File size: 5,472 Bytes
71de920
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c079558
 
 
 
 
71de920
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
96f3be4
71de920
 
 
 
 
 
96f3be4
71de920
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0fe1f96
71de920
 
 
 
 
 
 
 
 
1b8b833
 
 
 
 
71de920
5c704a8
 
71de920
 
 
 
 
 
 
5c704a8
71de920
5c704a8
71de920
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
import gradio as gr
from docling.document_converter import DocumentConverter, PdfFormatOption
from docling.datamodel.pipeline_options import PdfPipelineOptions
from docling_core.types.doc import ImageRefMode
from docx import Document as WordDocument
from docx.shared import Inches
import re
import io
import base64
import tempfile

def write_markdown_to_word(md_text: str, word_doc: WordDocument, include_images=False):
    image_pattern = re.compile(r'!\[.*?\]\((data:image\/png;base64,[^\)]+)\)')
    lines = md_text.splitlines()
    i = 0
    while i < len(lines):
        line = lines[i].rstrip()
         # -- NEW: REMOVE image placeholder tags if not including images <!-- image --> or <!-- image ... --> or embed image
        if not include_images and (line.strip() == "<!-- image -->" or line.strip().startswith("<!-- image")):
            i += 1
            continue
        # Images (base64-encoded)
        match = image_pattern.search(line)
        if match:
            if include_images:
                image_data = match.group(1)
                header, b64data = image_data.split(',', 1)
                img_stream = io.BytesIO(base64.b64decode(b64data))
                word_doc.add_picture(img_stream, width=Inches(4))
            i += 1
            continue
        if not line:
            word_doc.add_paragraph("")
            i += 1
            continue
        if line.startswith("#"):
            level = min(line.count("#"), 4)
            heading_text = line.lstrip("#").strip()
            word_doc.add_heading(heading_text, level=level)
            i += 1
            continue
        if line.startswith("```"):
            code_lines = []
            i += 1
            while i < len(lines) and not lines[i].startswith("```"):
                code_lines.append(lines[i])
                i += 1
            word_doc.add_paragraph("\n".join(code_lines))
            if i < len(lines) and lines[i].startswith("```"):
                i += 1
            continue
        if line.lstrip().startswith(("- ", "* ")):
            while i < len(lines) and lines[i].lstrip().startswith(("- ", "* ")):
                item = lines[i].lstrip()[2:].strip()
                word_doc.add_paragraph(f" - {item}")
                i += 1
            continue
        if "|" in line:
            if i + 1 < len(lines) and set(lines[i + 1].replace("|", "").replace(" ", "")) <= set("-:"):
                header_cells = [c.strip() for c in line.strip("|").split("|")]
                i += 2
                data_rows = []
                while i < len(lines) and "|" in lines[i]:
                    row_cells = [c.strip() for c in lines[i].strip("|").split("|")]
                    data_rows.append(row_cells)
                    i += 1
                n_cols = len(header_cells)
                n_rows = len(data_rows)
                tbl = word_doc.add_table(rows=n_rows + 1, cols=n_cols)
                tbl.style = "Table Grid"
                for c_idx, col_name in enumerate(header_cells):
                    tbl.cell(0, c_idx).text = str(col_name)
                for r_idx, row in enumerate(data_rows):
                    for c_idx in range(n_cols):
                        val = row[c_idx] if c_idx < len(row) else ""
                        tbl.cell(r_idx + 1, c_idx).text = str(val)
                word_doc.add_paragraph("")
                continue
        word_doc.add_paragraph(line)
        i += 1

def pdf_to_word(pdf_file, include_images_radio):
    try:
        include_images = (include_images_radio == "Yes")
        if include_images:
            pipeline_options = PdfPipelineOptions()
            pipeline_options.images_scale = 2.0
            pipeline_options.generate_page_images = True
            pipeline_options.generate_picture_images = True
            converter = DocumentConverter(
                format_options={
                    "pdf": PdfFormatOption(pipeline_options=pipeline_options)
                }
            )
        else:
            converter = DocumentConverter()
        with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp_pdf:
            tmp_pdf.write(pdf_file)
            pdf_path = tmp_pdf.name
        result = converter.convert(pdf_path)
        doc = result.document
        if include_images:
            md = doc.export_to_markdown(image_mode=ImageRefMode.EMBEDDED)
        else:
            md = doc.export_to_markdown()
        word_doc = WordDocument()
        write_markdown_to_word(md, word_doc, include_images=include_images)
        # Write Word doc to temp file and return file path
        with tempfile.NamedTemporaryFile(delete=False, suffix=".docx") as tmp_docx:
            word_doc.save(tmp_docx)
            docx_path = tmp_docx.name
        return docx_path, ""
    except Exception as e:
        # Return 'None' for the file, and the error message for the error component
        return (None, f"Error: {str(e)}")

with gr.Blocks() as demo:
    gr.Markdown("## PDF to Word (Text, Tables, and Optional Images) via Docling")
    pdf_input = gr.File(label="Upload PDF", type="binary")
    include_images = gr.Radio(["Yes", "No"], label="Include Images", value="No")
    btn = gr.Button("Convert and Download Word (.docx)")
    output = gr.File(label="Download Word File (.docx)")
    error_box = gr.Textbox(label="Error (if any)", interactive=False)

    btn.click(fn=pdf_to_word, inputs=[pdf_input, include_images], outputs=[output, error_box])

demo.launch()