OCR/openspec/changes/archive/2025-12-04-improve-translated-text-fitting/tasks.md

1. Backend Implementation

• 1.1 Create generate_reflow_pdf() method in pdf_generator_service.py
• 1.2 Implement _get_elements_in_reading_order() for both tracks
• 1.3 Implement reflow text rendering with consistent font sizes
• 1.4 Implement table rendering in reflow mode (Platypus Table)
• 1.5 Implement inline image embedding
• 1.6 Add format=reflow query parameter to tasks download endpoint
• 1.7 Update generate_translated_pdf() to use reflow mode

2. Frontend Implementation

• 2.1 Add "Download Reflow PDF" button for original documents
• 2.2 Update download logic to support format parameter
• 2.3 Remove MADLAD-400 badge (line 545)
• 2.4 Update translation description text to reflect Dify cloud service (line 652)

3. Testing

• 3.1 Test OCR track reflow PDF (with reading_order) - Basic smoke test passed
• 3.2 Test Direct track reflow PDF (implicit order) - No test data available
• 3.3 Test translated PDF (reflow mode) - Basic smoke test passed
• 3.4 Test documents with tables - SUCCESS (62294 bytes, 2 tables)
• 3.5 Test documents with images - SUCCESS (embedded img_in_table)
• 3.6 Test multi-page documents - SUCCESS (11451 bytes, 3 pages)
• 3.7 Verify layout PDF still works correctly - SUCCESS (104543 bytes)

4. Documentation

• 4.1 Update spec with reflow layout requirements