README.md

canon/tiers/02_Tier_2_Region_Pages/_ready_for_processing/README.md

Ready for Processing — book-processing inbox

Drop scanned books and downloaded digital sources here. The pipeline is LIVE (see

docs/BOOK_PROCESSING_WORKFLOW.md): harness/rag/drain_inbox.py ingests everything at this

folder's top level into the research corpus (text-layer gate, region auto-resolve, chunk,

embed, register). Processing is incremental — drop one book or fifty, whenever; there is no

"whole library first" gate.

The lifecycle (what being in this folder means)

• This folder's top level = pending. Every file here is awaiting ingest or a triage call.
• Ingested books LEAVE here → .rag/book_src/processed/<region>/SRC_NNNNN__<title>.pdf

(renamed with their T0_Source_Material_Registry id, organized by region).

• Scans with no text layer → quarantined to _needs_ocr/ until the OCR leg

(harness/book_processing/pdf_ocr.py) gives them one; OCR'd originals archive to

_needs_ocr/done_originals/.

• Gazetteer-unresolved files STAY here (the noise gate: never bulk-vectorized).

The triage lane either assigns a region (a .meta.json sidecar) or rejects them

to ../_processed/rejected_noise/.

• Completed staging trees (NatGeo discs, the sacred-texts mirror) move whole to

../_processed/ once their products are converted/ingested.

Scan spec (physical books)

300 DPI minimum (400 for small print or dense footnotes), PDF output, one file per book

or per volume. Color for plates and maps; grayscale is fine for plain text.

No text layer needed — the OCR leg handles that locally.

Name per the suggestion below so region auto-resolve has something to bite on.

Important — the scans are NOT committed to git

The source files themselves (PDF, images, EPUB) are gitignored. Large binary scans would bloat the repository, so only this README and the .gitignore are tracked. The scans stay local on the machine that processes them. Keep a separate backup of the raw scans; git is not their home.

Naming suggestion

<Author>_<ShortTitle>_<Chapter-or-region>.<ext> — for example Fernandez_Bwiti_Ch12.pdf, Schipper_TaoistBody_Ch21.pdf, Brown_MamaLola_Ch42.pdf.

What stages here

The minimized §17.1 primary-practitioner set (docs/BOOK_ACQUISITION_MINIMIZED.md §8.5), once scanned or downloaded:

• Fernandez, Bwiti — free digital version located, no physical purchase.
• Schipper, The Taoist Body (Ch 21).
• Lehtola, The Sami People (Ch 29).
• Brown, Mama Lola (Ch 42).
• Abimbola, Ifa (Yoruba, Ch 12) — care-call, recommended.
• Watson, Irene — Aboriginal legal scholarship (Ch 57) — confirm the exact title.
• Manker, Sami drum corpus (Ch 29) — digitized museum plates preferred over the antiquarian physical set.

Owned-library scans and the sacred-texts USB primaries can also stage here whenever a specific primary is wanted in the pipeline.

Generated by harness/site/structure_site.py — the URL path is the repo path. review root