- Update to version 3.0.0:
* drop python 3.7 support
* Update from upstream markdown-it 12.2.0 to 13.0.0
- A key change is the addition of a new Token type, text_special,
which is used to represent HTML entities and backslash escaped
characters.
- This ensures that (core) typographic transformation rules are not
incorrectly applied to these texts.
- The final core rule is now the new text_join rule, which joins
adjacent text/text_special tokens, and so no text_special tokens
should be present in the final token stream.
- Any custom typographic rules should be inserted before text_join.
- A new linkify rule has also been added to the inline chain, which
will linkify full URLs (e.g. https://example.com), and fixes
collision of emphasis and linkifier (so
http://example.org/foo._bar_-_baz is now a single link, not
emphasized).
- Emails and fuzzy links are not affected by this.
* REFACTOR: Replace most character code use with strings
- The use of StateBase.srcCharCode is deprecated (with
backward-compatibility), and all core uses are replaced by
StateBase.src.
- Conversion of source string characters to an integer
representing the Unicode character is prevalent in the upstream
JavaScript implementation, to improve performance. However, it
is unnecessary in Python and leads to harder to read code and
performance deprecations (during the conversion in the StateBase
initialisation).
* Centralise indented code block tests
- For CommonMark, the presence of indented code blocks prevent any
other block element from having an indent of greater than 4
spaces.
- Certain Markdown flavors and derivatives, such as mdx and djot,
disable these code blocks though, since it is more common to use
code fences and/or arbitrary indenting is desirable.
- Previously, disabling code blocks did not remove the indent
limitation, since most block elements had the 3 space limitation
hard-coded.
- This change centralised the logic of applying this limitation
(in StateBlock.is_code_block), and only applies it when indented
code blocks are enabled.
* Maintenance changes
- Strict type annotation checking has been applied to the whole
code base, ruff is now used for linting, and fuzzing tests have
been added to the CI, to integrate with Google OSS-Fuzz testing,
thanks to @DavidKorczynski.