jkj

Provider/openlibrary.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import base64
+import io
 from concurrent import futures
 import hashlib
 import json as json_module
@@ -34,6 +35,53 @@ except ImportError:
     tqdm = None  # type: ignore
 
 
+def _image_paths_to_pdf_bytes(images: List[str]) -> Optional[bytes]:
+    if not images:
+        return None
+    try:
+        from PIL import Image  # type: ignore
+    except Exception:
+        return None
+
+    pil_images: List[Any] = []
+    try:
+        for p in images:
+            img_path = Path(p)
+            if not img_path.is_file():
+                continue
+            with Image.open(img_path) as im:  # type: ignore[attr-defined]
+                # Ensure PDF-compatible mode.
+                if im.mode in {"RGBA", "LA", "P"}:
+                    im = im.convert("RGB")
+                else:
+                    im = im.convert("RGB")
+                pil_images.append(im.copy())
+    except Exception:
+        for im in pil_images:
+            try:
+                im.close()
+            except Exception:
+                pass
+        return None
+
+    if not pil_images:
+        return None
+
+    buf = io.BytesIO()
+    first, rest = pil_images[0], pil_images[1:]
+    try:
+        first.save(buf, format="PDF", save_all=True, append_images=rest)
+        return buf.getvalue()
+    except Exception:
+        return None
+    finally:
+        for im in pil_images:
+            try:
+                im.close()
+            except Exception:
+                pass
+
+
 def _looks_like_isbn(text: str) -> bool:
     t = (text or "").replace("-", "").strip()
     return t.isdigit() and len(t) in (10, 13)
@@ -941,32 +989,22 @@ class OpenLibrary(SearchProvider):
             try:
                 images = self._archive_download(session=session, n_threads=10, directory=temp_dir, links=links, scale=3, book_id=archive_id)
 
-                try:
-                    import img2pdf  # type: ignore
-
-                    pdf_bytes = img2pdf.convert(images) if images else None
-                    if not pdf_bytes:
-                        log("[openlibrary] PDF conversion failed", file=sys.stderr)
-                        try:
-                            shutil.rmtree(temp_dir)
-                        except Exception:
-                            pass
-                        return None
-
-                    pdf_path = unique_path(output_dir / f"{title}.pdf")
-                    with open(pdf_path, "wb") as f:
-                        f.write(pdf_bytes)
-
-                    try:
-                        shutil.rmtree(temp_dir)
-                    except Exception:
-                        pass
-                    return pdf_path
-
-                except ImportError:
-                    # Keep images folder.
+                pdf_bytes = _image_paths_to_pdf_bytes(images)
+                if not pdf_bytes:
+                    # Keep images folder for manual conversion.
+                    log("[openlibrary] PDF conversion failed; keeping images folder", file=sys.stderr)
                     return Path(temp_dir)
 
+                pdf_path = unique_path(output_dir / f"{title}.pdf")
+                with open(pdf_path, "wb") as f:
+                    f.write(pdf_bytes)
+
+                try:
+                    shutil.rmtree(temp_dir)
+                except Exception:
+                    pass
+                return pdf_path
+
             except Exception:
                 try:
                     shutil.rmtree(temp_dir)