1import json
2import sys
3 
4from pypdf import PdfReader, PdfWriter
5from pypdf.annotations import FreeText
6 
7 
8 
9 
10def transform_from_image_coords(bbox, image_width, image_height, pdf_width, pdf_height):
11    x_scale = pdf_width / image_width
12    y_scale = pdf_height / image_height
13 
14    left = bbox[0] * x_scale
15    right = bbox[2] * x_scale
16 
17    top = pdf_height - (bbox[1] * y_scale)
18    bottom = pdf_height - (bbox[3] * y_scale)
19 
20    return left, bottom, right, top
21 
22 
23def transform_from_pdf_coords(bbox, pdf_height):
24    left = bbox[0]
25    right = bbox[2]
26 
27    pypdf_top = pdf_height - bbox[1]      
28    pypdf_bottom = pdf_height - bbox[3]   
29 
30    return left, pypdf_bottom, right, pypdf_top
31 
32 
33def fill_pdf_form(input_pdf_path, fields_json_path, output_pdf_path):
34    
35    with open(fields_json_path, "r") as f:
36        fields_data = json.load(f)
37    
38    reader = PdfReader(input_pdf_path)
39    writer = PdfWriter()
40    
41    writer.append(reader)
42    
43    pdf_dimensions = {}
44    for i, page in enumerate(reader.pages):
45        mediabox = page.mediabox
46        pdf_dimensions[i + 1] = [mediabox.width, mediabox.height]
47    
48    annotations = []
49    for field in fields_data["form_fields"]:
50        page_num = field["page_number"]
51 
52        page_info = next(p for p in fields_data["pages"] if p["page_number"] == page_num)
53        pdf_width, pdf_height = pdf_dimensions[page_num]
54 
55        if "pdf_width" in page_info:
56            transformed_entry_box = transform_from_pdf_coords(
57                field["entry_bounding_box"],
58                float(pdf_height)
59            )
60        else:
61            image_width = page_info["image_width"]
62            image_height = page_info["image_height"]
63            transformed_entry_box = transform_from_image_coords(
64                field["entry_bounding_box"],
65                image_width, image_height,
66                float(pdf_width), float(pdf_height)
67            )
68        
69        if "entry_text" not in field or "text" not in field["entry_text"]:
70            continue
71        entry_text = field["entry_text"]
72        text = entry_text["text"]
73        if not text:
74            continue
75        
76        font_name = entry_text.get("font", "Arial")
77        font_size = str(entry_text.get("font_size", 14)) + "pt"
78        font_color = entry_text.get("font_color", "000000")
79 
80        annotation = FreeText(
81            text=text,
82            rect=transformed_entry_box,
83            font=font_name,
84            font_size=font_size,
85            font_color=font_color,
86            border_color=None,
87            background_color=None,
88        )
89        annotations.append(annotation)
90        writer.add_annotation(page_number=page_num - 1, annotation=annotation)
91        
92    with open(output_pdf_path, "wb") as output:
93        writer.write(output)
94    
95    print(f"Successfully filled PDF form and saved to {output_pdf_path}")
96    print(f"Added {len(annotations)} text annotations")
97 
98 
99if __name__ == "__main__":
100    if len(sys.argv) != 4:
101        print("Usage: fill_pdf_form_with_annotations.py [input pdf] [fields.json] [output pdf]")
102        sys.exit(1)
103    input_pdf = sys.argv[1]
104    fields_json = sys.argv[2]
105    output_pdf = sys.argv[3]
106    
107    fill_pdf_form(input_pdf, fields_json, output_pdf)
108