app.py

import gradio as gr
import tempfile
import re
import os 
import spacy
import pytesseract
import pdf2image
import subprocess
from pdf2image.exceptions import (
    PDFInfoNotInstalledError,
    PDFPageCountError,
    PDFSyntaxError
)
import fitz  # PyMuPDF
from PIL import Image, UnidentifiedImageError
import io
import base64

def clean_text(text):
    nlp = spacy.load("en_core_web_sm", disable=["tagger", "parser", "ner", "textcat"])
    text = re.sub(r'\n+', '\n', text) 
    text = re.sub(r'\s+', ' ', text)
    return text.strip()

def safe_base64_decode(s):
    # add missing padding if necessary
    missing_padding = len(s) % 4
    if missing_padding:
        s += '='* (4 - missing_padding)
    try:
        return base64.b64decode(s)
    except binascii.Error as e:
        print("Error decoding base64 string:", e)
        return None

def image_to_latex(image):
    image_path = "/tmp/equation.png"  # Modify as needed
    image.save(image_path)
    result = subprocess.run(["pix2tex", image_path], capture_output=True, text=True)
    return result.stdout

def pdf_to_text(file):
    doc = fitz.open(file.name)
    full_text = ''
    for i, page in enumerate(doc):
        # Extract text
        page_text = page.get_text()

        # Extract images and convert to LaTeX
        image_list = page.get_images(full=True)
        for img in image_list:
            xref, name, ext, color_space, width, height, bpc, image_data, image_mask, smask_data = img
            # Check if image_data is base64 encoded string
            if isinstance(image_data, str) and re.match(r'^[A-Za-z0-9+/]+[=]{0,2}$', image_data):
                image_data = safe_base64_decode(image_data)
            try:
                image = Image.open(io.BytesIO(image_data))
                latex_code = image_to_latex(image)
                page_text += "\n" + latex_code  # Add LaTeX code to page text
            except UnidentifiedImageError:
                print(f"Could not identify image on page {i+1}")

        page_text = clean_text(page_text)
        if len(page_text.split()) > 5: 
            page_number = i + 1
            page_text = "## Metadata: Page Number " + str(page_number) + "\n" + page_text
            full_text += page_text + "\n\n"

    base_name = os.path.splitext(os.path.basename(file.name))[0]
    output_file_name = base_name + ".txt"
    with open(output_file_name, 'w') as f:
        f.write(full_text)

    return output_file_name

iface = gr.Interface(fn=pdf_to_text,
                     inputs=gr.inputs.File(label="Your PDF"),
                     outputs=gr.outputs.File(label="Download TXT"),
                     title="PDF to TXT",
                     description="Convert your PDF files to clean text")
iface.launch()