1. import subprocess
2. from pathlib import Path
3. ​
4. from pikepdf import (
5. Page,
6. Operator,
7. Pdf,
8. Stream,
9. parse_content_stream,
10. unparse_content_stream,
11. )
12. ​
13. # input file
14. infile: Path = Path.cwd().joinpath("rtl8196e-vex-cg-datasheet-1.1.pdf")
15. # object matcher
16. wmark_colours = (0.8, 0.8, 0.8)
17. ​
18. # intermediary qdf file
19. qdf_file: Path = infile.with_suffix(".qdf")
20. # output file
21. outfile: Path = infile.with_stem(f"{infile.stem}.clean")
22. ​
23. # function to strip watermark text from a page
24. def strip_watermark_text(pdf: Pdf, page: Page, watermark_colors: tuple):
25. stream = []
26. in_text_obj = False
27. color_values = (0.0, 0.0, 0.0)
28. text_objects = []
29. ​
30. for operands, operator in parse_content_stream(page, ""):
31. if not in_text_obj:
32. if operator == Operator("BT"):
33. in_text_obj = True
34. color_values = (0.0, 0.0, 0.0)
35. text_objects.append((operands, operator))
36. else:
37. stream.append((operands, operator))
38. else:
39. if operator == Operator("rg"):
40. color_values = (float(operands[0]), float(operands[1]), float(operands[2]))
41. text_objects.append((operands, operator))
42. if operator == Operator("ET"):
43. in_text_obj = False
44. if color_values != watermark_colors:
45. stream.extend(text_objects)
46. text_objects.clear()
47. ​
48. content_stream = unparse_content_stream(stream)
49. page.Contents = Stream(pdf, content_stream)
50. ​
51. ​
52. if __name__ == "__main__":
53. # Convert to QDF
54. # qpdf --qdf --stream-data=uncompress infile outfile
55. print("Converting PDF to QDF...")
56. result = subprocess.call(["qpdf", "--qdf", "--stream-data=uncompress", infile, qdf_file])
57. if result != 0:
58. raise RuntimeError("qpdf failed parsing the PDF file to QDF")
59. ​
60. # open the QDF as a pikepdf object
61. print("Opening QDF file as pikepdf object...")
62. document: Pdf = Pdf.open(qdf_file)
63. ​
64. print("Processing pages...")
65. for page in document.pages:
66. print(f"{page.index}", end=" ")
67. strip_watermark_text(document, page, wmark_colours)
68. ​
69. print(f"Saving to {outfile} ...")
70. document.save(outfile)
71. ​
72. print(f"Removing {qdf_file} ...")
73. qdf_file.unlink()
74. ​
75. print("Done! Hopefully...")