app.py

import gradio as gr
import PyPDF2
from pdf2image import convert_from_path
import pytesseract
from PIL import Image
import io
import tempfile
import os

def extract_text_from_pdf(pdf_file):
    """
    Extract text from PDF. Uses direct text extraction if available,
    falls back to OCR if the PDF is image-based.
    """
    if pdf_file is None:
        return None, "Please upload a PDF file first."
    
    try:
        # Try direct text extraction first
        pdf_reader = PyPDF2.PdfReader(pdf_file)
        extracted_text = ""
        
        for page_num, page in enumerate(pdf_reader.pages):
            page_text = page.extract_text()
            extracted_text += page_text
        
        # Check if we got meaningful text (more than just whitespace)
        if extracted_text.strip():
            status = f"✓ Text extracted directly from PDF ({len(pdf_reader.pages)} pages)\nNo OCR needed - PDF contains searchable text."
            return create_txt_file(extracted_text), status
        
        # If no text found, use OCR
        status_msg = "⚠ PDF appears to be image-based. Running OCR...\n"
        
        # Save uploaded file temporarily for pdf2image
        with tempfile.NamedTemporaryFile(delete=False, suffix='.pdf') as tmp_file:
            tmp_file.write(pdf_file.read())
            tmp_path = tmp_file.name
        
        # Convert PDF pages to images
        images = convert_from_path(tmp_path)
        
        # Perform OCR on each page
        ocr_text = ""
        for i, image in enumerate(images):
            page_text = pytesseract.image_to_string(image)
            ocr_text += f"\n--- Page {i + 1} ---\n{page_text}\n"
        
        # Clean up temp file
        os.unlink(tmp_path)
        
        status_msg += f"✓ OCR completed on {len(images)} pages"
        return create_txt_file(ocr_text), status_msg
        
    except Exception as e:
        return None, f"Error processing PDF: {str(e)}"

def create_txt_file(text):
    """Create a downloadable text file from the extracted text."""
    txt_bytes = text.encode('utf-8')
    return txt_bytes

# Create Gradio interface
with gr.Blocks(title="PDF to Text Converter") as demo:
    gr.Markdown(
        """
        # 📄 PDF to Text Converter with Smart OCR
        
        Upload a PDF file and get downloadable text. The app automatically:
        - Extracts text directly if the PDF contains searchable text
        - Uses OCR (Optical Character Recognition) only for image-based PDFs
        """
    )
    
    with gr.Row():
        with gr.Column():
            pdf_input = gr.File(
                label="Upload PDF File",
                file_types=[".pdf"],
                type="binary"
            )
            convert_btn = gr.Button("Convert to Text", variant="primary")
        
        with gr.Column():
            status_output = gr.Textbox(
                label="Status",
                lines=3,
                interactive=False
            )
            txt_output = gr.File(
                label="Download Text File",
                type="binary"
            )
    
    gr.Markdown(
        """
        ### How it works:
        1. Upload your PDF file
        2. Click "Convert to Text"
        3. The app will extract text directly if possible, or use OCR if needed
        4. Download the resulting .txt file
        
        **Note:** OCR may take longer for large PDFs with many pages.
        """
    )
    
    convert_btn.click(
        fn=extract_text_from_pdf,
        inputs=pdf_input,
        outputs=[txt_output, status_output]
    )

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