Features
- Standalone Markdown conversion API
- Attach named exports as static properties for UMD/CJS consumers
Fixes
- Detect and handle table layouts used for positioning
- Sidenote-style footnotes now treated as proper footnotes
- Fix arXiv equation tables rendered as raw HTML instead of LaTeX (#141)
- Fix arXiv multi-citation groups losing all but first reference (#142)
- Fix arXiv cross-reference numbers ("Figure 1", "Table 3") being stripped (#143)
- Fix arXiv footnote marks doubled with "footnotemark:" text leaking (#144)
- Fix content scoring removing blocks with navigation-like words in article text (#112)
- Fix over-aggressive stripping of content (#107, #77)
- Fix
&encoding breaking LaTeX alignment in complex HTML tables - Fix
script[type="math/tex"]elements not extracting LaTeX - Fix wrong authors picked up from comments and related articles (#135)
- Fix links in Quartz sites
- Fix retry heuristic preferring cluttered results for short articles
- Fix
_getSchemaText()crash when schema data contains objects instead of strings - Fix JS-warning messages appearing in output
- Collapse newlines to spaces and wrap preformatted code blocks (#84, #24)
Dependencies
- Update temml to ^0.13.1, mathml-to-latex to ^1.5.0
Internal
- Eliminate innerHTML writes via centralized parseHTML/transferContent utilities
- Consolidate serialize HTML helpers
- Deduplicate math helpers