fix: skip invalid local sources in DoclingServeConverter

integrations/docling_serve/src/haystack_integrations/components/converters/docling_serve/converter.py

@@ -518,6 +518,12 @@ def run(
                         documents.append(Document(content=content, meta=merged_meta))
                     else:
                         logger.warning("No content returned for source {source}.", source=source)
+                except (OSError, TypeError) as e:
+                    logger.warning(
+                        "Could not read local source {source}. Skipping it. Error: {error}",
+                        source=source,
+                        error=e,
+                    )
                 except httpx.HTTPStatusError as e:
                     logger.warning(
                         "DoclingServe returned HTTP {status} for {source}: {body}",
@@ -592,6 +598,12 @@ async def run_async(
                         documents.append(Document(content=content, meta=merged_meta))
                     else:
                         logger.warning("No content returned for source {source}.", source=source)
+                except (OSError, TypeError) as e:
+                    logger.warning(
+                        "Could not read local source {source}. Skipping it. Error: {error}",
+                        source=source,
+                        error=e,
+                    )
                 except httpx.HTTPStatusError as e:
                     logger.warning(
                         "DoclingServe returned HTTP {status} for {source}: {body}",

integrations/docling_serve/tests/test_converter.py

@@ -325,6 +325,21 @@ def test_run_skips_failed_sync_file_response_details(self, tmp_path, caplog):
         assert result["documents"] == []
         assert "file conversion failed" in caplog.text
 
+    def test_run_skips_missing_local_file(self, tmp_path, caplog):
+        missing = tmp_path / "missing.pdf"
+        good = tmp_path / "good.pdf"
+        good.write_bytes(b"%PDF-test")
+        converter = DoclingServeConverter(api_key=None)
+        mock_resp = _mock_httpx_response("# Good content")
+
+        with patch("httpx.Client.post", return_value=mock_resp):
+            with caplog.at_level(logging.WARNING):
+                result = converter.run(sources=[missing, good])
+
+        assert len(result["documents"]) == 1
+        assert result["documents"][0].content == "# Good content"
+        assert "Could not read local source" in caplog.text
+
     def test_run_text_export(self):
         converter = DoclingServeConverter(export_type=ExportType.TEXT, api_key=None)
         mock_resp = _mock_httpx_response("plain text content", ExportType.TEXT)
@@ -604,6 +619,26 @@ async def test_run_async_skips_on_connection_error(self, caplog):
         assert result["documents"] == []
         assert "Could not connect" in caplog.text
 
+    @pytest.mark.asyncio
+    async def test_run_async_skips_missing_local_file(self, tmp_path, caplog):
+        missing = tmp_path / "missing.pdf"
+        good = tmp_path / "good.pdf"
+        good.write_bytes(b"%PDF-test")
+        converter = DoclingServeConverter(api_key=None)
+        mock_resp = httpx.Response(
+            200,
+            json={"document": {"md_content": "# Good content"}, "status": "success"},
+            request=httpx.Request("POST", "http://test"),
+        )
+
+        with patch("httpx.AsyncClient.post", new_callable=AsyncMock, return_value=mock_resp):
+            with caplog.at_level(logging.WARNING):
+                result = await converter.run_async(sources=[missing, good])
+
+        assert len(result["documents"]) == 1
+        assert result["documents"][0].content == "# Good content"
+        assert "Could not read local source" in caplog.text
+
     @pytest.mark.asyncio
     async def test_run_async_multiple_sources(self):
         converter = DoclingServeConverter(api_key=None)