feat: Image → Text (OCR) + Receipt Scanner (on-device) - #44
Merged
Conversation
fix(updater): generate latest.json/.sig + remove internal dev docs
fix(theme) + docs: dark-mode persistence, merged DEPLOYMENT, GA env clarity
feat(shell): Settings nav link desktop-only
chore: trigger production rebuild (SITE_GA_ID)
feat: hide desktop-only tools (Hotkey Test) from web
chore(release): desktop v1.0.0-beta.2
docs(release): fix macOS Gatekeeper guidance
Promote to production: FFmpeg bundling + lint cleanup
PP-OCR pipeline on ONNX Runtime Web (WebGPU/WASM), single image + PDF input, review-&-adjust preprocessing, reason-specific errors. Structured receipt parsing, multi-language, and batch deferred to later phases.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…rors ppu-paddle-ocr@6.2.0 (PP-OCRv5 on ONNX Runtime Web, WebGPU/WASM). SDK boundary isolated to ocr.engine.ts; ocr.lib holds tested logic (caching, row-grouped text assembly, OcrError reasons). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Single image + PDF input, review-&-adjust (rotate/cleanup/threshold), on-device inference, reason-specific errors with retry. Crop UI + box overlay deferred. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Heuristic parser over OCR text+box layout; pluggable ReceiptKeywords (English now, language-extensible); 'Parse as receipt' toggle + dedicated Receipt Scanner tool; editable fields, JSON/CSV export. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
parseReceipt over OCR text+box layout (merchant/date/currency/subtotal/tax/total); pluggable ReceiptKeywords (English now, language-extensible); receiptToJson/Csv. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Extract shared OcrWorkbench from ImageOcr (behavior-preserving); editable ReceiptFields with JSON/CSV/copy export; new image-receipt-scanner tool. Parsed data memoized per OCR result so edits survive re-renders. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Deploying with
|
| Status | Name | Latest Commit | Updated (UTC) |
|---|---|---|---|
| ✅ Deployment successful! View logs |
goodwebtools | 3f206d2 | Jul 30 2026, 04:52 PM |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds two on-device tools (nothing leaves the browser except the OCR model, fetched from CDN + cached).
Image → Text (OCR)
PP-OCR (PaddleOCR) via
ppu-paddle-ocron ONNX Runtime Web (WebGPU→WASM). Single image or PDF input, review-&-adjust (rotate / cleanup / threshold), reason-specific errors with retry, editable text + copy + download .txt.Receipt Scanner + "Parse as receipt"
Heuristic parser over the OCR text + box layout extracts merchant, date, currency, subtotal, tax, total — editable, with JSON/CSV export. Surfaced both as a dedicated Receipt Scanner tool and a Parse as receipt toggle in the OCR tool. Keyword anchors are English but pluggable (
ReceiptKeywords) for future languages.Architecture
Pure, unit-tested libs (
ocr.lib,ocr-preprocess.lib,ocr-pdf.lib,receipt.lib,receipt-export.lib); a sharedOcrWorkbenchcomponent drives both tools; islands stay thin.Verification
469 tests pass ·
npm run lint0 errors ·astro buildgreen (both/tools/image-ocrand/tools/image-receipt-scannerbuild). ORT wasm is runtime-fetched (pruned from the bundle), so no PWA precache bloat.Specs/plans under
docs/superpowers/.