Reset all fixes

app.py

@@ -1,71 +1,71 @@
-import gradio as gr
+import os
 import tempfile
-from io import BytesIO
-from PIL import Image
+from pathlib import Path
+
+import gradio as gr
 from ebooklib import epub
-from olmocr.model import process_pdf  # your forked olmocr model
+from olmocr import process_pdf  # your forked olmocr model
 
-def process_pdf_to_epub(pdf_file, title, author):
-    # Run the OCR + LLM pipeline from olmocr
-    print("Starting PDF processing...")
-    page_results = process_pdf(pdf_file.name)
+def process_pdf_to_epub(pdf_path, title="Untitled", author="Unknown"):
+    print(f"Processing PDF: {pdf_path}")
+    output_dir = tempfile.mkdtemp()
+    results = process_pdf(pdf_path, output_dir)
 
-    # Create the EPUB book
     book = epub.EpubBook()
     book.set_identifier("id123456")
-    book.set_title(title if title else "Untitled Document")
+    book.set_title(title)
     book.set_language("en")
-    if author:
-        book.add_author(author)
-
-    # Try to use the first page as cover
-    try:
-        with Image.open(pdf_file.name) as img:
-            img.convert("RGB").save("cover.jpg", "JPEG")
-            with open("cover.jpg", "rb") as f:
-                cover_data = f.read()
-                book.set_cover("cover.jpg", cover_data)
-    except Exception as e:
-        print("Could not generate cover:", e)
+    book.add_author(author)
 
-    # Add chapters from pages
     chapters = []
-    for i, page in enumerate(page_results):
-        text = page.get("decoded", {}).get("natural_text", "")
-        if not text.strip():
+
+    for i, result in enumerate(results):
+        text = result.get("decoded_content", {}).get("natural_text", "")
+        if not text:
             continue
-        safe_text = text.replace("\n", "<br/>")
+
         chapter = epub.EpubHtml(title=f"Page {i+1}", file_name=f"page_{i+1}.xhtml", lang="en")
+        # Avoid backslash in f-string expression
+        safe_text = text.replace("\n", "<br/>")
         chapter.content = f"<h1>Page {i+1}</h1><p>{safe_text}</p>"
         book.add_item(chapter)
         chapters.append(chapter)
 
-    # Define spine and table of contents
-    book.toc = chapters
-    book.spine = ["nav"] + chapters
+        print(f"Processed page {i+1}")
+
+    if not chapters:
+        raise ValueError("No content extracted from PDF.")
+
+    book.toc = tuple(chapters)
     book.add_item(epub.EpubNcx())
     book.add_item(epub.EpubNav())
+    book.spine = ["nav"] + chapters
 
-    # Write to in-memory buffer
-    output_buffer = BytesIO()
-    epub.write_epub(output_buffer, book)
-    output_buffer.seek(0)
-    return output_buffer
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".epub", dir="/tmp") as tmp:
+        epub.write_epub(tmp.name, book)
+        print(f"EPUB written to {tmp.name}")
+        return tmp.name
 
-with gr.Blocks() as demo:
-    gr.Markdown("# PDF to EPUB Converter\nPowered by `olmOCR`")
-    with gr.Row():
-        pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
-    title = gr.Textbox(label="EPUB Title", placeholder="Optional title")
-    author = gr.Textbox(label="Author", placeholder="Optional author name")
-    convert_button = gr.Button("Convert to EPUB")
-    epub_output = gr.File(label="Download EPUB", file_types=[".epub"])
+# Gradio UI
+title_input = gr.Textbox(label="EPUB Title", value="Untitled")
+author_input = gr.Textbox(label="Author", value="Unknown")
+file_input = gr.File(label="Upload PDF", file_types=[".pdf"])
 
-    convert_button.click(
-        fn=process_pdf_to_epub,
-        inputs=[pdf_input, title, author],
-        outputs=epub_output
-    )
+output_file = gr.File(label="Download EPUB")
+
+iface = gr.Interface(
+    fn=process_pdf_to_epub,
+    inputs=[file_input, title_input, author_input],
+    outputs=output_file,
+    title="PDF to EPUB Converter with olmOCR",
+    description="Upload a PDF to convert it into an EPUB. First page is used as the cover."
+)
 
 if __name__ == "__main__":
-    demo.launch()
+    iface.launch(
+        server_name="0.0.0.0",
+        server_port=7860,
+        share=True,
+        debug=True,
+        allowed_paths=["/tmp"]
+    )