Skip to content

pdf: bump to 0.5.0#2218

Merged
carlopi merged 1 commit into
duckdb:mainfrom
asubbarao:pdf-0.5.0
Jul 10, 2026
Merged

pdf: bump to 0.5.0#2218
carlopi merged 1 commit into
duckdb:mainfrom
asubbarao:pdf-0.5.0

Conversation

@asubbarao

Copy link
Copy Markdown
Contributor

Bumps the pdf extension 0.3.0 → 0.5.0, pinned to 9006e93 (all-platform CI green on the extension repo, including windows_amd64/MSVC).

New since 0.3.0 — the extension grows from a reader into a full PDF toolkit:

  • Structure & RAG: read_pdf_elements (layout elements with bounding boxes, deterministic geometry), pdf_chunks (heading-aware retrieval-ready chunking).
  • Inspection: pdf_info, pdf_outline, pdf_attachments, pdf_form_fields, pdf_annotations, pdf_revisions (incremental-update forensics), pdf_signatures (detection + cryptographic integrity verification over signed byte ranges), pdf_images (embedded raster extraction).
  • Transform (qpdf): pdf_merge / pdf_split / pdf_split_blank / pdf_rotate / pdf_pages / pdf_compress / pdf_encrypt / pdf_decrypt (AES-256 R6) / pdf_watermark / pdf_bates.
  • Robustness: ignore_errors on read_pdf / read_pdf_meta for fault-isolated folder scans.

qpdf (12.3.2, vcpkg, default-features=false) is compiled in an isolated C++17 translation unit so the extension still builds against the registry's default C++11 driver. Descriptor docs updated to cover the new surface.

🤖 Generated with Claude Code

@asubbarao asubbarao mentioned this pull request Jul 10, 2026
@carlopi

carlopi commented Jul 10, 2026

Copy link
Copy Markdown
Member

Thanks!

@carlopi carlopi merged commit 820b490 into duckdb:main Jul 10, 2026
13 checks passed
carlopi pushed a commit that referenced this pull request Jul 11, 2026
0.5.0 (#2218) was pinned to ref 9006e93, which is the batch-1 baseline: it
predates every feature the 0.5.0 description advertises. At that commit
pdf_images, pdf_split_blank, pdf_watermark, pdf_bates, pdf_redact, and pdf_sign
do not exist, so installing 0.5.0 and calling any of them fails with
'function does not exist'.

This bumps the ref to d33b1f3 (asubbarao/duckdb-pdf main), where all of those
functions are present and tested, plus two robustness fixes landed after the
feature merge:
- auto-OCR is now best-effort: a blank/scanned page in a plain read no longer
  hard-errors on machines without a Tesseract model (explicit ocr:=true still
  raises the actionable install error).
- pdf_revisions no longer fatally invalidates the DuckDB connection on a
  header-less input (an unescaped %PDF- in the error format string escalated a
  clean IOException to a fatal InternalException).

The description already documents these functions; only the pinned ref was
wrong. No description changes.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants