app.py

import gradio as gr
import torch
import base64
import fitz  # PyMuPDF
from io import BytesIO
from PIL import Image
from pathlib import Path
from transformers import AutoProcessor, Qwen2VLForConditionalGeneration

from olmocr.data.renderpdf import render_pdf_to_base64png
from olmocr.prompts import build_finetuning_prompt
from olmocr.prompts.anchor import get_anchor_text

from ebooklib import epub

# Load model and processor
model = Qwen2VLForConditionalGeneration.from_pretrained(
    "allenai/olmOCR-7B-0225-preview", torch_dtype=torch.bfloat16
).eval()
processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

def process_pdf_to_epub(pdf_file, title, author):
    pdf_path = pdf_file.name
    doc = fitz.open(pdf_path)
    num_pages = len(doc)

    # Create EPUB book
    book = epub.EpubBook()
    book.set_identifier("id123456")
    book.set_title(title)
    book.add_author(author)

    chapters = []

    for i in range(num_pages):
        page_num = i + 1

        try:
            # Render page to base64 image
            image_base64 = render_pdf_to_base64png(pdf_path, page_num, target_longest_image_dim=1024)
            anchor_text = get_anchor_text(pdf_path, page_num, pdf_engine="pdfreport", target_length=4000)
            prompt = build_finetuning_prompt(anchor_text)

            # Format prompt
            messages = [
                {
                    "role": "user",
                    "content": [
                        {"type": "text", "text": prompt},
                        {"type": "image_url", "image_url": {"url": f"data:image/png;base64,{image_base64}"}},
                    ],
                }
            ]
            text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
            image = Image.open(BytesIO(base64.b64decode(image_base64)))

            inputs = processor(
                text=[text],
                images=[image],
                padding=True,
                return_tensors="pt",
            )
            inputs = {k: v.to(device) for k, v in inputs.items()}

            output = model.generate(
                **inputs,
                temperature=0.8,
                max_new_tokens=512,
                num_return_sequences=1,
                do_sample=True,
            )
            prompt_length = inputs["input_ids"].shape[1]
            new_tokens = output[:, prompt_length:]

        try:
            decoded_list = processor.tokenizer.batch_decode(new_tokens, skip_special_tokens=True)
            decoded = decoded_list[0].strip() if decoded_list else "[No output generated]"
        except Exception as decode_error:
            decoded = f"[Decoding error on page {page_num}: {str(decode_error)}]"

        # Create chapter
        chapter = epub.EpubHtml(title=f"Page {page_num}", file_name=f"page_{page_num}.xhtml", lang="en")
        chapter.content = f"<h1>Page {page_num}</h1><p>{decoded}</p>"
        book.add_item(chapter)
        chapters.append(chapter)

        # Save cover image from page 1
        if page_num == 1:
            cover_image = Image.open(BytesIO(base64.b64decode(image_base64)))
            cover_io = BytesIO()
            cover_image.save(cover_io, format='PNG')
            book.set_cover("cover.png", cover_io.getvalue())

    # Assemble EPUB
    book.toc = tuple(chapters)
    book.add_item(epub.EpubNcx())
    book.add_item(epub.EpubNav())
    book.spine = ['nav'] + chapters

    output_path = "/tmp/output.epub"
    epub.write_epub(output_path, book)
    return output_path

# Gradio Interface
iface = gr.Interface(
    fn=process_pdf_to_epub,
    inputs=[
        gr.File(label="Upload PDF", file_types=[".pdf"]),
        gr.Textbox(label="EPUB Title"),
        gr.Textbox(label="Author(s)")
    ],
    outputs=gr.File(label="Download EPUB"),
    title="PDF to EPUB Converter (with olmOCR)",
    description="Uploads a PDF, extracts text from each page with vision + prompt, and builds an EPUB using the outputs. Sets the first page as cover.",
    allow_flagging="never"  # Add this line to avoid the flagged directory issue
)

if __name__ == "__main__":
    iface.launch(
        server_name="0.0.0.0",  # Required to make app publicly accessible
        server_port=7860,       # Can be changed if needed
        share=True,             # Optional: creates a public Gradio link if supported
        debug=True,             # Optional: helpful if you're troubleshooting
        allowed_paths=["/tmp"]  # Optional: makes it explicit that Gradio can write here
    )