Metadata enrichment from copyright page + back-cover blurb (depends on #33)

[Arkoniak]

Metadata enrichment from the book itself — split out from #23 (EPUB assembly).

Depends on: #33 (page classifier — to identify copyright page and back-cover page)

What this covers

1. Copyright/title page parsing — extract title, author(s), publisher, year, ISBN, LCCN from the OCR'd copyright page text. Needs Page-level classifier: content / chapter_opener / special_page / picture #33 to know which page to parse. Likely needs LLM assistance for reliable structured extraction (regex is too brittle against OCR noise).

2. Back-cover blurb — extract dc:description from the back-cover page OCR text. Again needs Page-level classifier: content / chapter_opener / special_page / picture #33 to locate the page.

3. Cross-check against Open Library — compare book's own fields vs. Open Library data (fetched in EPUB assembly via Pandoc (HTML canonical) + metadata from ISBN #23 by ISBN); on conflict prefer the book's text and flag the mismatch.

4. Series metadata — belongs-to-collection EPUB3 field from Open Library enrichment.

Out of scope

• ISBN lookup from Open Library → handled in EPUB assembly via Pandoc (HTML canonical) + metadata from ISBN #23.
• Pandoc pipeline → handled in EPUB assembly via Pandoc (HTML canonical) + metadata from ISBN #23.

Acceptance criteria

• meta.yaml produced for mindblast contains accurate title, both authors (Diane Duane and Peter Morwood — Open Library misses Morwood), publisher, year, ISBN, language, subjects, description.
• Mismatches between book text and Open Library are logged.