Source File: Handle network errors

airbyte-integrations/connectors/source-file/source_file/client.py

@@ -11,6 +11,7 @@
 from os import environ
 from typing import Iterable
 from urllib.parse import urlparse
+from urllib3.exceptions import ProtocolError
 
 import backoff
 import boto3
@@ -396,6 +397,12 @@ def read(self, fields: Iterable = None) -> Iterable[dict]:
             except ConnectionResetError:
                 logger.info(f"Catched `connection reset error - 104`, stream: {self.stream_name} ({self.reader.full_url})")
                 raise ConnectionResetError
+            except ProtocolError as err:
+                error_msg = (
+                    f"File {fp} can not be opened due to connection issues on provider side. Please check provided links and options"
+                )
+                logger.error(f"{error_msg}\n{traceback.format_exc()}")
+                raise ConfigurationError(error_msg) from err
 
     def _cache_stream(self, fp):
         """cache stream to file"""

airbyte-integrations/connectors/source-file/unit_tests/test_client.py

@@ -7,6 +7,7 @@
 
 import pytest
 from pandas import read_csv, read_excel
+from urllib3.exceptions import ProtocolError
 from source_file.client import Client, ConfigurationError, URLFile
 
 
@@ -150,3 +151,11 @@ def test_read(test_read_config):
         except ConnectionResetError:
             print("Exception has been raised correctly!")
         mock_method.assert_called()
+
+
+def test_read_network_issues(test_read_config):
+    test_read_config.update(format='excel')
+    client = Client(**test_read_config)
+    client.sleep_on_retry_sec = 0  # just for test
+    with patch.object(client, "_cache_stream", side_effect=ProtocolError), pytest.raises(ConfigurationError):
+        next(client.read(["date", "key"]))