Release 3.9.0

Changed

• Footnotes are now ordered by the occurrence of their references in the
  document. A new configuration option for the footnotes extension,
  USE_DEFINITION_ORDER, has been added to support restoring the previous
  behavior of ordering footnotes by the occurrence of definitions (#1367).

Fixed

• Ensure inline processing iterates through elements in document order (#1546).
• Fix handling of incomplete HTML tags in code spans in Python 3.14 (#1547).

Release 3.8.2

Fixed

• Fix codecs deprecation in Python 3.14.
• Fix issue with unclosed comment parsing in Python 3.14.
• Fix issue with unclosed declarations in Python 3.14.
• Fix issue with unclosed HTML tag <foo and Python 3.14.

Release 3.8.1

Fixed

• Ensure incomplete markup declaration in raw HTML doesn't crash parser (#1534).
• Fixed dropped content in md_in_html (#1526).
• Fixed HTML handling corner case that prevented some content from not being rendered (#1528).

Release 3.8

Changed

• DRY fix in abbr extension by introducing method create_element (#1483).
• Clean up test directory by removing some redundant tests and port
  non-redundant cases to the newer test framework.
• Improved performance of the raw HTML post-processor (#1510).

Fixed

• Backslash Unescape IDs set via attr_list on toc (#1493).
• Ensure md_in_html processes content inside "markdown" blocks as they are
  parsed outside of "markdown" blocks to keep things more consistent for
  third-party extensions (#1503).
• md_in_html handle tags within inline code blocks better (#1075).
• md_in_html fix handling of one-liner block HTML handling (#1074).
• Ensure <center> is treated like a block-level element (#1481).
• Ensure that abbr extension respects AtomicString and does not process
  perceived abbreviations in these strings (#1512).
• Ensure smarty extension correctly renders nested closing quotes (#1514).

Release 3.7

Changed

Refactor abbr Extension

A new AbbrTreeprocessor has been introduced, which replaces the now deprecated
AbbrInlineProcessor. Abbreviation processing now happens after Attribute Lists,
avoiding a conflict between the two extensions (#1460).

The AbbrPreprocessor class has been renamed to AbbrBlockprocessor, which
better reflects what it is. AbbrPreprocessor has been deprecated.

A call to Markdown.reset() now clears all previously defined abbreviations.

Abbreviations are now sorted by length before executing AbbrTreeprocessor
to ensure that multi-word abbreviations are implemented even if an abbreviation
exists for one of those component words. (#1465)

Abbreviations without a definition are now ignored. This avoids applying
abbr tags to text without a title value.

Added an optional glossary configuration option to the abbreviations extension.
This provides a simple and efficient way to apply a dictionary of abbreviations
to every page.

Abbreviations can now be disabled by setting their definition to "" or ''.
This can be useful when using the glossary option.

Fixed

• Fixed links to source code on GitHub from the documentation (#1453).

Release 3.6

Changed

Refactor TOC Sanitation

• All postprocessors are now run on heading content.
• Footnote references are now stripped from heading content. Fixes #660.
• A more robust striptags is provided to convert headings to plain text.
  Unlike, the markupsafe implementation, HTML entities are not unescaped.
• The plain text name, rich html, and unescaped raw data-toc-label are
  saved to toc_tokens, allowing users to access the full rich text content of
  the headings directly from toc_tokens.
• The value of data-toc-label is sanitized separate from heading content
  before being written to name. This fixes a bug which allowed markup through
  in certain circumstances. To access the raw unsanitized data, retrieve the
  value from token['data-toc-label'] directly.
• An html.unescape call is made just prior to calling slugify so that
  slugify only operates on Unicode characters. Note that html.unescape is
  not run on name, html, or data-toc-label.
• The functions get_name and stashedHTML2text defined in the toc extension
  are both deprecated. Instead, third party extensions should use some
  combination of the new functions run_postprocessors, render_inner_html and
  striptags.

Fixed

• Include scripts/*.py in the generated source tarballs (#1430).
• Ensure lines after heading in loose list are properly detabbed (#1443).
• Give smarty tree processor higher priority than toc (#1440).
• Permit carets (^) and square brackets (]) but explicitly exclude
  backslashes (\) from abbreviations (#1444).
• In attribute lists (attr_list, fenced_code), quoted attribute values are
  now allowed to contain curly braces (}) (#1414).

Release 3.5.2

Fixed

• Fix type annotations for convertFile - it accepts only bytes-based buffers.
  Also remove legacy checks from Python 2 (#1400)
• Remove legacy import needed only in Python 2 (#1403)
• Fix typo that left the attribute AdmonitionProcessor.content_indent unset
  (#1404)
• Fix edge-case crash in InlineProcessor with AtomicString (#1406).
• Fix edge-case crash in codehilite with an empty code tag (#1405).
• Improve and expand type annotations in the code base (#1401).
• Fix handling of bogus comments (#1425).

Release 3.5.1

Fixed

• Fix a performance problem with HTML extraction where large HTML input could
  trigger quadratic line counting behavior (#1392).
• Improve and expand type annotations in the code base (#1394).