readme-ssi

Update your markdown readme's ToC and snippets without seperate template or docs files, according to directives that remind of Server Side Includes.

Commands

• verbatim until
• verbatim lncnt
• toc
• echo json
• include file

All commands are given on a line of their own, in the format

<!--#command opt1="…" opt2="…" -->

with zero or more named options and one or more space characters before -->. The quotes are optional for integer values and the empty string value.

If you need quotes, brackets or ampersands in your option values, please use their XML named entity notation:

  " = &quot;     ' = &apos;     < = &lt;     > = &gt;     & = &amp;

CharRefs (e.g. ') are not currently supported. Whether and how they translate to an option value may change at any time. The same applies for named entities other than those listed above.

When commands replace text, they usually do that up to the next line that (ignoring space characters and tabs) begins with <!--.

• Recommendation: Use <!--/#command --> for the end mark, with the name of the command whose replacing is stopped.

verbatim until

Starting from the next line, text that looks like an SSI command will have no effect, the line will be printed verbatim. This effect is revoked, and magic re-activated, after the next line whose text is exactly the same as the value of the until option (required).

For some values, there are pre-defined shorthands that look like broken XML entities:

• &,: a line consisting of 3 accents grave.

verbatim lncnt

Where lncnt is a non-negative integer. The next lncnt lines won't have any special effect, even if they might look like an SSI command or a headline that toc would otherwise list.

toc

Replace text with a generated table of contents. It will include all lines that start with the headline prefix, up to the next stop mark or the end of file. The stop mark is:

<!--#toc stop="scan" -->

Options:

• pfx: The prefix used for headlines that shall appear in the ToC. Listing multiple levels of headlines isn't supoorted yet. Default: ### 
• fmt: The format of the generated ToC lines. Default:   * [&$caption;](#&$anchor;)
• cap-start, cap-end: Boundary markers for the caption part of your headline source line, in case it contains additional decoration like status symbols or version number hints.
  • Defaults: If cap-start is empty or unset, the caption starts right after pfx. If cap-end is empty or unset, the caption ends at the end of its source line.
  • The trimming on both sides is greedy. cap-start goes first, then cap-end.
  • Whitespace might be trimmed or not, don't rely on it.
  • You can provide multiple alternative texts for each boundary by separating them with the � CharRef. Greedy trimming will select the most hoggish alternative. A literal NUL (U+0000) might work as well, but don't rely on it.
• anchor: Template for what this ToC's anchor names shall look like. Any # is replaced with an ID generated from the caption part of the headline.
  • Helps avoid conflicts when you have multiple ToCs.
  • Why custom anchors? Trying to guess the auto-generated anchor names produced by various markdown parsers turned out to be too unreliable, as their rules for converting links, non-american letters and XML CharRefs might change at any time.
  • Default: toc-#

Work-arounds:

• After the ToC lines, a blank line will be added in order to help Github detect that the list ends, and treat the subsequent comment as a comment.

echo json

Replace text with a value from a JSON file. Options:

• json: Data source filename, relative to the readme. Required.
• key: The path to the property that shall be used. Empty = root object of the JSON source file. If the first character of the path is one of A-Z, a-z, 0-9, _, the path component seperator will be ., otherwise the first character will be used.
• wrap: Try to wrap lines to this number of characters. Default: 78
• before, after: Text that shall be inserted before or after the retrieved value. This is done before line wrapping. You can use it to add quotes or indentation.
• cut-head, cut-tail: Marks for where to chop off decorations from the value. They work like cap-start and cap-end for toc (see above), with these convenience exceptions:
  • With json="package.json" and key="description", cut-tail defaults to  [npm search keywords: .
• raw: If present without a value or set to 1, the data is included as found in the source. If unset or empty or set to 0, some characters are encoded as XML entities, most importantly including <, &, >. All other values are reserved for future use.

include file

Replace text with lines from another file. If the very next line after the include command is a verbatim command, all text covered by the latter is replaced.

Options:

• file: The source file's name, relative to the file that contains the include command. Required.
• start: If set, copying of lines from the source file will only start after (not: "at") the first line whose text is exactly the same as the value of the start option.
• stop: If set, copying will stop just before the first to-be-copied line with this option's text.
• maxln: If set to a positive number, copy at most this many lines.
• code: If set,
  • replace text up to and including the next line that consists of three accents grave.
  • The text included will be wrapped in a markdown code block denoted by these accents. The starting ones will be followed by the value of the code option, or text if the option value is empty.
  • The generated code block will be preceeded by a generated verbatim until="&," command in order to have the toc command ignore it.
• outdent: If a line copied from the source file starts with this text, remove the outdent text. At most one occurrence is removed.
• cut-head, cut-tail: Chop marks, same as with echo json.
• indent: This text will be prepended to any line that is copied from the source file.

Order of removal and decoration operations:

1. outdent
2. cut-head
3. cut-tail
4. indent

Caveats

• No attempt to protect your readme from failing partial writes to file. Stage it to git before updating.
• Trims trailing whitespace from each line. If you need it, use another tool.
• Trims trailing whitespace from the entire file. If you need some blank lines below the actual readme text, just add a comment after them, e.g. <!-- thanks for readme-ssi! -->
• Normalizes line endings to \n, including one after the last line. This is to annoy users of the original MS Windows notepad and save a few bytes for anyone else.
• Data included from external sources might be exempt from some of the whitespace normalization rules. Don't rely on it.

License

ISC