app.py

import gradio as gr
import torch
import base64
import fitz  # PyMuPDF
import tempfile
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.anchor import get_anchor_text

from mathml_utils import convert_inline_and_block_latex_to_mathml

from ebooklib import epub
import json
import html

# 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_html(pdf_file, title, author):
    import re
    import markdown2
    from latex2mathml.converter import convert as latex_to_mathml

    pdf_path = pdf_file.name
    doc = fitz.open(pdf_path)
    num_pages = len(doc)

    # Extract TOC as a dict: {page_number: [(level, title), ...]}
    toc_entries = doc.get_toc()
    toc_by_page = {}
    for level, title, page in toc_entries:
        toc_by_page.setdefault(page, []).append((level, title))

    all_text = ""
    cover_img_html = ""

    for i in range(num_pages):
        page_num = i + 1
        print(f"Processing page {page_num}...")

        try:
            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 = (
                "Below is the image of one page of a document, as well as some raw textual content that was previously "
                "extracted for it. Just return the plain text representation of this document as if you were reading it naturally.\n"
                "Do not hallucinate.\n"
                "RAW_TEXT_START\n"
                f"{anchor_text}\n"
                "RAW_TEXT_END"
            )

            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=5096,
                num_return_sequences=1,
                do_sample=True,
            )

            prompt_length = inputs["input_ids"].shape[1]
            new_tokens = output[:, prompt_length:].detach().cpu()

            decoded = "[No output generated]"
            if new_tokens is not None and new_tokens.shape[1] > 0:
                try:
                    decoded_list = processor.tokenizer.batch_decode(new_tokens, skip_special_tokens=True)
                    raw_output = decoded_list[0].strip() if decoded_list else "[No output generated]"
                    try:
                        parsed = json.loads(raw_output)
                        decoded = parsed.get("natural_text", raw_output)
                    except json.JSONDecodeError:
                        decoded = raw_output
                except Exception as decode_error:
                    decoded = f"[Decoding error on page {page_num}: {str(decode_error)}]"
            else:
                decoded = "[Model returned no new tokens]"

        except Exception as processing_error:
            decoded = f"[Processing error on page {page_num}: {str(processing_error)}]"

        print(f"Decoded content for page {page_num}: {decoded}")

        # Convert inline and block LaTeX math to MathML
        def convert_latex(text):
            def replacer(match):
                try:
                    return f"<math>{latex_to_mathml(match.group(1))}</math>"
                except:
                    return html.escape(match.group(0))
            text = re.sub(r'\\\((.*?)\\\)', replacer, text)
            text = re.sub(r'\\\[(.*?)\\\]', replacer, text)
            return text

        math_converted = convert_latex(decoded)
        markdown_converted = markdown2.markdown(math_converted)
        html_page = markdown_converted.replace("\n", "<br>")

        # Add TOC-derived headers if present on this page
        if page_num in toc_by_page:
            for level, header in toc_by_page[page_num]:
                tag = f"h{min(level, 6)}"  # Limit to h6
                html_page = f"<{tag}>{html.escape(header)}</{tag}>\n" + html_page

        all_text += f"<div>{html_page}</div>\n"

        if page_num == 1:
            cover_img_html = f'<img src="data:image/png;base64,{image_base64}" alt="cover" style="max-width:100%; height:auto;"><hr>'

    # MathJax fallback in case MathML fails (some browsers prefer it)
    mathjax_script = """
    <script type="text/javascript" id="MathJax-script" async
      src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js">
    </script>
    """

    full_html = f"""<!DOCTYPE html>
    <html>
    <head>
        <meta charset="utf-8">
        <title>{html.escape(title)}</title>
        {mathjax_script}
    </head>
    <body>
        <h1>{html.escape(title)}</h1>
        <h3>{html.escape(author)}</h3>
        {cover_img_html}
        {all_text}
    </body>
    </html>
    """

    with tempfile.NamedTemporaryFile(delete=False, suffix=".html", dir="/tmp", mode="w", encoding="utf-8") as tmp:
        tmp.write(full_html)
        return tmp.name

# Gradio Interface
iface = gr.Interface(
    fn=process_pdf_to_html,  # NEW FUNCTION
    inputs=[
        gr.File(label="Upload PDF", file_types=[".pdf"]),
        gr.Textbox(label="HTML Title"),
        gr.Textbox(label="Author(s)")
    ],
    outputs=gr.File(label="Download HTML"),
    title="PDF to HTML Converter (for Calibre/Kindle)",
    description="Uploads a PDF, extracts text via vision+prompt, embeds it in a styled HTML file with math support. Ready for Calibre.",
    allow_flagging="never"
)

if __name__ == "__main__":
    iface.launch(
        server_name="0.0.0.0",
        server_port=7860,
        share=True,
        debug=True,
        allowed_paths=["/tmp"]
    )