app.py

import os
import fitz  # PyMuPDF for PDF handling
import easyocr  # OCR for text extraction
import tempfile
import streamlit as st


def extract_text_with_ocr(pdf_path):
    """
    Extract text with bounding box positions using OCR for both English and Arabic text.
    :param pdf_path: Path to the input PDF file.
    :return: List of dictionaries containing text and positions for each page.
    """
    extracted_data = []
    doc = fitz.open(pdf_path)

    # Convert each PDF page to an image for OCR processing
    for page_num in range(len(doc)):
        page = doc.load_page(page_num)
        pix = page.get_pixmap(dpi=300)  # Convert PDF page to image
        image_path = f"temp_page_{page_num}.png"
        pix.save(image_path)

        # Perform OCR on the image
        reader = easyocr.Reader(['en'])  # Supports English (add 'ar' for Arabic if needed)
        results = reader.readtext(image_path, detail=1)  # detail=1 returns bounding box info

        # Extract text and positions
        page_data = []
        for (bbox, text, confidence) in results:
            (x0, y0), (x1, y1) = bbox[0], bbox[2]
            page_data.append({
                "text": text,
                "x0": x0,
                "y0": y1,  # Adjust to bottom-left corner (PDF coordinates)
                "font_size": y1 - y0,  # Approximate font size
                "confidence": confidence
            })

        extracted_data.append(page_data)

        # Cleanup temporary image
        os.remove(image_path)

    return extracted_data


def overlay_text_with_fonts(pdf_path, extracted_data, output_pdf_path):
    """
    Overlay extracted text onto the original PDF using fonts from different font families.
    :param pdf_path: Path to the input PDF file.
    :param extracted_data: List of extracted text with positions.
    :param output_pdf_path: Path to save the output PDF file.
    """
    doc = fitz.open(pdf_path)

    # Define default font settings
    default_font = "Helvetica"  # You can replace it with specific fonts like "Arial" or others.

    for page_num, page_data in enumerate(extracted_data):
        page = doc[page_num]

        for item in page_data:
            if item["confidence"] > 0.8:  # Only overlay high-confidence text
                page.insert_text(
                    (item["x0"], item["y0"]),
                    item["text"],
                    fontsize=item["font_size"],
                    fontname=default_font,
                    color=(0, 0, 0),  # Black text
                    render_mode=0  # Ensure text is not outlined
                )

    doc.save(output_pdf_path)
    print(f"PDF saved to: {output_pdf_path}")


def process_pdf(uploaded_pdf, output_pdf_path):
    """
    Process the uploaded PDF to extract text using OCR and overlay it as editable text.
    :param uploaded_pdf: The uploaded PDF file.
    :param output_pdf_path: Path to save the output PDF file.
    """
    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as temp_pdf:
        temp_pdf.write(uploaded_pdf.read())
        temp_pdf_path = temp_pdf.name

    # Step 1: Extract text using OCR
    extracted_data = extract_text_with_ocr(temp_pdf_path)

    # Step 2: Overlay extracted text onto the original PDF
    overlay_text_with_fonts(temp_pdf_path, extracted_data, output_pdf_path)

    # Cleanup temporary file
    if os.path.exists(temp_pdf_path):
        os.remove(temp_pdf_path)


# Streamlit App
def main():
    st.title("PDF Text Conversion Tool")
    st.write("Upload a PDF to convert vector text into regular, editable text.")

    uploaded_file = st.file_uploader("Upload PDF", type=["pdf"])
    if uploaded_file:
        output_pdf_path = "converted_output.pdf"

        with st.spinner("Processing your PDF..."):
            process_pdf(uploaded_file, output_pdf_path)

        st.success("PDF processing complete!")

        # Provide a download button for the processed PDF
        with open(output_pdf_path, "rb") as f:
            st.download_button(
                label="Download Converted PDF",
                data=f,
                file_name="converted_output.pdf",
                mime="application/pdf"
            )

        # Cleanup the processed output PDF
        if os.path.exists(output_pdf_path):
            os.remove(output_pdf_path)


if __name__ == "__main__":
    main()