Changelog

All notable changes to adoctra. The format follows Keep a Changelog, and the project uses Semantic Versioning. Binaries are on the releases page.

Unreleased

Parsing is 10–30% faster again: large, rare inline payloads are boxed, so ordinary inline nodes are copied as 112 instead of 392 bytes. Output is unchanged.
AST: AttributePart::OpenXref and AttributePart::OpenLink now hold a Box, as do the xref, image, footnote, ui and term fields of the corresponding Inline variants. Field access is unchanged; code that constructs or moves these values needs Box::new or *.

Inline substitutions are faster: replacements jump between their trigger characters, and the passthrough pass indexes markers by byte search and skips text without them. Quote and passthrough scans skip offsets that cannot start a match, and inline passes no longer scan every document attribute per paragraph. Parsing a 405 KiB document takes about half the time it did in 0.0.3; output is unchanged.

Windows: file names in messages and the docfile/docdir attributes no longer carry the \\?\ prefix (\\?\C:\Work\a.adoc); they are written like Asciidoctor’s, C:/Work/a.adoc.
DOCX: diagrams are embedded when the output goes to another directory (-o c:\tmp\out.docx), where their images are written, instead of warning "image not found or unreadable".

Windows: absolute paths with a drive letter or backslashes (-o c:\tmp\out.docx, -D, -B, includes and images) are resolved as absolute paths, like Asciidoctor on Windows, instead of being joined to the document’s directory (which failed with an invalid file name error @ dir_s_mkdir - …/C:).

First public release. An early version: the processor converts real-world

documents, but the API and output may still change between releases.

adoctra command-line tool with the options of the asciidoctor command for document conversion (-b, -d, -e, -o, -D, -R, -B, -S, -n, -a, --failure-level, -q, -v, -t).
Parser compatible with Asciidoctor 2.0.23: preprocessor directives and includes, document header and attributes, sections, all delimited blocks, lists, tables, images and media, cross references, footnotes, index terms, STEM, table of contents and the full substitution pipeline.
HTML5 backend with modern semantic markup and the bundled adoctra.css stylesheet, visually matching Asciidoctor’s default output.
Template-based DOCX backend (--to docx, docx-template attribute).
JSON backend with the document tree, attributes and diagnostics.
Server-side syntax highlighting: a port of Rouge with all 228 lexers (source-highlighter=rouge).
Diagram extension: all 58 diagram types of asciidoctor-diagram 3.2.1 through their tools, Kroki or a PlantUML server (-r asciidoctor-diagram).
Rust extension API: block, block macro, inline macro and tree processors.
WASM bindings for editors: error-tolerant parsing with byte spans, diagnostics and highlighting tokens, a virtual file system for includes, and diagrams through Kroki.