Match _index.json entries to .zh.md files by title fuzzy matching.

docling sometimes splits a single logical table into multiple consecutive Markdown table blocks separated by a blank line, each with differen

Merge new items into the framework's news JSON file.

Main migration entry point.

Main migration entry point.

Parse 2026/193 amendment to extract new standard entries.

Parse the Annex table from consolidated Implementing Decision HTML.

Parse EC MDCG guidance page HTML and extract structured entries. The EC page uses <tr class="ecl-table__row"> with cells: - data-ecl-table-

Parse EUR-Lex HTML and return (title, markdown_body). Handles two HTML formats: 1. Standard eli-container / document1 structure (most regulat

Process a single doc type (regulations or guidance).

Process a single entry: discover PDF, download, extract, save.

Process a single IMDRF index entry: download DOCX and convert to Markdown.

Translate a single file.

Process a single regulation entry.

Translate all fulltext files in a section (e.g. nmpa/guidance).

Process a single Markdown file.

Process update report and generate draft PRs. Returns number of PRs created.

Push branch to origin.

Convert inline footnote lines to a footnotes section at the end. Detects lines matching: ^<digits> <text> that appear after a blank line

Scan _unmatched/ files, try to match them to index entries missing fulltext. Moves matched files from _unmatched/ to fulltext/ with proper slug.

Remove TOC tables (and their preceding heading) from content.

Render a single news item as markdown.

Run version checks and return report.

Sanitize string for markdown table cell (no truncation).

Save a single WordPress doc as Markdown files. Returns (zh_file, fm).

Save a single post as a Markdown file.

Save content that couldn't be matched to any index entry.

Scan a directory for Markdown files and process images.

Select the appropriate content generator based on update category.

Determine if an image URL should be downloaded locally.

Check if a line should be kept as-is (not translated).

Step 4: Compare fetched data with current data layer.

Show translation statistics across all sections.

Copy data layer *.zh.md files into docs/zh/<section_key>/ for VitePress rendering. Strips the .zh suffix so the route becomes /zh/<section_ke

Copy data layer *.en.md files into docs/en/<section_key>/ for VitePress rendering. Strips the .en suffix so the route becomes /en/<section_ke

Generate slug from Chinese title.

Generate slug from title or entry ID.

Call the translation API to translate a batch of texts.

Update or create _index.json for a directory.

Update or create _index.json for insights directory.

Validate that extracted content is the actual guidance document. Returns (is_valid, reason).

Validate a _index.json file.

Validate a single Markdown file's front matter.

Validate all files under a path. Returns (checked, failed) counts.

Run all verification checks. Returns (errors, warnings).

Check which Markdown files are out of sync with JSON data. Returns list of out-of-sync files.