Log message:
textproc/ruby-asciidoctor: update to 2.0.23

2.0.23 (2024-05-17)

This is a patch release in the 2.0.x release line to fix bugs and deal with
compliance issues.

The primary bug fix in this release is to not duplicate the block attribute
line above a detached block that breaks a dlist, which fixes the duplicate
role warning on that detached block.  The main compliance improvement is to
encode spaces in mailto links using %20, in accordance with RFC 3986,
instead of +.

This release introduces several fixes and improvements related to parsing a
table that has jagged rows.  If the colspan of the last cell in a row,
including the last row, exceeds the specified number of columns, only that
row is dropped.  On the other hand, if the last row is missing cells, an
error message is logged to notify the author.

If a URL is used as the target of an xref shorthand, the processor now
treats the URL as a fragment identifier instead of crashing.

Finally, repeating spaces in verbatim content in the manpage output are
preserved.

Changelog

Compliance

* Encode spaces in mailto links as %20, in accordance with RFC 3986, instead
  of + (#4576)

Improvements

* Log error when an incomplete row is detected at the end of a table (#4573)

Bug Fixes

* Don't leave behind empty line inside skipped preprocessor conditional
  (#4580)
* Don't duplicate block attribute line above detached block that breaks a
  dlist; fixes duplicate role on detached block (#4565)
* Don't crash when parsing xref shorthand if target starts with URL protocol
  and text is offset by space (#4570)
* Only drop current row if colspan of last cell exceeds specified number of
  columns (#4587)
* Drop last row if colspan of last cell in table exceeds specified number of
  columns (#4587)
* Preserve repeating spaces in verbatim content in manpage output (#3583)

Credits

A very special thanks to all the awesome supporters of the Asciidoctor
OpenCollective campaign, who provide critical funding for the ongoing
development of this project.

Log message:
textproc/ruby-asciidoctor: update to 2.0.22

2.0.22 (2024-03-08)

This is a patch release in the 2.0.x release line to further compatibility
with Ruby 3.3.

The most critical fix in this release is to remove use of the base64 library
since it will no longer being provided as part of the Ruby language starting
with Ruby 3.4.  When using Ruby 3.3, users would see a warning if this
library is used, which happens when the data-uri attribute is set.  The
functionality has been replaced with a lower-level method on the String
object (String#pack), which the base64 library delegates to anyway.

This release also sets the cloaked-context attribute on a source block that
is defined using either a styled paragraph or a literal or fenced code
structural container.  This matches the functionality used for custom block
extensions (when the source context does not match the final parsed
context).

Distribution

* RubyGem (asciidoctor)

Asciidoctor is also packaged for Fedora, Debian, Ubuntu, Alpine Linux,
OpenSUSE, and Homebrew.  You can use the system's package manager to install
the package named asciidoctor.

Changelog

Improvements

* Set cloaked-context attribute on source block when context is not :listing
  (#4556)

Compliance

* Remove use of base64 library to prevent warning in Ruby >= 3.3 (#4561)

Log message:
textproc/ruby-asciidoctor: update to 2.0.21

2.0.21 (2024-02-20)

This is a patch release in the 2.0.x release line, primarily to address
compatibility with Ruby 3.3.  While development of Asciidoctor 2.1 is
beginning to pick up again, we extracted several other important
improvements and bug fixes from main to include in this patch release.

The most critical fix in this release is to make Asciidoctor's logger
compatible with the logger API in Ruby 3.3.  The code is now tested against
Ruby 3.3 on each change.  Nearly as important, this release also turns off
Ruby's system-dependent newline conversion when writing files so that
Asciidoctor's output is consistent between *nix and Windows.

Numerous changes were made to correct AsciiDoc parsing.  A nested dlist
attached using a list continuation no longer consumes a detached paragraph.
And an attached block is correctly attached to a nested dlist when that
dlist is offset from its parent by an empty line.  The secondary and
tertiary terms on the indexterm macro are preserved when the primary term is
quoted and contains an equals sign.  And a bare URL enclosed in angle
brackets acts as unconstrained syntax that stops at the first occurrence of
a closing angle bracket.  When includes are not enabled, any spaces in the
include target are preserved when generating the fallback link.

Several changes were made to the DocBook output to address compliance.  The
abstract is moved inside the <info> tag (applies to both the abstract in the
preamble as well an an abstract for a book part).  The scaledwidth and scale
attributes are now supported on an inline macro, like with a block image
macro.

A few changes to the manpage output were made.  Cells in the head row are
styled as bold.  Paragraph breaks in a normal table cell are preserved.

Finally, a few refinements to the default stylesheet were applied.  The
extra border below the doctitle when the sidebar toc is collapsed into main
content area was removed.  And a text decoration was added to the active
footnote number link in the footnotes list.

Compliance

* Turn off system-dependent newline conversion when writing files; don't
  convert line feeds to system-dependent newline (#4550)

* Support logger in Ruby 3.3 by instantiating super class (#4493) (@mtasaka)

* Add support for scaledwidth and scale attributes on inline image macro in
  DocBook output (#4552)

* Update latest Ruby to 3.3 in CI workflows

Improvements

* Change title to doctitle in warning message about use of abstract to make
  subject more clear

* Modify default stylesheet to add text decoration to active footnote number
  link in footnotes list (#4530) (@Larhzu)

Bug Fixes

* Nested dlist attached using list continuation should not consume detached
  paragraph (#3535)

* Don't break nested dlist with attached block if offset from parent list by \ 
empty line (#3693)

* Preserve paragraph breaks in normal table cell in manpage output (#4481)

* Style cells in head row as bold in manpage output (#4490)

* Escape spaces in include target (using inline passthrough) when generating
  link from include directive (#4461)

* Move abstract inside info tag in DocBook output (#3602)

* Honor secondary and tertiary terms on indexterm macro when primary term is
  quoted and contains an equals sign (#3652)

* Remove extra border below doctitle when sidebar toc is collapsed into main
  content area (#4523)

* Treat bare URL enclosed in angle brackets as unconstrained syntax; only
  match until closing angled bracket (#4468)

* Allow URL enclosed in angled brackets syntax to be escaped using backslash
  (#4468)

Credits

A very special thanks to all the awesome supporters of the Asciidoctor
OpenCollective campaign, who provide critical funding for the ongoing
development of this project.

Log message:
textproc/ruby-asciidoctor: update to 2.0.20

2.0.19 (2023-05-17)

Development on Asciidoctor 2.1 has been on pause while we focus on
progressing the AsciiDoc Language Specification.  In the meantime, we
extracted some of the minor improvements and bug fixes from main and put
them into this patch release in the 2.0.x release line.

The most impactful fix is a switch from recursion to a while loop to prevent
a stack limit error when looking for the next line to process.  Those using
JRuby will appreciate that fact that a uri:classloader: prefix is treated as
an absolute path rather than a URL, which will allow those paths to be read
correctly and avoid misleading security warnings.

A few refinements where made to how the AsciiDoc source is parsed.  First,
constrained passthroughs inside a monospace span are now always processed as
expected.  When parsing a dlist, a nested list that starts with at least one
block attribute line is now attached correctly.  The partintro block is now
generated consistently, despite how it's defined in the AsciiDoc source.
Numeric character references are passed over when searching for the URL
fragment in the xref target.  An inline ref is cataloged even if the
resolved reftext is empty.  An include target may not start or end with a
space (to distinguish it from a dlist entry).  Finally, square brackets are
not swallowed when parsing an escaped URL macro.

Several improvements have been made when converting to man pages.  When
generating DocBook, the root <reftext> tag is correctly enclosed in an
<article> tag and reftext substitutions are applied to the value of the
mantitle attribute.  When converting to the man page format, the alt text of
a block image is processed correctly (i.e., "manified").  In the HTML
output, a monospace span in a section title is not converted to uppercase.

The default stylesheet has been updated so that the margin on the first and
last child of a sidebar is collapsed correctly and so the font size of a
term in a horizontal list matches the font size of a term in a regular
dlist.

The code has been updated to be compatible with recent software releases,
including Ruby 3.2, Haml 6, and Rouge 4.1.

Distribution

Changelog

Improvements
* Return empty string instead of nil if raw or verbatim block has no lines
* Don't uppercase monospace span in section title in manpage output (#4402)
* Simplify processing of implicit link (i.e., autolink) by separating
  implicit and explicit match
* Generate partintro block consistently (#4450)
* Add Kiswahili translation for built-in labels (PR #4454) (@bkmgit)

Compliance
* Fix call order so use of an include file with invalid encoding continues
  to raise error when using Ruby >= 3.2.0
* Fix test assertion for fallback Rouge stylesheet to be compatible with
  Rouge 4.1 (#4406) (@tmzullinger)
* Support notitle option on section as alternative to untitled to hide title
  (#4437)
* Add support for Haml 6 to template converter (#4429)

Bug Fixes
* Process constrained inline passthrough inside monospace span (#4458)
* Catalog inline ref defined using anchor macro even when resolved reftext
  is empty
* Use while loop rather than recursion to locate next line to process;
  prevents stack limit error (#4368)
* Avoid matching numeric character references when searching for # in xref
  target (#4393)
* Use correct selector to collapse margin on first and last child of sidebar
* Don't allow target of include directive to start with a space (to
  distinguish it from a dlist item) or to end with a space
* Manify alt text of block image in manpage output (#4401)
* Adjust font size of term in horizontal dlist to match font size of term in
  regular dlist
* Implicitly attach nested list that starts with block attribute lines to
  dlist entry (#4268)
* Don't swallow square brackets when processing escaped URL macro
* Treat uri:classloader: as an absolute path prefix when running on JRuby
  (#3929)
* Apply reftext substitutions to value of mantitle attribute in DocBook
  output (#4448)
* Enclose <reftext> tag in <article> tag in DocBook output for man page
  (#4452)
* Correctly handle compat role on monospace and constrained passthrough when
  box attrlist or formatted text is escaped

2.0.20 (2023-05-18)

After the 2.0.19 release, we discovered that the man page was not being
updated during the release.  This release has no changes other than t fix
the release process.  For changes related to 2.0.19, see
https://github.com/asciidoctor/asciidoctor/releases/tag/v2.0.19.

Distribution

Changelog

Bug Fixes
* Update release-version attribute in READMEs and man page during release
* Rebuild man page during release

Log message:
Update PLIST caused by update of ruby-manpages.

Bump PKGREVISION.

Log message:
Bump PKGREVISION to reflect fix of ruby-manpages package.

Log message:
textproc/ruby-asciidoctor: proper use of RUBYGEM_MANPAGES

NFC.

Log message:
textproc/ruby-asciidoctor: add support to RUBYGEM_MANPAGES

Bump PKGREVISION since dependency might change.