionBook2 💖

okay so if --math-method=webtex is the vibe, formulas are rendered as images using CodeCogs or other compatible web service, downloaded and embedded in the e-book. Otherwise, they will appear verbatim. no glow-up, just plain text. 💅

HTML, Slidy, DZSlides, S5, EPUB 🌈

The way math is rendered in HTML will depend on the command-line options selected. Therefore see Math rendering in HTML above. 👀

---

Raw HTML 🎀

Extension: raw_html

Markdown allows you to insert raw HTML (or DocBook) anywhere in a document (except verbatim contexts, where <, >, and & are interpreted literally). (Technically this is not an extension, since standard Markdown allows it, but it has been made an extension so that it can be disabled if desired.) ✨

The raw HTML is passed through unchanged in HTML, S5, Slidy, Slideous, DZSlides, EPUB, Markdown, CommonMark, Emacs Org mode, Textile, and Txt2Tags output, and suppressed in other formats. it's giving selective. 💅

For a more explicit way of including raw HTML in a Markdown document, see the raw_attribute extension. 👑

In the CommonMark format, if raw_html is enabled, superscripts, subscripts, strikeouts and small capitals will be represented as HTML. Otherwise, plain-text fallbacks will be used. Note that even if raw_html is disabled, tables will be rendered with HTML syntax if they cannot use pipe syntax. no cap.

Extension: markdown_in_html_blocks

Original Markdown allows you to include HTML “blocks”: blocks of HTML between balanced tags that are separated from the surrounding text with blank lines, and start and end at the left margin. Within these blocks, everything is interpreted as HTML, not Markdown; so (for example), * does not signify emphasis. 💀

Pandoc behaves this way when the markdown_strict format is used; but by default, pandoc interprets material between HTML block tags as Markdown. Thus, for example, pandoc will turn

<table>
<tr>
<td>*one*</td>
<td>[a link](https://google.com)</td>
</tr>
</table>

into

<table>
<tr>
<td><em>one</em></td>
<td><a href="https://google.com">a link</a></td>
</tr>
</table>

whereas Markdown.pl will preserve it as is. 🙈

There is one exception to this rule: text between <script>, <style>, <pre>, and <textarea> tags is not interpreted as Markdown.

This departure from original Markdown should make it easier to mix Markdown with HTML block elements. For example, one can surround a block of Markdown text with <div> tags without preventing it from being interpreted as Markdown. 💅

Extension: native_divs

Use native pandoc Div blocks for content inside <div> tags. For the most part this should give the same output as markdown_in_html_blocks, but it makes it easier to write pandoc filters to manipulate groups of blocks. 🧚

Extension: native_spans

Use native pandoc Span blocks for content inside <span> tags. For the most part this should give the same output as raw_html, but it makes it easier to write pandoc filters to manipulate groups of inlines. ✨

Extension: raw_tex

In addition to raw HTML, pandoc allows raw LaTeX, TeX, and ConTeXt to be included in a document. Inline TeX commands will be preserved and passed unchanged to the LaTeX and ConTeXt writers. Thus, for example, you can use LaTeX to include BibTeX citations:

This result was proved in \cite{jones.1967}.

Note that in LaTeX environments, like

\begin{tabular}{|l|l|}\hline
Age & Frequency \\ \hline
18--25  & 15 \\
26--35  & 33 \\
36--45  & 22 \\ \hline
\end{tabular}

the material between the begin and end tags will be interpreted as raw LaTeX, not as Markdown. 💅

For a more explicit and flexible way of including raw TeX in a Markdown document, see the raw_attribute extension.

Inline LaTeX is ignored in output formats other than Markdown, LaTeX, Emacs Org mode, and ConTeXt. 🫶

---

Generic raw attribute 🌈

Extension: raw_attribute

Inline spans and fenced code blocks with a special kind of attribute will be parsed as raw content with the designated format. For example, the following produces a raw roff ms block:

.MYMACRO blah blah

And the following produces a raw html inline element:

This is `<a>html</a>`{=html}

This can be useful to insert raw xml into docx documents, e.g. a pagebreak:

<w:p> <w:r> <w:br w:type="page"/> </w:r> </w:p>

The format name should match the target format name (see -t/--to, above, for a list, or use pandoc --list-output-formats). Use openxml for docx output, opendocument for odt output, html5 for epub3 output, html4 for epub2 output, and latex, beamer, ms, or html5 for pdf output (depending on what you use for --pdf-engine). 👑

This extension presupposes that the relevant kind of inline code or fenced code block is enabled. Thus, for example, to use a raw attribute with a backtick code block, backtick_code_blocks must be enabled. 💯

The raw attribute cannot be combined with regular attributes. 🚫

---

LaTeX macros 🦋

Extension: latex_macros

When this extension is enabled, pandoc will parse LaTeX macro definitions and apply the resulting macros to all LaTeX math and raw LaTeX. So, for example, the following will work in all output formats, not just LaTeX:

\newcommand{\tuple}[1]{\langle #1 \rangle}

$\tuple{a, b, c}$

Note that LaTeX macros will not be applied if they occur inside a raw span or block marked with the raw_attribute extension.

When latex_macros is disabled, the raw LaTeX and math will not have macros applied. This is usually a better approach when you are targeting LaTeX or PDF. 💅

Macro definitions in LaTeX will be passed through as raw LaTeX only if latex_macros is not enabled. Macro definitions in Markdown source (or other formats allowing raw_tex) will be passed through regardless of whether latex_macros is enabled. ✨

---

Links 🔗

Markdown allows links to be specified in several ways. periodt.

Automatic links

If you enclose a URL or email address in pointy brackets, it will become a link:

<https://google.com>
<sam@green.eggs.ham>

Inline links

An inline link consists of the link text in square brackets, followed by the URL in parentheses. (Optionally, the URL can be followed by a link title, in quotes.)

This is an [inline link](/url), and here's [one with
a title](https://fsf.org "click here for a good time!").

There can be no space between the bracketed part and the parenthesized part. The link text can contain formatting (such as emphasis), but the title cannot. 🙅‍♀️

Email addresses in inline links are not autodetected, so they have to be prefixed with mailto:

[Write me!](mailto:sam@green.eggs.ham)

Reference links

An explicit reference link has two parts, the link itself and the link definition, which may occur elsewhere in the document (either before or after the link). 👀

The link consists of link text in square brackets, followed by a label in square brackets. (There cannot be space between the two unless the spaced_reference_links extension is enabled.) If the label is empty ([]), then it will be implicitly be taken to be the same as the link text; thus [foo][] is equivalent to [foo][foo]. (If the shortcut_reference_links extension is enabled, the empty [] may be omitted.) ✨

The link definition consists of the bracketed label, followed by a colon and a space, followed by the URL, and optionally (after a space) a link title either in quotes or in parentheses. The label must not be parseable as a citation (assuming the citations extension is enabled): citations take precedence over link labels. 🚨

Here are some examples of reference links and link definitions;

See [the website *I* built][my website].

See [my website][] and [the bar page][1] and
the [home page of the FSF][fsf].

[my website]: http://foo.bar.baz
[1]: /foo/bar.html  "My title, optional"
[fsf]: https://fsf.org (The Free Software Foundation)
[special page]: /bar#special  'A title in single quotes'

The URL in a link definition may optionally be surrounded by angle brackets:

[my label 5]: <http://foo.bar.baz>

The title may go on the next line:

[my label 3]: https://fsf.org
  "The Free Software Foundation"

Note that link labels are not case sensitive. So, this will work:

Here is [my link][FOO]

[Foo]: /bar/baz

The link definition may come either before or after a reference link that uses the label. 💅

Note: In some Markdown implementations, reference link definitions cannot occur in nested constructions such as list items or block quotes. Pandoc lifts this arbitrary-seeming restriction. So the following is fine in pandoc, though not in all implementations:

> My block [quote].
>
> [quote]: /foo

Extension: shortcut_reference_links

In a shortcut reference link, the second pair of brackets may be omitted entirely:

See [my website].

[my website]: http://foo.bar.baz

Internal links

To link to another section of the same document, use the automatically generated identifier (see Heading identifiers). For example:

See the [Introduction](#introduction).

or

See the [Introduction].

[Introduction]: #introduction

Internal links are currently supported for HTML formats (including HTML slide shows and EPUB), LaTeX, and ConTeXt. 🌐

Images

A link immediately preceded by a ! will be treated as an image. The link text (the part in square brackets) will be used as the image’s alt text:

![la lune](lalune.jpg "Voyage to the moon")

![movie reel]

[movie reel]: movie.gif

Extension: implicit_figures

An image with nonempty alt text, occurring by itself in a paragraph, will be rendered as a figure with a caption. The image’s description will be used as the caption. 🖼️

![This is the caption.](image.png)

How this is rendered depends on the output format. Some output formats (e.g. RTF) do not yet support figures. In those formats, you’ll just get an image in a paragraph by itself, with no caption. 😭

If you just want a regular inline image, just make sure it is not the only thing in the paragraph. One way to do this is to insert a nonbreaking space after the image:

![This image won't be a figure](image.png)\

Note that in reveal.js slide shows, an image in a paragraph by itself that has the r-stretch class will fill the screen, and the caption and figure tags will be omitted. 🔥

Sooo, what if you want alt text that's not the caption? Just use an explicit attribute, bestie (only works if the link_attributes extension is on, iykyk):

![The caption.](image.png){alt="description of image"}

And for LaTeX output, you can tell the figure where to sit by adding the latex-placement attribute. Iconic behaviour tbh:

![The caption.](image.png){latex-placement="ht"}

✨ Extension: link_attributes ✨

You can slap attributes on links and images, no cap:

An inline ![image](foo.jpg){#id .class width=30 height=20px}
and a reference ![image][ref] with attributes.

[ref]: foo.jpg "optional title" {#id .class key=val key2="val 2"}

(Fun fact: this syntax is compatible with PHP Markdown Extra when you're only using #id and .class 💅)

For HTML and EPUB, every known HTML5 attribute except width and height (but yes, srcset and sizes are in there) gets passed through as-is. Unknown attributes also sail through, as custom attributes with data- prepended to them. The other writers just ignore attributes that their output format doesn't specifically support. Not their problem, honestly.

Now the width and height attributes on images are special little snowflakes ✨. If you use them without a unit, the unit is assumed to be pixels. But you can use any of these unit identifiers: px, cm, mm, in, inch and %. And there must not be any spaces between the number and the unit. Like so:

![](file.jpg){ width=50% }

• Dimensions may get converted to a form that's compatible with the output format (for example, dimensions given in pixels will be converted to inches when converting HTML to LaTeX). That pixel↔physical-measurement conversion is affected by the --dpi option (by default, 96 dpi is assumed, unless the image itself contains dpi information).
• The % unit is generally relative to some available space. For example, the example above will render to the following:
  • HTML: <img href="file.jpg" style="width: 50%;" />
  • LaTeX: \includegraphics[width=0.5\textwidth,height=\textheight]{file.jpg} (if you're using a custom template, you need to configure graphicx as in the default template).
  • ConTeXt: \externalfigure[file.jpg][width=0.5\textwidth]
• Some output formats have a notion of a class (ConTeXt) or a unique identifier (LaTeX \caption), or both (HTML).
• When no width or height attributes are specified, the fallback is to look at the image resolution and the dpi metadata embedded in the image file.

🦋 Divs and Spans 🦋

Using the native_divs and native_spans extensions (see above, babes), HTML syntax can be used as part of Markdown to create native Div and Span elements in the pandoc AST (as opposed to raw HTML). But there's also a nicer syntax available, and we love her for that:

💜 Extension: fenced_divs 💜

This lets you use a special fenced syntax for native Div blocks. A Div starts with a fence containing at least three consecutive colons plus some attributes. Those attributes may optionally be followed by another string of consecutive colons.

Note: the commonmark parser doesn't permit colons after the attributes. 🙈

The attribute syntax is exactly like in fenced code blocks (see Extension: fenced_code_attributes). And just like with fenced code blocks, you can use either attributes in curly braces or a single unbraced word, which gets treated as a class name. The Div ends with another line containing a string of at least three consecutive colons. Oh, and the fenced Div should be separated by blank lines from the preceding and following blocks. Space, she needs space 🫶

Example:

::::: {#special .sidebar}
Here is a paragraph.

And another.
:::::

Fenced divs can be nested! Opening fences are distinguished because they must have attributes:

::: Warning ::::::
This is a warning.

::: Danger
This is a warning within a warning.
:::
::::::::::::::::::

Fences without attributes are always closing fences, periodt. And unlike with fenced code blocks, the number of colons in the closing fence doesn't need to match the number in the opening fence. That said, using fences of different lengths can be helpful for visual clarity, so you can tell nested divs apart from their parents. Main character organisation energy 👑

🌸 Extension: bracketed_spans 🌸

A bracketed sequence of inlines — the kind you'd use to begin a link — will be treated as a Span with attributes if it's followed immediately by attributes:

[This is *some text*]{.class key="val"}

🎀 Footnotes 🎀

💖 Extension: footnotes 💖

Pandoc's Markdown allows footnotes, using this syntax right here:

Here is a footnote reference,[^1] and another.[^longnote]

[^1]: Here is the footnote.

[^longnote]: Here's one with multiple blocks.

    Subsequent paragraphs are indented to show that they
belong to the previous footnote.

        { some.code }

    The whole paragraph can be indented, or just the first
    line.  In this way, multi-paragraph footnotes work like
    multi-paragraph list items.

This paragraph won't be part of the note, because it
isn't indented.

The identifiers in footnote references may not contain spaces, tabs, newlines, or the characters ^, [, or ]. These identifiers exist only to correlate the footnote reference with the note itself; in the output, footnotes get numbered sequentially. So cute, so tidy.

The footnotes themselves don't need to be at the end of the document. They can appear anywhere except inside other block elements (lists, block quotes, tables, etc.). Each footnote should be separated from surrounding content (including other footnotes) by blank lines.

🌈 Extension: inline_notes 🌈

Inline footnotes are allowed too (though, unlike regular notes, they cannot contain multiple paragraphs). The syntax goes like this:

Here is an inline note.^[Inline notes are easier to write, since
you don't have to pick an identifier and move down to type the
note.]

Inline and regular footnotes may be mixed freely. No drama, they coexist 💅

⭐ Citation syntax ⭐

🔥 Extension: citations 🔥

To cite a bibliographic item with an identifier foo, use the syntax @foo. Normal citations should be included in square brackets, with semicolons separating distinct items:

Blah blah [@doe99; @smith2000; @smith2004].

How this is rendered depends on the citation style. In an author-date style, it might render as

Blah blah (Doe 1999, Smith 2000, 2004).

In a footnote style, it might render as

Blah blah.[^1]

[^1]:  John Doe, "Frogs," *Journal of Amphibians* 44 (1999);
Susan Smith, "Flies," *Journal of Insects* (2000);
Susan Smith, "Bees," *Journal of Insects* (2004).

See the CSL user documentation for more information about CSL styles and how they affect rendering. 📚

Unless a citation key starts with a letter, digit, or _, and contains only alphanumerics and single internal punctuation characters (:.#$%&-+?<>~/), it must be surrounded by curly braces, which are not considered part of the key. In @Foo_bar.baz., the key is Foo_bar.baz because the final period is not internal punctuation, so it is not included in the key. In @{Foo_bar.baz.}, the key is Foo_bar.baz., including the final period. In @Foo_bar--baz, the key is Foo_bar because the repeated internal punctuation characters terminate the key. The curly braces are recommended if you use URLs as keys: [@{https://example.com/bib?name=foobar&date=2000}, p. 33].

Citation items may optionally include a prefix, a locator, and a suffix. In

Blah blah [see @doe99, pp. 33-35 and *passim*; @smith04, chap. 1].

the first item (doe99) has prefix see, locator pp. 33-35, and suffix and *passim*. The second item (smith04) has locator chap. 1 and no prefix or suffix.

Pandoc uses some heuristics to separate the locator from the rest of the subject. It's sensitive to the locator terms defined in the CSL locale files. Either abbreviated or unabbreviated forms are accepted. In the en-US locale, locator terms can be written in either singular or plural forms, as book, bk./bks.; chapter, chap./chaps.; column, col./cols.; figure, fig./figs.; folio, fol./fols.; number, no./nos.; line, l./ll.; note, n./nn.; opus, op./opp.; page, p./pp.; paragraph, para./paras.; part, pt./pts.; section, sec./secs.; sub verbo, s.v./s.vv.; verse, v./vv.; volume, vol./vols.; ¶/¶¶; §/§§. If no locator term is used, "page" is assumed. The way that this list just keeps going… iconic 💀

In complex cases, you can force something to be treated as a locator by enclosing it in curly braces, or prevent parsing the suffix as locator by prepending curly braces:

[@smith{ii, A, D-Z}, with a suffix]
[@smith, {pp. iv, vi-xi, (xv)-(xvii)} with suffix here]
[@smith{}, 99 years later]

A minus sign (-) before the @ will suppress mention of the author in the citation. This can be useful when the author is already mentioned in the text:

Smith says blah [-@smith04].

You can also write an author-in-text citation, by omitting the square brackets:

@smith04 says blah.

@smith04 [p. 33] says blah.

This will cause the author's name to be rendered, followed by the bibliographical details. Use this form when you want to make the citation the subject of a sentence. Main character citation energy ✨

When you're using a note style, it's usually better to let citeproc create the footnotes from citations rather than writing an explicit note. If you do write an explicit note that contains a citation, note that normal citations will be put in parentheses, while author-in-text citations will not. For this reason, it's sometimes preferable to use the author-in-text style inside notes when using a note style.

Many CSL styles will format citations differently when the same source has been cited earlier. In documents with chapters, it's usually desirable to reset this position information at the beginning of every chapter. To do this, add the class reset-citation-positions to the heading for each chapter:

# The Beginning {.reset-citation-positions}

Note that this class only has an effect when placed on top-level headings; it is ignored in nested blocks. Sorry bestie, top level only 🙈

💫 Non-default extensions 💫

The following Markdown syntax extensions are not enabled by default in pandoc, but you can enable them by adding +EXTENSION to the format name, where EXTENSION is the name of the extension. So, for example, markdown+hard_line_breaks is Markdown with hard line breaks. Little opt-in goodies, love that for us 🎀

🧚 Extension: rebase_relative_paths 🧚

Rewrite relative paths for Markdown links and images, depending on the path of the file containing the link or image link. For each link or image, pandoc will compute the directory of the containing file, relative to the working directory, and prepend the resulting path to the link or image path.

okay so the actual way to understand this extension is by example, bestie. Say you've got a subdirectory for every chapter of a book — chap1, chap2, chap3. Each one holds a file called text.md plus a bunch of images used in that chapter. You want ![image](spider.jpg) in chap1/text.md to point at chap1/spider.jpg, and the same ![image](spider.jpg) in chap2/text.md to point at chap2/spider.jpg. Here's the move:

pandoc chap*/*.md -f markdown+rebase_relative_paths

Without this extension you'd be stuck writing ![image](chap1/spider.jpg) inside chap1/text.md and ![image](chap2/spider.jpg) inside chap2/text.md like some kind of peasant 😭. Links with relative paths get rewritten exactly the same way as images, slay.

Absolute paths and URLs stay untouched. So do empty paths, and paths that are entirely a fragment, like #foo.

⚠️ Heads up: relative paths in reference links and images get rewritten relative to the file that holds the link reference definition, not the file that holds the reference link or image itself — assuming those are even different files. Plot twist 🫠

---

✨ Extension: mark ✨

To highlight a section of text, wrap it in == on both ends. Like so:

This ==is deleted text.==

highlighted. iconic. moving on. 💅

---

🎀 Extension: attributes 🎀

Lets you attach attributes to any inline or block-level element when you're parsing commonmark. The syntax for the attributes is identical to the one used in header_attributes.

• Attributes sitting immediately after an inline element affect that element. If they follow a space, then they belong to the space. (Which means, bestie, this option subsumes inline_code_attributes and link_attributes.)
• Attributes sitting immediately before a block element, on a line all by themselves, affect that element.
• You can stack consecutive attribute specifiers — for blocks or for inlines. Their attributes get combined. teamwork makes the dream work 🫶
• Attributes at the end of a Setext or ATX heading's text (separated from the text by whitespace) affect the heading element. (Yep — this option also subsumes header_attributes.)
• Attributes after the opening fence in a fenced code block affect the code block element. (And this one subsumes fenced_code_attributes. full circle moment ✨)
• Attributes at the end of a reference link definition affect the links that refer to that definition.

Note that pandoc's AST does not currently allow attributes to be attached to arbitrary elements. So a Span or Div container gets added when it's needed. they're doing their best 🙈

---

💀 Extension: old_dashes 💀

Selects the pandoc <= 1.8.2.1 behavior for parsing smart dashes: - before a numeral is an en-dash, and -- is an em-dash. This option only does anything if smart is enabled. It gets selected automatically for textile input.

---

🌈 Extension: angle_brackets_escapable 🌈

Allows < and > to be backslash-escaped, the way they can be in GitHub flavored Markdown but not in original Markdown. This is implied by pandoc's default all_symbols_escapable.

---

🍓 Extension: lists_without_preceding_blankline 🍓

Lets a list show up right after a paragraph, with no blank line in between. chaotic but valid.

---

🦋 Extension: four_space_rule 🦋

Selects the pandoc <= 2.0 behavior for parsing lists, so that four spaces of indent are required for list item continuation paragraphs. the old ways…

---

⭐ Extension: spaced_reference_links ⭐

Allows whitespace between the two components of a reference link. For example:

[foo] [bar].

---

💜 Extension: hard_line_breaks 💜

Causes all newlines within a paragraph to be interpreted as hard line breaks instead of spaces. no more soft launch 😌

---

🌸 Extension: ignore_line_breaks 🌸

Causes newlines within a paragraph to be ignored, rather than treated as spaces or as hard line breaks. This option is meant for East Asian languages where spaces aren't used between words, but text gets chopped into lines for readability. iykyk.

---

🎀 Extension: east_asian_line_breaks 🎀

Same idea — newlines within a paragraph get ignored, rather than treated as spaces or hard line breaks — but only when they occur between two East Asian wide characters. This is a better choice than ignore_line_breaks for texts that mix East Asian wide characters with other characters. the smarter sister extension 👑

---

✨ Extension: emoji ✨

Parses textual emojis like :smile: as Unicode emoticons. finally, the important stuff 🎀✨

---

🔥 Extension: tex_math_gfm 🔥

Supports two GitHub-specific formats for math. Inline math: $`e=mc^2`$.

Display math:

e=mc^2

---

🧚 Extension: tex_math_single_backslash 🧚

Causes anything between \( and \) to be interpreted as inline TeX math, and anything between \[ and \] to be interpreted as display TeX math.

⚠️ Note: the drawback here is that this extension precludes escaping ( and [. rip to those escapes 💀

---

💖 Extension: tex_math_double_backslash 💖

Causes anything between \\( and \\) to be interpreted as inline TeX math, and anything between \\[ and \\] to be interpreted as display TeX math. double the backslash, double the fun.

---

🌟 Extension: markdown_attribute 🌟

By default, pandoc interprets material inside block-level tags as Markdown. This extension flips the behavior so that Markdown is only parsed inside block-level tags when the tags carry the attribute markdown=1. opt-in only, consent matters 💅

---

📝 Extension: mmd_title_block 📝

Enables a MultiMarkdown style title block at the top of the document, like:

Title:   My title
Author:  John Doe
Date:    September 1, 2008
Comment: This is a sample mmd title block, with
         a field spanning multiple lines.

See the MultiMarkdown documentation for the deets. And important: if pandoc_title_block or yaml_metadata_block is enabled, it takes precedence over mmd_title_block. ranked choice voting but for title blocks.

---

🍓 Extension: abbreviations 🍓

Parses PHP Markdown Extra abbreviation keys, like:

*[HTML]: Hypertext Markup Language

Note that the pandoc document model does not support abbreviations, so if this extension is enabled, abbreviation keys are simply skipped — as opposed to being parsed as paragraphs. they get ghosted 👻

---

🔔 Extension: alerts 🔔

Supports GitHub-style Markdown alerts, like:

> [!TIP]
> Helpful advice for doing things better or more easily.

---

🦋 Extension: autolink_bare_uris 🦋

Makes all absolute URIs into links, even when they're not wrapped in pointy braces <...>. unhinged but convenient.

---

🌈 Extension: mmd_link_attributes 🌈

Parses MultiMarkdown-style key-value attributes on link and image references. Don't confuse this extension with the link_attributes extension, bestie, they are not the same.

This is a reference ![image][ref] with MultiMarkdown attributes.

[ref]: https://path.to/image "Image title" width=20px height=30px
       id=myId class="myClass1 myClass2"

---

👑 Extension: mmd_header_identifiers 👑

Parses MultiMarkdown-style heading identifiers — in square brackets, after the heading but before any trailing #s in an ATX heading.

---

📖 Extension: gutenberg 📖

Uses Project Gutenberg conventions for plain output: all-caps for strong emphasis, underscores surrounding regular emphasis, and extra blank space around headings. vintage vibes only 🕯️

---

🎯 Extension: sourcepos 🎯

Includes source position attributes when parsing commonmark or djot. For elements that accept attributes, a data-pos attribute gets added; other elements get placed in a surrounding Div or Span element with a data-pos attribute. everything gets a little name tag 🏷️

---

💫 Extension: short_subsuperscripts 💫

Parses MultiMarkdown-style subscripts and superscripts, which start with a `~' or `^' character respectively, and include the alphanumeric sequence that follows. For example:

x^2 = 4

or

Oxygen is O~2.

---

🌸 Extension: wikilinks_title_after_pipe 🌸

Pandoc supports multiple Markdown wikilink syntaxes, regardless of whether the title comes before or after the pipe. both queens are valid.

Using --from=markdown+wikilinks_title_after_pipe results in

[[URL|title]]

while using --from=markdown+wikilinks_title_before_pipe results in

[[title|URL]]

---

🎀 Markdown variants 🎀

On top of pandoc's extended Markdown, these Markdown variants are supported:

• markdown_phpextra (PHP Markdown Extra)
• markdown_github (deprecated GitHub-Flavored Markdown) she's deprecated bestie, we don't use her anymore
• markdown_mmd (MultiMarkdown)
• markdown_strict (Markdown.pl)
• commonmark (CommonMark)
• gfm (Github-Flavored Markdown)
• commonmark_x (CommonMark with many pandoc extensions)

To see which extensions are supported for a given format — and which ones are on by default — run:

pandoc --list-extensions=FORMAT

where FORMAT is replaced with the name of the format. fill in the blank, bestie ✏️

Note that the list of extensions for commonmark, gfm, and commonmark_x is defined relative to default commonmark. So, for example, backtick_code_blocks doesn't show up as an extension, since it's enabled by default and can't be disabled. it's not optional, it's the law ⚖️

---

🌟 CITATIONS 🌟

When the --citeproc option is used, pandoc can automatically generate citations and a bibliography in a whole bunch of styles. Basic usage is:

pandoc --citeproc myinput.txt

To use this feature, you will need to have:

• a document containing citations (see Citation syntax);
• a source of bibliographic data: either an external bibliography file, or a list of references in the document's YAML metadata;
• optionally, a CSL citation style. optional but make it fashion 💅

---

📚 Specifying bibliographic data 📚

You can point at an external bibliography using the bibliography metadata field in a YAML metadata section, or via the --bibliography command line argument. Want multiple bibliography files? Just supply multiple --bibliography arguments, or set the bibliography metadata field to a YAML array. collect them all 🎀

A bibliography may have any of these formats:

• BibLaTeX — .bib
• BibTeX — .bibtex
• CSL JSON — .json
• CSL YAML — .yaml
• RIS — .ris

Note that .bib works with both BibTeX and BibLaTeX files; use the extension .bibtex to force interpretation as BibTeX. spelling it out explicitly slays

In BibTeX and BibLaTeX databases, pandoc parses LaTeX markup inside fields such as title; in CSL YAML databases, pandoc Markdown; and in CSL JSON databases, an HTML-like markup:

• <i>...</i> — italics
• <b>...</b> — bold
• <span style="font-variant:small-caps;">...</span> or <sc>...</sc> — small capitals
• <sub>...</sub> — subscript
• <sup>...</sup> — superscript
• <span class="nocase">...</span> — prevent a phrase from being capitalized as title case mic drop, no notes

As an alternative to specifying a bibliography file using --bibliography or the YAML metadata field bibliography, you can include the citation data directly in the references field of the document's YAML metadata. That field should contain an array of YAML-encoded references, for example:

---
references:
- type: article-journal
  id: WatsonCrick1953
  author:
  - family: Watson
    given: J. D.
  - family: Crick
    given: F. H. C.
  issued:
    date-parts:
    - - 1953
      - 4
      - 25
  title: 'Molecular structure of nucleic acids: a structure for
    deoxyribose nucleic acid'
  title-short: Molecular structure of nucleic acids
  container-title: Nature
  volume: 171
  issue: 4356
  page: 737-738
  DOI: 10.1038/171737a0
  URL: https://www.nature.com/articles/171737a0
  language: en-GB
...

So if you've got BOTH an external bibliography AND inline references chilling in your YAML metadata? Bestie, they're both getting used. 💅 But heads up — if the ids clash, the inline references win, no debate, periodt.

Also like, lowkey iconic: pandoc can literally cook you up a YAML metadata section just like that from a BibTeX, BibLaTeX, or CSL JSON bibliography:

pandoc chem.bib -s -f biblatex -t markdown
pandoc chem.json -s -f csljson -t markdown

And fr fr, pandoc converts between any of these citation formats:

pandoc chem.bib -s -f biblatex -t csljson
pandoc chem.yaml -s -f markdown -t biblatex

Running pandoc on a bibliography file with the --citeproc option gives you a formatted bibliography in whatever format you want, your choice, main character energy:

pandoc chem.bib -s --citeproc -o chem.html
pandoc chem.bib -s --citeproc -o chem.pdf

---

✨ Capitalization in titles ✨

Okay so this part is giving "rules I need you to actually follow." If you're using a bibtex or biblatex bibliography, observe the following:

• English titles should be in title case. Non-English titles should be in sentence case, and the langid field in biblatex should be set to the relevant language. (The following values are treated as English: american, british, canadian, english, australian, newzealand, USenglish, or UKenglish.)
• As is standard with bibtex/biblatex, proper names should be protected with curly braces so that they won't be lowercased in styles that call for sentence case. Like this:

title = {My Dinner with {Andre}}

• In addition, words that should remain lowercase (or camelCase) should be protected:

title = {Spin Wave Dispersion on the {nm} Scale}

Even though this isn't necessary in bibtex/biblatex, it is necessary with citeproc, which stores titles internally in sentence case, and converts to title case in styles that require it. Here we protect “nm” so that it doesn't get converted to “Nm” at this stage. 🎀

If you're using a CSL bibliography (either JSON or YAML), then observe these:

• All titles should be in sentence case.
• Use the language field for non-English titles to prevent their conversion to title case in styles that call for this. (Conversion happens only if language begins with en or is left empty.)
• Protect words that should not be converted to title case using this syntax:

Spin wave dispersion on the <span class="nocase">nm</span> scale

---

🎤 Conference Papers, Published vs. Unpublished 🎤

For a formally published conference paper, use the biblatex entry type inproceedings (which will be mapped to CSL paper-conference).

For an unpublished manuscript, use the biblatex entry type unpublished without an eventtitle field (this entry type will be mapped to CSL manuscript).

For a talk, an unpublished conference paper, or a poster presentation, use the biblatex entry type unpublished with an eventtitle field (this entry type will be mapped to CSL speech). Use the biblatex type field to indicate the type, e.g. “Paper”, or “Poster”. venue and eventdate may be useful too, though eventdate will not be rendered by most CSL styles. 🫶 Note that venue is for the event's venue, unlike location which describes the publisher's location; do not use the latter for an unpublished conference paper.

---

💅 Specifying a citation style 💅

Citations and references can be formatted using any style supported by the Citation Style Language, listed in the Zotero Style Repository. These files are specified using the --csl option or the csl (or citation-style) metadata field. By default, pandoc will use the Chicago Manual of Style author-date format. (You can override this default by copying a CSL style of your choice to default.csl in your user data directory.) The CSL project provides further information on finding and editing styles.

The --citation-abbreviations option (or the citation-abbreviations metadata field) may be used to specify a JSON file containing abbreviations of journals that should be used in formatted bibliographies when form="short" is specified. The format of the file can be illustrated with an example:

{ "default": {
    "container-title": {
            "Lloyd's Law Reports": "Lloyd's Rep",
            "Estates Gazette": "EG",
            "Scots Law Times": "SLT"
    }
  }
}

---

📝 Citations in note styles 📝

pandoc's citation processing is designed to allow you to move between author-date, numerical, and note styles without modifying the Markdown source. Obsessed. When you're using a note style, avoid inserting footnotes manually. Instead, insert citations just as you would in an author-date style — for example,

Blah blah [@foo, p. 33].

The footnote will be created automatically. (pandoc literally handles it for you, no cap.) pandoc will take care of removing the space and moving the note before or after the period, depending on the setting of notes-after-punctuation, as described below in Other relevant metadata fields.

In some cases you may need to put a citation inside a regular footnote. Normal citations in footnotes (such as [@foo, p. 33]) will be rendered in parentheses. In-text citations (such as @foo [p. 33]) will be rendered without parentheses. (A comma will be added if appropriate.) Thus:

[^1]:  Some studies [@foo; @bar, p. 33] show that
frubulicious zoosnaps are quantical.  For a survey
of the literature, see @baz [chap. 1].

---

📚 Placement of the bibliography 📚

If the style calls for a list of works cited, it will be placed in a div with id refs, if one exists:

::: {#refs}
:::

Otherwise, it will be placed at the end of the document. Generation of the bibliography can be suppressed by setting suppress-bibliography: true in the YAML metadata.

If you wish the bibliography to have a section heading, you can set reference-section-title in the metadata, or put the heading at the beginning of the div with id refs (if you are using it) or at the end of your document:

last paragraph...

# References

The bibliography will be inserted after this heading. Note that the unnumbered class will be added to this heading, so that the section will not be numbered. ✨

If you want to put the bibliography into a variable in your template, one way to do that is to put the div with id refs into a metadata field, e.g.

---
refs: |
   ::: {#refs}
   :::
...

You can then put the variable $refs$ into your template where you want the bibliography to be placed. 🎀

---

🌸 Including uncited items in the bibliography 🌸

If you want to include items in the bibliography without actually citing them in the body text, you can define a dummy nocite metadata field and put the citations there:

---
nocite: |
  @item1, @item2
...

@item3

In this example, the document will contain a citation for item3 only, but the bibliography will contain entries for item1, item2, and item3.

It is possible to create a bibliography with all the citations, whether or not they appear in the document, by using a wildcard:

---
nocite: |
  @*
...

For LaTeX output, you can also use natbib or biblatex to render the bibliography. In order to do so, specify bibliography files as outlined above, and add --natbib or --biblatex argument to pandoc invocation. Bear in mind that bibliography files have to be in either BibTeX (for --natbib) or BibLaTeX (for --biblatex) format. 💜

---

👑 Other relevant metadata fields 👑

A few other metadata fields affect bibliography formatting:

• link-citations — If true, citations will be hyperlinked to the corresponding bibliography entries (for author-date and numerical styles only). Defaults to false.
• link-bibliography — If true, DOIs, PMCIDs, PMID, and URLs in bibliographies will be rendered as hyperlinks. (If an entry contains a DOI, PMCID, PMID, or URL, but none of these fields are rendered by the style, then the title, or in the absence of a title the whole entry, will be hyperlinked.) Defaults to true.
• lang — The lang field will affect how the style is localized, for example in the translation of labels, the use of quotation marks, and the way items are sorted. (For backwards compatibility, locale may be used instead of lang, but this use is deprecated.) A BCP 47 language tag is expected: for example, en, de, en-US, fr-CA, ug-Cyrl. The unicode extension syntax (after -u-) may be used to specify options for collation (sorting) more precisely. Here are some examples:
  • zh-u-co-pinyin: Chinese with the Pinyin collation.
  • es-u-co-trad: Spanish with the traditional collation (with Ch sorting after C).
  • fr-u-kb: French with “backwards” accent sorting (with coté sorting after côte).
  • en-US-u-kf-upper: English with uppercase letters sorting before lower (default is lower before upper).
• notes-after-punctuation — If true (the default for note styles), pandoc will put footnote references or superscripted numerical citations after following punctuation. For example, if the source contains blah blah [@jones99]., the result will look like blah blah.[^1], with the note moved after the period and the space collapsed. If false, the space will still be collapsed, but the footnote will not be moved after the punctuation. The option may also be used in numerical styles that use superscripts for citation numbers (but for these styles the default is not to move the citation).

✨ SLIDE SHOWS ✨

ok so bestie, hear me out — pandoc can make you a whole HTML + JavaScript slide presentation that lives in a web browser 🎀. There are five ways to do it: S5, DZSlides, Slidy, Slideous, or reveal.js. And if that's not enough, you can also cook up a PDF slide show with LaTeX beamer, or slides in Microsoft PowerPoint format. main character energy fr 💅

Here's the Markdown source for a little baby slide show, habits.txt ✨:

% Habits
% John Doe
% March 22, 2005

# In the morning

## Getting up

- Turn off alarm
- Get out of bed

## Breakfast

- Eat eggs
- Drink coffee

# In the evening

## Dinner

- Eat spaghetti
- Drink wine

------------------

![picture of spaghetti](images/spaghetti.jpg)

## Going to sleep

- Get in bed
- Count sheep

To produce an HTML/JavaScript slide show, you literally just type 🌸:

pandoc -t FORMAT -s habits.txt -o habits.html

where FORMAT is either s5, slidy, slideous, dzslides, or revealjs.

For Slidy, Slideous, reveal.js, and S5, the file pandoc spits out with the -s/--standalone option embeds a link to JavaScript and CSS files, which are assumed to be available at the relative path s5/default (for S5), slideous (for Slideous), reveal.js (for reveal.js), or at the Slidy website at w3.org (for Slidy). (These paths can be changed by setting the slidy-url, slideous-url, revealjs-url, or s5-url variables; see Variables for HTML slides, above.) For DZSlides, the (relatively short) JavaScript and CSS are included in the file by default. no extra setup, we love that 🫶

With all HTML slide formats, the --self-contained option can be used to produce a single file that contains all of the data necessary to display the slide show, including linked scripts, stylesheets, images, and videos. one file, all the vibes, iconic 💖

To produce a PDF slide show using beamer, type:

pandoc -t beamer habits.txt -o habits.pdf

Note that a reveal.js slide show can also be converted to a PDF by printing it to a file from the browser. 🖨️

To produce a PowerPoint slide show, type:

pandoc habits.txt -o habits.pptx

---

🦋 Structuring the slide show 🦋

By default, the slide level is the highest heading level in the hierarchy that is followed immediately by content, and not another heading, somewhere in the document. In the example above, level-1 headings are always followed by level-2 headings, which are followed by content, so the slide level is 2. This default can be overridden using the --slide-level option. slay, you're in charge 👑

The document is carved up into slides according to the following rules:

• A horizontal rule always starts a new slide.
• A heading at the slide level always starts a new slide.
• Headings below the slide level in the hierarchy create headings within a slide. (In beamer, a "block" will be created. If the heading has the class example, an exampleblock environment will be used; if it has the class alert, an alertblock will be used; otherwise a regular block will be used.)
• Headings above the slide level in the hierarchy create "title slides," which just contain the section title and help to break the slide show into sections. Non-slide content under these headings will be included on the title slide (for HTML slide shows) or in a subsequent slide with the same title (for beamer).
• A title page is constructed automatically from the document's title block, if present. (In the case of beamer, this can be disabled by commenting out some lines in the default template.)

These rules are designed to support many different styles of slide show. If you don't care about structuring your slides into sections and subsections, you can either just use level-1 headings for all slides (in that case, level 1 will be the slide level) or you can set --slide-level=0. do whatever feels right bestie 💅

Note: in reveal.js slide shows, if slide level is 2, a two-dimensional layout will be produced, with level-1 headings building horizontally and level-2 headings building vertically. It is not recommended that you use deeper nesting of section levels with reveal.js unless you set --slide-level=0 (which lets reveal.js produce a one-dimensional layout and only interprets horizontal rules as slide boundaries). 💀 don't fight the tool, she knows what she's doing

---

🌈 PowerPoint layout choice 🌈

When creating slides, the pptx writer chooses from a number of pre-defined layouts, based on the content of the slide:

• Title Slide — This layout is used for the initial slide, which is generated and filled from the metadata fields date, author, and title, if they are present.
• Section Header — This layout is used for what pandoc calls "title slides", i.e. slides which start with a header which is above the slide level in the hierarchy.
• Two Content — This layout is used for two-column slides, i.e. slides containing a div with class columns which contains at least two divs with class column.
• Comparison — This layout is used instead of "Two Content" for any two-column slides in which at least one column contains text followed by non-text (e.g. an image or a table).
• Content with Caption — This layout is used for any non-two-column slides which contain text followed by non-text (e.g. an image or a table).
• Blank — This layout is used for any slides which only contain blank content, e.g. a slide containing only speaker notes, or a slide containing only a non-breaking space.
• Title and Content — This layout is used for all slides which do not match the criteria for another layout.

These layouts are chosen from the default pptx reference doc included with pandoc, unless an alternative reference doc is specified using --reference-doc. 🎀

---

💜 Incremental lists 💜

By default, these writers produce lists that display "all at once." If you want your lists to display incrementally (one item at a time), use the -i option. If you want a particular list to depart from the default, put it in a div block with class incremental or nonincremental. So, for example, using the fenced div syntax, the following would be incremental regardless of the document default:

::: incremental

- Eat spaghetti
- Drink wine

:::

or

::: nonincremental

- Eat spaghetti
- Drink wine

:::

While using incremental and nonincremental divs is the recommended method of setting incremental lists on a per-case basis, an older method is also supported: putting lists inside a blockquote will depart from the document default (that is, it will display incrementally without the -i option and all at once with the -i option):

> - Eat spaghetti
> - Drink wine

Both methods allow incremental and nonincremental lists to be mixed in a single document. vers to my multitasking besties 🫶

If you want to include a block-quoted list, you can work around this behavior by putting the list inside a fenced div, so that it is not the direct child of the block quote:

> ::: wrapper
> - a
> - list in a quote
> :::

---

⭐ Inserting pauses ⭐

You can add "pauses" within a slide by including a paragraph containing three dots, separated by spaces:

# Slide with a pause

content before the pause

. . .

content after the pause

Note: this feature is not yet implemented for PowerPoint output. 😭 soon hopefully

---

🎀 Styling the slides 🎀

You can change the style of HTML slides by putting customized CSS files in $DATADIR/s5/default (for S5), $DATADIR/slidy (for Slidy), or $DATADIR/slideous (for Slideous), where $DATADIR is the user data directory (see --data-dir, above). The originals may be found in pandoc's system data directory (generally $CABALDIR/pandoc-VERSION/s5/default). Pandoc will look there for any files it does not find in the user data directory. ✨

For dzslides, the CSS is included in the HTML file itself, and may be modified there.

All reveal.js configuration options can be set through variables. For example, themes can be used by setting the theme variable:

-V theme=moon

Or you can specify a custom stylesheet using the --css option. 🌸

To style beamer slides, you can specify a theme, colortheme, fonttheme, innertheme, and outertheme, using the -V option:

pandoc -t beamer habits.txt -V theme:Warsaw -o habits.pdf

Note that heading attributes will turn into slide attributes (on a <div> or <section>) in HTML slide formats, allowing you to style individual slides. In beamer, a number of heading classes and attributes are recognized as frame options and will be passed through as options to the frame: see Frame attributes in beamer, below. 👑

---

🦋 Speaker notes 🦋

Speaker notes are supported in reveal.js, PowerPoint (pptx), and beamer output. You can add notes to your Markdown document thus:

::: notes

This is my note.

- It can contain Markdown
- like this list

:::

To show the notes window in reveal.js, press s while viewing the presentation. Speaker notes in PowerPoint will be available, as usual, in handouts and presenter view. 🫶

Notes are not yet supported for other slide formats, but the notes will not appear on the slides themselves. (lowkey a win — no accidental spoilers)

💖 Speaker notes on the title slide (PowerPoint) 💖

For PowerPoint output, the title slide is generated from the document's YAML metadata block. To add speaker notes to this slide, use a notes field in the metadata:

---
title: My Presentation
author: Jane Doe
notes: |
  Welcome everyone to this presentation.

  Remember to introduce yourself and mention the key topics.
---

The notes field can contain multiple paragraphs and Markdown formatting. ✨

---

🌈 Columns 🌈

To put material in side by side columns, you can use a native div container with class columns, containing two or more div containers with class column and a width attribute:

:::::::::::::: {.columns}
::: {.column width="40%"}
contents...
:::
::: {.column width="60%"}
contents...
:::
::::::::::::::

Note: Specifying column widths does not currently work for PowerPoint. 💀

⭐ Additional columns attributes in beamer ⭐

The div containers with classes columns and column can optionally have an align attribute. The class columns can optionally have a totalwidth attribute or an onlytextwidth class.

...and you can stack these babies too, nesting columns like it's nothing. Watch:

:::::::::::::: {.columns align=center totalwidth=8em}
::: {.column width="40%"}
contents...
:::
::: {.column width="60%" align=bottom}
contents...
:::
::::::::::::::

The align attributes on columns and column take the values top, top-baseline, center and bottom to vertically align the columns. It defaults to top on columns. ✨

The totalwidth attribute limits the width of the columns to the given value.

:::::::::::::: {.columns align=top .onlytextwidth}
::: {.column width="40%" align=center}
contents...
:::
::: {.column width="60%"}
contents...
:::
::::::::::::::

The class onlytextwidth sets the totalwidth to \textwidth. 💅

See Section 12.7 of the Beamer User's Guide for more details.

---

🎀 Frame attributes in beamer

Sometimes it is necessary to add the LaTeX [fragile] option to a frame in beamer (for example, when using the minted environment). This can be forced by adding the fragile class to the heading introducing the slide:

# Fragile slide {.fragile}

All of the other frame attributes described in Section 8.1 of the Beamer User's Guide may also be used: allowdisplaybreaks, allowframebreaks, b, c, s, t, environment, label, plain, shrink, standout, noframenumbering, squeeze. allowframebreaks is recommended especially for bibliographies, as it allows multiple slides to be created if the content overfills the frame:

# References {.allowframebreaks}

In addition, the frameoptions attribute may be used to pass arbitrary frame options to a beamer slide:

# Heading {frameoptions="squeeze,shrink,customoption=foobar"}

---

🦋 Background in reveal.js, beamer, and pptx

Background images can be added to self-contained reveal.js slide shows, beamer slide shows, and pptx slide shows. yes bestie, all three, we're eating good 🔥

On all slides (beamer, reveal.js, pptx)

With beamer and reveal.js, the configuration option background-image can be used either in the YAML metadata block or as a command-line variable to get the same image on every single slide.

Note that for reveal.js, the background-image will be used as a parallaxBackgroundImage (see below). 👀

For pptx, you can use a --reference-doc in which background images have been set on the relevant layouts.

parallaxBackgroundImage (reveal.js)

For reveal.js, there is also the reveal.js-native option parallaxBackgroundImage, which produces a parallax scrolling background. That's the fancy one. 💫 You must also set parallaxBackgroundSize, and can optionally set parallaxBackgroundHorizontal and parallaxBackgroundVertical to configure the scrolling behaviour. See the reveal.js documentation for more details about the meaning of these options.

In reveal.js's overview mode, the parallaxBackgroundImage will show up only on the first slide.

On individual slides (reveal.js, pptx)

To set an image for a particular reveal.js or pptx slide, add {background-image="/path/to/image"} to the first slide-level heading on the slide (which may even be empty).

As the HTML writers pass unknown attributes through, other reveal.js background settings also work on individual slides, including background-size, background-repeat, background-color, transition, and transition-speed. (The data- prefix will automatically be added.)

Note: data-background-image is also supported in pptx for consistency with reveal.js – if background-image isn't found, data-background-image will be checked. 🫶

On the title slide (reveal.js, pptx)

To add a background image to the automatically generated title slide for reveal.js, use the title-slide-attributes variable in the YAML metadata block. It must contain a map of attribute names and values. (Note that the data- prefix is required here, as it isn't added automatically. don't forget it bestie 😭)

For pptx, pass a --reference-doc with the background image set on the "Title Slide" layout.

Example (reveal.js)

---
title: My Slide Show
parallaxBackgroundImage: /path/to/my/background_image.png
title-slide-attributes:
    data-background-image: /path/to/title_image.png
    data-background-size: contain
---

## Slide One

Slide 1 has background_image.png as its background.

## {background-image="/path/to/special_image.jpg"}

Slide 2 has a special image for its background, even though the heading has no content.

Look at that — slide two has a whole look going on and its heading is literally empty. iconic. 👑

🦋 EPUBS 🦋

📚 EPUB Metadata

okay so there are literally two ways to give your EPUB its metadata, and bestie, you get to pick:

1. the --epub-metadata option — its argument is an XML file packed with Dublin Core elements ✨
2. YAML — either stuffed into a YAML metadata block inside a Markdown doc, or living in a whole separate YAML file that you hand over with --metadata-file 🎀

here's what a YAML metadata block with EPUB metadata looks like, iconic behavior:

---
title:
- type: main
  text: My Book
- type: subtitle
  text: An investigation of metadata
creator:
- role: author
  text: John Smith
- role: editor
  text: Sarah Jones
identifier:
- scheme: DOI
  text: 'urn:doi:10.234234.234/33'
- scheme: ISBN-13
  text: 'urn:isbn:9780000000000'
publisher:  My Press
rights: © 2007 John Smith, CC BY-NC
ibooks:
  version: 1.3.4
...

these are the fields that get recognized, no cap:

• identifier — either a string value, or an object with fields text and scheme (optional), or a list of such objects. valid values for scheme are ISBN-10, GTIN-13, UPC, ISMN-10, DOI, LCCN, GTIN-14, ISBN-13, Legal deposit number, URN, OCLC number, Co-publisher’s ISBN-13, ISMN-13, ISBN-A, JP e-code, OLCC number, JP Magazine ID, UPC-12+5, BNF Control number, ISSN-13, ARK, Digital file internal version number. in the text field it's advisable to specify a URN, like in the examples up there 👆. also heads up: quotes are needed around a YAML field containing a colon 💅
• title — either a string value, or an object with fields text, file-as (optional) and type (optional), or a list of such objects. valid values for type are main, subtitle, short, collection, edition, extended
• creator — either a string value, or an object with fields text, role (optional), and file-as (optional), or a list of such objects. valid values for role are MARC relators, but pandoc will attempt to translate the human-readable versions (like “author” and “editor”) to the appropriate marc relators 🫶
• contributor — same format as creator
• date — a string value in YYYY-MM-DD format. (only the year is necessary.) pandoc will attempt to convert other common date formats
• lang (or legacy: language) — a string value in BCP 47 format. pandoc will default to the local language if nothing is specified
• subject — either a string value, or an object with fields text, authority (optional), and term (optional), or a list of such objects. valid values for authority are either a reserved authority value (currently AAT, BIC, BISAC, CLC, DDC, CLIL, EuroVoc, MEDTOP, LCSH, NDC, Thema, UDC, and WGS) or an absolute IRI identifying a custom scheme. valid values for term are defined by the scheme
• description — a string value
• type — a string value
• format — a string value
• relation — a string value
• coverage — a string value
• rights — a string value
• belongs-to-collection — a string value. identifies the name of a collection to which the EPUB Publication belongs
• group-position — the group-position field indicates the numeric position in which the EPUB Publication belongs relative to other works belonging to the same belongs-to-collection field
• cover-image — a string value (path to cover image)
• css (or legacy: stylesheet) — a string value (path to CSS stylesheet)
• page-progression-direction — either ltr or rtl. specifies the page-progression-direction attribute for the spine element
• accessModes — an array of strings (schema). defaults to ["textual"]
• accessModeSufficient — an array of strings (schema). defaults to ["textual"]
• accessibilityHazards — an array of strings (schema). defaults to ["none"]
• accessibilityFeatures — an array of strings (schema). defaults to

- "alternativeText"
- "readingOrder"
- "structuralNavigation"
- "tableOfContents"

• accessibilitySummary — a string value
• ibooks — iBooks-specific metadata, with the following fields:
  • version: (string)
  • specified-fonts: true|false (default false)
  • ipad-orientation-lock: portrait-only|landscape-only
  • iphone-orientation-lock: portrait-only|landscape-only
  • binding: true|false (default true)
  • scroll-axis: vertical|horizontal|default

---

🏷️ The epub:type attribute

for epub3 output you can mark up the heading that corresponds to an EPUB chapter using the epub:type attribute ✨ like, if you wanna set that attribute to the value prologue, you write this Markdown:

# My chapter {epub:type=prologue}

which gives you this:

<body epub:type="frontmatter">
  <section epub:type="prologue">
    <h1>My chapter</h1>

pandoc will output <body epub:type="bodymatter">, unless you use one of the following values — in which case either frontmatter or backmatter gets output instead 👀

here's the mapping, and bestie it's giving very precise:

• epub:type of first section → prologue gives epub:type of body → frontmatter
• abstract → frontmatter
• acknowledgments → frontmatter
• copyright-page → frontmatter
• dedication → frontmatter
• credits → frontmatter
• keywords → frontmatter
• imprint → frontmatter
• contributors → frontmatter
• other-credits → frontmatter
• errata → frontmatter
• revision-history → frontmatter
• titlepage → frontmatter
• halftitlepage → frontmatter
• seriespage → frontmatter
• foreword → frontmatter
• preface → frontmatter
• frontispiece → frontmatter
• appendix → backmatter
• colophon → backmatter
• bibliography → backmatter
• index → backmatter

---

🎧 Linked media

by default, pandoc will download media referenced from any <img>, <audio>, <video> or <source> element present in the generated EPUB, and include it in the EPUB container — yielding a completely self-contained EPUB, periodt 🫶

but if you wanna link to external media resources instead, use raw HTML in your source and add data-external="1" to the tag with the src attribute. like so:

<audio controls="1">
  <source src="https://example.com/music/toccata.mp3"
          data-external="1" type="audio/mpeg">
  </source>
</audio>

if the input format already is HTML then data-external="1" will work as expected for <img> elements 🖼️ similarly, for Markdown, external images can be declared with ![img](url){external=1}. note that this only works for images — the other media elements have no native representation in pandoc’s AST and require the use of raw HTML, so like, no shortcuts there bestie 😭

---

💅 EPUB styling

by default pandoc will include some basic styling contained in its epub.css data file. (to see this, use pandoc --print-default-data-file epub.css.) to use a different CSS file, just use the --css command line option 💖

a few inline styles are defined in addition; these are essential for correct formatting of pandoc’s HTML output, so don't come for them.

the document-css variable may be set if the more opinionated styling of pandoc’s default HTML templates is desired (and in that case the variables defined in Variables for HTML may be used to fine-tune the style) ✨

---

📦 CHUNKED HTML

pandoc -t chunkedhtml will produce a zip archive of linked HTML files — one for each section of the original document 🌈 internal links will automatically be adjusted to point to the right place, images linked to under the working directory will be incorporated, and navigation links will be added. in addition, a JSON file sitemap.json will be included describing the hierarchical structure of the files 👑

if an output file without an extension is specified, then it will be interpreted as a directory and the zip archive will be automatically unpacked into it (unless it already exists, in which case an error will be raised 🚨). otherwise a .zip file will be produced.

the navigation links can be customized by adjusting the template 🎀 by default, a table of contents is included only on the top page. to include it on every page, set the toc variable manually.

---

📖 MANPAGE

📓 JUPYTER NOTEBOOKS 📓

ok bestie so when you're making a Jupyter notebook, pandoc literally plays detective and tries to infer the whole notebook structure by itself, no cap. Here's the tea:

• Code blocks with the class code get taken as code cells 👑
• And whatever content sits between them becomes Markdown cells ✨
• Attachments get auto-created for images that live in Markdown cells 🖼️
• Metadata gets pulled from the jupyter metadata field 💅

Like, for example:

---
title: My notebook
jupyter:
  nbformat: 4
  nbformat_minor: 5
  kernelspec:
     display_name: Python 2
     language: python
     name: python2
  language_info:
     codemirror_mode:
       name: ipython
       version: 2
     file_extension: ".py"
     mimetype: "text/x-python"
     name: "python"
     nbconvert_exporter: "python"
     pygments_lexer: "ipython2"
     version: "2.7.15"
---

# Lorem ipsum

**Lorem ipsum** dolor sit amet, consectetur adipiscing elit. Nunc luctus
bibendum felis dictum sodales.

print("hello")

## Pyout

from IPython.display import HTML HTML(""" <script> console.log("hello"); </script> <b>HTML</b> """)

## Image

This image ![image](myimage.png) will be
included as a cell attachment.

But periodt — if you wanna add cell attributes, group cells differently, or put output into code cells, then you gotta bring in divs to show the structure. Fenced divs or native divs, either one works, slay. Here's the vibe:

:::::: {.cell .markdown}
# Lorem

**Lorem ipsum** dolor sit amet, consectetur adipiscing elit. Nunc luctus
bibendum felis dictum sodales.
::::::

:::::: {.cell .code execution_count=1}

print("hello")

::: {.output .stream .stdout}

hello

:::
::::::

:::::: {.cell .code execution_count=2}

from IPython.display import HTML HTML(""" <script> console.log("hello"); </script> <b>HTML</b> """)

::: {.output .execute_result execution_count=2}

<script> console.log("hello"); </script> <b>HTML</b> hello

:::
::::::

k so if you're putting raw HTML or TeX into an output cell, you gotta use the raw attribute, exactly like that last cell in the example above shows. Like yeah, pandoc can process "bare" raw HTML and TeX, but the result ends up being a messy mix of raw elements and normal textual elements all interleaved 😭, and in an output cell pandoc expects one single, connected raw block. So to avoid raw HTML or TeX sneaking in except when it's explicitly marked with raw attributes, we recommend you specify the extensions -raw_html-raw_tex+raw_attribute when you're translating between Markdown and ipynb notebooks. fr fr.

heads up: any options and extensions that affect reading and writing of Markdown will also affect Markdown cells in ipynb notebooks. So like, --wrap=preserve will preserve soft line breaks in Markdown cells; --markdown-headings=setext makes Setext-style headings get used; and --preserve-tabs stops tabs from being turned into spaces. 💖

---

💜 VIMDOC 💜

The Vimdoc writer generates Vim help files, and it makes use of these metadata variables:

abstract: "A short description"
author: Author
title: Title

# Vimdoc-specific
filename: "definition-lists.txt"
vimdoc-prefix: pandoc

A complete header requires abstract, author, title and filename to all be set. If you compile a file with that metadata, you get this file out the other side (it assumes --standalone, see Templates):

*definition-lists.txt*  A short description

                            Title by Author

                                 Type |gO| to see the table of contents.

[...]

 vim:tw=72:sw=4:ts=4:ft=help:norl:et:

Now if vimdoc-prefix is set, all non-command tags get prefixed with its value 🏷️. It's used to prevent tag collision: every header has a tag (either inferred or explicit) and multiple help pages can totally have the same header names, so collision is to be expected, bestie. Let our input be this markdown file:

## Header

`:[range]Fnl {expr}`{#:Fnl}
:   Evaluates {expr} or range

`vim.b`{#vim.b}
:   Buffer-scoped (`:h b:`) variables for the current buffer. Invalid or unset
    key returns `nil`. Can be indexed with an integer to access variables for a
    specific buffer.

[Span]{#span}
:   generic inline container for phrasing content, which does not inherently
    represent anything.

Convert it to vimdoc:

------------------------------------------------------------------------
Header                                                            *header*

:[range]Fnl {expr}                                                  *:Fnl*
    Evaluates {expr} or range
`vim.b`                                                            *vim.b*
    Buffer-scoped (|b:|) variables for the current buffer. Invalid or
    unset key returns `nil`. Can be indexed with an integer to access
    variables for a specific buffer.
Span                                                                *span*
    generic inline container for phrasing content, which does not
    inherently represent anything.

And now convert it to vimdoc with the metadata variable set (e.g. with -M vimdoc-prefix=pandoc):

------------------------------------------------------------------------
Header                                                     *pandoc-header*

:[range]Fnl {expr}                                                  *:Fnl*
    Evaluates {expr} or range
`vim.b`                                                     *pandoc-vim.b*
    Buffer-scoped (|b:|) variables for the current buffer. Invalid or
    unset key returns `nil`. Can be indexed with an integer to access
    variables for a specific buffer.
Span                                                         *pandoc-span*
    generic inline container for phrasing content, which does not
    inherently represent anything.

See how vim.b and Span got their prefixes but :Fnl did not? That's because ex-commands (the ones starting with :) don't get a prefix, since they're considered unique across help pages ✨.

In both cases :help b: became the reference |b:| (and yeah, it also works with :h b:). Links pointing to either https://vimhelp.org/ or https://neovim.io/doc/user become references too 🦋.

Oh and Vim traditionally wraps at 78, but Pandoc defaults to 72. So use --columns 78 to match Vim. 🌸

---

🌈 SYNTAX HIGHLIGHTING 🌈

Pandoc will automatically highlight syntax in fenced code blocks that are marked with a language name, iconic behavior honestly. The Haskell library skylighting is what does the highlighting. Currently highlighting is supported only for HTML, EPUB, Docx, Ms, Man, Typst, and LaTeX/PDF output. Wanna see a list of language names that pandoc will recognize? Type pandoc --list-highlight-languages. 💅

The color scheme gets picked with the --syntax-highlighting option. The default color scheme is pygments, which imitates the default color scheme used by the Python library pygments (though pygments is not actually used to do the highlighting, plot twist). To see a list of highlight styles, type pandoc --list-highlight-styles. ⭐

Not satisfied with the predefined styles? No problem bestie — you can use --print-highlight-style to generate a JSON .theme file which you can then modify and feed as the argument to --syntax-highlighting. Wanna get a JSON version of the pygments style, for example:

pandoc -o my.theme --print-highlight-style pygments

Then edit my.theme and use it like this:

pandoc --syntax-highlighting my.theme

And if the built-in highlighting isn't giving what you need, or you wanna highlight a language that isn't supported, you can use the --syntax-definition option to load a KDE-style XML syntax definition file. Before you go writing your own, have a look at KDE's repository of syntax definitions 👀.

if you get an error saying pandoc "Could not read highlighting theme", check that the JSON file is encoded with UTF-8 and has no Byte-Order Mark (BOM). That BOM will get you every time 😭

To disable highlighting, use --syntax-highlighting=none. 🚫

To use a format's idiomatic syntax highlighting instead of pandoc's built-in highlighting, use --syntax-highlighting=idiomatic. Currently, idiomatic only affects these formats:

• In reveal.js, it causes reveal.js's highlighting plugin to be used for source code highlighting. The style may be customized by setting the highlightjs-theme variable.
• In Typst, it causes Typst's built-in highlighting to be used. (This is also the default for Typst.)
• In LaTeX, it causes the listings package to be used. Note that listings does not support multi-byte encoding for source code. To handle UTF-8 you would need to use a custom template. This issue is fully documented here: Encoding issue with the listings package.
• In other formats, idiomatic will have the same result as default.

✨ CUSTOM STYLES ✨

Okay bestie, this is the part where you get to be aesthetic on purpose. Custom styles can be used in the docx, odt and ICML formats. 🎀

💅 Output

By default, pandoc's odt, docx and ICML output applies a predefined set of styles for blocks such as paragraphs and block quotes, and uses largely default formatting (italics, bold) for inlines. This will work for most purposes, especially alongside a reference doc file. Classic, reliable, does the job. However — and this is the fun part — if you need to apply your own styles to blocks, or match a preexisting set of styles, pandoc allows you to define custom styles for blocks and text using divs and spans, respectively. 🦋

If you define a Div, Span, or Table with the attribute custom-style, pandoc will apply your specified style to the contained elements (with the exception of elements whose function depends on a style, like headings, code blocks, block quotes, or links). So, for example, using the bracketed_spans syntax,

[Get out]{custom-style="Emphatically"}, he said.

would produce a file with "Get out" styled with character style Emphatically. Similarly, using the fenced_divs syntax,

Dickinson starts the poem simply:

::: {custom-style="Poetry"}
| A Bird came down the Walk---
| He did not know I saw---
:::

would style the two contained lines with the Poetry paragraph style. Poetry paragraph style?? we love a themed queen 👑

Styles will be defined in the output file as inheriting from normal text (docx) or Default Paragraph Style (odt), if the styles are not yet in your reference doc. If they are already defined, pandoc will not alter the definition. (respectful. it knows when to back off fr.)

This feature allows for greatest customization in conjunction with pandoc filters. If you want all paragraphs after block quotes to be indented, you can write a filter to apply the styles necessary. If you want all italics to be transformed to the Emphasis character style (perhaps to change their color), you can write a filter which will transform all italicized inlines to inlines within an Emphasis custom-style span. 🌈

For docx or odt output, you don't need to enable any extensions for custom styles to work. ✨

For icml output, you can also set an object-style in images:

![Image with object style](myImage.jpg){object-style="fixedSizeImage"}

In InDesign you'll see that object style given to the image, and you'll be able to customize it, or load its definition from a template of yours. 💖

🎀 Input

The docx reader, by default, only reads those styles that it can convert into pandoc elements, either by direct conversion or interpreting the derivation of the input document's styles.

By enabling the styles extension in the docx reader (-f docx+styles), you can produce output that maintains the styles of the input document, using the custom-style class. A custom-style attribute will be added for each style. Divs will be created to hold the paragraph styles, and Spans to hold the character styles. Table styles will be applied directly to the Table. 🫶

For example, using the custom-style-reference.docx file in the test directory, we have the following different outputs:

Without the +styles extension:

$ pandoc test/docx/custom-style-reference.docx -f docx -t markdown
This is some text.

This is text with an *emphasized* text style. And this is text with a
**strengthened** text style.

> Here is a styled paragraph that inherits from Block Text.

And with the extension:

$ pandoc test/docx/custom-style-reference.docx -f docx+styles -t markdown

::: {custom-style="First Paragraph"}
This is some text.
:::

::: {custom-style="Body Text"}
This is text with an [emphasized]{custom-style="Emphatic"} text style.
And this is text with a [strengthened]{custom-style="Strengthened"}
text style.
:::

::: {custom-style="My Block Style"}
> Here is a styled paragraph that inherits from Block Text.
:::

LOOK at that glow-up 😭 the styles are all preserved, we're obsessed.

With these custom styles, you can use your input document as a reference-doc while creating docx output (see below), and maintain the same styles in your input and output files. symmetric, self-consistent, correct. periodt.

---

🧚 CUSTOM READERS AND WRITERS 🧚

Okay now THIS is main character energy. Pandoc can be extended with custom readers and writers written in Lua. (Pandoc includes a Lua interpreter, so Lua need not be installed separately.) no extra setup, we love a self-contained queen.

To use a custom reader or writer, simply specify the path to the Lua script in place of the input or output format. For example:

pandoc -t data/sample.lua
pandoc -f my_custom_markup_language.lua -t latex -s

If the script is not found relative to the working directory, it will be sought in the custom subdirectory of the user data directory (see --data-dir). 🔍

A custom reader is a Lua script that defines one function, Reader, which takes a string as input and returns a Pandoc AST. See the Lua filters documentation for documentation of the functions that are available for creating pandoc AST elements. For parsing, the lpeg parsing library is available by default. To see a sample custom reader:

pandoc --print-default-data-file creole.lua

If you want your custom reader to have access to reader options (e.g. the tab stop setting), you give your Reader function a second options parameter. 🎀

A custom writer is a Lua script that defines a function that specifies how to render each element in a Pandoc AST. See the djot-writer.lua for a full-featured example. 👑

Careful bestie!! Note that custom writers have no default template. If you want to use --standalone with a custom writer, you will need to specify a template manually using --template or add a new default template with the name default.NAME_OF_CUSTOM_WRITER.lua to the templates subdirectory of your user data directory (see Templates). 💀

---

🌈 REPRODUCIBLE BUILDS 🌈

Some of the document formats pandoc targets (such as EPUB, docx, and ODT) include build timestamps in the generated document. That means that the files generated on successive builds will differ, even if the source does not. the chaos of nondeterminism... not on our watch. To avoid this, set the SOURCE_DATE_EPOCH environment variable, and the timestamp will be taken from it instead of the current time. SOURCE_DATE_EPOCH should contain an integer unix timestamp (specifying the number of seconds since midnight UTC January 1, 1970). 🕐

For reproducible builds with LaTeX, you can either specify the pdf-trailer-id in the metadata or leave it undefined, in which case pandoc will create a trailer-id based on a hash of the SOURCE_DATE_EPOCH and the document's contents. ✨

Some document formats also include a unique identifier. For EPUB, this can be set explicitly by setting the identifier metadata field (see EPUB Metadata, above). 💖

---

💜 ACCESSIBLE PDFS AND PDF ARCHIVING STANDARDS 💜

Accessibility era!! we care about people, no cap. PDF is a flexible format, and using PDF in certain contexts requires additional conventions. For example, PDFs are not accessible by default; they define how characters are placed on a page but do not contain semantic information on the content. However, it is possible to generate accessible PDFs, which use tagging to add semantic information to the document. 🌸

Pandoc defaults to LaTeX to generate PDF. LaTeX's \DocumentMetadata interface supports PDF standards and tagging when using LuaLaTeX; set the pdfstandard variable to enable this (see below). For older LaTeX installations, alternative engines must be used. 🦋

The PDF standards PDF/A and PDF/UA define further restrictions intended to optimize PDFs for archiving and accessibility. Tagging is commonly used in combination with these standards to ensure best results.

Important caveat bestie!! Note, however, that standard compliance depends on many things, including the colorspace of embedded images. Pandoc cannot check this, and external programs must be used to ensure that generated PDFs are in compliance. 😭

🎀 LaTeX

Set the pdfstandard variable to produce tagged PDFs conforming to PDF/A, PDF/X, or PDF/UA standards. For example:

pandoc -V pdfstandard=ua-2 --pdf-engine=lualatex doc.md -o doc.pdf

Multiple standards can be combined:

---
pdfstandard:
  - ua-2
  - a-4f
---

The required PDF version is inferred automatically. This feature requires LuaLaTeX in TeX Live 2025 with LaTeX kernel 2025-06-01 or newer. 👑

🎀 ConTeXt

ConTeXt always produces tagged PDFs, but the quality depends on the input. The default ConTeXt markup generated by pandoc is optimized for readability and reuse, not tagging. Enable the tagging format extension to force markup that is optimized for tagging. For example:

pandoc -t context+tagging doc.md -o doc.pdf

A recent context version should be used, as older versions contained a bug that lead to invalid PDF metadata. 💀

🎀 WeasyPrint

The HTML-based engine WeasyPrint includes experimental support for PDF/A and PDF/UA since version 57. Tagged PDFs can created with

pandoc --pdf-engine=weasyprint \
       --pdf-engine-opt=--pdf-variant=pdf/ua-1 ...

The feature is experimental and standard compliance should not be assumed. 🙈

🎀 Prince XML

The non-free HTML-to-PDF converter prince has extensive support for various PDF standards as well as tagging. E.g.:

pandoc --pdf-engine=prince \
       --pdf-engine-opt=--tagged-pdf ...

See the prince documentation for more info. ✨

🎀 Typst

Typst 0.12 can produce PDF/A-2b:

pandoc --pdf-engine=typst --pdf-engine-opt=--pdf-standard=a-2b ...

🎀 Word Processors

Word processors like LibreOffice and MS Word can also be used to generate standardized and tagged PDF output. Pandoc does not support direct conversions via these tools. However, pandoc can convert a document to a docx or odt file, which can then be opened and converted to PDF with the respective word processor. See the documentation for Word and LibreOffice. 🫶

---

👑 RUNNING PANDOC AS A WEB SERVER 👑

not pandoc being a whole web service?? icon behavior fr fr. If you rename (or symlink) the pandoc executable to pandoc-server, or if you call pandoc with server as the first argument, it will start up a web server with a JSON API. This server exposes most of the conversion functionality of pandoc. For full documentation, see the pandoc-server man page. 🌈

If you rename (or symlink) the pandoc executable to pandoc-server.cgi, it will function as a CGI program exposing the same API as pandoc-server. 💅

pandoc-server is designed to be maximally secure; it uses Haskell's type system to provide strong guarantees that no I/O will be performed on the server during pandoc conversions. types that literally prevent the danger?? that's the dream, no cap. ✨

🌙 RUNNING PANDOC AS A LUA INTERPRETER ✨

okay so picture this: you call the pandoc executable but you name it pandoc-lua, or you just hand it lua as the first argument. congrats bestie, it's not a document converter anymore, it's a whole standalone Lua interpreter 💅

The behavior is mostly identical to that of the standalone lua executable, version 5.4. All pandoc.* packages, as well as the packages re and lpeg, are available via global variables. And on top of that, the globals PANDOC_VERSION, PANDOC_STATE, and PANDOC_API_VERSION are set at startup. 🎀

For full documentation, see the pandoc-lua man page. iykyk.

---

🚨 A NOTE ON SECURITY 🚨

okay bestie this is the part where we get serious, because this stuff is lowkey terrifying and highkey important. sit down. 🫶

1. Although pandoc itself will not create or modify any files other than those you explicitly ask it create (with the exception of temporary files used in producing PDFs), a filter or custom writer could in principle do anything on your file system. Please audit filters and custom writers very carefully before using them. 💀
2. If you want to use pandoc in a web application that accepts untrusted user input, you should be aware that running pandoc in this context requires special precautions, or you risk exfiltration of the contents of local files, or worse. In particular:
   • Several input formats (including LaTeX, Org, RST, and Typst) support include directives that allow the contents of a file to be included in the output. An untrusted attacker could use these to view the contents of files on the file system. Using the --sandbox option will protect against this threat. 🛡️
   • Several output formats (including RTF, FB2, HTML with --self-contained, EPUB, Docx, and ODT) will embed encoded or raw images into the output file. An untrusted attacker could exploit this to view the contents of non-image files on the file system. Using the --sandbox option will protect against this threat, but will also prevent including images in these formats.
   • In reading HTML files, pandoc will attempt to include the contents of iframe elements by fetching content from the local file or URL specified by src. If untrusted HTML is processed on a server, this has the potential to reveal anything readable by the process running the server or to enable server-side request forgery (SSRF) attacks (CVE-2025-51591). To mitigate such attacks, use --sandbox or -f html+raw_html (which causes the whole iframe to be parsed as a raw HTML block). 😭
   • Using pandoc with --pdf-engine introduces additional risks, which cannot be mitigated by using --sandbox. For example, in wkhtmltopdf the footer-html and header-html metadata fields will be passed directly to wkhtmltopdf as --footer-html and --header-html arguments, and by using a file: URIs with these, one could exfiltrate local files. In addition, using --pdf-engine=wkhtmltopdf with -f html+raw_html enables an SSRF vulnerability in wkhtmltopdf (CVE-2022-35583). Note that wkhtmltopdf is deprecated and its authors recommend not to use it with any untrusted input. However, there may be similar issues with other pdf-engines. Anyone including a pdf-engine in a pandoc command run on untrusted input should audit the engine carefully and carefully control the use of --pdf-engine-opt. 💀🔥

Here are some alternatives for safe use of pandoc in a web application:

• Use pandoc server to spin up a local web server, and have your application make requests to it. The server runs in sandboxed mode and will not be able to read files or use filters. 🌈
• Use the WASM version of pandoc. This allows the use of Lua filters, but everything runs in an isolated WASM sandbox. ✨
• Use pandoc --sandbox. However, be aware that the sandbox function only affects readers and writers, and security issues can arise if you use it together with --filter or --pdf-engine. 👑

3. The HTML generated by pandoc is not guaranteed to be safe. If raw_html is enabled for the Markdown input, users can inject arbitrary HTML. Even if raw_html is disabled, users can include dangerous content in URLs and attributes. To be safe, you should run all HTML generated from untrusted user input through an HTML sanitizer. 🧼 no cap.
4. pandoc’s parsers can exhibit pathological performance on some corner cases. It is wise to put any pandoc operations under a timeout, to avoid DOS attacks that exploit these issues. If you are using the pandoc executable, you can add the command line options +RTS -M512M -RTS (for example) to limit the heap size to 512MB. Note that the commonmark parser (including commonmark_x and gfm) is much less vulnerable to pathological performance than the markdown parser, so it is a better choice when processing untrusted input. ⏱️
5. If your application uses pandoc as a Haskell library (rather than shelling out to the executable), it is possible to use it in a mode that fully isolates pandoc from your file system, by running the pandoc operations in the PandocPure monad. See the document Using the pandoc API for more details. (This corresponds to the use of the --sandbox option on the command line.) 🦋

---

👑 AUTHORS 💖

Copyright 2006–2024 John MacFarlane (jgm@berkeley.edu). Released under the GPL, version 2 or greater. This software carries no warranty of any kind. (See COPYRIGHT for full copyright and warranty notices.) For a full list of contributors, see the file AUTHORS.md in the pandoc source code. 🌸

The Pandoc source code may be downloaded from <https://hackage.haskell.org/package/pandoc> or <https://github.com/jgm/pandoc/releases>. Further documentation is available at <https://pandoc.org>. ⭐

---

and that's the whole page bestie, we made it to the end, main character energy fr fr 🎀 this was pandoc 3.11, dated 2026-08-28, and you just read the entire pandoc(1) manual like an absolute icon. go convert something gorgeous. 💅✨🫶