acryl-datahub 0.15.0.1rc7__py3-none-any.whl → 0.15.0.1rc8__py3-none-any.whl

{acryl_datahub-0.15.0.1rc7.dist-info → acryl_datahub-0.15.0.1rc8.dist-info}/RECORD

@@ -1,4 +1,4 @@
-datahub/__init__.py,sha256=dj0h5Hq8a33nXbLNFmlqql5K3OaWumjRX8IsgKQUCfs,576
+datahub/__init__.py,sha256=Ed_HdiA9eGLLG0fhJKPwruUxl4bgAPR8p2MDlRHqts8,576
 datahub/__main__.py,sha256=pegIvQ9hzK7IhqVeUi1MeADSZ2QlP-D3K0OQdEg55RU,106
 datahub/entrypoints.py,sha256=3-qSfXAx3Z0FEkBV5tlO8fQr4xk4ySeDRMVTpS5Xd6A,7793
 datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -197,7 +197,7 @@ datahub/ingestion/source/glue_profiling_config.py,sha256=vpMJH4Lf_qgR32BZy58suab
 datahub/ingestion/source/ldap.py,sha256=Vnzg8tpwBYeyM-KBVVsUJvGZGBMJiCJ_i_FhxaFRQ9A,18627
 datahub/ingestion/source/metabase.py,sha256=oemiMdzjfr82Hx6rdwTNBzFM8962LDkosYh7SD_I5cY,31717
 datahub/ingestion/source/mlflow.py,sha256=-yWUuAEVBiNN-elz8Pgn0UeGsC3fVB20z1zKNIr4LXI,12309
-datahub/ingestion/source/mode.py,sha256=fuDTByENqcbxViFyYjU70B86FyAYr3Pk9usIBI0Vl1U,63384
+datahub/ingestion/source/mode.py,sha256=cq1KIpLxuplETF7sUW0hoMQIZG1cgga5BGHP54a28wE,63467
 datahub/ingestion/source/mongodb.py,sha256=vZue4Nz0xaBoCUsQr3_0OIRkWRxeE_IH_Y_QKZ1s7S0,21077
 datahub/ingestion/source/nifi.py,sha256=ttsjZ9aRUvINmewvKFIQD8Rwa4jcl35WFG-F-jPGPWQ,56146
 datahub/ingestion/source/openapi.py,sha256=3ea2ORz1cuq4e7L2hSjxG9Cw3__pVoJ5UNYTJS3EnKU,17386
@@ -882,7 +882,7 @@ datahub/sql_parsing/sql_parsing_common.py,sha256=h_V_m54hJ9EUh5kczq7cYOIeNeo4bgf
 datahub/sql_parsing/sql_parsing_result_utils.py,sha256=prwWTj1EB2fRPv1eMB4EkpFNafIYAt-X8TIK0NWqank,796
 datahub/sql_parsing/sqlglot_lineage.py,sha256=gUVq3NwZUzQByJs43JZXz8lZf0ZVzVt0FzaW5wZOwK4,47460
 datahub/sql_parsing/sqlglot_utils.py,sha256=n6yufzEGwSlFeCSU540hEldIuab0q8KGqm9x0vSawkc,14699
-datahub/sql_parsing/tool_meta_extractor.py,sha256=7tY4FAClhFcqwc23lGVlnT6Dequ_5Xcpbt0hDvnlLzM,6670
+datahub/sql_parsing/tool_meta_extractor.py,sha256=qEPq8RFWyK0tmSPNlluvd5cxgwbd2v6m9ViSY4hm2QM,6822
 datahub/telemetry/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/telemetry/stats.py,sha256=YltbtC3fe6rl1kcxn1A-mSnVpECTPm5k-brrUt7QxTI,967
 datahub/telemetry/telemetry.py,sha256=gzla-QGNsynGg2FqFxiDDFQ0emG53MJ9lhOA2-UUg-Y,15047
@@ -982,8 +982,8 @@ datahub_provider/operators/datahub_assertion_operator.py,sha256=uvTQ-jk2F0sbqqxp
 datahub_provider/operators/datahub_assertion_sensor.py,sha256=lCBj_3x1cf5GMNpHdfkpHuyHfVxsm6ff5x2Z5iizcAo,140
 datahub_provider/operators/datahub_operation_operator.py,sha256=aevDp2FzX7FxGlXrR0khoHNbxbhKR2qPEX5e8O2Jyzw,174
 datahub_provider/operators/datahub_operation_sensor.py,sha256=8fcdVBCEPgqy1etTXgLoiHoJrRt_nzFZQMdSzHqSG7M,168
-acryl_datahub-0.15.0.1rc7.dist-info/METADATA,sha256=hl14lRgFU4pk8d2s_Qxx1Xtkbd2TQp6gEek2gpkea1o,173642
-acryl_datahub-0.15.0.1rc7.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-acryl_datahub-0.15.0.1rc7.dist-info/entry_points.txt,sha256=xnPSPLK3bJGADxe4TDS4wL4u0FT_PGlahDa-ENYdYCQ,9512
-acryl_datahub-0.15.0.1rc7.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
-acryl_datahub-0.15.0.1rc7.dist-info/RECORD,,
+acryl_datahub-0.15.0.1rc8.dist-info/METADATA,sha256=mW2V4Czvd-ZE_mUJX8XkNZxNwnBa-gLJxebl0KWsM2A,173642
+acryl_datahub-0.15.0.1rc8.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+acryl_datahub-0.15.0.1rc8.dist-info/entry_points.txt,sha256=xnPSPLK3bJGADxe4TDS4wL4u0FT_PGlahDa-ENYdYCQ,9512
+acryl_datahub-0.15.0.1rc8.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
+acryl_datahub-0.15.0.1rc8.dist-info/RECORD,,

datahub/__init__.py

@@ -3,7 +3,7 @@ import warnings
 
 # Published at https://pypi.org/project/acryl-datahub/.
 __package_name__ = "acryl-datahub"
-__version__ = "0.15.0.1rc7"
+__version__ = "0.15.0.1rc8"
 
 
 def is_dev_mode() -> bool:

datahub/ingestion/source/mode.py

@@ -5,6 +5,7 @@ import time
 from dataclasses import dataclass
 from datetime import datetime, timezone
 from functools import lru_cache
+from json import JSONDecodeError
 from typing import Dict, Iterable, List, Optional, Set, Tuple, Union
 
 import dateutil.parser as dp
@@ -193,6 +194,9 @@ class HTTPError429(HTTPError):
     pass
 
 
+ModeRequestError = (HTTPError, JSONDecodeError)
+
+
 @dataclass
 class ModeSourceReport(StaleEntityRemovalSourceReport):
     filtered_spaces: LossyList[str] = dataclasses.field(default_factory=LossyList)
@@ -328,11 +332,11 @@ class ModeSource(StatefulIngestionSourceBase):
         # Test the connection
         try:
             self._get_request_json(f"{self.config.connect_uri}/api/verify")
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_failure(
                 title="Failed to Connect",
                 message="Unable to verify connection to mode.",
-                context=f"Error: {str(http_error)}",
+                context=f"Error: {str(e)}",
             )
 
         self.workspace_uri = f"{self.config.connect_uri}/api/{self.config.workspace}"
@@ -521,11 +525,11 @@ class ModeSource(StatefulIngestionSourceBase):
                 if self.config.owner_username_instead_of_email
                 else user_json.get("email")
             )
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_warning(
                 title="Failed to retrieve Mode creator",
                 message=f"Unable to retrieve user for {href}",
-                context=f"Reason: {str(http_error)}",
+                context=f"Reason: {str(e)}",
             )
         return user
 
@@ -571,11 +575,11 @@ class ModeSource(StatefulIngestionSourceBase):
                     logging.debug(f"Skipping space {space_name} due to space pattern")
                     continue
                 space_info[s.get("token", "")] = s.get("name", "")
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_failure(
                 title="Failed to Retrieve Spaces",
                 message="Unable to retrieve spaces / collections for workspace.",
-                context=f"Workspace: {self.workspace_uri}, Error: {str(http_error)}",
+                context=f"Workspace: {self.workspace_uri}, Error: {str(e)}",
             )
 
         return space_info
@@ -721,11 +725,11 @@ class ModeSource(StatefulIngestionSourceBase):
         try:
             ds_json = self._get_request_json(f"{self.workspace_uri}/data_sources")
             data_sources = ds_json.get("_embedded", {}).get("data_sources", [])
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_failure(
                 title="Failed to retrieve Data Sources",
                 message="Unable to retrieve data sources from Mode.",
-                context=f"Error: {str(http_error)}",
+                context=f"Error: {str(e)}",
             )
 
         return data_sources
@@ -812,11 +816,11 @@ class ModeSource(StatefulIngestionSourceBase):
                 if definition.get("name", "") == definition_name:
                     return definition.get("source", "")
 
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_failure(
                 title="Failed to Retrieve Definition",
                 message="Unable to retrieve definition from Mode.",
-                context=f"Definition Name: {definition_name}, Error: {str(http_error)}",
+                context=f"Definition Name: {definition_name}, Error: {str(e)}",
             )
         return None
 
@@ -1382,11 +1386,11 @@ class ModeSource(StatefulIngestionSourceBase):
                 f"{self.workspace_uri}/spaces/{space_token}/reports"
             )
             reports = reports_json.get("_embedded", {}).get("reports", {})
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_failure(
                 title="Failed to Retrieve Reports for Space",
                 message="Unable to retrieve reports for space token.",
-                context=f"Space Token: {space_token}, Error: {str(http_error)}",
+                context=f"Space Token: {space_token}, Error: {str(e)}",
             )
         return reports
 
@@ -1400,11 +1404,11 @@ class ModeSource(StatefulIngestionSourceBase):
             url = f"{self.workspace_uri}/spaces/{space_token}/datasets"
             datasets_json = self._get_request_json(url)
             datasets = datasets_json.get("_embedded", {}).get("reports", [])
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_failure(
                 title="Failed to Retrieve Datasets for Space",
                 message=f"Unable to retrieve datasets for space token {space_token}.",
-                context=f"Error: {str(http_error)}",
+                context=f"Error: {str(e)}",
             )
         return datasets
 
@@ -1416,11 +1420,11 @@ class ModeSource(StatefulIngestionSourceBase):
                 f"{self.workspace_uri}/reports/{report_token}/queries"
             )
             queries = queries_json.get("_embedded", {}).get("queries", {})
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_failure(
                 title="Failed to Retrieve Queries",
                 message="Unable to retrieve queries for report token.",
-                context=f"Report Token: {report_token}, Error: {str(http_error)}",
+                context=f"Report Token: {report_token}, Error: {str(e)}",
             )
         return queries
 
@@ -1433,11 +1437,11 @@ class ModeSource(StatefulIngestionSourceBase):
                 f"{self.workspace_uri}/reports/{report_token}/runs/{report_run_id}/query_runs{query_run_id}"
             )
             queries = queries_json.get("_embedded", {}).get("queries", {})
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_failure(
                 title="Failed to Retrieve Queries for Report",
                 message="Unable to retrieve queries for report token.",
-                context=f"Report Token:{report_token}, Error: {str(http_error)}",
+                context=f"Report Token:{report_token}, Error: {str(e)}",
             )
             return {}
         return queries
@@ -1451,13 +1455,13 @@ class ModeSource(StatefulIngestionSourceBase):
                 f"/queries/{query_token}/charts"
             )
             charts = charts_json.get("_embedded", {}).get("charts", {})
-        except HTTPError as http_error:
+        except ModeRequestError as e:
             self.report.report_failure(
                 title="Failed to Retrieve Charts",
                 message="Unable to retrieve charts from Mode.",
                 context=f"Report Token: {report_token}, "
                 f"Query token: {query_token}, "
-                f"Error: {str(http_error)}",
+                f"Error: {str(e)}",
             )
         return charts
 
@@ -1477,6 +1481,8 @@ class ModeSource(StatefulIngestionSourceBase):
                 response = self.session.get(
                     url, timeout=self.config.api_options.timeout
                 )
+                if response.status_code == 204:  # No content, don't parse json
+                    return {}
                 return response.json()
             except HTTPError as http_error:
                 error_response = http_error.response

datahub/sql_parsing/tool_meta_extractor.py

@@ -40,6 +40,7 @@ def _get_last_line(query: str) -> str:
 class ToolMetaExtractorReport(Report):
     num_queries_meta_extracted: Dict[str, int] = field(default_factory=int_top_k_dict)
     failures: List[str] = field(default_factory=list)
+    looker_user_mapping_missing: Optional[bool] = None
 
 
 class ToolMetaExtractor:
@@ -108,7 +109,9 @@ class ToolMetaExtractor:
             PlatformResource.search_by_filters(query=query, graph_client=graph)
         )
 
-        if len(platform_resources) > 1:
+        if len(platform_resources) == 0:
+            report.looker_user_mapping_missing = True
+        elif len(platform_resources) > 1:
             report.failures.append(
                 "Looker user metadata extraction failed. Found more than one looker user id mappings."
             )