Add/update documentation across backend, Python OCR service, and frontend
for receipt scanning, manual extraction, and Gemini integration. Create
new CLAUDE.md files for engines/, fuel-logs/, documents/, and maintenance/
features.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add POST /api/ocr/extract/manual endpoint that proxies to the Python
OCR service's manual extraction pipeline. Includes Pro tier gating via
document.scanMaintenanceSchedule, PDF-only validation, 200MB file size
limit, and async 202 job response for polling via existing job status
endpoint.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace traditional OCR pipeline (table_detector, table_parser,
maintenance_patterns) with GeminiEngine for semantic PDF extraction.
Map Gemini serviceName values to 27 maintenance subtypes via
ServiceMapper fuzzy matching. Add 8 unit tests covering normal
extraction, unusual names, empty response, and error handling.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add standalone GeminiEngine class for maintenance schedule extraction
from PDF owners manuals using Vertex AI Gemini 2.5 Flash with structured
JSON output enforcement, 20MB size limit, and lazy initialization.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add Google Places Text Search to match receipt merchant names (e.g.
"Shell", "COSTCO #123") to real gas stations. Backend exposes
POST /api/stations/match endpoint. Frontend calls it after OCR
extraction and pre-fills locationData with matched station's placeId,
name, and address. Users can clear the match in the review modal.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Free tier users see locked button with upgrade prompt dialog.
Pro+ users can capture receipts normally. Works on mobile and desktop.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add POST /api/ocr/extract/receipt endpoint that proxies to the Python
OCR service's /extract/receipt for receipt-specific field extraction.
- ReceiptExtractionResponse type with receiptType, extractedFields, rawText
- OcrClient.extractReceipt() with optional receipt_type form field
- OcrService.extractReceipt() with 10MB max, image-only validation
- OcrController.extractReceipt() with file upload and error mapping
- Route with auth middleware
- 9 unit tests covering normal, edge, and error scenarios
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The set -e + curl --fail-with-body inside $() caused the script to exit
with code 22 and empty stderr, hiding the actual Auth0 error. Switch to
writing the body to a temp file and checking HTTP status manually so the
error response is visible in logs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The google-wif-config.json was never synced to the deploy path, so the
Docker bind mount created a directory artifact instead of a file. Vision
client initialization failed on every request, silently falling back to
PaddleOCR.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Switch OCR engine config to google_vision primary / paddleocr fallback
- Mount Auth0 OCR secrets and WIF config into all OCR containers
- Add WIF config to repo (not a secret, contains no credentials)
- Remove obsolete google-vision-key.json.example
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add AUTH0_OCR_CLIENT_ID and AUTH0_OCR_CLIENT_SECRET to inject-secrets.sh
- Add new secrets to staging and production workflow env blocks
- Create .example files for new secret documentation
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Create fetch-auth0-token.sh for Auth0 M2M -> GCP WIF token exchange
- Add jq to Dockerfile system dependencies
- Ensure script is executable in container image
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add VISION_MONTHLY_LIMIT config setting (default 1000)
- Update CloudEngine to use WIF credential config via ADC
- Rewrite HybridEngine to support cloud-primary with Redis counter
- Pass monthly_limit through engine factory
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Redis only supports debug|verbose|notice|warning -- not info or error.
The command was using ${LOG_LEVEL:-info} which resolved to INFO in
production (from workflow env), causing Redis to crash loop. Hardcode
the correct Redis-native levels (debug for dev, warning for prod) and
add available log level comments above every container's log setting.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
VIN OCR confidence now reflects recognition accuracy only (not match quality).
Review modal replaces read-only fields with editable cascade dropdowns
pre-populated from NHTSA decode, with NHTSA reference hints for unmatched fields.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The retake button failed because the stream tracks could become inactive
during the crop phase, but handleRetake never re-acquired the camera.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Bridge guidance overlay position to crop tool initial coordinates so the
crop box appears centered matching the viewfinder guide. Increase handle
touch targets to 44px (32px on compact viewports) for mobile usability.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add engine abstraction tests and update docs to reflect PaddleOCR primary
architecture with optional Google Vision cloud fallback.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Three bugs fixed in the draw-first crop tool introduced by PR #114:
1. Stale cropAreaRef: replaced useEffect-based ref sync with direct
synchronous updates in handleMove and handleDrawStart. The useEffect
ran after browser paint, so handleDragEnd read stale values (often
{width:0, height:0}), preventing cropDrawn from being set.
2. Aspect ratio minSize: when aspectRatio=6 (VIN mode), height=width/6
required width>=60% to pass the height>=10% check. Now only checks
width>=minSize when aspect ratio constrains height.
3. Bounds clamping: aspect-ratio-forced height could push crop area
past 100% of container. Now clamps y position to keep within bounds.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Replace libtesseract-dev with libgomp1 (OpenMP for PaddlePaddle)
- Pre-download PP-OCRv4 models during Docker build
- Add OCR engine env vars to all compose files (base, staging, prod)
- Add optional Google Vision secret mount (commented, enable on demand)
- Create google-vision-key.json.example placeholder
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
CloudEngine wraps Google Vision TEXT_DETECTION with lazy init.
HybridEngine runs primary engine, falls back to cloud when confidence
is below threshold. Disabled by default (OCR_FALLBACK_ENGINE=none).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace direct pytesseract calls with OcrEngine interface in vin_extractor.py,
receipt_extractor.py, and ocr_service.py. PSM mode fallbacks replaced with
engine-agnostic single-line/single-word configs. Dead _process_ocr_data removed.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Introduce pluggable OcrEngine ABC with PaddleOCR PP-OCRv4 as primary
engine and Tesseract wrapper for backward compatibility. Engine factory
reads OCR_PRIMARY_ENGINE config to instantiate the correct engine.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When OCR reads extra characters (e.g. sticker border as 'C', spurious
'Z' insertion), the raw text exceeds 17 chars and the old first-17
trim produced wrong VINs. New strategy tries all 17-char sliding
windows and single/double character deletions, validating each via
check digit. For 'CWVGGNPE2Z4NP069500', this finds the correct VIN
'WVGGNPE24NP069500' (valid check digit) instead of 'CWVGGNPE2Z4NP0695'
(invalid).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
tessedit_char_whitelist does not work with OEM 1 (LSTM engine) and
causes empty/erratic output. This was the root cause of Tesseract
returning empty text despite clear, well-preprocessed images.
Character filtering is already handled post-OCR by the VIN validator's
correct_ocr_errors() method (I->1, O->0, Q->0, etc).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The min-channel correctly extracts contrast (white text=255 vs green
sticker bg=130), but Tesseract expects dark text on light background.
Without inversion, the grayscale-only path returned empty text for
every PSM mode because Tesseract couldn't see bright-on-dark text.
Invert via bitwise_not: text becomes 0 (black), sticker bg becomes
125 (gray). Fixes all three OCR paths (adaptive, grayscale, Otsu).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
