BPL Manuscript Catalog — raw Internet Archive mirror

Raw mirror of the Boston Public Library Manuscript Catalog card collection from the Internet Archive. Mirrored for the FineBooks / FineCollections pipeline (archival index cards as a source of structured catalog data + OCR-difficulty material).

• Source collection: https://archive.org/details/bplmanuscriptcatalog
• Items: 123 (each IA item ≈ one drawer/box of cards; ~246 MB of images on average)
• Approx. size: ~30.2 GB
• License: Public Domain Mark 1.0 (per Internet Archive item metadata)
• Uploaded to IA by: Boston Public Library
• OCR: Tesseract (hOCR), as produced by the Internet Archive

What's in here

Each IA item is mirrored under raw/{ia_item_id}/:

• {item}_jp2.zip — page images (JP2), zipped
• {item}_hocr.html — Tesseract hOCR (word-level OCR text + bounding boxes)

This is a raw mirror — the zipped JP2 + hOCR exactly as held on the Internet Archive. It is not yet unpacked into per-card images or a parquet dataset; those are downstream steps.

Provenance

Mirrored from the Internet Archive (2026-06-27) with download_to_bucket.py, which streams one item at a time: download from IA → sync_bucket to this bucket → delete the local copy. Re-running resumes from whatever is already present, so the mirror is idempotent.

Related BPL index-card buckets

Part of the Boston Public Library × Hugging Face index-card mirroring effort. See also the shelf list cards mirror.