Source Google Sheets: handle config errors (#26097)

* Source Google Sheets: handle config errors

* Source Google Sheets: update docs

* Source Google Sheets: fix test

* Source Google Sheets: add unit tests

* auto-bump connector version

* Automated Change

---------

Co-authored-by: Octavia Squidington III <octavia-squidington-iii@users.noreply.github.com>
Co-authored-by: artem1205 <artem1205@users.noreply.github.com>

airbyte-integrations/connectors/source-google-sheets/Dockerfile

@@ -34,5 +34,5 @@ COPY source_google_sheets ./source_google_sheets
 ENV AIRBYTE_ENTRYPOINT "python /airbyte/integration_code/main.py"
 ENTRYPOINT ["python", "/airbyte/integration_code/main.py"]
 
-LABEL io.airbyte.version=0.2.37
+LABEL io.airbyte.version=0.2.38
 LABEL io.airbyte.name=airbyte/source-google-sheets

airbyte-integrations/connectors/source-google-sheets/acceptance-test-config.yml

@@ -15,7 +15,7 @@ acceptance_tests:
     - config_path: secrets/service_config.json
       status: succeed
     - config_path: integration_tests/invalid_config.json
-      status: failed
+      status: exception
   discovery:
     tests:
     - config_path: secrets/service_config.json

airbyte-integrations/connectors/source-google-sheets/metadata.yaml

@@ -5,7 +5,7 @@ data:
   connectorSubtype: file
   connectorType: source
   definitionId: 71607ba1-c0ac-4799-8049-7f4b90dd50f7
-  dockerImageTag: 0.2.37
+  dockerImageTag: 0.2.38
   dockerRepository: airbyte/source-google-sheets
   githubIssueLabel: source-google-sheets
   icon: google-sheets.svg

airbyte-integrations/connectors/source-google-sheets/source_google_sheets/source.py

@@ -8,6 +8,7 @@
 from typing import Any, Generator, List, MutableMapping, Union
 
 from airbyte_cdk.logger import AirbyteLogger
+from airbyte_cdk.models import FailureType
 from airbyte_cdk.models.airbyte_protocol import (
     AirbyteCatalog,
     AirbyteConnectionStatus,
@@ -18,6 +19,7 @@
     Type,
 )
 from airbyte_cdk.sources.source import Source
+from airbyte_cdk.utils import AirbyteTracedException
 from apiclient import errors
 from google.auth import exceptions as google_exceptions
 from requests.status_codes import codes as status_codes
@@ -41,9 +43,6 @@ class SourceGoogleSheets(Source):
     Spreadsheets API Reference: https://developers.google.com/sheets/api/reference/rest/v4/spreadsheets
     """
 
-    def __init__(self):
-        super().__init__()
-
     def check(self, logger: AirbyteLogger, config: json) -> AirbyteConnectionStatus:
         # Check involves verifying that the specified spreadsheet is reachable with our credentials.
         try:
@@ -56,18 +55,22 @@ def check(self, logger: AirbyteLogger, config: json) -> AirbyteConnectionStatus:
         try:
             spreadsheet = client.get(spreadsheetId=spreadsheet_id, includeGridData=False)
         except errors.HttpError as err:
-            reason = str(err)
+            message = "Config error: "
             # Give a clearer message if it's a common error like 404.
             if err.resp.status == status_codes.NOT_FOUND:
-                reason = "Requested spreadsheet was not found."
-            logger.error(f"Formatted error: {reason}")
-            return AirbyteConnectionStatus(
-                status=Status.FAILED, message=f"Unable to connect with the provided credentials to spreadsheet. Error: {reason}"
-            )
+                message += "The spreadsheet link is not valid. Enter the URL of the Google spreadsheet you want to sync."
+            raise AirbyteTracedException(
+                message=message,
+                internal_message=message,
+                failure_type=FailureType.config_error,
+            ) from err
         except google_exceptions.GoogleAuthError as err:
-            return AirbyteConnectionStatus(
-                status=Status.FAILED, message=f"Unable to connect with the provided credentials to spreadsheet. Authentication Error: {err}"
-            )
+            message = "Access to the spreadsheet expired or was revoked. Re-authenticate to restore access."
+            raise AirbyteTracedException(
+                message=message,
+                internal_message=message,
+                failure_type=FailureType.config_error,
+            ) from err
 
         # Check for duplicate headers
         spreadsheet_metadata = Spreadsheet.parse_obj(spreadsheet)

airbyte-integrations/connectors/source-google-sheets/unit_tests/conftest.py

@@ -0,0 +1,19 @@
+#
+# Copyright (c) 2023 Airbyte, Inc., all rights reserved.
+#
+
+import pytest
+
+
+@pytest.fixture
+def invalid_config():
+    return {
+        "spreadsheet_id": "invalid_spreadsheet_id",
+        "credentials":
+            {
+                "auth_type": "Client",
+                "client_id": "fake_client_id",
+                "client_secret": "fake_client_secret",
+                "refresh_token": "fake_refresh_token"
+            }
+    }

airbyte-integrations/connectors/source-google-sheets/unit_tests/test_stream.py

@@ -0,0 +1,30 @@
+#
+# Copyright (c) 2023 Airbyte, Inc., all rights reserved.
+#
+
+import pytest
+import requests
+from airbyte_cdk.utils import AirbyteTracedException
+from apiclient import errors
+from source_google_sheets import SourceGoogleSheets
+from source_google_sheets.client import GoogleSheetsClient
+from source_google_sheets.helpers import SCOPES
+
+
+def test_invalid_credentials_error_message(invalid_config):
+    source = SourceGoogleSheets()
+    with pytest.raises(AirbyteTracedException) as e:
+        source.check(logger=None, config=invalid_config)
+    assert e.value.args[0] == 'Access to the spreadsheet expired or was revoked. Re-authenticate to restore access.'
+
+
+def test_invalid_link_error_message(mocker, invalid_config):
+    source = SourceGoogleSheets()
+    resp = requests.Response()
+    resp.status = 404
+    mocker.patch.object(GoogleSheetsClient, "__init__", lambda s, credentials, scopes=SCOPES: None)
+    mocker.patch.object(GoogleSheetsClient, "get", side_effect=errors.HttpError(resp=resp, content=b''))
+    with pytest.raises(AirbyteTracedException) as e:
+        source.check(logger=None, config=invalid_config)
+    expected_message = 'Config error: The spreadsheet link is not valid. Enter the URL of the Google spreadsheet you want to sync.'
+    assert e.value.args[0] == expected_message

docs/integrations/sources/google-sheets.md

@@ -75,7 +75,8 @@ The [Google API rate limit](https://developers.google.com/sheets/api/limits) is
 ## Changelog
 
 | Version | Date       | Pull Request                                             | Subject                                                                       |
-|---------| ---------- | -------------------------------------------------------- |-------------------------------------------------------------------------------|
+|---------|------------|----------------------------------------------------------|-------------------------------------------------------------------------------|
+| 0.2.38  | 2023-05-16 | [26097](https://github.com/airbytehq/airbyte/pull/26097) | Refactor config error                                                         |
 | 0.2.37  | 2023-02-21 | [23292](https://github.com/airbytehq/airbyte/pull/23292) | Skip non grid sheets.                                                         |
 | 0.2.36  | 2023-02-21 | [23272](https://github.com/airbytehq/airbyte/pull/23272) | Handle empty sheets gracefully.                                               |
 | 0.2.35  | 2023-02-23 | [23057](https://github.com/airbytehq/airbyte/pull/23057) | Slugify column names                                                          |