- Release 0.19 (2022-07-05)
  - General
    - Dropped support for Python 2.7, 3.5, and 3.6. and removed
      compatibility hacks from code and tests.
    - Code cleanup, check PEP 8 conformity with flake8
      (exceptions in file tox.ini).
    - New module. Support for python -m docutils. Also used for
      the docutils console script entry point.
    - Let Publisher.publish() print info and prompt
      when waiting for input from a terminal (cf.
      https://clig.dev/#interactivity).
    - Respect "input_encoding_error_handler" setting when opening
      a source.
    - New function error_string() obsoletes
      utils.error_reporting.ErrorString.
    - Class ErrorOutput moved here from utils/error_reporting.
    - Use "utf-8-sig" instead of Python's default encoding if the
      "input_encoding" setting is None.
    - Fix error when reading of UTF-16 encoded source without
      trailing newline.
    - Aliases "markdown" and "commonmark" point to
      "commonmark_wrapper".
    - Alias for the "myst" parser
      (https://pypi.org/project/myst-docutils).
    - Use absolute module names in _parser_aliases instead
      of two import attempts. (Keeps details if the
      recommonmark_wrapper.py module raises an ImportError.)
    - Prepend parser name to ImportError if importing a parser
      class fails.
    - New module for parsing CommonMark input. Selects a
      locally installed 3rd-party parser (pycmark, myst, or
      recommonmark).
    - Raise ImportError, if import of the upstream parser
      module fails. If called from an "include" directive, the
      system-message now has source/line info.
    - Adapt to and test with "recommonmark" versions 0.6.0 and
      0.7.1.
    - Update PEP base URL (fixes bug #445), use "https:" scheme
      in RFC base URL.
    - Add reporter to Directive class attributes.
    - parser_name() keeps details if converting ImportError to
      ValueError.
    - Don't use mutable default values for function
      arguments. Fixes bug #430.
    - Fix bug #435: invalid references in problematic nodes with
      report_level=4.
    - decode_path() returns str instance instead of
      nodes.reprunicode.
    - Add deprecation warning.
    - Add "html writers" to config_section_dependencies. Fixes
      bug #443.
    - Write table column widths with 3 digits precision. Fixes
      bug #444.
    - Add space before "charset" meta tag closing sequence.
    - Remove class value "controls" from an image node with video
      content after converting it to a "control" attribute of the
      <video> tag.
    - Wrap groups of footnotes in an <aside> for easier styling.
    - Use "https:" scheme in "python_home" URL default.
    - Fix links in template.txt.
    - New "docutils" console script entry point. Fixes bug #447.
    - Always encode the log file "alltests.out" using 'utf-8'.
    - exception_data() now returns None if no exception was
      raised.
    - recommonmark_wrapper only imported if upstream parser is
      present.
    - Fix bug #436: Null char valid in CSV since Python 3.11.
    - Allow 3rd-party drop-in components for reader and parser,
      too.
    - Fix help output.
    - Actual code moved to docutils.__main__.py.
    - Options -h and --help print short usage message.
- Release 0.18.1 (2021-11-23)
    - Node.traverse() returns a list again to restore backwards
      compatibility. Fixes bug #431.
    - New method Node.findall(): like Node.traverse() but returns
      an iterator. Obsoletes Node.traverse().
    - Fix behaviour of get_stylesheet_list(): do not look up
      stylesheets given as "stylesheet" setting. Cf. bug #434.
    - Fix handling of footnote_backlinks==False (report Alan G
      Isaac).
    - Fix typo (bug #432).
    - Fix spurious output with Windows (bug #350).
    - Fix a false positive (bug #434).
- Release 0.18 (2021-10-26)
    - mark as provisional (will switch from using "optparse" to
      "argparse").
    - remove hack for the now obsolete "mod_python" Apache
      module.
    - new function get_default_settings()
    - Don't change a list while looping over it (in
      document.set_name_id_map()). Thanks to Mickey Endito.
    - Test and update to work with recommonmark version
      0.6.0. Still provisional.
    - Unfortunately, recommonmark is no longer maintained.
    - Fix bug #424 Wrong circular inclusion detection. Use a
      "magic" comment instead of line numbers to keep a log of
      recursive inclusions.
    - Use a "magic" comment to update the log of recursive
      inclusions.
    - New option `image_loading`_. Support "lazy" loading of
      images. Obsoletes "embed_images".
    - Fix spelling of option "detailed".
    - Read settings from standard configuration files.
- Release 0.18b1 (2021-10-05)
  - The default value for the "auto_id_prefix" setting changed to
    "%": auto-generated IDs use the tag name as prefix.
  - Make meta a standard Docutils doctree node. Writers may
    ignore "meta" nodes if they are not supported by the output
    format.
  - document.make_id(): Do not strip leading number and hyphen
    characters from name if the id_prefix setting is non-empty.
  - Node.traverse() returns an iterator instead of a list.
  - Removed. (Meta directive moved to misc.py.)
  - Meta directive class (moved from html.py) inserts meta
    (instead of pending) nodes.
  - Add class option to Raw directive.
  - Unify behaviour of "widths" option: check that the length of
    an integer list equals the number of table columns also for
    the "table" directive.
  - Fork from elyxer and remove code that is not required for
    math conversion.
  - Scale variable sized operators and big delimiters with CSS
  - Support more commands, fix mapping of commands to Unicode
    characters (cf. LaTeX syntax for mathematics).
  - Fix bug #244 Wrong subscript/superscript order.
  - Don't use <tt> element (deprecated in HTML5).
  - Use STIX fonts if available.
  - Fix source location (line number) for attribution
    elements. Patch by Mickey Endito.
  - Add line, source, and rawsource internal attributes for
    blockquote elements. Patch by Mickey Endito.
  - Skip system_messages when propagating targets. Fixes bug
    #425.
  - Removed unique_combinations() (obsoleted by
    itertools.combinations()).
  - Major update (fixes and support for additional commands and
    symbols). Fixes bug #407.
  - Write footnote brackets and field term colons to HTML, so
    that they are present also without CSS and when copying
    text. Adapt minimal.css.
  - Use semantic tags <aside> for footnote text, topics,
    admonitions, and system-messages and <nav> for the table of
    contents. Use <div> for citations.
  - Only specify table column widths, if the "widths" option
    is set and is not "auto" (fixes bug #426). The table_style
    setting "colwidths-grid" restores the current default.
  - Use ARIA roles to enable accessible HTML for abstract,
    dedication, the table of contents, footnote, references,
    footnotes, citations, and backlinks.
  - Use "aria-level" attribute instead of invalid tags <h7>,
    <h8>, ... for headings of deeply nested sections.
  - Do not set classes "compound-first", "compound-middle",
    or "compound-last" to elements nested in a compound. Use
    class value "backrefs" instead of "fn-backref" for a span of
    back-references.
  - Do not write class values handled by the HTML writer
    ("colwidths-auto", "colwidths-given", "colwidths-grid") to
    the output.
  - Move space character between section number and heading into
    "sectnum" span.
  - Removed attribute HTMLTranslator.topic_classes
  - Items of a definition list with class argument "details" are
    converted to details disclosure elements.
  - Overwrite methods in _html_base.HTMLTranslator that use
    HTML5 tags (details, aside, nav, ...) and attributes (role,
    aria-level).
  - The setting legacy_class_functions now defaults to
    "False". Adapt stylesheets modifying \DUadmonition and/or
    \DUtitle.
  - Apply patch #181 "Fix tocdepth when chapter/part in use" by
    John Thorvald Wodder II.
  - Fix newlines after/before ids_to_labels() (cf. patch #183).
  - Refactor/revise ToC writing.
  - Don't add \phantomsection to labels in math-blocks.
  - Improve spacing and allow customization of Docutils-generated
    table of contents.
  - New algorithm for table column widths. Fixes bug #422. New
    configuration setting legacy_column_widths.
  - Table.set_table_style() arguments changed.
  - Only write "continued on next page..." if it fits without
    making the table columns wider.
  - Table width option overrides conflicting "auto" column
    widths.
  - Fix excessive padding above sidebar titles.
  - Fix option "detailed" under Python 2.7.
  - Remove IE 6 compatibility workarounds iepngfix.htc and
    blank.gif (fixes bug #169).
  - Fix: double quotes need to be escaped on macro
    invocation. Done everywhere.

• Files Changed
• Browse Source