egg/OCR
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: correct orientation detection to use OCR pixel coordinates

Browse Source 
Fixed two issues in PDF orientation detection:

1. Unit mismatch: Orientation detection was comparing content bboxes
   (in pixels) against PDF page dimensions (in points). Now correctly
   uses OCR dimensions (pixels) for detection.

2. Priority override: When orientation adjustment is needed, now also
   updates original_page_sizes dict so per-page processing uses the
   adjusted dimensions instead of the original PDF dimensions.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
egg
2025-11-30 13:37:03 +08:00
parent c65e4f98d4
commit f27b4d9710
1 changed files with 25 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
backend/app/services/pdf_generator_service.py
View File

@@ -1064,17 +1064,39 @@ class PDFGeneratorService:
# Step 4: Detect content orientation mismatch # Step 4: Detect content orientation mismatch
# This handles rotated scans where content bbox exceeds page dimensions # This handles rotated scans where content bbox exceeds page dimensions
needs_rotation, adjusted_width, adjusted_height = self._detect_content_orientation( # IMPORTANT: Use OCR dimensions (pixels) for detection, not PDF points
target_width, target_height, ocr_data # because content bboxes are in the same coordinate system as OCR dimensions
needs_rotation, adjusted_ocr_width, adjusted_ocr_height = self._detect_content_orientation(
ocr_width, ocr_height, ocr_data
) )
# If orientation change detected, calculate the adjusted target dimensions
if needs_rotation:
# Swap target dimensions to match the detected orientation
adjusted_width = target_height
adjusted_height = target_width
elif adjusted_ocr_width != ocr_width or adjusted_ocr_height != ocr_height:
# Content extends beyond OCR dimensions, scale target proportionally
scale_w = adjusted_ocr_width / ocr_width if ocr_width > 0 else 1.0
scale_h = adjusted_ocr_height / ocr_height if ocr_height > 0 else 1.0
adjusted_width = target_width * scale_w
adjusted_height = target_height * scale_h
else:
adjusted_width = target_width
adjusted_height = target_height
if needs_rotation or (adjusted_width != target_width or adjusted_height != target_height): if needs_rotation or (adjusted_width != target_width or adjusted_height != target_height):
logger.info(f"頁面尺寸調整: {target_width:.1f}x{target_height:.1f} -> " logger.info(f"頁面尺寸調整: {target_width:.1f}x{target_height:.1f} -> "
f"{adjusted_width:.1f}x{adjusted_height:.1f} (旋轉={needs_rotation})") f"{adjusted_width:.1f}x{adjusted_height:.1f} (旋轉={needs_rotation})")
target_width, target_height = adjusted_width, adjusted_height target_width, target_height = adjusted_width, adjusted_height
# Also update page_dimensions for consistency in per-page processing # Update BOTH page_dimensions AND original_page_sizes for consistency
# This ensures per-page processing uses the adjusted dimensions
if 0 in page_dimensions: if 0 in page_dimensions:
page_dimensions[0] = {'width': target_width, 'height': target_height} page_dimensions[0] = {'width': target_width, 'height': target_height}
if 0 in original_page_sizes:
# Override original file dimensions with content-based dimensions
original_page_sizes[0] = (target_width, target_height)
logger.info(f"覆蓋原始文件尺寸以適應內容方向")
# Create PDF canvas with initial page size (will be updated per page) # Create PDF canvas with initial page size (will be updated per page)
pdf_canvas = canvas.Canvas(str(output_path), pagesize=(target_width, target_height)) pdf_canvas = canvas.Canvas(str(output_path), pagesize=(target_width, target_height))