Skip to content

Add OCR review pipeline and dual-edition builds - #1

Draft
cwyalpha wants to merge 2 commits into
mainfrom
codex/ocr-v0.2.0
Draft

Add OCR review pipeline and dual-edition builds#1
cwyalpha wants to merge 2 commits into
mainfrom
codex/ocr-v0.2.0

Conversation

@cwyalpha

Copy link
Copy Markdown
Owner

What changed

  • add fixed-hash four-way page orientation and expose OCR pipeline controls
  • add PDF/image review UI with OCR boxes, confidence highlighting, paging, editing, and rerun
  • add immutable OCR/full editions and a four-platform, eight-asset build matrix
  • add generated fixture gates, searchable-PDF regression tests, and recursive sensitive-data release auditing

Why

This improves rotated and scanned PDF accuracy while supporting lightweight OCR-only packages alongside full Office-conversion builds.

Validation

  • 30 pytest tests passed
  • real PP-OCRv6 tests passed for 0/90/180/270 degree inputs and an uncopyable scanned PDF converted to searchable PDF
  • macOS arm64 OCR/full frozen-package smoke tests passed
  • Kylin V10 ARM64 OCR/full clean-container CLI tests passed, including bundled LibreOffice 6.0.6.1
  • source and all four locally buildable assets passed recursive sensitive-data audit

Remaining release gates

  • build Windows x86_64 OCR/full packages
  • build Kylin V10 x86_64 OCR/full packages
  • grant the new macOS bundles Screen Recording permission and run the final real selection/OCR regression
  • generate the unified eight-asset SHA256SUMS.txt and publish the Release

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.

1 participant