Let httpx errors flow through DoclingServeClient

haiku_rag_slim/haiku/rag/ingester/workers/pipeline.py

@@ -52,10 +52,9 @@ def _classify(exc: BaseException) -> Exception:
             return TransientError(f"HTTP {status}: {exc}")
         return PermanentError(f"HTTP {status}: {exc}")
 
-    if isinstance(
-        exc,
-        httpx.ConnectError | httpx.ReadTimeout | httpx.WriteTimeout | httpx.PoolTimeout,
-    ):
+    if isinstance(exc, httpx.TransportError):
+        # Umbrella for ConnectError, NetworkError, TimeoutException, ProtocolError,
+        # ProxyError — every transport-layer failure that's worth retrying.
         return TransientError(f"network: {exc}")
 
     if isinstance(exc, asyncio.TimeoutError | TimeoutError | OSError):

haiku_rag_slim/haiku/rag/providers/docling_serve.py

@@ -118,50 +118,22 @@ class DoclingServeClient:
     ) -> dict[str, Any]:
         """Submit a task and poll until completion; fetch result as JSON.
 
-        Args:
-            endpoint: The async endpoint path (e.g., "/v1/convert/file/async")
-            files: Files to upload
-            data: Form data parameters
-            name: Name for error messages
-
-        Returns:
-            The result dictionary from the completed task
-
-        Raises:
-            ValueError: If the task fails or service is unavailable
+        httpx exceptions (ConnectError, HTTPStatusError, TimeoutException,
+        etc.) propagate so the ingester's pipeline classifier can route
+        4xx → PermanentError and 5xx/network → TransientError. ValueError
+        is raised by `_submit_and_wait` when docling-serve reports a task
+        failure or returns no task_id.
         """
         headers = self._get_headers()
         base_url = self._pick_url()
-
-        try:
-            async with self._httpx_client() as client:
-                task_id = await self._submit_and_wait(
-                    client, base_url, endpoint, files, data, headers, name
-                )
-                result_url = f"{base_url}/v1/result/{task_id}"
-                result_response = await client.get(result_url, headers=headers)
-                result_response.raise_for_status()
-                return result_response.json()
-
-        except httpx.ConnectError as e:
-            raise ValueError(
-                f"Could not connect to docling-serve at {base_url}. "
-                f"Ensure the service is running and accessible. Error: {e}"
-            ) from e
-        except httpx.TimeoutException as e:
-            raise ValueError(
-                f"Request to docling-serve timed out after {self.timeout}s. Error: {e}"
-            ) from e
-        except httpx.HTTPStatusError as e:
-            if e.response.status_code == 401:
-                raise ValueError(
-                    "Authentication failed. Check your API key configuration."
-                ) from e
-            raise ValueError(f"HTTP error from docling-serve: {e}") from e
-        except ValueError:
-            raise
-        except Exception as e:
-            raise ValueError(f"Failed to process via docling-serve: {e}") from e
+        async with self._httpx_client() as client:
+            task_id = await self._submit_and_wait(
+                client, base_url, endpoint, files, data, headers, name
+            )
+            result_url = f"{base_url}/v1/result/{task_id}"
+            result_response = await client.get(result_url, headers=headers)
+            result_response.raise_for_status()
+            return result_response.json()
 
     async def submit_and_poll_zip(
         self,
@@ -179,33 +151,11 @@ class DoclingServeClient:
         """
         headers = self._get_headers()
         base_url = self._pick_url()
-
-        try:
-            async with self._httpx_client() as client:
-                task_id = await self._submit_and_wait(
-                    client, base_url, endpoint, files, data, headers, name
-                )
-                result_url = f"{base_url}/v1/result/{task_id}"
-                result_response = await client.get(result_url, headers=headers)
-                result_response.raise_for_status()
-                return result_response.content
-
-        except httpx.ConnectError as e:
-            raise ValueError(
-                f"Could not connect to docling-serve at {base_url}. "
-                f"Ensure the service is running and accessible. Error: {e}"
-            ) from e
-        except httpx.TimeoutException as e:
-            raise ValueError(
-                f"Request to docling-serve timed out after {self.timeout}s. Error: {e}"
-            ) from e
-        except httpx.HTTPStatusError as e:
-            if e.response.status_code == 401:
-                raise ValueError(
-                    "Authentication failed. Check your API key configuration."
-                ) from e
-            raise ValueError(f"HTTP error from docling-serve: {e}") from e
-        except ValueError:
-            raise
-        except Exception as e:
-            raise ValueError(f"Failed to process via docling-serve: {e}") from e
+        async with self._httpx_client() as client:
+            task_id = await self._submit_and_wait(
+                client, base_url, endpoint, files, data, headers, name
+            )
+            result_url = f"{base_url}/v1/result/{task_id}"
+            result_response = await client.get(result_url, headers=headers)
+            result_response.raise_for_status()
+            return result_response.content

tests/ingester/test_pipeline.py

@@ -196,6 +196,27 @@ async def test_connect_error_classified_transient():
         await run_job(client, _job())
 
 
+@pytest.mark.parametrize(
+    "exc_factory",
+    [
+        lambda: httpx.ConnectTimeout("slow connect"),
+        lambda: httpx.ReadTimeout("slow read"),
+        lambda: httpx.WriteTimeout("slow write"),
+        lambda: httpx.PoolTimeout("pool"),
+        lambda: httpx.ProxyError("proxy"),
+    ],
+)
+@pytest.mark.asyncio
+async def test_transport_subclasses_classified_transient(exc_factory):
+    """The classifier umbrellas on httpx.TransportError so every transport-
+    layer subclass routes to TransientError, not the generic 'unexpected'
+    fallback."""
+    client = _mock_client()
+    client.create_document_from_source.side_effect = exc_factory()
+    with pytest.raises(TransientError, match="network"):
+        await run_job(client, _job())
+
+
 @pytest.mark.asyncio
 async def test_unknown_exception_classified_transient():
     client = _mock_client()

tests/test_chunker.py

@@ -443,7 +443,7 @@ class TestDoclingServeChunker:
         converter = get_converter(Config)
         doc = await converter.convert_text("# Test", name="test.md")
 
-        with pytest.raises(ValueError, match="Could not connect to docling-serve"):
+        with pytest.raises(httpx.ConnectError):
             await chunker.chunk(doc)
 
     @pytest.mark.asyncio
@@ -459,13 +459,15 @@ class TestDoclingServeChunker:
         converter = get_converter(Config)
         doc = await converter.convert_text("# Test", name="test.md")
 
-        with pytest.raises(ValueError, match="timed out"):
+        with pytest.raises(httpx.TimeoutException):
             await chunker.chunk(doc)
 
     @pytest.mark.asyncio
     @patch("haiku.rag.providers.docling_serve.httpx.AsyncClient")
     async def test_chunk_auth_error(self, mock_client_class, chunker):
-        """Test handling of authentication errors."""
+        """Auth failures surface as httpx.HTTPStatusError(401) so the
+        ingester's pipeline classifier can route them to PermanentError —
+        retrying a bad token is wasted work."""
         import httpx
 
         mock_request = Mock()
@@ -482,8 +484,9 @@ class TestDoclingServeChunker:
         converter = get_converter(Config)
         doc = await converter.convert_text("# Test", name="test.md")
 
-        with pytest.raises(ValueError, match="Authentication failed"):
+        with pytest.raises(httpx.HTTPStatusError) as exc_info:
             await chunker.chunk(doc)
+        assert exc_info.value.response.status_code == 401
 
     @pytest.mark.asyncio
     @patch("haiku.rag.providers.docling_serve.httpx.AsyncClient")

tests/test_converters.py

@@ -1130,7 +1130,7 @@ class TestDoclingServeConverter:
             mock_client.__aexit__ = AsyncMock(return_value=None)
             mock_client_class.return_value = mock_client
 
-            with pytest.raises(ValueError, match="Could not connect to docling-serve"):
+            with pytest.raises(httpx.ConnectError):
                 await converter.convert_text("# Test")
 
     @pytest.mark.asyncio
@@ -1143,12 +1143,14 @@ class TestDoclingServeConverter:
             mock_client.__aexit__ = AsyncMock(return_value=None)
             mock_client_class.return_value = mock_client
 
-            with pytest.raises(ValueError, match="timed out"):
+            with pytest.raises(httpx.TimeoutException):
                 await converter.convert_text("# Test")
 
     @pytest.mark.asyncio
     async def test_convert_text_auth_error(self, converter):
-        """Test handling of authentication errors."""
+        """Auth failures surface as httpx.HTTPStatusError(401) so the
+        ingester's pipeline classifier can route them to PermanentError —
+        retrying a bad token is wasted work."""
         mock_response = Mock()
         mock_response.status_code = 401
 
@@ -1163,8 +1165,9 @@ class TestDoclingServeConverter:
             mock_client.__aexit__ = AsyncMock(return_value=None)
             mock_client_class.return_value = mock_client
 
-            with pytest.raises(ValueError, match="Authentication failed"):
+            with pytest.raises(httpx.HTTPStatusError) as exc_info:
                 await converter.convert_text("# Test")
+            assert exc_info.value.response.status_code == 401
 
     @pytest.mark.asyncio
     async def test_convert_file_pdf(self, converter):