explicit conversion timeout and handling

python/packages/officeconvert/src/officeconvert/conversion.py

@@ -57,13 +57,19 @@ def convert_pptx_to_pdf(pptx_path: Path, pdf_path: Path, *, timeout_s: int = 120
         str(output_dir),
         str(pptx_path.resolve()),
     ]
-    completed = subprocess.run(
-        command,
-        check=False,
-        capture_output=True,
-        text=True,
-        timeout=timeout_s,
-    )
+    try:
+        completed = subprocess.run(
+            command,
+            check=False,
+            capture_output=True,
+            text=True,
+            timeout=timeout_s,
+        )
+    except subprocess.TimeoutExpired as exc:
+        raise RuntimeError(
+            "LibreOffice conversion timed out after "
+            f"{timeout_s} seconds while rendering {pptx_path.name}"
+        ) from exc
     if completed.returncode != 0:
         raise RuntimeError(
             f"LibreOffice conversion failed: {completed.stderr.strip() or completed.stdout.strip()}"
@@ -116,13 +122,20 @@ def render_pdf_to_images(
         str(pdf_path.resolve()),
         str(prefix_path),
     ]
-    completed = subprocess.run(
-        command,
-        check=False,
-        capture_output=True,
-        text=True,
-        timeout=timeout_s,
-    )
+    try:
+        completed = subprocess.run(
+            command,
+            check=False,
+            capture_output=True,
+            text=True,
+            timeout=timeout_s,
+        )
+    except subprocess.TimeoutExpired as exc:
+        raise RuntimeError(
+            "Poppler rasterization timed out after "
+            f"{timeout_s} seconds while rendering {pdf_path.name}; "
+            "increase conversion PDF render timeout or lower image DPI"
+        ) from exc
     if completed.returncode != 0:
         raise RuntimeError(
             f"Poppler rasterization failed: {completed.stderr.strip() or completed.stdout.strip()}"
@@ -165,6 +178,8 @@ def convert_pptx_to_slidedeck(
     *,
     dpi: int = 180,
     image_format: str = "png",
+    pptx_to_pdf_timeout_s: int = 180,
+    pdf_to_images_timeout_s: int = 600,
 ) -> SlideDeckResult:
     """Convert a PPTX into rendered images and extracted notes.
 
@@ -177,6 +192,8 @@ def convert_pptx_to_slidedeck(
         work_dir: Scratch directory for generated outputs.
         dpi: Rasterization DPI for output slide images.
         image_format: Output image format accepted by `pdftoppm`.
+        pptx_to_pdf_timeout_s: Timeout in seconds for the LibreOffice subprocess.
+        pdf_to_images_timeout_s: Timeout in seconds for the Poppler subprocess.
 
     Returns:
         Fully materialized `SlideDeckResult` with local image paths.
@@ -189,12 +206,13 @@ def convert_pptx_to_slidedeck(
     pdf_path = work_dir / f"{pptx_path.stem}.pdf"
     image_dir = work_dir / "slides"
 
-    convert_pptx_to_pdf(pptx_path, pdf_path)
+    convert_pptx_to_pdf(pptx_path, pdf_path, timeout_s=pptx_to_pdf_timeout_s)
     image_paths = render_pdf_to_images(
         pdf_path,
         image_dir,
         dpi=dpi,
         image_format=image_format,
+        timeout_s=pdf_to_images_timeout_s,
     )
     notes = extract_slide_notes(pptx_path)