app.py

import gradio as gr
import os
from docx import Document
from utils import extract_text_from_pdf, parse_pdf_to_dict

from docx.shared import RGBColor

from docx import Document
from docx.shared import RGBColor

def replace_red_text_with_data(word_path, data_dict):
    doc = Document(word_path)

    for para in doc.paragraphs:
        full_text = para.text
        for i, run in enumerate(para.runs):
            if run.font.color and run.font.color.rgb == RGBColor(255, 0, 0):
                # Search for the key (label) before this red text
                preceding_text = ''.join(r.text for r in para.runs[:i]).lower()
                for key in data_dict:
                    if key.lower() in preceding_text:
                        run.text = data_dict[key]
                        break

    return doc

def process_files(pdf_file, template_docx):
    pdf_path = pdf_file
    doc_path = template_docx
    output_path = "filled_output.docx"

    # Extract and parse PDF
    raw_text = extract_text_from_pdf(pdf_path)
    data_dict = parse_pdf_to_dict(raw_text)

    # Replace red text with data
    final_doc = replace_red_text_with_data(doc_path, data_dict)

    # Save final document
    final_doc.save(output_path)
    return output_path

demo = gr.Interface(
    fn=process_files,
    inputs=[
        gr.File(label="Upload PDF Report", file_types=[".pdf"]),
        gr.File(label="Upload Word Template (.docx)", file_types=[".docx"])
    ],
    outputs=gr.File(label="Download Filled Report (.docx)"),
    title="Audit Report Generator",
    description="Upload a PDF and a Word template. This tool will auto-fill red-highlighted fields with data from the PDF."
)

if __name__ == "__main__":
    demo.launch()