Extract multiple candidate title strings from content, skipping boilerplate. Used for deeper rematch attempts.

Extract the actual guidance title from an _unmatched/ file, skipping boilerplate headers and metadata comments.

Join multi-line alt text in Markdown image syntax into a single line.

Custom mammoth image converter with compression.

Fetch EC page, parse all MDCG refs, compare with DB.

Check ASTM page for withdrawal/historical status.

Check IEC webstore page for newer version signal. IEC webstore does NOT mark old publication pages as superseded. Detection strategy:

Check ISO page for revision signals. ISO marks old standard pages with 'Withdrawn' status and 'Revised by' links. The lifecycle section i

Check iTeh/CEN page for withdrawn status.

Check if URL returns 200 and is not a generic redirect.

Format a single document entry as a Markdown list item with link.

Generate EN and ZH draft Markdown for a new MDCG guidance document.

Generate EN and ZH draft Markdown for FDA guidance updates.

Generate a generic EN and ZH draft for any update type.

Generate EN and ZH draft Markdown for NMPA/CMDE updates.

Determine which schema to use based on file path.

Return (front_matter_block, body) where front_matter_block includes --- delimiters.