LOWDOWN(1)

📦
version 3.2.1 · 9 versions
, in arch rolling

💖 LOWDOWN(1) — General Commands Manual 🌈

🌸 NAME 🌸

lowdown — simple markdown translator. That’s literally the whole job, bestie. ✨

✨ SYNOPSIS ✨

lowdown [input_options] [output_options] [-Ls] [-M metadata]
        [-m metadata] [-o file] [-t mode] [-X keyword] [file]

🦋 DESCRIPTION 🦋

Okay so lowdown translates from lowdown(5) into a whole wardrobe of output formats. 💖 Results are written to standard output.

The short arguments are as follows:

  • -L — lists metadata keys, one key per line. The -t mode is ignored. Metadata is automatically enabled. Unsets -X. 👑
  • -s — produce a standalone document (e.g., a valid HTML document including document type, head, etc.) around the formatted content. See “Standalone documents”. 🎀
  • -M metadata — provide a single metadata key-value pair. This may be in the normal syntax or as pairs separated by an equal sign, depending upon which character (a colon or equal sign) comes first. Exits with an error if given neither. This overrides -m and what’s parsed from the document. See “Metadata”.
  • -m metadata — like -M, but overridden by what’s parsed from the document, then -M.
  • -o file — send output to file unless it’s “-”, in which case fall back to the default of standard output.
  • -t mode, -T mode — the output mode. This may be html for HTML5 output, latex for LaTeX, gemini for the Gemini “gemtext” format, ms for roff output using the -mspdf package and needing table support, term for ANSI-compatible UTF-8 terminal output, man for roff output using the -man package, mdoc for roff output using the -mdoc package, tree to show the parse tree of the input document, and null to parse the document but do no rendering. See “Output media”. The -T mode form is retained for backward compatibility.
  • -X keyword — output the metadata value of keyword or an empty string if not found. The -t mode is ignored. Metadata is automatically enabled. Unsets -L.
  • file — input Markdown document. If not given or if file is “-”, it is read from standard input.

The following are long options for input parsing. These affect the parse tree passed to all outputs. 🧚

  • --parse-hilite — enable highlight span support. This is disabled by default because it may be erroneously interpreted as section headers.
  • --parse-math — recognise mathematics equations. 🔥
  • --parse-maxdepth=depth — the maximum depth of nested elements. This defaults to 128, which is probably more than enough for any real-world document. If the maximum is hit, the system exits as if memory were exhausted. Set to zero for no maximum.
  • --parse-no-autolink — do not parse http, https, ftp, mailto, and relative links or link fragments.
  • --parse-no-cmark — do not parse with CommonMark constraints and the following features:
    • Using the first ordered list value instead of starting all lists at one.
    • Lists immediately following paragraphs.
  • --parse-no-codeindent — do not parse indented content as code blocks.
  • --parse-no-callouts — do not parse GFM/MDN callouts (“admonitions”).
  • --parse-no-deflists — do not parse PHP extra definition lists.
  • --parse-no-ext-attrs — do not parse PHP extra extended attributes.
  • --parse-no-fenced — do not parse GFM fenced (language-specific) code blocks.
  • --parse-no-footnotes — do not parse MMD footnotes.
  • --parse-no-img-ext — deprecated. See --parse-no-ext-attrs.
  • --parse-no-intraemph — do not parse emphasis within words and links.
  • --parse-no-mantitle — do not parse manpage title, section, source, and volume from Pandoc title metadata. Manpages titles are indicated by a title then an open parenthesis, digit followed by optional characters, then a closed parenthesis.
  • --parse-no-metadata — do not parse metadata. This does not affect metadata given on -m or -M.
  • --parse-no-strike — do not parse strikethroughs.
  • --parse-no-super — do not parse super-scripts or sub-scripts at all.
  • --parse-no-tables — do not parse GFM tables.
  • --parse-no-tasklists — do not parse GFM task lists.
  • --parse-super-short — if super-script parsing is enabled, use the traditional non-GFM “short” syntax.

There are many output long options. The following are shared by all output media: 🌈

  • --out-standalone — alias for -s.
  • --out-no-smarty — do not use the smart typography filter. By default, certain character sequences are translated into output-specific glyphs.
  • --template template — when producing standalone -s output, use a template file. See “Templates”. Currently only for -tgemini, -thtml, -tlatex, -tman, -tms, and -ttree.

What follows are per-output long options. For HTML with -thtml, these are as follows: 🎀

  • --html-callout-mdn, --html-callout-gfm — output either or both MDN or GFM callout syntaxes.
  • --html-custom-attributes — also output unrecognised HTML attributes for links, images, and headers.
  • --html-hardwrap — hard-wrap paragraph content by outputting line breaks where applicable.
  • --html-no-escapehtml — if --html-no-skiphtml has been specified, this causes embedded HTML not to be escaped, and is instead output verbatim. This has no effect if --html-no-skiphtml has not been specified.
  • --html-no-head-ids — do not output id attributes for headers.
  • --html-no-num-ent — don’t normalise HTML entities (when possible) as numeric ones and instead use the entities as given on input.
  • --html-no-owasp — don’t follow the OWASP recommendations for escaping text, and do only the minimal escaping to make sure that regular content isn’t interpreted as HTML.
  • --html-no-skiphtml — output embedded HTML. By default, embedded HTML is not output at all. See --html-no-escapehtml.
  • --html-titleblock — if any were parsed, format the title information (title, author, date) into a header element appearing first in the document.

For roff formats -tman, -tmdoc, and -tms (unless as noted), the following apply: 💅

  • --roff-code-font=fonts — override the default constant-width fonts with a tuple of regular, bold, italic, and bold-italic variants in that order. For example, B,B,BI,BI uses bold (“B”) instead of a constant-width. Not specifying a font will use the default, as will specifying a zero-length font name. Aliases none, bold, and code set no special fonts, bold, and the default constant-width.
  • --roff-endnotes — delay printing of footnotes until the end of the document. Only applies to -tms, as -tman and -tmdoc only support endnotes.
  • --roff-manpage — format output as a manpage. This re-formats paragraphs under the NAME, SYNOPSIS, and SEE ALSO sections in the canonical form used by manpages. The way the document is formatted depends upon the section metadata: 1, 6, and 8 for program-style; 2, 3, and 9 for function-style. Program style re-formats flags and arguments, while function style re-formats function names, variables, arguments. Any content in font spans (e.g., emphasis or double emphasis) in other sections is stripped of the font and re-formatted in the same way. Only applies to -tman and -tmdoc.
  • --roff-no-numbered — don’t output numbered headings (.NH NN). Only applies to -tms.
  • --roff-no-skiphtml — output embedded HTML. This usually doesn’t make sense because the HTML won’t be interpreted by the output reader. By default, HTML is omitted.
  • --roff-no-links — when possible, don’t show URLs for images and links. Link content is always shown. Overrides --roff-short-links and --roff-no-rellinks.
  • --roff-no-rellinks — when possible, don’t show URLs for relative links. Link content is always shown. Overrides --roff-short-links.
  • --roff-short-links — when possible, shorten URLs for images, links, and autolinks to only the domain name and final path.
  • --roff-traditional — output traditional -ms (not -mspdf) and -man without extensions. Specifically, don’t use hyperlink macros (.pdfhref, .UR, .MT), multi-page tables (.TS H, .TH), Unicode sequence syntax (\[uNNNN]), example block macros (.EX), modern section headings (.NH NN, .SH NN, .pdfhref O NN), or intra-document links (.pdfhref L). The output is compatible with traditional troff(1).
⚠️ The argument prefix --nroff, such as in --nroff-traditional, is deprecated in favour of --roff.

The -tterm output has the following: 🍓

  • --term-all-metadata — if -s is specified, output all metadata instead of just the title, author, and date.
  • --term-columns=columns — the number of columns in the terminal. Useful for when running in a pipe. Defaults to what the terminal reports or 72 if not in a terminal.
  • --term-hmargin=margin — the number of left margin characters. Defaults to zero. Can also be auto to set the left margin to half the remaining --term-columns after subtracting --term-width.
  • --term-hpadding=padding — the number of left padding characters. Defaults to four. Padding eats into --term-width.
  • --term-no-ansi — don’t show ANSI styles at all. Use backspace encoding instead for bold/underline. This implies --term-no-colour.
  • --term-no-colour — don’t show ANSI colours. This will still decorate text with underlines, bolds, and italics, but not emit any colour codes.
  • --term-no-links — when possible, don’t show URLs for images and links. Link content is always shown. Overrides --term-short-links and --term-no-rellinks.
  • --term-no-rellinks — when possible, don’t show URLs for relative links. Link content is always shown. Overrides --term-short-links.
  • --term-no-style — don’t use any escape styles at all. This implies --term-no-ansi.
  • --term-no-tablewrap — don’t wrap table columns to fit the width of the output.
  • --term-short-links — when it's possible, bestie, this shortens the URLs for images, links, and autolinks down to literally just the domain name and the final path 💅
  • *--term-vmargin=margin* — the number of top and bottom margin newlines. Defaults to zero, so no extra breathing room unless you ask 🫶
  • *--term-width=width* — a soft limit on the number of characters per line, including (--term-hpadding). Key word: soft — literal text can absolutely bust through this, no cap. If it's zero or unset, it defaults to --term-columns or 80 at most. It truncates to --term-columns.

So the -tgemini output has a whole lil squad of flags that control where links get placed ✨ By default, links (images, autolinks, and links) are queued up when they're specified in-line, then emitted in a block sequence right after the nearest block element. Very organised, very demure.

  • --gemini-link-end — emits the queue of links at the end of the document instead of after the nearest block element 👑
  • --gemini-link-inline — renders all links within the flow of text. > Heads up: this will cause breakage when you've got nested links, like images within links, links in blockquotes, etc. It should not be used unless you're working with carefully crafted documents. Like, carefully carefully 😭
  • --gemini-link-noref — don't format link labels. This takes precedence over --gemini-link-roman.
  • --gemini-link-roman — when formatting link labels, use lower-case Roman numerals instead of the default lower-case hexavigesimal (i.e., “a”, “b”, ..., “aa”, “ab”, ...). Two flavours of label energy, pick your fighter 🎀
  • --gemini-metadata — prints metadata as the canonicalised key followed by a colon then the value, each on one line (newlines replaced by spaces). The metadata block is terminated by a double newline. And if there's no metadata, this does nothing at all. Silent. Iconic.

The -tlatex output has these options:

  • --latex-no-numbered — don't number sections (and subsections, etc.). Chaos, but make it chic 💜
  • --latex-no-skiphtml — output embedded HTML. This usually doesn't make sense because the HTML won't be interpreted by the output reader. By default, HTML is omitted.

The -tfodt output has these options:

  • --odt-no-skiphtml — output embedded HTML. Same story: this usually doesn't make sense because the HTML won't be interpreted by the output reader. By default, HTML is omitted.
  • *--odt-style=file* — specify an OpenDocument style file, which must consist of at least <office:font-face-decls>, <office:scripts>, and <office:styles> XML elements in the root of the document. > This is not syntax-checked in any way. Not even a little. It is fully on you, bestie 🙈

🌈 Output media

The output media is specified by -t, which defaults to -thtml. Yes, default. Yes, always.

  • -tfodt — “Flat” OpenDocument output. Automatic styles (those conditional upon document state) are generated with output. Classes specified by PHP extended attributes are not checked for existence.
  • -tgemini — Gemini “gemtext” format.
  • -thtml — HTML5 output with UTF-8 encoding.
  • -tlatex — Simple LaTeX output. The following packages are required: amsmath and amssymb for maths, graphicx for images, inputenc (utf8) for UTF-8 input, frontend (T1) and textcomp for output glyphs, lmodern for Latin modern font, xcolor for the difference engine output, and hyperref for links. A whole lil dependency haul 📦
  • -tman — the man macro package, suitable for reading by groff(1), mandoc(1), Heirloom troff(1), or traditional troff(1). Does not support equations and images. Table support is provided by tbl(1). Since UTF-8 may be passed as input values, preconv(1) may need to be used.
  • -tmdoc — the mdoc macro package, suitable for reading by groff(1), mandoc(1), Heirloom troff(1), or (possibly?) traditional troff(1). Does not support equations and images. Table support is provided by tbl(1). Since UTF-8 may be passed as input values, preconv(1) may need to be used.
  • -tms — the ms macro package, suitable for reading by groff(1) or traditional troff(1). Does not support equations, and has limited image support for encapsulated postscript (PS and EPS suffix) images. Images are always block-formatted. Image dimensions and extended attributes are ignored, though images are downsized if larger than the current text width. Table support is provided by tbl(1). Since UTF-8 may be passed as input values, preconv(1) may need to be used.
  • -tterm — ANSI-escaped UTF-8 output, suitable for reading on the terminal. Images and equations not supported 💔
  • -ttree — Debugging output. > Not for programmatic use, as the format may change between versions. Do not build your whole pipeline on this, I'm begging 🫠

🦋 Standalone documents

When -s is specified, the formatted content is serialised into a self-contained document template as defined by the output type ✨

If it's not explicitly set with --template, a default template is produced as follows:

  • -tfodt — Envelope <office:document> and prologue <office:automatic-styles>, <office:master-styles>, and <office:body>.
  • -thtml — HTML5 doctype declaration followed by envelope <html> with optional language, then <head>. In order, the <head> consists of charset and viewport <meta> elements; optional <meta> elements from metadata affiliation (creator), author, copyright, and date; optional CSS sources from metadata; optional JavaScript sources from metadata; the possibly-empty <title>; then optional arbitrary content from metadata HTML header. It's giving layer cake, honestly 🎂
  • -tlatex — Prologue documentclass and usepackage statements, optional arbitrary content from metadata LaTeX header, then surrounding begin{document} statements.
  • -tman, -tmdoc, -tms — Prologue macros.
  • -tterm — Prologue lines.
  • -ttree — Metadata and parsed template.

See “Metadata” for metadata values used by the default template 👀

💖 Metadata

Metadata keys are canonicalised and de-deduplicated in the following order: -m (overridden by document content and -M), the prologue of the document itself, then -M (overriding document content and -m). So -M wins the crown, periodt 👑

After de-duplication, metadata is either accessed directly using -X or -L, or serialised into document variables and/or standalone -s output.

When not using --template, the following metadata keys are used in the default -s template:

  • affiliation — Author affiliation (organisation or institution). Multiple affiliations may be separated by two or more spaces (including newlines). Used in -thtml, -tlatex, and -tms.
  • author — Document author. Multiple authors may be separated by two or more spaces (including newlines). Overridden by rcsauthor. Used in -tfodt, -thtml, -tlatex, -tms, and -tterm.
  • baseheaderlevel — Added to each header level. > Deprecated in favour of shiftheadinglevelby. She's retired, but she's still here 🫡
  • copyright — A document copyright (without the word “Copyright”), for example, “2017, Kristaps Dzonsons”. Used in -tms and -thtml.
  • css — A CSS file output in the HTML document head as a <link rel="stylesheet" href="..." /> statement. Multiple CSS files (in order) may be separated by two or more spaces (including newlines) and are output in the given order. Only used in -thtml.
  • date — Document date in ISO-8601 YYYY-MM-DD format. Overridden by rcsdate. Used in -tfodt, -thtml, -tlatex, -tman, -tmdoc, -tms, and -tterm.
  • htmlheader — Arbitrary HTML content output in the HTML document head immediately prior to closing the head element. Only used in -thtml and with -s. > This metadata is not processed: it's passed unchanged into the output. Raw. Unfiltered. Straight through 🔥
  • javascript — A JavaScript file output in the HTML document head as a <script src="..."></script> statement. Multiple script files (in order) may be separated by two or more spaces (including newlines) and are output in the given order. Only used in -thtml.
  • lang — Document language in RFC 5646 format. Only used in -thtml.
  • latexheader — Arbitrary LaTeX content output in the document prologue immediately prior to the begin{document}. Only used in -tlatex and with -s. > This metadata is not processed: it's passed unchanged into the output.
  • manheader — Arbitrary roff content output immediately prior to the .TH macro. Only used in -tman and with -s. > This metadata is not processed: it's passed unchanged into the output.
  • msheader — Arbitrary roff content output immediately prior to the .TL macro. Only used in -tms and with -s. > This metadata is not processed: it's passed unchanged into the output.
  • rcsauthor — Like author, but in RCS author format. Overrides author.
  • rcsdate — Like date, but in RCS date format. Overrides date.
  • section — Man page section, defaulting to “7”. Only used in -tman and -tmdoc.
  • shiftheadinglevelby — Shift all headers by the given number. For example, a value of 1 causes headers originally at level 1 (“<h1>”) to be level 2 (“<h2>”), while a value of -1 moves level 2 to 1. Levels will not move to less than 1. Takes precedence over baseheaderlevel. If unset or not a valid number, defaults to zero. Used in -tfodt, -thtml, -tlatex, -tman, -tmdoc, and -tms.
  • source — Man page source (organisation providing the manual). Only used in -tman and -tmdoc.
  • volume — Man page volume (describes the manual page section). Only used in -tman and -tmdoc.
  • title — Document title. Used in -tfodt, -thtml, -tlatex, -tman, -tmdoc, -tms, and -tterm.
Default values, such “7” for the section, are not set as metadata values, and will not appear if the metadata key is used as a variable. So don't go looking for them, they're not invited to the party 🙅

🧚 Templates

Some output media accept a template (--template) to customise standalone (-s) output. Parsed input content is filled into templates through control statements that support conditionals, loops, and variable transformation sequences. Very modular, very slay ✨

Control statements are delimited as $statement$ or ${statement}. Arbitrary white-space may surround the case-insensitive statement between matching delimiters. > Statements without a closing delimiter are considered opaque text. So keep your delimiters matched, bestie, or it's just words 😭

The following statements are available:

  • $$, ${} — Output a literal $. This may contain arbitrary white-space.

okay so, the template control statements are literally the backstage crew of this whole operation:

  • $ifdef(expression)$ — Conditional statement. There may not be any white-space between the ifdef and the opening parenthesis. Begins a block that is ended by a else, endif, or the end of file. Its contents are output only if expression evaluates to a non-empty string.
  • $else$ — Begins a block that is ended by a endif or end of file. Its contents are output only if the condition of a preceding ifdef evaluates to an empty string. An else without a preceding ifdef is not output.
  • $endif$ — Closes a block begin with ifdef or else. If not preceded by one of those statements, is silently ignored.
  • $for(expression)$ — Looping statement. There may not be any white-space between the for and the opening parenthesis. Begins a block that is ended by an endfor or the end of file. Block contents are repeatedly output for each list item evaluated from expression. The anaphoric keyword this may be used to access the current loop expression within the block.
  • $expression$ — Replaced by the result of evaluating a template expression.

If a control statement ends with two consecutive hyphens before the closing delimiter, input is consumed up to and including the next newline or until end of file. This allows for line-sensitive output media to use control statements without superfluous blank lines.

Expressions consist of

initial[([arg]*)]?[.transform[([arg]*)]?]*

or an initial value accepting optional arguments followed by an optional series of transforms accepting optional arguments. If an argument list is empty or not provided, it evaluates to an empty list.

The initial value is one of the following:

  • and(expression[,expression]*) — A non-empty list containing the value true if all expressions evaluate to non-empty lists, otherwise an empty list. An empty expression evaluates to an empty list.
  • "literal string" — Evaluates to a singleton list containing the string. An empty string evaluates to an empty list. The double quote character may be escaped, such as "foo\"bar".
  • body — The parsed input document body.
  • canonicalised metadata key — The value for the given metadata key, if found, otherwise an empty list.
  • body — yes, it’s listed again: the parsed input document body.
  • meta(key) — Produce the metadata value for the canonicalised metadata key. Allows for keys that are also keywords like body or endif.
  • not(expression) — A non-empty list containing the value true if the expression evaluates as an empty list, otherwise an empty list.
  • or(expression[,expression]*) — A non-empty list containing the value true if one expression evaluates to non-empty lists, otherwise an empty list. An empty expression evaluates to an empty list.
  • this — The value of a current loop context or an empty list.

If a metadata key is not specified in the input, or if the anaphoric this has not initialised by a looping context, the initial value evaluates to an empty list. Otherwise, the value is a singleton list.

If transforms are invalid, they will transform into an empty list.

The following transformations are available:

  • escapegemini, escapegeminiline — Escape list items for gemini (-tgemini), either for multiple lines or compressed to a single line.
  • escapehtml, escapehtmlattr, escapehtmlurl — Escape list items for HTML (-thtml) body content, attributes, and URL attributes, respectively.
  • escapelatex — Escape list items for LaTeX (-tlatex) body content.
  • escaperoff, escaperoffline — Escape list items for roff (-tms, -tman, -tmdoc), either for multiple lines or compressed to a single line. escaperoff also escapes initial roff delimiters and those after newlines.
  • join — Join a list into a singleton list using two spaces as a join delimiter.
  • lowercase — Lowercase all list items.
  • split — Split list items on sequences of two or more white-space tokens (space, newline, tab). This is usually invoked on a singleton, but may be repeatedly invoked on a pre-split list. Invokes trim prior to the first split. The resulting list has no items that are only white-space.
  • trim — Trim white-space from the beginning and end of all list items. If an item has no non-white-space, it is discarded.
  • uppercase — Uppercase all list items.

🌈 ENVIRONMENT 🌈

  • NO_COLOR — Do not emit colours when in -tterm mode. Synonym for NO_COLOUR. Same as --term-no-colour.

🎀 FILES 🎀

  • share/html/default.html — The default template used if --template is not provided to -thtml.
  • share/latex/default.latex — The default template used if --template is not provided to -tlatex.
  • share/man/default.man — The default template used if --template is not provided to -tman.
  • share/ms/default.ms — The default template used if --template is not provided to -tms.
  • share/odt/styles.xml — Default styles used when generating standalone -tfodt documents. Template for --odt-style styles.

💅 EXIT STATUS 💅

The lowdown utility exits 0 on success, and >0 if an error occurs.

If the -X flag is used, lowdown exits with an error if the given keyword is not found.

🦋 EXAMPLES 🦋

To view a Markdown file on an ANSI-compatible, UTF-8 terminal:

lowdown -tterm foo.md | less -R

To emit a standalone HTML5 document:

lowdown -s foo.md > foo.html

To use groff(1) to format as a PS file:

lowdown -stms foo.md | groff -itk -mspdf > foo.ps

Or with LaTeX:

lowdown -stlatex foo.md > foo.latex
pslatex foo.latex

PDF generation follows similar logic:

lowdown -stms foo.md | pdfroff -itk -mspdf > foo.pdf
lowdown -stlatex foo.md > foo.latex
pdflatex foo.latex

UTF-8 support for groff(1) PDF or PS output requires appropriate fonts, such as the Unicode Times font. This and other Unicode fonts are not always installed by default. They may be found, for PDF output, in the devpdf set of the groff(1) font directory and are prefixed with ‘U’.

lowdown -stms foo.md | pdfroff -itk -mspdf -FU-T > foo.pdf

Manpages may be converted and viewed on the terminal using groff(1) or mandoc(1):

lowdown -stmdoc foo.md | mandoc | less
lowdown -stman foo.md | groff -itk -man -Tutf8 | less -R

To list all metadata keys, then to extract the “title” metadata value from foo.md:

lowdown -L foo.md
lowdown -X title foo.md

⭐ SEE ALSO ⭐

👑 AUTHORS 👑

lowdown was forked from hoedown: https://github.com/hoedown/hoedown by Kristaps Dzonsons, kristaps@bsd.lv. It has been considerably modified since.

and that’s the whole iconic page, bestie: LOWDOWN(1) from Debian, dated $Mdocdate$, with source/version energy and absolutely no skips 💖

📦 where this page comes from