Dropped --output-links option so that we could also drop the
beautifulsoup4 requirement. There is now an example script
supplied in eg/script that shows how such can be implemented.
The parser module/layer no longer calls the Markdown renderer.
(The Document and Section models still do, in their methods,
which are still called during parsing; but there is an argument
that this could/should be delayed using lazy evaluation.)
Dropped the Markdown package requirement as well; the marko
HTML renderer is used, when HTML output is requested. (This did
necessitate adding python-slugify as a dependency, however.)
When writing out Markdown, the single link reference block at
the end of the document is written out in sorted order.
0.16
Support for Python 2.x has been dropped.
Parsing of Markdown is now done with (a bundled fork of)
Frost Ming's marko
package, with the aim of improved compliance with the major
Markdown formats in use today (CommonMark, GFM).
When writing out Markdown, all link references are collected in
a single block at the end of the document.
Each Feedmark document can now have a "header comment" section
that appears just before the document properties.
Some internal refactoring, and reformatting using black.
0.15
Updated dependency on beautifulsoup4, to support recent Pythons.
Python 2.7 is still technically supported, but no longer tested.
Fixed how anchors are generated, handling . and ' characters
more correctly.
0.14
Generated anchors are now compatible with those on GitHub
(multiple spaces are not coalesced into a single hyphen.)
Fixed bug with schema checking where, if multiple documents
were given on command line, not all documents were checked.
0.13
The default interpreter for the feedmark script is now
python3.
The test harness runs tests under Python 2, or Python 3,
or both, depending on what's available.
0.12
Added support for hyperlinks on images on entries.
Added test cases for correct handling of images on entries.
Format of entry images in output JSON has changed: each image
is now represented as a dictionary with two or three keys
rather than as a list with two elements.
0.11
Fixed a bug with --rewrite-markdown under Python 3.x.
0.10
Format of refdex files has changed: an entry can now have a
key filenames, which is like filename, but can be a list.
This is backwards-compatible on input, and you can pass the
flag --output-refdex-single-filename to cause the output
from --output-refdex to strip all but the last filename
and produce only filename entries on output.
Parser now allows trailing ### on h3-level section headers.
0.9-2019.1015
Minor release to work out issues with setup.py.
0.9
Added conventional --version option to feedmark.
Added setup.py so that Feedmark can be treated as a Python
package.
0.8
Added --htmlized-json option which, when using --output-json,
converts Markdown to HTML5 every place it is present in the JSON.
Added --ordered-json option which, when using --output-json,
serializes the properties (of the document and of each section)
as an array of two-element arrays, instead of a JSON object.
0.7
Renamed link-to-anchors-on property to link-target-url.
Improved unit tests.
0.6
Uses the toc extension that ships with markdown to generate
heading-anchors (instead of a home-rolled function for that.)
Heading anchors are generated on all header levels, not just h3.
Format in which links extracted from documents are output in JSON
has changed. Each link is a dictionary, includes more info.
Able to output links that don't appear in any Feedmark section.
Ability to archive targets of links has been split off into a
separate utility (called yastasoti and hosted in its own repo)
which consumes the new link JSON format.
Tests are an actual Python unittest suite, which passes when run
under Python 3.
0.5
Format of JSON written by --output-json cleaned up; documents
are given in a list, sections in a document are also in lists.
Functionality for generating a table of contents has been removed.
Example script that generates a ToC from feedmark --output-json
is provided in the eg directory instead.
0.4
Checking links now works for local files; it checks they are in
the directory specified as --article-root.
Archiving targets of links is "more idempotent":
if the file has not changed, it is not changed on disk
if it has changed, the old file is renamed to a datestamped name
--archive-missing-only only downloads if not already have it
Writing the anchor id of h3 elements is now much faster, as it
is now done with a Markdown extension instead of BeautifulSoup.
preamble of a document is now formatted the same way as the body
of each section in the document.
HISTORY document.
Other small fixes.
0.3
Output JSON with --output-json. Also --by-publication-date.
Output links with --output-links.
Multiple input refdexes can be read with --input-refdexes.
A prefix can be virtually appended to every filename in the
input refdexes using --input-refdex-filename-prefix.
Output a table of contents.
Some internal refactoring.
Rudimentary test suite.
Other small fixes.
0.2
Check documents against schema.
Check, and archive the targets of, links in documents.
Output Markdown and HTML from input Feedmark documents.
Rewrite Feedmark documents in-place.
Create and use a refdex file (reference-style links.)
Other small fixes.
0.1
Initial release.
Ability to create an Atom feed from a Markdown document.