castor-extractor 0.24.9__tar.gz → 0.24.10__tar.gz

{castor_extractor-0.24.9 → castor_extractor-0.24.10}/CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## 0.24.10 - 2025-04-30
+
+* Tableau - skip warnings instead of raising an error
+
 ## 0.24.9 - 2025-04-16
 
 * Introduce API client for **Coalesce**

{castor_extractor-0.24.9 → castor_extractor-0.24.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: castor-extractor
-Version: 0.24.9
+Version: 0.24.10
 Summary: Extract your metadata assets.
 Home-page: https://www.castordoc.com/
 License: EULA
@@ -210,6 +210,10 @@ For any questions or bug report, contact us at [support@castordoc.com](mailto:su
 
 # Changelog
 
+## 0.24.10 - 2025-04-30
+
+* Tableau - skip warnings instead of raising an error
+
 ## 0.24.9 - 2025-04-16
 
 * Introduce API client for **Coalesce**

{castor_extractor-0.24.9 → castor_extractor-0.24.10}/castor_extractor/visualization/tableau/client/client_metadata_api.py

@@ -29,24 +29,62 @@ _RETRY_BASE_MS = 10_000
 _RETRY_COUNT = 4
 
 
+def _is_timeout(error: dict) -> bool:
+    error_message = error.get("message")
+    return error_message == _TIMEOUT_MESSAGE
+
+
+def _is_warning(error: dict) -> bool:
+    extensions = error.get("extensions")
+    if not extensions:
+        return False
+
+    severity = extensions.get("severity")
+    if not severity:
+        return False
+
+    return severity.lower() == "warning"
+
+
 def _check_errors(answer: dict) -> None:
     """
-    handle errors in graphql response:
+    Handle errors in graphql response:
     - return None when there's no errors in the answer
-    - TableauApiTimeout if any of the errors is a timeout
-    - TableauApiError (generic) otherwise
+    - raise TableauApiTimeout if any of the errors is a timeout
+    - else raise TableauApiError if any of the errors is critical
+    - return None otherwise
+    More info about Tableau errors:
+    https://help.tableau.com/current/api/metadata_api/en-us/docs/meta_api_errors.html#other-errors
     """
     if "errors" not in answer:
         return
 
     errors = answer["errors"]
 
-    for error in errors:
-        if error.get("message") == _TIMEOUT_MESSAGE:
-            # we need specific handling for timeout issues (retry strategy)
-            raise TableauApiTimeout(errors)
+    has_timeout_errors = False
+    has_critical_errors = False
 
-    raise TableauApiError(answer["errors"])
+    for error in errors:
+        if _is_timeout(error):
+            has_timeout_errors = True
+            continue
+
+        if _is_warning(error):
+            # in this case, the answer contains the data anyway
+            # just display the warning
+            logger.warning(error)
+            continue
+
+        # at this point, it's not a timeout error
+        # besides, it's not a warning (severity is either "error" or Unknown)
+        has_critical_errors = True
+
+    if has_timeout_errors:
+        raise TableauApiTimeout(errors)
+    if has_critical_errors:
+        raise TableauApiError(errors)
+
+    return None
 
 
 def gql_query_scroll(

castor_extractor-0.24.10/castor_extractor/visualization/tableau/client/client_metadata_api_test.py

@@ -0,0 +1,111 @@
+import pytest
+
+from .client_metadata_api import _TIMEOUT_MESSAGE, _check_errors, _deduplicate
+from .errors import TableauApiError, TableauApiTimeout
+
+
+def test__deduplicate():
+    result_pages = iter(
+        [
+            [
+                {"id": 1, "name": "workbook_1"},
+                {"id": 2, "name": "workbook_2"},
+            ],
+            [
+                {"id": 1, "name": "workbook_1"},
+                {"id": 3, "name": "workbook_3"},
+                {"id": 4, "name": "workbook_4"},
+            ],
+            [
+                {"id": 4, "name": "workbook_4"},
+                {"id": 5, "name": "workbook_5"},
+                {"id": 5, "name": "workbook_5"},
+                {"id": 5, "name": "workbook_5"},
+            ],
+            [
+                {"id": 1, "name": "workbook_1"},
+                {"id": 3, "name": "workbook_3"},
+            ],
+        ]
+    )
+    deduplicated = _deduplicate(result_pages)
+    assert len(deduplicated) == 5
+    deduplicated_keys = {item["id"] for item in deduplicated}
+    assert deduplicated_keys == {1, 2, 3, 4, 5}
+
+
+def _timeout_error():
+    return {"message": _TIMEOUT_MESSAGE}
+
+
+def _critical_error():
+    return {
+        "message": "critical error",
+        "extensions": {"severity": "Error"},
+    }
+
+
+def _unknown_severity_error():
+    return {
+        "message": "unknown severity error",
+        "extensions": {"severity": None},
+    }
+
+
+def _warning():
+    return {
+        "message": "critical error",
+        "extensions": {"severity": "Warning"},
+    }
+
+
+def test__check_errors():
+    answer = {"data": []}  # no errors
+    assert _check_errors(answer) is None
+
+    # only warnings
+    answer = {
+        "data": [],
+        "errors": [
+            _warning(),
+            _warning(),
+            _warning(),
+        ],
+    }
+    assert _check_errors(answer) is None
+
+    # timeout issues should be prio (because they can be retried)
+    answer = {
+        "data": [],
+        "errors": [
+            _warning(),
+            _critical_error(),
+            _timeout_error(),
+            _warning(),
+        ],
+    }
+    with pytest.raises(TableauApiTimeout):
+        _check_errors(answer)
+
+    # expect critical error
+    answer = {
+        "data": [],
+        "errors": [
+            _warning(),
+            _warning(),
+            _critical_error(),
+        ],
+    }
+    with pytest.raises(TableauApiError):
+        _check_errors(answer)
+
+    # unknown severity is considered as critical
+    answer = {
+        "data": [],
+        "errors": [
+            _unknown_severity_error(),
+            _warning(),
+        ],
+    }
+    with pytest.raises(TableauApiError):
+        _check_errors(answer)

{castor_extractor-0.24.9 → castor_extractor-0.24.10}/pyproject.toml

@@ -3,7 +3,7 @@ requires = ["setuptools>=61.2"]
 
 [tool.poetry]
 name = "castor-extractor"
-version = "0.24.9"
+version = "0.24.10"
 description = "Extract your metadata assets."
 authors = ["Castor <support@castordoc.com>"]
 license = "EULA"

castor_extractor-0.24.9/castor_extractor/visualization/tableau/client/client_metadata_api_test.py

@@ -1,31 +0,0 @@
-from .client_metadata_api import _deduplicate
-
-
-def test__deduplicate():
-    result_pages = iter(
-        [
-            [
-                {"id": 1, "name": "workbook_1"},
-                {"id": 2, "name": "workbook_2"},
-            ],
-            [
-                {"id": 1, "name": "workbook_1"},
-                {"id": 3, "name": "workbook_3"},
-                {"id": 4, "name": "workbook_4"},
-            ],
-            [
-                {"id": 4, "name": "workbook_4"},
-                {"id": 5, "name": "workbook_5"},
-                {"id": 5, "name": "workbook_5"},
-                {"id": 5, "name": "workbook_5"},
-            ],
-            [
-                {"id": 1, "name": "workbook_1"},
-                {"id": 3, "name": "workbook_3"},
-            ],
-        ]
-    )
-    deduplicated = _deduplicate(result_pages)
-    assert len(deduplicated) == 5
-    deduplicated_keys = {item["id"] for item in deduplicated}
-    assert deduplicated_keys == {1, 2, 3, 4, 5}