git @ Cat's Eye Technologies Feedmark / 0.2

Tree @0.2 (Download .tar.gz)


Version 0.2. Subject to change in backwards-incompatible ways without notice.

Feedmark is a format for embedding entities in Markdown files with accompanying metadata in a way which is both human-readable and machine-extractable. Feedmark is a subset of Markdown.

To this end, it is not dissimilar to Falderal, however it has different goals. It is more oriented for "curational" tasks. The Dossier is written in Feedmark format.

Informally, the format says that every h3-level heading in the Markdown file gives the title of an entity, and may be followed immediately by the entity's "plaque", which is a bullet list where every item is prefixed by an identifier and a colon.

Example Feedmark documents can be found in the eg/ directory. Further examples can be found in The Dossier.


This repository contains a Python program, feedmark, which is a reference implementation of a processor for the Feedmark format. It is currently able to do the following things:

Parse Feedmark documents

which will check that they are minimally well-formed.

bin/feedmark eg/*.md

Archive all web objects linked to from the documents

bin/feedmark --archive-links-to=downloads eg/Recent\ Llama\

If it is only desired that the links be checked, --check-links will make HEAD requests and will not save any of the responses.

Convert Feedmark documents to an Atom (née RSS) feed

bin/feedmark "eg/Recent Llama" --output-atom=feed.xml
python -m SimpleHTTPServer 7000 &
python -m webbrowser http://localhost:7000/feed.xml

Check entries against a schema

A Feedmark schema is simply another Feedmark document, one in which each entry describes a property that entries should have.

bin/feedmark eg/Video\ --check-against=eg/Video\ games\

Note that this facility is still under development.

Rewrite documents in-place

They will be parsed as Feedmark, and then output as Markdown, to the same files that were read in as input. (This is destructive, but it is recommended that the original files be under version control such as git, which will easily allow the changes to be reverted.)

bin/feedmark --rewrite-markdown eg/*.md

Note that this facility is still under development.

Markdown supports "reference-style" links, which are not inline with the text.

feedmark can rewrite reference-style links that match the name of an entry in a previously-created "refdex", so that they can be kept current and point to the canonical document in which the entry exists, since it may exist in multiple, or be moved over time.

bin/feedmark eg/*.md --output-refdex >refdex.json
bin/feedmark --input-refdex=refdex.json --rewrite-markdown eg/*.md

Note that this facility is still under development.

Write out to miscellaneous formats

Output entries as JSON, indexed by entry, or by property.

bin/feedmark --dump-entries eg/*.md
bin/feedmark --by-property eg/*.md

Output entries as Markdown, or HTML, or a snippet of HTML

bin/feedmark --output-markdown eg/*.md
bin/feedmark --output-html eg/*.md
bin/feedmark --output-html-snippet eg/*.md


Why is Feedmark desirable? Because if your structured data format is a subset of Markdown, the effort to format it into something nicely human-readable is very small. YAML and Markdown are both fairly easy to read as raw text, but Github, for example, automatically formats Markdown as HTML, making it that much nicer.

Or, if you like the transitivity: in the same way that a Markdown file is still a readable text file, which is nice, a Feedmark file is still a readable Markdown file, which is still a readable text file, which is nice.


"common" properties on document which all entries within inherit.

Sub-entries. Somehow. For individual games in a series, implementations or variations on a programming language, etc.

Allow trailing ### on h3-level headings.

Index creation from refdex, for permalinks.