Update app.py

app.py

@@ -60,39 +60,35 @@ def ocr_page(pdf_path, page_num):
     return decoded[0] if decoded else ""
 
 def convert_pdf_to_epub(pdf_file, title, author, language):
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp_pdf:
-        tmp_pdf.write(pdf_file.read())
-        tmp_pdf_path = tmp_pdf.name
+    # Save the uploaded file to a temporary path
+    tmp_pdf_path = "/tmp/uploaded.pdf"
+    with open(tmp_pdf_path, "wb") as f:
+        f.write(pdf_file.read())  # This ensures the file isn't empty
 
+    # Now it's safe to read it
     reader = PdfReader(tmp_pdf_path)
-    num_pages = len(reader.pages)
-
-    book = epub.EpubBook()
-    book.set_title(title)
-    book.add_author(author)
-    book.set_language(language)
-
-    # Set cover from page 1
-    cover_image_b64 = render_pdf_to_base64png(tmp_pdf_path, 1, target_longest_image_dim=1024)
-    cover_image_bytes = base64.b64decode(cover_image_b64)
-    book.set_cover("cover.jpg", cover_image_bytes)
-
-    # Add OCR'd pages as chapters
-    for i in range(num_pages):
-        text = ocr_page(tmp_pdf_path, i)
-        chapter = epub.EpubHtml(title=f"Page {i+1}", file_name=f"page_{i+1}.xhtml", lang=language)
-        chapter.content = f"<h1>Page {i+1}</h1><p>{text}</p>"
-        book.add_item(chapter)
-        book.spine.append(chapter)
-
-    # Finalize EPUB
-    book.add_item(epub.EpubNcx())
-    book.add_item(epub.EpubNav())
-    epub_path = os.path.join(tempfile.gettempdir(), "output.epub")
-    epub.write_epub(epub_path, book, {})
-
-    with open(epub_path, "rb") as f:
-        return epub_path, f.read()
+
+    # Extract the first page for the cover (if needed)
+    first_page = reader.pages[0]
+    cover_path = "/tmp/cover.jpg"
+    images = convert_from_path(tmp_pdf_path, first_page=1, last_page=1)
+    images[0].save(cover_path, "JPEG")
+
+    # Run OCR and get text from olmocr
+    ocr_text = olmocr.process(tmp_pdf_path)
+
+    # Use metadata
+    epub_path = "/tmp/output.epub"
+    create_epub_from_text(
+        text=ocr_text,
+        output_path=epub_path,
+        title=title,
+        author=author,
+        language=language,
+        cover_image=cover_path
+    )
+
+    return epub_path, cover_path
 
 def interface_fn(pdf, title, author, language):
     epub_path, _ = convert_pdf_to_epub(pdf, title, author, language)