inspect_pdf_pages.py 869 B

123456789101112131415161718192021222324
  1. import fitz
  2. doc = fitz.open(r"c:/Users/lanfr144/Documents/DOPRO1/Antigravity/Food/docs/Installation_Guide.pdf")
  3. print(f"Total pages: {len(doc)}")
  4. page = doc[1] # Page 2 (0-indexed)
  5. print("--- PAGE 2 DRAWINGS ---")
  6. drawings = page.get_drawings()
  7. print(f"Number of drawings: {len(drawings)}")
  8. for idx, draw in enumerate(drawings[:10]):
  9. print(f"Drawing {idx}: rect={draw['rect']}, type={draw['type']}, fill={draw.get('fill')}, color={draw.get('color')}")
  10. import sys
  11. if hasattr(sys.stdout, 'reconfigure'):
  12. sys.stdout.reconfigure(encoding='utf-8')
  13. print("\n--- PAGE 2 TEXT BLOCKS ---")
  14. blocks = page.get_text("blocks")
  15. print(f"Number of text blocks: {len(blocks)}")
  16. for idx, block in enumerate(blocks):
  17. text_clean = block[4].encode('utf-8', errors='replace').decode('utf-8')
  18. print(f"Block {idx}: rect={block[:4]}, text={repr(text_clean)}")
  19. doc.close()