File size: 3,348 Bytes
9f0f92b
 
 
 
 
 
 
 
 
 
5e2965a
9f0f92b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5e2965a
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
import fitz
import pytesseract
from PIL import Image
import io
import os
import shutil
import pandas as pd
from docling.document_converter import DocumentConverter, PdfFormatOption
from docling.datamodel.base_models import InputFormat
from docling.datamodel.pipeline_options import PdfPipelineOptions
import pandas as pd

class RedactoParser:
    def __init__(self):
        pipeline_options = PdfPipelineOptions()
        pipeline_options.do_ocr = False
        pipeline_options.do_table_structure = True
        pipeline_options.generate_page_images = False
        
        self.converter = DocumentConverter(
            format_options={InputFormat.PDF: PdfFormatOption(pipeline_options=pipeline_options)}
        )

    def is_scanned_pdf(self, pdf_path, sample_pages=3):
        doc = fitz.open(pdf_path)
        total_text_length = 0
        pages_to_check = min(sample_pages, len(doc))
        for i in range(pages_to_check):
            total_text_length += len(doc[i].get_text("text").strip())
        doc.close()
        return total_text_length < 100

    def make_searchable(self, input_pdf_path, output_pdf_path, dpi=300):
        print(f"🔍 OCRing Scanned PDF: {input_pdf_path}")
        src_doc = fitz.open(input_pdf_path)
        out_doc = fitz.open()

        for i, page in enumerate(src_doc):
            print(f"  -> Processing page {i+1} / {len(src_doc)}...")
            pix = page.get_pixmap(dpi=dpi)
            img = Image.open(io.BytesIO(pix.tobytes("png")))
            pdf_bytes = pytesseract.image_to_pdf_or_hocr(img, extension='pdf', lang='eng')
            page_pdf = fitz.open("pdf", pdf_bytes)
            out_doc.insert_pdf(page_pdf)
            page_pdf.close()

        out_doc.save(output_pdf_path, garbage=4, deflate=True)
        out_doc.close()
        src_doc.close()
        return output_pdf_path

    def extract_layout(self, input_path, working_path):
        if self.is_scanned_pdf(input_path):
            self.make_searchable(input_path, working_path)
        else:
            shutil.copy(input_path, working_path)

        print(f"📄 Extracting structure from: {working_path}")
        result = self.converter.convert(working_path)
        
        layout_data = []
        current_h1, current_h2 = "Document Start", ""

        for item, level in result.document.iterate_items():
            label = getattr(item, "label", None)
            if label in ['page_header', 'page_footer']: continue

            if label == 'section_header':
                if level == 1: current_h1, current_h2 = item.text, ""
                elif level == 2: current_h2 = item.text

            elif label in ['text', 'list_item', 'paragraph', 'table']:
                bbox = item.prov[0].bbox.as_tuple() if hasattr(item, 'prov') and item.prov else None
                page_no = item.prov[0].page_no if hasattr(item, 'prov') and item.prov else 1
                
                text_content = item.export_to_markdown() if label == 'table' else item.text
                
                layout_data.append({
                    "context_h1": current_h1,
                    "context_h2": current_h2,
                    "text": text_content,
                    "label": label,
                    "page": page_no,
                    "bbox": bbox
                })

        return pd.DataFrame(layout_data)