Changelog

All notable changes to Apex will be documented in this file.

0.1.43 - 2025-12-31

Changed

• Reference image attribute expansion now converts IAL attributes (ID, classes) to key=value format for compatibility with inline image parsing
• Test suite refactored: split test_runner.c into multiple files (test_basic.c, test_extensions.c, test_helpers.c, test_ial.c, test_links.c, test_metadata.c, test_output.c, test_tables.c) for better organization
• Test fixtures reorganized: moved all .md test files from tests/ to tests/fixtures/ with subdirectories (basic/, demos/, extensions/, ial/, images/, output/, tables/)

New

• Support for Pandoc-style table captions using : Caption syntax (in addition to existing [Caption] and Table: Caption formats)
• IAL attributes in table captions are now extracted and applied to the table element (e.g., : Caption {#id .class} applies id and class to the table)
• Support for Pandoc-style IAL syntax without colon prefix ({#id .class} in addition to Kramdown {:#id .class} format)
• Support Pandoc-style IAL syntax ({#id .class}) in addition to Kramdown-style ({: #id .class}) for all IAL contexts including block-level, inline, and paragraph IALs
• Extract_ial_from_text function now recognizes both {: and {# or {. formats when extracting IALs from text
• Extract_ial_from_paragraph function now accepts Pandoc-style IALs for pure IAL paragraphs
• Process_span_ial_in_container function now processes Pandoc-style IALs for inline elements like links, images, and emphasis
• Is_ial_line function now detects Pandoc-style IAL-only lines in addition to Kramdown format
• Support Pandoc-style IAL syntax ({#id .class}) in addition to Kramdown-style ({: #id .class}) for all IAL contexts including block-level elements, paragraphs, inline elements, and headings
• Add support for Pandoc fenced divs syntax (::::: {#id .class} … :::::) in unified mode, enabled by default
• Add –divs and –no-divs command-line flags to control fenced divs processing
• Add comprehensive test suite for Pandoc fenced divs covering basic divs, nested divs, attributes, and edge cases
• Add bracketed spans feature that converts [text]{IAL} syntax to HTML span elements with attributes, enabled by default in unified mode
• Add –spans and –no-spans command-line flags to enable/disable bracketed spans in other modes
• Bracketed spans support all IAL attribute types (IDs, classes, key-value pairs) and process markdown inside spans
• Reference link definitions take precedence over bracketed spans - if [text] matches a reference link, it remains a link
• Add comprehensive test suite for bracketed spans including reference link precedence, nested brackets, and markdown processing
• Add bracketed spans examples and documentation
• Add automatic width/height attribute conversion: percentages and non-integer/non-px values convert to style attributes, Xpx values convert to integer width/height attributes (strips px suffix), bare integers remain as width/height attributes
• Add support for Pandoc/Kramdown IAL syntax on images: inline images support IAL after closing paren like {#id .class width=50%}
• Add support for IAL syntax after titles in reference image definitions: [ref]: url “title” {#id .class width=50%}
• Add support for Pandoc-style IAL with space prefix: { width=50% } syntax works for images
• Add comprehensive test suite for width/height conversion covering percentages, pixels, integers, mixed cases, and edge cases like decimals and viewport units
• Reference image definitions now preserve and include title attributes from definitions like [ref]: url “title” {#id}

Improved

• Table caption preprocessing now converts : Caption format to [Caption] format before definition list processing to avoid conflicts
• HTML renderer now extracts and injects IAL attributes from table captions while excluding internal attributes like data-caption
• IAL parsing automatically detects format and adjusts content offset accordingly (2 chars for {: format, 1 char for {# or {. format)
• HTML markdown extension now uses CMARK_OPT_UNSAFE to allow raw HTML including nested divs in processed content
• Width/height conversion properly merges with existing style attributes when both are present
• IAL detection now properly handles whitespace before IAL syntax for both inline and reference images
• Reference image expansion now correctly skips IAL even when closing paren is not found

Fixed

• Table caption test assertions now correctly match table tags with attributes by using <table instead of <table>
• Extract_ial_from_paragraph now allows newline character after closing brace in IAL syntax
• List items with key:value format (e.g., “- Foo: Bar”) are no longer incorrectly parsed as MMD metadata in unified mode
• Fenced divs now add markdown=”1” attribute so content inside divs is properly parsed as markdown
• HTML markdown extension now preserves all attributes (id, class, custom attributes) when processing divs with markdown=”1”
• HTML markdown extension now recursively processes nested divs with markdown=”1” attributes
• HTML markdown extension now adds newline after closing div tags to ensure following markdown headers and content are parsed correctly
• Bracketed spans now correctly handle nested brackets by matching outer brackets instead of first closing bracket
• Remove test assertions checking for markdown attribute on bracketed spans which is correctly removed by html_markdown extension
• IAL syntax with spaces (e.g., { width=50% }) now correctly detected and processed for images
• IAL syntax is now properly stripped from output even when parsing fails, preventing raw IAL from appearing in HTML
• Reference image definitions with IAL after URL (no title) now correctly detected and processed
• Test string length issues: replaced hardcoded lengths with strlen() calls to ensure full IAL syntax is processed in width/height conversion tests
• Removed unused style_attr_index variable to eliminate compiler warning

0.1.42 - 2025-12-30

New

• ALD references can now be combined with additional attributes in the same IAL (e.g., {:id .class3} where id is an ALD reference and .class3 is an additional class)

Improved

• When merging ALD attributes with additional attributes, duplicate key-value pairs are now replaced instead of duplicated (e.g., if ALD defines rel=”x” and IAL includes rel=”y”, the result is rel=”y”)
• Classes from additional attributes are appended to ALD classes, and IDs in IALs override ALD IDs when specified
• Enhanced merge_attributes function to properly handle attribute key conflicts by replacing existing values rather than creating duplicates

0.1.41 - 2025-12-30

New

• Inline Attribute Lists (IALs) can now appear immediately after inline elements within paragraphs, not just at the end of paragraphs
• IALs can be applied to strong (bold), emphasis (italic), and code elements in addition to links and images
• IALs now work with nested inline elements, allowing attributes to be applied to italic text inside bold text and similar nested structures
• Added comprehensive test suite for inline IAL functionality covering links, strong, emphasis, code, multiple IALs, and edge cases
• Added IAL demo markdown file (tests/ial_demo.md) demonstrating all supported IAL features
• Added script (tests/generate_ial_demo.sh) to automatically generate an HTML file with interactive attribute inspection tooltips

Fixed

• IALs (Inline Attribute Lists) are now correctly applied to the intended link element when multiple links in a document share the same URL
• IALs are now correctly applied to the intended element when multiple elements share the same URL or content, using separate element counters for each inline element type
• Block-level HTML elements now correctly include a space between the tag name and attributes when IAL attributes are injected

0.1.40 - 2025-12-23

Changed

• Table captions now default to below the table instead of above
• Tfoot row detection in AST now separates the logic for marking the === row itself versus rows that come after it, improving accuracy of tfoot section identification.
• Disallow using –combine and –mmd-merge together to avoid ambiguous multi-file behavior
• Update CSV include and inline table handling so both share the same CSV-to-table conversion and alignment behavior

New

• Added –captions option to control caption position (above or below)
• Added default CSS styling for figcaption elements in standalone output (centered, bold, 0.8em)
• Added CSS styling for table figures to align captions with tables (fit-content width)
• Support for tables that start with alignment rows (separator rows) without header rows. Column alignment specified in the separator row is automatically applied to all data columns.
• Support for individual cell alignment in tables using colons, similar to Jekyll Spaceship. Cells can be aligned independently with :Text (left), Text: (right), or :Text: (center). Colons are removed from output and alignment is applied via CSS text-align styles.
• Support per-cell alignment markers inside table cells
• Support multiline table cells using trailing backslash markers
• Support header and footer colspans based on empty cells
• Add –combine CLI mode to concatenate Markdown files with include expansion and GitBook-style SUMMARY.md index support.
• Add –mmd-merge CLI mode to merge MultiMarkdown index files into a single Markdown stream
• Support indentation-based header level shifting when merging mmd_merge index entries
• Support inline CSV/TSV tables using ```table fenced blocks with automatic CSV/TSV delimiter detection
• Support markers that convert following CSV/TSV lines into Markdown tables until a blank line
• Add –aria command-line option to enable ARIA labels and accessibility attributes in HTML output
• Add aria-label=”Table of contents” to TOC navigation elements when –aria is enabled
• Add role=”figure” to figure elements when –aria is enabled
• Add role=”table” to table elements when –aria is enabled
• Generate id attributes for figcaption elements in table figures when –aria is enabled
• Add aria-describedby attributes linking tables to their captions when –aria is enabled

Improved

• Removed unused variables to eliminate compiler warnings
• Empty thead sections from headerless tables are now removed from HTML output instead of rendering empty header cells.
• Table row mapping now better handles the relationship between HTML row indices and AST row indices, accounting for separator rows that are removed from HTML output.
• Added safeguards to prevent rows that should be in tbody from being skipped, including protection for the first few rows (header and first two data rows) when a === separator is present.
• Make table captions positionable above or below tables
• Center and style figcaptions in standalone HTML output
• Support optional alignment keyword rows (left, right, center, auto) and headless tables for both included CSV files and inline CSV/TSV data
• Preserve ```table fences without commas or tabs by leaving them as code blocks so users can show literal CSV/TSV without conversion

Fixed

• Removed unused variable ‘row_idx’ from advanced_tables.c to eliminate compiler warnings.
• Rows before the === separator are now correctly placed in tbody instead of being incorrectly placed in tfoot or skipped entirely. The fix includes HTML position verification to ensure rows that appear before the === row in the rendered HTML are always in tbody, regardless of AST marking.
• Prevent === separator rows from appearing as table content
• Ensure footer rows render in tfoot without losing body rows
• Preserve legitimate empty cells such as missing Q4 values
• Apply ^^ rowspans correctly for all table sections
• Apply ^^ rowspans correctly across table sections without leaking into unrelated rows
• Support footer colspans so footer cells can span multiple columns like headers and body rows
• Preserve legitimate empty table cells that are not part of colspans or rowspans
• KaTeX auto-render now properly configures delimiters and manually renders math spans to prevent plain text from appearing after rendered equations
• Relaxed table header conversion now only runs when relaxed_tables option is enabled
• HTML document wrapping now strips existing </body></html> tags to prevent duplicates when content already contains them
• KaTeX auto-render now properly configures delimiters and manually renders math spans to prevent plain text from appearing after rendered equations
• Relaxed table header conversion now only runs when relaxed_tables option is enabled
• HTML document wrapping now strips existing </body></html> tags to prevent duplicates when content already contains them
• Table captions appearing after their tables now correctly link via aria-describedby attributes

0.1.39 - 2025-12-19

Changed

• Table post-processing now tracks all cells (including removed ones) to correctly map HTML column positions to original AST column indices, ensuring attributes are applied to the correct cells.
• Added content-based detection for ^^ marker cells during HTML post-processing to ensure they are properly removed even when attribute matching fails.

New

• Added content verification to prevent false matches when cells are covered by rowspans
• Added tracking of previous cell’s colspan to detect and remove empty cells after colspan
• Added detection and removal of === marker rows in tfoot sections

Improved

• Rowspan cell tracking now uses a per-column active cell approach (inspired by Jekyll Spaceship) for more reliable rowspan calculation across complex table structures.
• Cell matching now uses position-based fallback to previous row when current row match fails
• Row mapping accounts for rowspan coverage to correctly identify visible columns
• Tfoot row detection now uses AST row indices instead of HTML row indices for accuracy

Fixed

• Table rowspan rendering now correctly handles rows where most cells use ^^ markers. Previously, cells like “Beta” and “Gamma” in rows with multiple ^^ cells would be missing or appear in the wrong position. The fix includes proper mapping between HTML row indices and AST row indices, accounting for separator rows that are removed from HTML output.
• Missing cells after colspan (e.g., “92.00” cell was missing when “Absent” had colspan=”2”)
• Rowspan not applying correctly when HTML row mapping was off by one
• Footer alignment rows (=== markers) appearing in output instead of being removed
• Empty cells after colspan not being removed from rendered HTML

0.1.38 - 2025-12-18

Changed

• In standalone mode, insert script tags just before </body>
• In snippet mode, append script tags at the end of the HTML fragment
• When –embed-css is used with –css, replace the stylesheet tag with an inline

New

• Support Pandoc-style “Table: Caption” syntax and
• Add –script CLI flag to inject scripts into HTML output
• Support shorthands for common JS libraries (mermaid, mathjax, katex, highlightjs, prism, htmx, alpine)
• Add –embed-css option to inline CSS files into the standalone document head

Improved

• Compress extraneous newlines between HTML elements
• Remove unused apex_remote_trim helper to eliminate compiler warnings

Fixed

• Prevent caption paragraphs from being reused across
• Skip URL encoding for footnote definitions ([^id]: …) so footnote

0.1.37 - 2025-12-17

Changed

• Image attributes are mode-dependent: work in Unified and MultiMarkdown modes only
• URL encoding is mode-dependent: works in Unified, MultiMarkdown, and Kramdown modes
• Improved caption detection to check all table rows for caption markers, not just the last row, to handle cases where captions come after tfoot rows.

New

• Support for MultiMarkdown-style image attributes in unified and MultiMarkdown modes
• Inline image attributes:
• Reference-style image attributes: ![][ref] with [ref]: url width=300
• Automatic URL encoding for links with spaces in unified, MultiMarkdown, and Kramdown modes
• URLs with spaces are automatically percent-encoded (e.g., “path with spaces.png” becomes “path%20with%20spaces.png”)
• Added support for MultiMarkdown-style image attributes in reference-style images. Reference definitions can now include attributes: [img1]: image.png width=300 style=”float:left”
• Added support for inline image attributes:
• Added automatic URL encoding for all link URLs (images and regular links). URLs with spaces are automatically percent-encoded (e.g., “path to/image.png” becomes “path%20to/image.png”)
• Added detection and removal of table alignment separator rows that were incorrectly being rendered as table rows.
• Added test cases for table captions appearing before and after tables.
• Added support for tfoot sections in tables using === row markers. Rows containing === markers are now placed in <tfoot> sections, and all subsequent rows after the first === row are also placed in tfoot.
• Added comprehensive table feature tests that validate rowspan,

Improved

• Improved attribute injection in HTML renderer to correctly place attributes before closing > or /> in img and link tags
• Enhanced URL parsing to distinguish between spaces within URLs vs spaces before attributes using forward-scanning pattern detection
• Self-closing img tags now consistently use “ />” (space before slash) when attributes are injected, matching the format used by cmark-gfm for img tags without injected attributes
• Rowspan and colspan attribute handling now properly appends to existing attributes instead of replacing them, allowing multiple attributes to coexist on table cells.
• Alignment rows (rows containing only ‘’ characters) are now detected and marked for removal, preventing them from appearing in HTML output.

Fixed

• Fixed bug where image prefix “![” was incorrectly removed during preprocessing of expanded reference-style images
• URL encoding now only encodes unsafe characters (space, control chars, non-ASCII). Valid URL characters like /, :, ?, #, ~, etc. are preserved and no longer incorrectly encoded.
• Titles in links and images are now correctly detected and excluded from URL encoding. Supports quoted titles (“title”, ‘title’) and parentheses titles ((title)). URLs with parentheses (like Wikipedia links) are correctly distinguished from titles based on whether a space precedes the opening parenthesis.
• Reference-style images with attributes now render correctly. Reference definitions with image attributes are removed from output, while those without attributes are preserved (with URL encoding) so cmark can resolve the references.
• Spacing between attributes in HTML output. Attributes injected into img and link tags now have proper spacing, preventing malformed HTML like alt=”text”width=”100”.
• Table attributes now render correctly with proper spacing. Fixed missing space in table tag when id attribute immediately follows (e.g., <tableid=”…” now renders as <table id=”…”).
• Rowspan and colspan injection now works correctly in all cases. Fixed bug where table tracking variables weren’t set when fixing missing space in table tag (e.g., <tableid=”…”), causing row and cell processing to be skipped. Table tracking is now properly initialized even when correcting tag spacing.
• Captions after tables were not being detected when tables had IAL attributes, as IAL processing replaced the caption data stored in user_data. Added fallback logic to check for caption paragraphs directly in the AST when user_data lookup fails.
• Rows containing only === markers are now properly skipped entirely rather than rendering as empty cells in tfoot sections.
• Caption paragraphs before tables are now properly removed,

0.1.36 - 2025-12-16

Fixed

• Resolve CMake error when building framework where file(GLOB) returns multiple dylib files, causing semicolon-concatenated paths in file(COPY) command. Now extracts first file from glob result before copying.
• Homebrew installation now correctly links to system libyaml instead of hardcoded CI path

0.1.35 - 2025-12-16

Changed

• Update Homebrew formula to install from a precompiled macOS universal binary instead of building from source with cmake.
• Allow –install-plugin to accept a Git URL or GitHub shorthand (user/repo) in addition to directory IDs when installing plugins.

Improved

• Simplify Homebrew installation so users no longer need cmake or Xcode build tools to install apex.
• Add an interactive security confirmation when installing plugins from a direct Git URL or GitHub repo name, reminding users that plugins execute unverified code.

0.1.34 - 2025-12-16

0.1.33 - 2025-12-16

0.1.32 - 2025-12-16

0.1.31 - 2025-12-16

0.1.30 - 2025-12-16

Changed

• Make –list-plugins show installed plugins before remote ones.
• Prevent remote plugins that are already installed from being listed under Available Plugins.
• Build system now detects libyaml via multiple methods (yaml-0.1, yaml, libyaml) for better cross-platform support.
• Homebrew formula now includes libyaml as a dependency to ensure full YAML support.
• Suppressed unused-parameter warnings from vendored cmark-gfm extensions to reduce build noise.

New

• Add –uninstall-plugin CLI flag to remove installed plugins.
• Run optional post_install command from plugin.yml after cloning a plugin.
• Full YAML parsing support using libyaml for arrays and nested structures in metadata and plugin manifests.
• Plugin bundle support allowing multiple plugins to be defined in a single plugin.yml manifest.
• Expose APEX_FILE_PATH to external plugins so scripts can see the original input path or base directory when processing.

Improved

• Split() metadata transform now accepts regular expressions as delimiters (for example split(,\s*)).
• YAML arrays are automatically normalized to comma-separated strings for backward compatibility with existing metadata transforms.
• External plugin environment now includes the source file path (when available) alongside APEX_PLUGIN_DIR and APEX_SUPPORT_DIR.

Fixed

• Tighten mutual-exclusion checks between install and uninstall plugin flags.
• Ensure CMake policy version is compatible with vendored cmark-gfm on newer CMake releases.
• Install the Apex framework with its public apex.h header correctly embedded in Apex.framework/Headers for Xcode use.
• Bundle libcmark-gfm and libcmark-gfm-extensions dylibs into Apex.framework so dependent apps no longer hit missing library errors at runtime.

0.1.29 - 2025-12-15

Changed

• Make –list-plugins show installed plugins before remote ones.
• Prevent remote plugins that are already installed from being listed under Available Plugins.

New

• Initial planning for a remote plugin directory and install features
• Add –uninstall-plugin CLI flag to remove installed plugins.

Fixed

• Superscript/subscript no longer process content inside Liquid {% %} tags.
• Autolink detection skips Liquid {% %} tags so emails and URLs are not rewritten there.
• Fix directory url for --list-plugins

0.1.28 - 2025-12-15

Changed

• Default wikilink URLs now replace spaces with dashes (e.g. [[Home Page]] -> href=”Home-Page”).

New

• Add –wikilink-space and –wikilink-extension flags to control how [[WikiLink]] hrefs are generated.
• Allow wikilink space and extension configuration via metadata keys wikilink-space and wikilink-extension.
• Support Kramdown-style {:toc …} markers mapped to Apex TOC generation.
• Add tests for {:toc} syntaxes
• MMD includes support full glob patterns like {{*.md}} and {{c?de.py}}.
• Add plugin discovery from .apex/plugins and ~/.config/apex/plugins.
• Allow external handler plugins in any language via JSON stdin/stdout.
• Support declarative regex plugins for pre_parse and post_render phases.
• Add --no-plugins CLI flag to disable all plugins for a run.
• Support plugins: true/false metadata to enable or disable plugins.
• Initial planning for a remote plugin directory and install features

Improved

• Exclude headings with .no_toc class from generated tables of contents for finer-grained TOC control.
• MMD-style {{file.*}} now resolves preferred extensions before globbing.
• Transclusion respects brace-style patterns such as {{{intro,part1}.md}} where supported.
• Provide APEX_PLUGIN_DIR and APEX_SUPPORT_DIR for plugin code and data.
• Add profiling (APEX_PROFILE_PLUGINS=1) for plugins

0.1.27 - 2025-12-15

Changed

• Default wikilink URLs now replace spaces with dashes (e.g. [[Home Page]] -> href=”Home-Page”).

New

• Add –wikilink-space and –wikilink-extension flags to control how [[WikiLink]] hrefs are generated.
• Allow wikilink space and extension configuration via metadata keys wikilink-space and wikilink-extension.
• Support Kramdown-style {:toc …} markers mapped to Apex TOC generation.
• Add tests for {:toc} syntaxes
• MMD includes support full glob patterns like {{*.md}} and {{c?de.py}}.
• Add plugin discovery from .apex/plugins and ~/.config/apex/plugins.
• Allow external handler plugins in any language via JSON stdin/stdout.
• Support declarative regex plugins for pre_parse and post_render phases.
• Add --no-plugins CLI flag to disable all plugins for a run.
• Support plugins: true/false metadata to enable or disable plugins.
• Initial planning for a remote plugin directory and install features

Improved

• Exclude headings with .no_toc class from generated tables of contents for finer-grained TOC control.
• MMD-style {{file.*}} now resolves preferred extensions before globbing.
• Transclusion respects brace-style patterns such as {{{intro,part1}.md}} where supported.
• Provide APEX_PLUGIN_DIR and APEX_SUPPORT_DIR for plugin code and data.
• Add profiling (APEX_PROFILE_PLUGINS=1) for plugins

0.1.26 - 2025-12-14

Changed

• Change --enable-includes to --[no-]includes, allowing --no-includes to disable includes in unified mode and shortening the flag
• Integrate metadata-to-options application into CLI after metadata merging
• Preserve bibliography files array when metadata mode resets options structure

New

• Add apex_apply_metadata_to_options() function to apply metadata values to apex_options structure
• Support controlling boolean flags via metadata (indices, wikilinks, includes, relaxed-tables, alpha-lists, mixed-lists, sup-sub, autolink, transforms, unsafe, tables, footnotes, smart, math, ids, header-anchors, embed-images, link-citations, show-tooltips, suppress-bibliography, suppress-index, group-index-by-letter, obfuscate-emails, pretty, standalone, hardbreaks)
• Support controlling string options via metadata (bibliography, csl, title, style/css, id-format, base-dir, mode)
• Boolean metadata values accept true/false, yes/no, or 1/0 (case-insensitive, downcased)
• String metadata values used directly for options that take arguments
• Metadata mode option resets options to mode defaults before applying other metadata
• Comprehensive tests for metadata control of command line options

0.1.25 - 2025-12-13

New

• Add citation processing with support for Pandoc, MultiMarkdown, and mmark syntaxes
• Add bibliography loading from BibTeX, CSL JSON, and CSL YAML formats
• Add –bibliography CLI option to specify bibliography files (can be used multiple times)
• Add –csl CLI option to specify citation style file
• Add –no-bibliography CLI option to suppress bibliography output
• Add –link-citations CLI option to link citations to bibliography entries
• Add –show-tooltips CLI option for citation tooltips
• Add bibliography generation and insertion at marker
• Add support for bibliography specified in document metadata
• Added missing docs and man page for citation support
• Add support for transclude base metadata to control file transclusion paths
• Add Base Header Level and HTML Header Level metadata to adjust heading levels
• Add CSS metadata to link external stylesheets in standalone HTML documents
• Add HTML Header and HTML Footer metadata to inject custom HTML
• Add Language metadata to set HTML lang attribute in standalone documents
• Add Quotes Language metadata to control smart quote styles (French, German, Spanish, etc.)
• Add –css CLI flag as alias for –style with metadata override precedence
• Add metadata key normalization: case-insensitive matching with spaces removed (e.g., “HTML Header Level” matches “htmlheaderlevel”)
• Add index extension supporting mmark syntax (!item), (!item, subitem), and (!!item, subitem) for primary entries
• Add TextIndex syntax support with {^}, [term]{^}, and {^params} patterns
• Add automatic index generation at end of document or at marker
• Add alphabetical sorting and optional grouping by first letter for index entries
• Add hierarchical sub-item support in generated index
• Add –indices CLI flag to enable index processing
• Add –no-indices CLI flag to disable index processing
• Add –no-index CLI flag to suppress index generation while keeping markers
• Add comprehensive test suite with 40 index tests covering both syntaxes

Improved

• Only process citations when bibliography is actually provided for better performance
• Add comprehensive tests for MultiMarkdown metadata keys
• Add comprehensive performance profiling system (APEX_PROFILE=1) to measure processing time for all extensions and CLI operations
• Add early exit checks for IAL processing when no {: markers are present
• Add early exit checks for index processing when no index patterns are found
• Add early exit checks for citation processing when no citation patterns are found
• Add early exit checks for definition list processing when no : patterns are found
• Optimize alpha lists postprocessing with single-pass algorithm replacing O(n*m) strstr() loops
• Add early exit check for alpha lists postprocessing when no markers are present
• Optimize file I/O by using fwrite() with known length instead of fputs()
• Add markdown syntax detection in definition lists to skip parser creation for plain text
• Optimize definition lists by selectively extracting only needed reference definitions instead of prepending all
• Add profiling instrumentation for all preprocessing, parsing, rendering, and post-processing steps
• Add profiling instrumentation for CLI operations (file I/O, metadata processing)

Fixed

• Prevent autolinking of @ symbols in citation syntax (e.g., [@key])
• Handle HTML comments in autolinker to preserve citation placeholders
• Fix quote language adjustment to handle Unicode curly quotes in addition to HTML entities
• Fix bibliography_files assignment to remove unnecessary cast
• Fix heap-buffer-overflow in html_renderer.c when writing null terminator (allocate capacity+1)
• Fix use-after-free in ial.c by deferring node unlinking until after iteration completes
• Fix buffer overflow in definition_list.c HTML entity escaping (correct length calculation for & and ")

0.1.24 - 2025-12-13

New

• Add citation processing with support for Pandoc, MultiMarkdown, and mmark syntaxes
• Add bibliography loading from BibTeX, CSL JSON, and CSL YAML formats
• Add –bibliography CLI option to specify bibliography files (can be used multiple times)
• Add –csl CLI option to specify citation style file
• Add –no-bibliography CLI option to suppress bibliography output
• Add –link-citations CLI option to link citations to bibliography entries
• Add –show-tooltips CLI option for citation tooltips
• Add bibliography generation and insertion at marker
• Add support for bibliography specified in document metadata

Improved

• Only process citations when bibliography is actually provided for better performance

Fixed

• Raw HTML tags and comments are now preserved in definition lists by default in unified mode. Previously, HTML content in definition list definitions was being replaced with “raw HTML omitted” even when using –unsafe or in unified mode.
• Unified mode now explicitly sets unsafe=true by default to ensure raw HTML is allowed.
• Prevent autolinking of @ symbols in citation syntax (e.g., [@key])
• Handle HTML comments in autolinker to preserve citation placeholders

0.1.23 - 2025-12-12

Changed

• Remove remote image embedding support (curl dependency removed)

New

• Add metadata variable transforms with [%key:transform] syntax
• Add –transforms and –no-transforms flags to enable/disable transforms
• Add 19 text transforms: upper, lower, title, capitalize, trim, slug, replace (with regex support), substring, truncate, default, html_escape, basename, urlencode, urldecode, prefix, suffix, remove, repeat, reverse, format, length, pad, contains
• Add array transforms: split, join, first, last, slice
• Add date/time transform: strftime with date parsing
• Add transform chaining support (multiple transforms separated by colons)
• Add –meta-file flag to load metadata from external files (YAML, MMD, or Pandoc format, auto-detected)
• Add –meta KEY=VALUE flag to set metadata from command line (supports multiple flags and comma-separated pairs)
• Add metadata merging with proper precedence: command-line > document > file
• Add –embed-images flag to embed local images as base64 data URLs in HTML output
• Add –base-dir flag to set base directory for resolving relative paths (images, includes, wiki links)
• Add automatic base directory detection from input file directory when reading from file
• Add base64 encoding utility for image data
• Add MIME type detection from file extensions (supports jpg, png, gif, webp, svg, bmp, ico)
• Add image embedding function that processes HTML and replaces local image src attributes with data URLs
• Add test suite for image embedding functionality

Improved

• Wiki link scanner now processes all links in a text node in a single pass instead of recursively processing one at a time, significantly improving performance for documents with multiple wiki links per text node.
• Added early-exit optimization to skip wiki link AST traversal entirely when no wiki link markers are present in the document.
• Improve error handling in transform execution to return original value instead of NULL on failure
• Add comprehensive test coverage for all transforms including edge cases
• Relative path resolution for images now uses base_directory option
• Base directory is automatically set from input file location when not specified

Fixed

• Fix bracket handling in regex patterns - properly match closing brackets in [%…] syntax when patterns contain brackets
• Fix YAML metadata parsing to strip quotes from quoted string values
• Raw HTML tags and comments are now preserved in definition lists by default in unified mode. Previously, HTML content in definition list definitions was being replaced with “raw HTML omitted” even when using –unsafe or in unified mode.
• Unified mode now explicitly sets unsafe=true by default to ensure raw HTML is allowed.

0.1.20 - 2025-12-11

NEW

• Added man page generation and installation support. Man pages can be generated from Markdown source using pandoc or go-md2man, with pre-generated man pages included in the repository as fallback. CMake build system now handles man page installation, and Homebrew formula installs the man page.
• Added comprehensive test suite for MMD 6 features including multi-line setext headers and link/image titles with different quote styles (single quotes, double quotes, parentheses). Tests verify these features work in both MultiMarkdown and unified modes.
• Added build-test man_page_copy target for man page installation.
• Added –obfuscate-emails flag to hex-encode mailto links.

IMPROVED

• Superscript processing now stops at sentence terminators (. , ; : ! ?) instead of including them in the superscript content. This prevents punctuation from being incorrectly included in superscripts.
• Enhanced subscript and underline detection logic. The processor now correctly differentiates between subscript (tildes within a word, e.g., H~2~O) and underline (tildes at word boundaries, e.g., ~text~) by checking if tildes are within alphanumeric words or at word boundaries.
• Expanded test coverage for superscript, subscript, underline, strikethrough, and highlight features with additional edge case tests.
• Email autolink detection trims trailing punctuation.

FIXED

• Autolink now only wraps real URLs/emails instead of every word.
• Email autolinks now use mailto: hrefs instead of bare text.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

0.1.19 - 2025-12-09

CHANGED

• HTML comments now replaced with “raw HTML omitted” in CommonMark and GFM modes by default
• Added enable_sup_sub flag to apex_options struct
• Updated mode configurations to enable sup/sub in appropriate modes
• Added sup_sub.c to CMakeLists.txt build configuration
• Removed unused variables to resolve compiler warnings
• Tag filter (GFM security feature) now only applies in GFM mode, not Unified mode, allowing raw HTML and autolinks in Unified mode as intended.
• Autolink extension registration now respects the enable_autolink option flag.

NEW

• Added MultiMarkdown-style superscript (^text^) and subscript (~text~) syntax support
• Added –[no-]sup-sub command-line option to enable/disable superscript/subscript
• Superscript/subscript enabled by default in unified and MultiMarkdown modes
• Created sup_sub extension (sup_sub.c and sup_sub.h) for processing ^ and ~ syntax
• Added –[no-]unsafe command-line option to control raw HTML handling
• Added test_sup_sub() function with 13 tests covering superscript and
• Added test_mixed_lists() function with 10 tests covering mixed list
• Added test_unsafe_mode() function with 8 tests covering raw HTML
• Added preprocessing for angle-bracket autolinks (http://...) to convert them to explicit markdown links, ensuring they work correctly with custom rendering paths.
• Added –[no-]autolink CLI option to control automatic linking of URLs and email addresses. Autolinking is enabled by default in GFM, MultiMarkdown, Kramdown, and unified modes, and disabled in CommonMark mode.
• Added enable_autolink field to apex_options structure to control autolink behavior programmatically.
• Added underline syntax support: ~text~ now renders as text when there’s a closing ~ with no space before it.

IMPROVED

• Test suite now includes 36 additional tests, increasing total test
• Autolink preprocessing now skips processing inside code spans (...) and code blocks (...), preventing URLs from being converted to links when they appear in code examples.
• Metadata replacement retains HTML edge-case handling and properly cleans up intermediate buffers.

FIXED

• Unified mode now correctly enables mixed list markers and alpha lists by default when no –mode is specified
• ^ marker now properly separates lists by creating a paragraph break instead of just blank lines
• Empty paragraphs created by ^ marker are now removed from final HTML output
• Superscript and subscript processing now skips ^ and ~ characters
• Superscript processing now skips ^ when part of footnote reference
• Subscript processing now skips ~ when part of critic markup patterns
• Setext headers are no longer broken when followed by highlight syntax (==text==). Highlight processing now stops at line breaks to prevent interference with header parsing.
• Metadata parser no longer incorrectly treats URLs and angle-bracket autolinks as metadata. Lines containing < or URLs (http://, https://, mailto:) are now skipped during metadata extraction.
• Superscript/subscript processor now correctly differentiates between ~text~ (underline), ~word (subscript), and ~~text~~ (strikethrough). Double-tilde sequences are skipped so strikethrough extension can handle them.
• Subscript processing now stops at sentence terminators (. , ; : ! ?) instead of including them in the subscript content.
• Metadata variable replacement now runs before autolinking so [%key] values containing URLs are turned into links when autolinking is enabled.
• MMD metadata parsing no longer incorrectly rejects entries with URL values; only URL-like keys or ‘<’ characters in keys are rejected, allowing “URL: https://example.com” as valid metadata.
• Headers starting with # are now correctly recognized instead of being treated as autolinks. The autolink preprocessor now skips # at the start of a line when followed by whitespace.
• Math processor now validates that \(...\) sequences contain actual math content (letters, numbers, or operators) before processing them. This prevents false positives like \(%\) from being treated as math when they only contain special characters.

0.1.18 - 2025-12-06

Fixed

• GitHub Actions workflow now properly builds separate Linux x86_64 and ARM64 binaries

0.1.17 - 2025-12-06

Fixed

• Relaxed tables now disabled by default for CommonMark, GFM, and MultiMarkdown modes (only enabled for Kramdown and Unified modes)
• Header ID extraction no longer incorrectly parses metadata variables like [%title] as MMD-style header IDs
• Tables with alignment/separator rows now correctly generate <thead> even when relaxed table mode is enabled
• Relaxed tables preprocessor preserves input newline behavior in output
• Memory management bug in IAL preprocessing removed unnecessary free call

0.1.16 - 2025-12-06

Fixed

• IAL (Inline Attribute List) markers appearing immediately after content without a blank line are now correctly parsed
• Added apex_preprocess_ial() function to ensure Kramdown-style IAL syntax works correctly with cmark-gfm parser

0.1.15 - 2025-12-06

Fixed

• Homebrew formula updated with correct version and commit hash

0.1.10 - 2025-12-06

Changed

• License changed to MIT

Added

• Homebrew formula update scripts

0.1.9 - 2025-12-06

Fixed

• Shell syntax in Linux checksum step for GitHub Actions

0.1.8 - 2025-12-06

Fixed

• Link order for Linux static builds

0.1.7 - 2025-12-06

Fixed

• Added write permissions for GitHub releases

0.1.6 - 2025-12-06

Fixed

• .gitignore pattern fixed to properly include apex headers (was incorrectly matching include/apex/)

0.1.5 - 2025-12-06

Changed

• Added verbose build output for CI debugging

0.1.4 - 2025-12-06

Fixed

• CMake build rules updated

0.1.3 - 2025-12-06

Fixed

• CMake policy version for cmark-gfm compatibility

0.1.2 - 2025-12-06

Fixed

• GitHub Actions workflow fixes

0.1.1 - 2025-12-04

Added

• CMake setup documentation

0.1.0 - 2025-12-04

Added

Core Features: - Initial release of Apex unified Markdown processor - Based on cmark-gfm for CommonMark + GFM support - Support for 5 processor modes: CommonMark, GFM, MultiMarkdown, Kramdown, Unified

Metadata: - YAML front matter parsing - MultiMarkdown metadata format - Pandoc title block format - Metadata variable replacement with [%key] syntax

Extended Syntax: - Wiki-style links: [[Page]], [[Page|Display]], [[Page#Section]] - Math support: $inline$ and $$display$$ with LaTeX - Critic Markup: All 5 types ({++add++}, {–del–}, {~~sub~~}, {==mark==}, {»comment«}) - GFM tables, strikethrough, task lists, autolinks - Reference-style footnotes - Smart typography (smart quotes, dashes, ellipsis)

Build System: - CMake build system for cross-platform support - Builds shared library, static library, CLI binary, and macOS framework - Clean compilation on macOS with Apple Clang

CLI Tool: - apex command-line binary - Support for all processor modes via --mode flag - Stdin/stdout support for Unix pipes - Comprehensive help and version information

Integration: - Objective-C wrapper (NSString+Apex) for Marked integration - macOS framework with proper exports - Detailed integration documentation and examples

Testing: - Automated test suite with 31 tests - 90% pass rate across all feature areas - Manual testing validated

Documentation: - Comprehensive user guide - Complete API reference - Architecture documentation - Integration guides - Code examples

Known Issues

• Critic Markup substitutions have edge cases with certain inputs
• Definition lists not yet implemented
• Kramdown attributes not yet implemented
• Inline footnotes not yet implemented

Performance

• Small documents (< 10KB): < 10ms
• Medium documents (< 100KB): < 100ms
• Large documents (< 1MB): < 1s

Credits

• Based on cmark-gfm by GitHub
• Developed for Marked by Brett Terpstra