openmetadata-ingestion 1.6.2.0__py3-none-any.whl → 1.10.14.4__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- _openmetadata_testutils/helpers/login_user.py +22 -0
- airflow_provider_openmetadata/__init__.py +3 -3
- airflow_provider_openmetadata/hooks/openmetadata.py +3 -3
- airflow_provider_openmetadata/lineage/backend.py +23 -4
- airflow_provider_openmetadata/lineage/callback.py +3 -3
- airflow_provider_openmetadata/lineage/config/commons.py +3 -3
- airflow_provider_openmetadata/lineage/config/loader.py +17 -3
- airflow_provider_openmetadata/lineage/operator.py +3 -3
- airflow_provider_openmetadata/lineage/runner.py +3 -3
- airflow_provider_openmetadata/lineage/status.py +3 -3
- metadata/__init__.py +33 -0
- metadata/__main__.py +3 -3
- metadata/__version__.py +3 -3
- metadata/antlr/split_listener.py +3 -3
- metadata/applications/example.py +74 -0
- metadata/automations/execute_runner.py +49 -0
- metadata/automations/extended_runner.py +13 -0
- metadata/automations/runner.py +48 -39
- metadata/cli/app.py +3 -3
- metadata/cli/classify.py +8 -8
- metadata/cli/common.py +25 -0
- metadata/cli/dataquality.py +8 -8
- metadata/cli/ingest.py +5 -7
- metadata/cli/ingest_dbt.py +367 -0
- metadata/cli/lineage.py +3 -3
- metadata/cli/profile.py +5 -7
- metadata/cli/restore.py +3 -3
- metadata/cli/usage.py +5 -7
- metadata/clients/aws_client.py +32 -11
- metadata/clients/azure_client.py +3 -3
- metadata/clients/domo_client.py +5 -5
- metadata/cmd.py +30 -3
- metadata/config/common.py +3 -3
- metadata/data_quality/api/models.py +3 -3
- metadata/data_quality/builders/validator_builder.py +3 -3
- metadata/data_quality/interface/pandas/pandas_test_suite_interface.py +3 -3
- metadata/data_quality/interface/sqlalchemy/databricks/test_suite_interface.py +3 -3
- metadata/data_quality/interface/sqlalchemy/snowflake/test_suite_interface.py +3 -3
- metadata/data_quality/interface/sqlalchemy/sqa_test_suite_interface.py +3 -3
- metadata/data_quality/interface/sqlalchemy/unity_catalog/test_suite_interface.py +3 -3
- metadata/data_quality/interface/test_suite_interface.py +12 -6
- metadata/data_quality/processor/test_case_runner.py +11 -22
- metadata/data_quality/runner/base_test_suite_source.py +16 -10
- metadata/data_quality/runner/core.py +3 -3
- metadata/data_quality/source/test_suite.py +49 -7
- metadata/data_quality/validations/base_test_handler.py +3 -3
- metadata/data_quality/validations/column/base/columnValueLengthsToBeBetween.py +3 -3
- metadata/data_quality/validations/column/base/columnValueMaxToBeBetween.py +3 -3
- metadata/data_quality/validations/column/base/columnValueMeanToBeBetween.py +3 -3
- metadata/data_quality/validations/column/base/columnValueMedianToBeBetween.py +3 -3
- metadata/data_quality/validations/column/base/columnValueMinToBeBetween.py +3 -3
- metadata/data_quality/validations/column/base/columnValueStdDevToBeBetween.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesMissingCount.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesSumToBeBetween.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesToBeAtExpectedLocation.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesToBeBetween.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesToBeInSet.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesToBeNotInSet.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesToBeNotNull.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesToBeUnique.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesToMatchRegex.py +3 -3
- metadata/data_quality/validations/column/base/columnValuesToNotMatchRegex.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValueLengthsToBeBetween.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValueMaxToBeBetween.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValueMeanToBeBetween.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValueMedianToBeBetween.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValueMinToBeBetween.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValueStdDevToBeBetween.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesMissingCount.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesSumToBeBetween.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesToBeAtExpectedLocation.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesToBeBetween.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesToBeInSet.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesToBeNotInSet.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesToBeNotNull.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesToBeUnique.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesToMatchRegex.py +3 -3
- metadata/data_quality/validations/column/pandas/columnValuesToNotMatchRegex.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValueLengthsToBeBetween.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValueMaxToBeBetween.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValueMeanToBeBetween.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValueMedianToBeBetween.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValueMinToBeBetween.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValueStdDevToBeBetween.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValuesMissingCount.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValuesSumToBeBetween.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeAtExpectedLocation.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeBetween.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeInSet.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeNotInSet.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeNotNull.py +3 -3
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeUnique.py +17 -5
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToMatchRegex.py +4 -4
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToNotMatchRegex.py +3 -3
- metadata/data_quality/validations/mixins/pandas_validator_mixin.py +3 -3
- metadata/data_quality/validations/mixins/sqa_validator_mixin.py +3 -3
- metadata/data_quality/validations/models.py +16 -3
- metadata/data_quality/validations/runtime_param_setter/base_diff_params_setter.py +219 -0
- metadata/data_quality/validations/runtime_param_setter/param_setter.py +2 -2
- metadata/data_quality/validations/runtime_param_setter/param_setter_factory.py +11 -2
- metadata/data_quality/validations/runtime_param_setter/table_custom_sql_query_params_setter.py +31 -0
- metadata/data_quality/validations/runtime_param_setter/table_diff_params_setter.py +56 -85
- metadata/data_quality/validations/table/base/tableColumnCountToBeBetween.py +3 -3
- metadata/data_quality/validations/table/base/tableColumnCountToEqual.py +3 -3
- metadata/data_quality/validations/table/base/tableColumnNameToExist.py +3 -3
- metadata/data_quality/validations/table/base/tableColumnToMatchSet.py +3 -3
- metadata/data_quality/validations/table/base/tableCustomSQLQuery.py +144 -5
- metadata/data_quality/validations/table/base/tableRowCountToBeBetween.py +3 -3
- metadata/data_quality/validations/table/base/tableRowCountToEqual.py +3 -3
- metadata/data_quality/validations/table/base/tableRowInsertedCountToBeBetween.py +3 -3
- metadata/data_quality/validations/table/pandas/tableColumnCountToBeBetween.py +3 -3
- metadata/data_quality/validations/table/pandas/tableColumnCountToEqual.py +3 -3
- metadata/data_quality/validations/table/pandas/tableColumnNameToExist.py +3 -3
- metadata/data_quality/validations/table/pandas/tableColumnToMatchSet.py +3 -3
- metadata/data_quality/validations/table/pandas/tableCustomSQLQuery.py +41 -3
- metadata/data_quality/validations/table/pandas/tableRowCountToBeBetween.py +3 -3
- metadata/data_quality/validations/table/pandas/tableRowCountToEqual.py +3 -3
- metadata/data_quality/validations/table/pandas/tableRowInsertedCountToBeBetween.py +20 -5
- metadata/data_quality/validations/table/sqlalchemy/tableColumnCountToBeBetween.py +3 -3
- metadata/data_quality/validations/table/sqlalchemy/tableColumnCountToEqual.py +3 -3
- metadata/data_quality/validations/table/sqlalchemy/tableColumnNameToExist.py +3 -3
- metadata/data_quality/validations/table/sqlalchemy/tableColumnToMatchSet.py +3 -3
- metadata/data_quality/validations/table/sqlalchemy/tableCustomSQLQuery.py +323 -3
- metadata/data_quality/validations/table/sqlalchemy/tableDiff.py +34 -3
- metadata/data_quality/validations/table/sqlalchemy/tableRowCountToBeBetween.py +3 -3
- metadata/data_quality/validations/table/sqlalchemy/tableRowCountToEqual.py +3 -3
- metadata/data_quality/validations/table/sqlalchemy/tableRowInsertedCountToBeBetween.py +3 -3
- metadata/examples/workflows/airbyte.yaml +8 -2
- metadata/examples/workflows/bigquery.yaml +1 -0
- metadata/examples/workflows/cassandra.yaml +35 -0
- metadata/examples/workflows/cockroach.yaml +24 -0
- metadata/examples/workflows/confluent_cdc.yaml +70 -0
- metadata/examples/workflows/confluent_cdc_cloud.yaml +86 -0
- metadata/examples/workflows/confluent_cdc_local.yaml +120 -0
- metadata/examples/workflows/databricks.yaml +9 -1
- metadata/examples/workflows/databricks_usage.yaml +7 -1
- metadata/examples/workflows/exasol.yaml +25 -0
- metadata/examples/workflows/grafana.yaml +115 -0
- metadata/examples/workflows/kinesisfirehose.yaml +99 -0
- metadata/examples/workflows/metabase.yaml +1 -0
- metadata/examples/workflows/my_confluent_cloud.yaml +49 -0
- metadata/examples/workflows/nifi.yaml +51 -0
- metadata/examples/workflows/opensearch.yaml +20 -0
- metadata/examples/workflows/rest.yaml +6 -2
- metadata/examples/workflows/tableau.yaml +1 -3
- metadata/examples/workflows/unity_catalog.yaml +7 -1
- metadata/examples/workflows/unity_catalog_usage.yaml +7 -1
- metadata/examples/workflows/wherescape.yaml +28 -0
- metadata/generated/antlr/EntityLinkLexer.py +419 -353
- metadata/generated/schema/analytics/__init__.py +1 -1
- metadata/generated/schema/analytics/basic.py +1 -1
- metadata/generated/schema/analytics/reportData.py +1 -1
- metadata/generated/schema/analytics/reportDataType/__init__.py +1 -1
- metadata/generated/schema/analytics/reportDataType/aggregatedCostAnalysisReportData.py +1 -1
- metadata/generated/schema/analytics/reportDataType/entityReportData.py +1 -1
- metadata/generated/schema/analytics/reportDataType/rawCostAnalysisReportData.py +1 -1
- metadata/generated/schema/analytics/reportDataType/webAnalyticEntityViewReportData.py +1 -1
- metadata/generated/schema/analytics/reportDataType/webAnalyticUserActivityReportData.py +1 -1
- metadata/generated/schema/analytics/webAnalyticEvent.py +9 -5
- metadata/generated/schema/analytics/webAnalyticEventData.py +1 -1
- metadata/generated/schema/analytics/webAnalyticEventType/__init__.py +1 -1
- metadata/generated/schema/analytics/webAnalyticEventType/customEvent.py +1 -1
- metadata/generated/schema/analytics/webAnalyticEventType/pageViewEvent.py +1 -1
- metadata/generated/schema/api/__init__.py +1 -1
- metadata/generated/schema/api/addGlossaryToAssetsRequest.py +3 -16
- metadata/generated/schema/api/addTagToAssetsRequest.py +1 -10
- metadata/generated/schema/api/analytics/__init__.py +1 -1
- metadata/generated/schema/api/analytics/createWebAnalyticEvent.py +6 -5
- metadata/generated/schema/api/automations/__init__.py +1 -1
- metadata/generated/schema/api/automations/createWorkflow.py +18 -7
- metadata/generated/schema/api/bulkAssets.py +1 -1
- metadata/generated/schema/api/classification/__init__.py +1 -1
- metadata/generated/schema/api/classification/createClassification.py +63 -7
- metadata/generated/schema/api/classification/createTag.py +37 -6
- metadata/generated/schema/api/classification/createTagWithRecognizers.py +67 -0
- metadata/generated/schema/api/classification/loadTags.py +1 -1
- metadata/generated/schema/api/configuration/__init__.py +3 -0
- metadata/generated/schema/api/configuration/rdfConfiguration.py +47 -0
- metadata/generated/schema/api/createBot.py +5 -5
- metadata/generated/schema/api/createEventPublisherJob.py +1 -1
- metadata/generated/schema/api/createType.py +6 -5
- metadata/generated/schema/api/data/__init__.py +1 -1
- metadata/generated/schema/api/data/createAPICollection.py +4 -4
- metadata/generated/schema/api/data/createAPIEndpoint.py +4 -4
- metadata/generated/schema/api/data/createChart.py +12 -4
- metadata/generated/schema/api/data/createContainer.py +4 -4
- metadata/generated/schema/api/data/createCustomProperty.py +5 -1
- metadata/generated/schema/api/data/createDashboard.py +4 -4
- metadata/generated/schema/api/data/createDashboardDataModel.py +8 -4
- metadata/generated/schema/api/data/createDataContract.py +102 -0
- metadata/generated/schema/api/data/createDatabase.py +4 -4
- metadata/generated/schema/api/data/createDatabaseSchema.py +4 -4
- metadata/generated/schema/api/data/createDirectory.py +105 -0
- metadata/generated/schema/api/data/createEntityProfile.py +36 -0
- metadata/generated/schema/api/data/createFile.py +104 -0
- metadata/generated/schema/api/data/createGlossary.py +4 -4
- metadata/generated/schema/api/data/createGlossaryTerm.py +1 -1
- metadata/generated/schema/api/data/createMetric.py +15 -4
- metadata/generated/schema/api/data/createMlModel.py +4 -4
- metadata/generated/schema/api/data/createPipeline.py +8 -4
- metadata/generated/schema/api/data/createQuery.py +5 -4
- metadata/generated/schema/api/data/createQueryCostRecord.py +35 -0
- metadata/generated/schema/api/data/createSearchIndex.py +4 -4
- metadata/generated/schema/api/data/createSpreadsheet.py +119 -0
- metadata/generated/schema/api/data/createStoredProcedure.py +4 -4
- metadata/generated/schema/api/data/createTable.py +5 -4
- metadata/generated/schema/api/data/createTableProfile.py +1 -1
- metadata/generated/schema/api/data/createTopic.py +5 -4
- metadata/generated/schema/api/data/createWorksheet.py +105 -0
- metadata/generated/schema/api/data/loadGlossary.py +1 -1
- metadata/generated/schema/api/data/restoreEntity.py +1 -1
- metadata/generated/schema/api/data/updateColumn.py +49 -0
- metadata/generated/schema/api/dataInsight/__init__.py +1 -1
- metadata/generated/schema/api/dataInsight/createDataInsightChart.py +5 -4
- metadata/generated/schema/api/dataInsight/custom/__init__.py +1 -1
- metadata/generated/schema/api/dataInsight/custom/createDataInsightCustomChart.py +1 -1
- metadata/generated/schema/api/dataInsight/kpi/__init__.py +1 -1
- metadata/generated/schema/api/dataInsight/kpi/createKpiRequest.py +5 -5
- metadata/generated/schema/api/docStore/__init__.py +1 -1
- metadata/generated/schema/api/docStore/createDocument.py +6 -5
- metadata/generated/schema/api/domains/__init__.py +1 -1
- metadata/generated/schema/api/domains/createDataProduct.py +15 -6
- metadata/generated/schema/api/domains/createDomain.py +13 -2
- metadata/generated/schema/api/entityRelationship/__init__.py +3 -0
- metadata/generated/schema/api/entityRelationship/entityRelationshipDirection.py +12 -0
- metadata/generated/schema/api/entityRelationship/esEntityRelationshipData.py +57 -0
- metadata/generated/schema/api/entityRelationship/relationshipRef.py +32 -0
- metadata/generated/schema/api/entityRelationship/searchEntityRelationshipRequest.py +47 -0
- metadata/generated/schema/api/entityRelationship/searchEntityRelationshipResult.py +47 -0
- metadata/generated/schema/api/entityRelationship/searchSchemaEntityRelationshipResult.py +20 -0
- metadata/generated/schema/api/events/__init__.py +3 -0
- metadata/generated/schema/api/events/createNotificationTemplate.py +61 -0
- metadata/generated/schema/api/events/notificationTemplateValidationRequest.py +30 -0
- metadata/generated/schema/api/events/notificationTemplateValidationResponse.py +36 -0
- metadata/generated/schema/api/feed/__init__.py +1 -1
- metadata/generated/schema/api/feed/closeTask.py +1 -1
- metadata/generated/schema/api/feed/createPost.py +1 -1
- metadata/generated/schema/api/feed/createSuggestion.py +1 -1
- metadata/generated/schema/api/feed/createThread.py +6 -2
- metadata/generated/schema/api/feed/resolveTask.py +1 -1
- metadata/generated/schema/api/feed/threadCount.py +25 -1
- metadata/generated/schema/api/governance/__init__.py +1 -1
- metadata/generated/schema/api/governance/createWorkflowDefinition.py +20 -5
- metadata/generated/schema/api/governance/createWorkflowInstanceState.py +1 -1
- metadata/generated/schema/api/lineage/__init__.py +1 -1
- metadata/generated/schema/api/lineage/addLineage.py +1 -1
- metadata/generated/schema/api/lineage/entityCountLineageRequest.py +66 -0
- metadata/generated/schema/api/lineage/esLineageData.py +88 -0
- metadata/generated/schema/api/lineage/lineageDirection.py +12 -0
- metadata/generated/schema/api/lineage/lineagePaginationInfo.py +57 -0
- metadata/generated/schema/api/lineage/nodeInformation.py +30 -0
- metadata/generated/schema/api/lineage/searchLineageRequest.py +45 -0
- metadata/generated/schema/api/lineage/searchLineageResult.py +46 -0
- metadata/generated/schema/api/mcp/__init__.py +3 -0
- metadata/generated/schema/api/mcp/mcpSearchResponse.py +118 -0
- metadata/generated/schema/api/mcp/mcpToolDefinition.py +54 -0
- metadata/generated/schema/api/openMetadataServerVersion.py +1 -1
- metadata/generated/schema/api/policies/__init__.py +1 -1
- metadata/generated/schema/api/policies/createPolicy.py +6 -5
- metadata/generated/schema/api/rdf/__init__.py +3 -0
- metadata/generated/schema/api/rdf/sparqlQuery.py +57 -0
- metadata/generated/schema/api/rdf/sparqlResponse.py +56 -0
- metadata/generated/schema/api/scim/__init__.py +3 -0
- metadata/generated/schema/api/scim/scimGroup.py +52 -0
- metadata/generated/schema/api/scim/scimPatchOp.py +32 -0
- metadata/generated/schema/api/scim/scimUser.py +83 -0
- metadata/generated/schema/api/search/__init__.py +3 -0
- metadata/generated/schema/api/search/previewSearchRequest.py +47 -0
- metadata/generated/schema/api/services/__init__.py +1 -1
- metadata/generated/schema/api/services/createApiService.py +12 -5
- metadata/generated/schema/api/services/createDashboardService.py +12 -5
- metadata/generated/schema/api/services/createDatabaseService.py +12 -5
- metadata/generated/schema/api/services/createDriveService.py +65 -0
- metadata/generated/schema/api/services/createMessagingService.py +12 -5
- metadata/generated/schema/api/services/createMetadataService.py +13 -5
- metadata/generated/schema/api/services/createMlModelService.py +12 -5
- metadata/generated/schema/api/services/createPipelineService.py +12 -5
- metadata/generated/schema/api/services/createSearchService.py +12 -5
- metadata/generated/schema/api/services/createSecurityService.py +67 -0
- metadata/generated/schema/api/services/createStorageService.py +12 -5
- metadata/generated/schema/api/services/ingestionPipelines/__init__.py +1 -1
- metadata/generated/schema/api/services/ingestionPipelines/createIngestionPipeline.py +28 -5
- metadata/generated/schema/api/setOwner.py +1 -1
- metadata/generated/schema/api/teams/__init__.py +1 -1
- metadata/generated/schema/api/teams/createPersona.py +12 -4
- metadata/generated/schema/api/teams/createRole.py +5 -4
- metadata/generated/schema/api/teams/createTeam.py +8 -1
- metadata/generated/schema/api/teams/createUser.py +11 -1
- metadata/generated/schema/api/tests/__init__.py +1 -1
- metadata/generated/schema/api/tests/createCustomMetric.py +1 -1
- metadata/generated/schema/api/tests/createLogicalTestCases.py +1 -1
- metadata/generated/schema/api/tests/createTestCase.py +13 -6
- metadata/generated/schema/api/tests/createTestCaseResolutionStatus.py +1 -1
- metadata/generated/schema/api/tests/createTestCaseResult.py +1 -1
- metadata/generated/schema/api/tests/createTestDefinition.py +5 -4
- metadata/generated/schema/api/tests/createTestSuite.py +17 -5
- metadata/generated/schema/api/tests/moveGlossaryTermRequest.py +27 -0
- metadata/generated/schema/api/validateGlossaryTagsRequest.py +31 -0
- metadata/generated/schema/api/voteRequest.py +1 -1
- metadata/generated/schema/auth/__init__.py +1 -1
- metadata/generated/schema/auth/basicAuth.py +1 -1
- metadata/generated/schema/auth/basicLoginRequest.py +1 -1
- metadata/generated/schema/auth/changePasswordRequest.py +1 -1
- metadata/generated/schema/auth/createPersonalToken.py +1 -1
- metadata/generated/schema/auth/emailRequest.py +1 -1
- metadata/generated/schema/auth/emailVerificationToken.py +3 -4
- metadata/generated/schema/auth/generateToken.py +1 -1
- metadata/generated/schema/auth/jwtAuth.py +1 -1
- metadata/generated/schema/auth/loginRequest.py +1 -1
- metadata/generated/schema/auth/logoutRequest.py +1 -1
- metadata/generated/schema/auth/passwordResetRequest.py +1 -1
- metadata/generated/schema/auth/passwordResetToken.py +1 -1
- metadata/generated/schema/auth/personalAccessToken.py +1 -1
- metadata/generated/schema/auth/refreshToken.py +1 -1
- metadata/generated/schema/auth/registrationRequest.py +1 -1
- metadata/generated/schema/auth/revokePersonalToken.py +1 -1
- metadata/generated/schema/auth/revokeToken.py +1 -1
- metadata/generated/schema/auth/serviceTokenEnum.py +1 -1
- metadata/generated/schema/auth/ssoAuth.py +1 -1
- metadata/generated/schema/auth/supportToken.py +35 -0
- metadata/generated/schema/auth/tokenRefreshRequest.py +1 -1
- metadata/generated/schema/configuration/__init__.py +1 -1
- metadata/generated/schema/configuration/aiPlatformConfiguration.py +58 -0
- metadata/generated/schema/configuration/appsPrivateConfiguration.py +4 -1
- metadata/generated/schema/configuration/assetCertificationSettings.py +1 -1
- metadata/generated/schema/configuration/authConfig.py +1 -1
- metadata/generated/schema/configuration/authenticationConfiguration.py +49 -13
- metadata/generated/schema/configuration/authorizerConfiguration.py +41 -9
- metadata/generated/schema/configuration/cacheConfiguration.py +137 -0
- metadata/generated/schema/configuration/changeEventConfiguration.py +1 -1
- metadata/generated/schema/configuration/dataQualityConfiguration.py +1 -1
- metadata/generated/schema/configuration/elasticSearchConfiguration.py +80 -1
- metadata/generated/schema/configuration/entityRulesSettings.py +24 -0
- metadata/generated/schema/configuration/eventHandlerConfiguration.py +1 -1
- metadata/generated/schema/configuration/fernetConfiguration.py +1 -1
- metadata/generated/schema/configuration/jwtTokenConfiguration.py +1 -1
- metadata/generated/schema/configuration/kafkaEventConfiguration.py +1 -1
- metadata/generated/schema/configuration/ldapConfiguration.py +1 -1
- metadata/generated/schema/configuration/ldapTrustStoreConfig/__init__.py +1 -1
- metadata/generated/schema/configuration/ldapTrustStoreConfig/customTrustManagerConfig.py +1 -1
- metadata/generated/schema/configuration/ldapTrustStoreConfig/hostNameConfig.py +1 -1
- metadata/generated/schema/configuration/ldapTrustStoreConfig/jvmDefaultConfig.py +1 -1
- metadata/generated/schema/configuration/ldapTrustStoreConfig/trustAllConfig.py +1 -1
- metadata/generated/schema/configuration/ldapTrustStoreConfig/truststoreConfig.py +1 -1
- metadata/generated/schema/configuration/limitsConfiguration.py +1 -1
- metadata/generated/schema/configuration/lineageSettings.py +9 -1
- metadata/generated/schema/configuration/logStorageConfiguration.py +105 -0
- metadata/generated/schema/configuration/loginConfiguration.py +1 -1
- metadata/generated/schema/configuration/logoConfiguration.py +1 -1
- metadata/generated/schema/configuration/openMetadataBaseUrlConfiguration.py +22 -0
- metadata/generated/schema/configuration/opertionalConfiguration.py +24 -0
- metadata/generated/schema/configuration/opsConfig.py +25 -0
- metadata/generated/schema/configuration/pipelineServiceClientConfiguration.py +9 -2
- metadata/generated/schema/configuration/profilerConfiguration.py +2 -1
- metadata/generated/schema/configuration/searchSettings.py +508 -4
- metadata/generated/schema/configuration/securityConfiguration.py +26 -0
- metadata/generated/schema/configuration/slackAppConfiguration.py +1 -1
- metadata/generated/schema/configuration/taskNotificationConfiguration.py +1 -1
- metadata/generated/schema/configuration/testResultNotificationConfiguration.py +1 -1
- metadata/generated/schema/configuration/themeConfiguration.py +1 -1
- metadata/generated/schema/configuration/uiThemePreference.py +1 -1
- metadata/generated/schema/configuration/workflowSettings.py +24 -3
- metadata/generated/schema/dataInsight/__init__.py +1 -1
- metadata/generated/schema/dataInsight/custom/__init__.py +1 -1
- metadata/generated/schema/dataInsight/custom/dataInsightCustomChart.py +5 -1
- metadata/generated/schema/dataInsight/custom/dataInsightCustomChartResult.py +1 -1
- metadata/generated/schema/dataInsight/custom/dataInsightCustomChartResultList.py +1 -1
- metadata/generated/schema/dataInsight/custom/formulaHolder.py +1 -1
- metadata/generated/schema/dataInsight/custom/lineChart.py +19 -1
- metadata/generated/schema/dataInsight/custom/summaryCard.py +1 -1
- metadata/generated/schema/dataInsight/dataInsightChart.py +9 -5
- metadata/generated/schema/dataInsight/dataInsightChartResult.py +1 -1
- metadata/generated/schema/dataInsight/kpi/__init__.py +1 -1
- metadata/generated/schema/dataInsight/kpi/basic.py +1 -1
- metadata/generated/schema/dataInsight/kpi/kpi.py +8 -4
- metadata/generated/schema/dataInsight/type/__init__.py +1 -1
- metadata/generated/schema/dataInsight/type/aggregatedUnusedAssetsCount.py +1 -1
- metadata/generated/schema/dataInsight/type/aggregatedUnusedAssetsSize.py +1 -1
- metadata/generated/schema/dataInsight/type/aggregatedUsedVsUnusedAssetsCount.py +1 -1
- metadata/generated/schema/dataInsight/type/aggregatedUsedVsUnusedAssetsSize.py +1 -1
- metadata/generated/schema/dataInsight/type/dailyActiveUsers.py +1 -1
- metadata/generated/schema/dataInsight/type/mostActiveUsers.py +1 -1
- metadata/generated/schema/dataInsight/type/mostViewedEntities.py +1 -1
- metadata/generated/schema/dataInsight/type/pageViewsByEntities.py +1 -1
- metadata/generated/schema/dataInsight/type/unusedAssets.py +1 -1
- metadata/generated/schema/email/__init__.py +1 -1
- metadata/generated/schema/email/emailRequest.py +1 -1
- metadata/generated/schema/email/emailTemplate.py +1 -1
- metadata/generated/schema/email/emailTemplatePlaceholder.py +1 -1
- metadata/generated/schema/email/smtpSettings.py +6 -5
- metadata/generated/schema/email/templateValidationReponse.py +1 -1
- metadata/generated/schema/entity/__init__.py +1 -1
- metadata/generated/schema/entity/applications/__init__.py +1 -1
- metadata/generated/schema/entity/applications/app.py +40 -7
- metadata/generated/schema/entity/applications/appExtension.py +1 -1
- metadata/generated/schema/entity/applications/appRunRecord.py +46 -6
- metadata/generated/schema/entity/applications/configuration/__init__.py +1 -1
- metadata/generated/schema/entity/applications/configuration/applicationConfig.py +21 -6
- metadata/generated/schema/entity/applications/configuration/external/__init__.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/__init__.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/addCustomProperties.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/addDataProductAction.py +41 -0
- metadata/generated/schema/entity/applications/configuration/external/automator/addDescriptionAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/addDomainAction.py +5 -5
- metadata/generated/schema/entity/applications/configuration/external/automator/addOwnerAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/addTagsAction.py +7 -2
- metadata/generated/schema/entity/applications/configuration/external/automator/addTermsAction.py +48 -0
- metadata/generated/schema/entity/applications/configuration/external/automator/addTestCaseAction.py +74 -0
- metadata/generated/schema/entity/applications/configuration/external/automator/addTierAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/lineagePropagationAction.py +28 -2
- metadata/generated/schema/entity/applications/configuration/external/automator/mlTaggingAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/propagationStopConfig.py +44 -0
- metadata/generated/schema/entity/applications/configuration/external/automator/removeCustomPropertiesAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/removeDataProductAction.py +33 -0
- metadata/generated/schema/entity/applications/configuration/external/automator/removeDescriptionAction.py +10 -2
- metadata/generated/schema/entity/applications/configuration/external/automator/removeDomainAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/removeOwnerAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/removeTagsAction.py +27 -3
- metadata/generated/schema/entity/applications/configuration/external/automator/removeTermsAction.py +59 -0
- metadata/generated/schema/entity/applications/configuration/external/automator/removeTestCaseAction.py +45 -0
- metadata/generated/schema/entity/applications/configuration/external/automator/removeTierAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automatorAppConfig.py +26 -1
- metadata/generated/schema/entity/applications/configuration/external/collateAIAppConfig.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/metadataExporterAppConfig.py +103 -0
- metadata/generated/schema/entity/applications/configuration/external/metadataExporterConnectors/__init__.py +3 -0
- metadata/generated/schema/entity/applications/configuration/external/metadataExporterConnectors/bigQueryConnection.py +90 -0
- metadata/generated/schema/entity/applications/configuration/external/metadataExporterConnectors/databricksConnection.py +101 -0
- metadata/generated/schema/entity/applications/configuration/external/metadataExporterConnectors/redshiftConnection.py +76 -0
- metadata/generated/schema/entity/applications/configuration/external/metadataExporterConnectors/snowflakeConnection.py +115 -0
- metadata/generated/schema/entity/applications/configuration/external/slackAppTokenConfiguration.py +1 -1
- metadata/generated/schema/entity/applications/configuration/internal/__init__.py +1 -1
- metadata/generated/schema/entity/applications/configuration/internal/autoPilotAppConfig.py +48 -0
- metadata/generated/schema/entity/applications/configuration/internal/collateAIQualityAgentAppConfig.py +46 -0
- metadata/generated/schema/entity/applications/configuration/internal/collateAITierAgentAppConfig.py +46 -0
- metadata/generated/schema/entity/applications/configuration/internal/dataInsightsAppConfig.py +111 -2
- metadata/generated/schema/entity/applications/configuration/internal/dataInsightsReportAppConfig.py +1 -1
- metadata/generated/schema/entity/applications/configuration/internal/dataRetentionConfiguration.py +46 -0
- metadata/generated/schema/entity/applications/configuration/internal/helloPipelinesConfiguration.py +18 -0
- metadata/generated/schema/entity/applications/configuration/internal/searchIndexingAppConfig.py +9 -1
- metadata/generated/schema/entity/applications/configuration/private/__init__.py +3 -0
- metadata/generated/schema/entity/applications/configuration/private/external/__init__.py +1 -1
- metadata/generated/schema/entity/applications/configuration/private/external/collateAIAppPrivateConfig.py +3 -22
- metadata/generated/schema/entity/applications/configuration/private/internal/__init__.py +3 -0
- metadata/generated/schema/entity/applications/configuration/private/internal/collateAITierAgentAppPrivateConfig.py +24 -0
- metadata/generated/schema/entity/applications/configuration/private/limits.py +33 -0
- metadata/generated/schema/entity/applications/createAppRequest.py +14 -6
- metadata/generated/schema/entity/applications/jobStatus.py +1 -1
- metadata/generated/schema/entity/applications/liveExecutionContext.py +1 -1
- metadata/generated/schema/entity/applications/marketplace/__init__.py +1 -1
- metadata/generated/schema/entity/applications/marketplace/appMarketPlaceDefinition.py +32 -12
- metadata/generated/schema/entity/applications/marketplace/createAppMarketPlaceDefinitionReq.py +35 -5
- metadata/generated/schema/entity/applications/scheduledExecutionContext.py +1 -1
- metadata/generated/schema/entity/automations/__init__.py +1 -1
- metadata/generated/schema/entity/automations/queryRunnerRequest.py +46 -0
- metadata/generated/schema/entity/automations/response/__init__.py +3 -0
- metadata/generated/schema/entity/automations/response/queryRunnerResponse.py +51 -0
- metadata/generated/schema/entity/automations/testServiceConnection.py +12 -1
- metadata/generated/schema/entity/automations/testSparkEngineConnection.py +31 -0
- metadata/generated/schema/entity/automations/workflow.py +30 -9
- metadata/generated/schema/entity/bot.py +9 -5
- metadata/generated/schema/entity/classification/__init__.py +1 -1
- metadata/generated/schema/entity/classification/classification.py +69 -7
- metadata/generated/schema/entity/classification/tag.py +53 -8
- metadata/generated/schema/entity/data/__init__.py +1 -1
- metadata/generated/schema/entity/data/apiCollection.py +15 -4
- metadata/generated/schema/entity/data/apiEndpoint.py +15 -4
- metadata/generated/schema/entity/data/chart.py +26 -4
- metadata/generated/schema/entity/data/container.py +14 -4
- metadata/generated/schema/entity/data/dashboard.py +13 -4
- metadata/generated/schema/entity/data/dashboardDataModel.py +23 -4
- metadata/generated/schema/entity/data/dataContract.py +403 -0
- metadata/generated/schema/entity/data/database.py +21 -4
- metadata/generated/schema/entity/data/databaseSchema.py +25 -8
- metadata/generated/schema/entity/data/directory.py +178 -0
- metadata/generated/schema/entity/data/file.py +189 -0
- metadata/generated/schema/entity/data/glossary.py +13 -4
- metadata/generated/schema/entity/data/glossaryTerm.py +15 -15
- metadata/generated/schema/entity/data/metric.py +25 -5
- metadata/generated/schema/entity/data/mlmodel.py +13 -4
- metadata/generated/schema/entity/data/pipeline.py +156 -5
- metadata/generated/schema/entity/data/query.py +13 -4
- metadata/generated/schema/entity/data/queryCostRecord.py +47 -0
- metadata/generated/schema/entity/data/queryCostSearchResult.py +56 -0
- metadata/generated/schema/entity/data/report.py +13 -4
- metadata/generated/schema/entity/data/searchIndex.py +15 -4
- metadata/generated/schema/entity/data/spreadsheet.py +215 -0
- metadata/generated/schema/entity/data/storedProcedure.py +22 -4
- metadata/generated/schema/entity/data/table.py +179 -4
- metadata/generated/schema/entity/data/topic.py +13 -4
- metadata/generated/schema/entity/data/worksheet.py +173 -0
- metadata/generated/schema/entity/datacontract/__init__.py +3 -0
- metadata/generated/schema/entity/datacontract/dataContractResult.py +74 -0
- metadata/generated/schema/entity/datacontract/qualityValidation.py +31 -0
- metadata/generated/schema/entity/datacontract/schemaValidation.py +31 -0
- metadata/generated/schema/entity/datacontract/semanticsValidation.py +39 -0
- metadata/generated/schema/entity/datacontract/slaValidation.py +32 -0
- metadata/generated/schema/entity/docStore/__init__.py +1 -1
- metadata/generated/schema/entity/docStore/document.py +9 -5
- metadata/generated/schema/entity/domains/__init__.py +1 -1
- metadata/generated/schema/entity/domains/dataProduct.py +171 -6
- metadata/generated/schema/entity/domains/domain.py +22 -3
- metadata/generated/schema/entity/events/__init__.py +1 -1
- metadata/generated/schema/entity/events/notificationTemplate.py +111 -0
- metadata/generated/schema/entity/events/webhook.py +7 -1
- metadata/generated/schema/entity/feed/__init__.py +1 -1
- metadata/generated/schema/entity/feed/assets.py +1 -1
- metadata/generated/schema/entity/feed/customProperty.py +1 -1
- metadata/generated/schema/entity/feed/description.py +1 -1
- metadata/generated/schema/entity/feed/domain.py +8 -8
- metadata/generated/schema/entity/feed/entityInfo.py +1 -1
- metadata/generated/schema/entity/feed/owner.py +1 -1
- metadata/generated/schema/entity/feed/suggestion.py +1 -1
- metadata/generated/schema/entity/feed/tag.py +1 -1
- metadata/generated/schema/entity/feed/testCaseResult.py +1 -1
- metadata/generated/schema/entity/feed/thread.py +4 -3
- metadata/generated/schema/entity/policies/__init__.py +1 -1
- metadata/generated/schema/entity/policies/accessControl/__init__.py +1 -1
- metadata/generated/schema/entity/policies/accessControl/resourceDescriptor.py +6 -1
- metadata/generated/schema/entity/policies/accessControl/resourcePermission.py +1 -1
- metadata/generated/schema/entity/policies/accessControl/rule.py +1 -1
- metadata/generated/schema/entity/policies/filters.py +1 -1
- metadata/generated/schema/entity/policies/policy.py +8 -4
- metadata/generated/schema/entity/services/__init__.py +1 -1
- metadata/generated/schema/entity/services/apiService.py +18 -3
- metadata/generated/schema/entity/services/connections/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/api/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/api/restConnection.py +15 -1
- metadata/generated/schema/entity/services/connections/common/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/common/sslCertPaths.py +1 -1
- metadata/generated/schema/entity/services/connections/common/sslCertValues.py +1 -1
- metadata/generated/schema/entity/services/connections/common/sslConfig.py +1 -1
- metadata/generated/schema/entity/services/connections/connectionBasicType.py +7 -1
- metadata/generated/schema/entity/services/connections/dashboard/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/customDashboardConnection.py +39 -2
- metadata/generated/schema/entity/services/connections/dashboard/domoDashboardConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/dashboard/grafanaConnection.py +77 -0
- metadata/generated/schema/entity/services/connections/dashboard/hexConnection.py +76 -0
- metadata/generated/schema/entity/services/connections/dashboard/lightdashConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/dashboard/lookerConnection.py +46 -2
- metadata/generated/schema/entity/services/connections/dashboard/metabaseConnection.py +50 -4
- metadata/generated/schema/entity/services/connections/dashboard/microStrategyConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/dashboard/modeConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/dashboard/powerBIConnection.py +50 -1
- metadata/generated/schema/entity/services/connections/dashboard/powerBIReportServerConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/dashboard/powerbi/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/powerbi/azureConfig.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/powerbi/bucketDetails.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/powerbi/gcsConfig.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/powerbi/s3Config.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/qlikCloudConnection.py +51 -2
- metadata/generated/schema/entity/services/connections/dashboard/qlikSenseConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/dashboard/quickSightConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/dashboard/redashConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/dashboard/sigmaConnection.py +40 -1
- metadata/generated/schema/entity/services/connections/dashboard/supersetConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/dashboard/tableauConnection.py +51 -11
- metadata/generated/schema/entity/services/connections/dashboard/thoughtSpotConnection.py +77 -0
- metadata/generated/schema/entity/services/connections/database/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/database/athenaConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/azureSQLConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/bigQueryConnection.py +48 -1
- metadata/generated/schema/entity/services/connections/database/bigTableConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/cassandra/__init__.py +3 -0
- metadata/generated/schema/entity/services/connections/database/cassandra/cloudConfig.py +61 -0
- metadata/generated/schema/entity/services/connections/database/cassandraConnection.py +106 -0
- metadata/generated/schema/entity/services/connections/database/clickhouseConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/cockroachConnection.py +136 -0
- metadata/generated/schema/entity/services/connections/database/common/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/database/common/azureConfig.py +1 -1
- metadata/generated/schema/entity/services/connections/database/common/basicAuth.py +1 -1
- metadata/generated/schema/entity/services/connections/database/common/iamAuthConfig.py +1 -1
- metadata/generated/schema/entity/services/connections/database/common/jwtAuth.py +1 -1
- metadata/generated/schema/entity/services/connections/database/common/noConfigAuthenticationTypes.py +1 -1
- metadata/generated/schema/entity/services/connections/database/couchbaseConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/customDatabaseConnection.py +31 -2
- metadata/generated/schema/entity/services/connections/database/databricks/__init__.py +3 -0
- metadata/generated/schema/entity/services/connections/database/databricks/azureAdSetup.py +37 -0
- metadata/generated/schema/entity/services/connections/database/databricks/databricksOAuth.py +30 -0
- metadata/generated/schema/entity/services/connections/database/databricks/personalAccessToken.py +23 -0
- metadata/generated/schema/entity/services/connections/database/databricksConnection.py +48 -8
- metadata/generated/schema/entity/services/connections/database/datalake/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/database/datalake/azureConfig.py +1 -1
- metadata/generated/schema/entity/services/connections/database/datalake/gcsConfig.py +1 -1
- metadata/generated/schema/entity/services/connections/database/datalake/s3Config.py +1 -1
- metadata/generated/schema/entity/services/connections/database/datalakeConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/db2Connection.py +34 -1
- metadata/generated/schema/entity/services/connections/database/deltaLakeConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/deltalake/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/database/deltalake/metastoreConfig.py +1 -1
- metadata/generated/schema/entity/services/connections/database/deltalake/storageConfig.py +1 -1
- metadata/generated/schema/entity/services/connections/database/domoDatabaseConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/dorisConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/druidConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/dynamoDBConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/epicConnection.py +80 -0
- metadata/generated/schema/entity/services/connections/database/exasolConnection.py +29 -1
- metadata/generated/schema/entity/services/connections/database/glueConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/greenplumConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/hiveConnection.py +41 -1
- metadata/generated/schema/entity/services/connections/database/iceberg/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/database/iceberg/dynamoDbCatalogConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/iceberg/glueCatalogConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/iceberg/hiveCatalogConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/iceberg/icebergCatalog.py +1 -1
- metadata/generated/schema/entity/services/connections/database/iceberg/icebergFileSystem.py +1 -1
- metadata/generated/schema/entity/services/connections/database/iceberg/restCatalogConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/icebergConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/impalaConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/mariaDBConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/mongoDBConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/mssqlConnection.py +54 -1
- metadata/generated/schema/entity/services/connections/database/mysqlConnection.py +37 -1
- metadata/generated/schema/entity/services/connections/database/oracleConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/pinotDBConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/postgresConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/prestoConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/redshiftConnection.py +30 -1
- metadata/generated/schema/entity/services/connections/database/salesforceConnection.py +50 -6
- metadata/generated/schema/entity/services/connections/database/sapErpConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/sapHana/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/database/sapHana/sapHanaHDBConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/sapHana/sapHanaSQLConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/sapHanaConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/sasConnection.py +32 -1
- metadata/generated/schema/entity/services/connections/database/serviceNowConnection.py +90 -0
- metadata/generated/schema/entity/services/connections/database/singleStoreConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/snowflakeConnection.py +62 -1
- metadata/generated/schema/entity/services/connections/database/sqliteConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/ssasConnection.py +47 -0
- metadata/generated/schema/entity/services/connections/database/synapseConnection.py +52 -2
- metadata/generated/schema/entity/services/connections/database/teradataConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/database/timescaleConnection.py +162 -0
- metadata/generated/schema/entity/services/connections/database/trinoConnection.py +34 -1
- metadata/generated/schema/entity/services/connections/database/unityCatalogConnection.py +39 -6
- metadata/generated/schema/entity/services/connections/database/verticaConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/drive/__init__.py +3 -0
- metadata/generated/schema/entity/services/connections/drive/customDriveConnection.py +78 -0
- metadata/generated/schema/entity/services/connections/drive/googleDriveConnection.py +115 -0
- metadata/generated/schema/entity/services/connections/drive/sharePointConnection.py +112 -0
- metadata/generated/schema/entity/services/connections/messaging/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/messaging/customMessagingConnection.py +15 -2
- metadata/generated/schema/entity/services/connections/messaging/kafkaConnection.py +18 -1
- metadata/generated/schema/entity/services/connections/messaging/kinesisConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/messaging/pulsarConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/messaging/redpandaConnection.py +18 -1
- metadata/generated/schema/entity/services/connections/messaging/saslMechanismType.py +1 -1
- metadata/generated/schema/entity/services/connections/metadata/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/metadata/alationConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/metadata/alationSinkConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/metadata/amundsenConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/metadata/atlasConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/metadata/collibraConnection.py +73 -0
- metadata/generated/schema/entity/services/connections/metadata/metadataESConnection.py +26 -1
- metadata/generated/schema/entity/services/connections/metadata/openMetadataConnection.py +26 -2
- metadata/generated/schema/entity/services/connections/mlmodel/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/mlmodel/customMlModelConnection.py +15 -2
- metadata/generated/schema/entity/services/connections/mlmodel/mlflowConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/mlmodel/sageMakerConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/mlmodel/sklearnConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/mlmodel/vertexaiConnection.py +15 -1
- metadata/generated/schema/entity/services/connections/pipeline/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/airbyte/__init__.py +3 -0
- metadata/generated/schema/entity/services/connections/pipeline/airbyte/basicAuth.py +26 -0
- metadata/generated/schema/entity/services/connections/pipeline/airbyte/oauthClientAuth.py +30 -0
- metadata/generated/schema/entity/services/connections/pipeline/airbyteConnection.py +25 -10
- metadata/generated/schema/entity/services/connections/pipeline/airflowConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/pipeline/backendConnection.py +11 -1
- metadata/generated/schema/entity/services/connections/pipeline/customPipelineConnection.py +15 -2
- metadata/generated/schema/entity/services/connections/pipeline/dagsterConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/pipeline/databricksPipelineConnection.py +28 -1
- metadata/generated/schema/entity/services/connections/pipeline/datafactoryConnection.py +15 -1
- metadata/generated/schema/entity/services/connections/pipeline/dbtCloudConnection.py +24 -1
- metadata/generated/schema/entity/services/connections/pipeline/domoPipelineConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/pipeline/fivetranConnection.py +11 -2
- metadata/generated/schema/entity/services/connections/pipeline/flinkConnection.py +15 -1
- metadata/generated/schema/entity/services/connections/pipeline/gluePipelineConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/pipeline/kafkaConnectConnection.py +16 -1
- metadata/generated/schema/entity/services/connections/pipeline/kinesisFirehoseConnection.py +58 -0
- metadata/generated/schema/entity/services/connections/pipeline/matillion/__init__.py +3 -0
- metadata/generated/schema/entity/services/connections/pipeline/matillion/matillionETL.py +39 -0
- metadata/generated/schema/entity/services/connections/pipeline/matillionConnection.py +13 -25
- metadata/generated/schema/entity/services/connections/pipeline/nifi/__init__.py +3 -0
- metadata/generated/schema/entity/services/connections/pipeline/nifi/basicAuth.py +40 -0
- metadata/generated/schema/entity/services/connections/pipeline/nifi/clientCertificateAuth.py +38 -0
- metadata/generated/schema/entity/services/connections/pipeline/nifiConnection.py +12 -2
- metadata/generated/schema/entity/services/connections/pipeline/openLineageConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/pipeline/snowplowConnection.py +93 -0
- metadata/generated/schema/entity/services/connections/pipeline/sparkConnection.py +16 -1
- metadata/generated/schema/entity/services/connections/pipeline/splineConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/pipeline/ssisConnection.py +56 -0
- metadata/generated/schema/entity/services/connections/pipeline/stitchConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/pipeline/wherescapeConnection.py +51 -0
- metadata/generated/schema/entity/services/connections/search/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/search/customSearchConnection.py +15 -2
- metadata/generated/schema/entity/services/connections/search/elasticSearch/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/search/elasticSearch/apiAuth.py +1 -1
- metadata/generated/schema/entity/services/connections/search/elasticSearch/basicAuth.py +1 -1
- metadata/generated/schema/entity/services/connections/search/elasticSearchConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/search/openSearchConnection.py +37 -29
- metadata/generated/schema/entity/services/connections/security/__init__.py +3 -0
- metadata/generated/schema/entity/services/connections/security/ranger/__init__.py +3 -0
- metadata/generated/schema/entity/services/connections/security/ranger/basicAuth.py +26 -0
- metadata/generated/schema/entity/services/connections/security/rangerConnection.py +39 -0
- metadata/generated/schema/entity/services/connections/serviceConnection.py +7 -1
- metadata/generated/schema/entity/services/connections/storage/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/storage/adlsConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/storage/customStorageConnection.py +15 -2
- metadata/generated/schema/entity/services/connections/storage/gcsConnection.py +10 -1
- metadata/generated/schema/entity/services/connections/storage/s3Connection.py +19 -2
- metadata/generated/schema/entity/services/connections/testConnectionDefinition.py +9 -5
- metadata/generated/schema/entity/services/connections/testConnectionResult.py +1 -1
- metadata/generated/schema/entity/services/dashboardService.py +27 -3
- metadata/generated/schema/entity/services/databaseService.py +36 -3
- metadata/generated/schema/entity/services/driveService.py +143 -0
- metadata/generated/schema/entity/services/ingestionPipelines/__init__.py +1 -1
- metadata/generated/schema/entity/services/ingestionPipelines/ingestionPipeline.py +50 -4
- metadata/generated/schema/entity/services/ingestionPipelines/pipelineServiceClientResponse.py +1 -1
- metadata/generated/schema/entity/services/ingestionPipelines/reverseIngestionResponse.py +47 -0
- metadata/generated/schema/entity/services/ingestionPipelines/status.py +1 -1
- metadata/generated/schema/entity/services/messagingService.py +18 -3
- metadata/generated/schema/entity/services/metadataService.py +21 -3
- metadata/generated/schema/entity/services/mlmodelService.py +18 -3
- metadata/generated/schema/entity/services/pipelineService.py +30 -3
- metadata/generated/schema/entity/services/searchService.py +19 -4
- metadata/generated/schema/entity/services/securityService.py +135 -0
- metadata/generated/schema/entity/services/serviceType.py +3 -1
- metadata/generated/schema/entity/services/storageService.py +18 -3
- metadata/generated/schema/entity/teams/__init__.py +1 -1
- metadata/generated/schema/entity/teams/persona.py +20 -5
- metadata/generated/schema/entity/teams/role.py +9 -5
- metadata/generated/schema/entity/teams/team.py +12 -1
- metadata/generated/schema/entity/teams/teamHierarchy.py +1 -1
- metadata/generated/schema/entity/teams/user.py +39 -3
- metadata/generated/schema/entity/type.py +9 -11
- metadata/generated/schema/entity/utils/__init__.py +1 -1
- metadata/generated/schema/entity/utils/entitiesCount.py +1 -1
- metadata/generated/schema/entity/utils/servicesCount.py +1 -1
- metadata/generated/schema/entity/utils/supersetApiConnection.py +1 -1
- metadata/generated/schema/events/__init__.py +1 -1
- metadata/generated/schema/events/alertMetrics.py +1 -1
- metadata/generated/schema/events/api/__init__.py +1 -1
- metadata/generated/schema/events/api/createEventSubscription.py +13 -4
- metadata/generated/schema/events/api/eventSubscriptionDiagnosticInfo.py +1 -1
- metadata/generated/schema/events/api/eventsRecord.py +1 -1
- metadata/generated/schema/events/api/testEventSubscriptionDestination.py +1 -1
- metadata/generated/schema/events/api/typedEvent.py +1 -1
- metadata/generated/schema/events/emailAlertConfig.py +1 -1
- metadata/generated/schema/events/eventFilterRule.py +1 -1
- metadata/generated/schema/events/eventSubscription.py +32 -6
- metadata/generated/schema/events/eventSubscriptionOffset.py +1 -1
- metadata/generated/schema/events/failedEvent.py +1 -1
- metadata/generated/schema/events/failedEventResponse.py +1 -1
- metadata/generated/schema/events/filterResourceDescriptor.py +1 -1
- metadata/generated/schema/events/statusContext.py +1 -1
- metadata/generated/schema/events/subscriptionResourceDescriptor.py +2 -1
- metadata/generated/schema/events/subscriptionStatus.py +1 -1
- metadata/generated/schema/events/testDestinationStatus.py +1 -1
- metadata/generated/schema/governance/workflows/__init__.py +1 -1
- metadata/generated/schema/governance/workflows/elements/__init__.py +1 -1
- metadata/generated/schema/governance/workflows/elements/edge.py +2 -2
- metadata/generated/schema/governance/workflows/elements/nodeSubType.py +7 -1
- metadata/generated/schema/governance/workflows/elements/nodeType.py +2 -1
- metadata/generated/schema/governance/workflows/elements/nodes/__init__.py +1 -1
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/__init__.py +1 -1
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/checkEntityAttributesTask.py +28 -8
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/createAndRunIngestionPipelineTask.py +86 -0
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/dataCompletenessTask.py +124 -0
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/rollbackEntityTask.py +54 -0
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/runAppTask.py +73 -0
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/setEntityAttributeTask.py +66 -0
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/setEntityCertificationTask.py +27 -6
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/setGlossaryTermStatusTask.py +29 -9
- metadata/generated/schema/governance/workflows/elements/nodes/endEvent/__init__.py +1 -1
- metadata/generated/schema/governance/workflows/elements/nodes/endEvent/endEvent.py +9 -4
- metadata/generated/schema/governance/workflows/elements/nodes/gateway/__init__.py +3 -0
- metadata/generated/schema/governance/workflows/elements/nodes/gateway/parallelGateway.py +30 -0
- metadata/generated/schema/governance/workflows/elements/nodes/startEvent/__init__.py +1 -1
- metadata/generated/schema/governance/workflows/elements/nodes/startEvent/startEvent.py +9 -4
- metadata/generated/schema/governance/workflows/elements/nodes/userTask/__init__.py +1 -1
- metadata/generated/schema/governance/workflows/elements/nodes/userTask/userApprovalTask.py +45 -7
- metadata/generated/schema/governance/workflows/elements/triggers/__init__.py +1 -1
- metadata/generated/schema/governance/workflows/elements/triggers/eventBasedEntityTrigger.py +103 -8
- metadata/generated/schema/governance/workflows/elements/triggers/noOpTrigger.py +20 -0
- metadata/generated/schema/governance/workflows/elements/triggers/periodicBatchEntityTrigger.py +94 -8
- metadata/generated/schema/governance/workflows/workflowDefinition.py +34 -7
- metadata/generated/schema/governance/workflows/workflowInstance.py +11 -8
- metadata/generated/schema/governance/workflows/workflowInstanceState.py +8 -8
- metadata/generated/schema/jobs/__init__.py +1 -1
- metadata/generated/schema/jobs/backgroundJob.py +12 -3
- metadata/generated/schema/jobs/enumCleanupArgs.py +1 -1
- metadata/generated/schema/metadataIngestion/__init__.py +1 -1
- metadata/generated/schema/metadataIngestion/apiServiceMetadataPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/application.py +8 -1
- metadata/generated/schema/metadataIngestion/applicationPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/dashboardServiceMetadataPipeline.py +4 -4
- metadata/generated/schema/metadataIngestion/dataInsightPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/databaseServiceAutoClassificationPipeline.py +3 -3
- metadata/generated/schema/metadataIngestion/databaseServiceMetadataPipeline.py +20 -4
- metadata/generated/schema/metadataIngestion/databaseServiceProfilerPipeline.py +29 -3
- metadata/generated/schema/metadataIngestion/databaseServiceQueryLineagePipeline.py +42 -2
- metadata/generated/schema/metadataIngestion/databaseServiceQueryUsagePipeline.py +9 -1
- metadata/generated/schema/metadataIngestion/dbtPipeline.py +16 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/__init__.py +1 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/dbtAzureConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/dbtBucketDetails.py +1 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/dbtCloudConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/dbtGCSConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/dbtHttpConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/dbtLocalConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/dbtS3Config.py +1 -1
- metadata/generated/schema/metadataIngestion/driveServiceMetadataPipeline.py +165 -0
- metadata/generated/schema/metadataIngestion/engine/__init__.py +3 -0
- metadata/generated/schema/metadataIngestion/engine/nativeEngineConfig.py +23 -0
- metadata/generated/schema/metadataIngestion/engine/sparkEngineConfig.py +43 -0
- metadata/generated/schema/metadataIngestion/messagingServiceMetadataPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/metadataToElasticSearchPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/mlmodelServiceMetadataPipeline.py +29 -2
- metadata/generated/schema/metadataIngestion/pipelineServiceMetadataPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/reverseIngestionPipeline.py +71 -0
- metadata/generated/schema/metadataIngestion/reverseingestionconfig/__init__.py +3 -0
- metadata/generated/schema/metadataIngestion/reverseingestionconfig/descriptionConfig.py +24 -0
- metadata/generated/schema/metadataIngestion/reverseingestionconfig/ownerConfig.py +28 -0
- metadata/generated/schema/metadataIngestion/reverseingestionconfig/tagsConfig.py +28 -0
- metadata/generated/schema/metadataIngestion/searchServiceMetadataPipeline.py +2 -2
- metadata/generated/schema/metadataIngestion/securityServiceMetadataPipeline.py +27 -0
- metadata/generated/schema/metadataIngestion/storage/__init__.py +1 -1
- metadata/generated/schema/metadataIngestion/storage/containerMetadataConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/storage/manifestMetadataConfig.py +17 -1
- metadata/generated/schema/metadataIngestion/storage/storageBucketDetails.py +1 -1
- metadata/generated/schema/metadataIngestion/storage/storageMetadataADLSConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/storage/storageMetadataGCSConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/storage/storageMetadataHttpConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/storage/storageMetadataLocalConfig.py +1 -1
- metadata/generated/schema/metadataIngestion/storage/storageMetadataS3Config.py +1 -1
- metadata/generated/schema/metadataIngestion/storageServiceMetadataPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/testSuitePipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/workflow.py +19 -1
- metadata/generated/schema/monitoring/__init__.py +1 -1
- metadata/generated/schema/monitoring/eventMonitorProvider.py +1 -1
- metadata/generated/schema/scim/__init__.py +3 -0
- metadata/generated/schema/scim/scimConfiguration.py +32 -0
- metadata/generated/schema/search/__init__.py +3 -0
- metadata/generated/schema/search/aggregationRequest.py +80 -0
- metadata/generated/schema/search/searchRequest.py +120 -0
- metadata/generated/schema/security/__init__.py +1 -1
- metadata/generated/schema/security/client/__init__.py +1 -1
- metadata/generated/schema/security/client/auth0SSOClientConfig.py +1 -1
- metadata/generated/schema/security/client/azureSSOClientConfig.py +1 -1
- metadata/generated/schema/security/client/customOidcSSOClientConfig.py +1 -1
- metadata/generated/schema/security/client/googleSSOClientConfig.py +1 -1
- metadata/generated/schema/security/client/oidcClientConfig.py +22 -10
- metadata/generated/schema/security/client/oktaSSOClientConfig.py +1 -1
- metadata/generated/schema/security/client/openMetadataJWTClientConfig.py +1 -1
- metadata/generated/schema/security/client/samlSSOClientConfig.py +6 -6
- metadata/generated/schema/security/credentials/__init__.py +1 -1
- metadata/generated/schema/security/credentials/accessTokenAuth.py +1 -1
- metadata/generated/schema/security/credentials/apiAccessTokenAuth.py +1 -1
- metadata/generated/schema/security/credentials/awsCredentials.py +1 -1
- metadata/generated/schema/security/credentials/azureCredentials.py +1 -1
- metadata/generated/schema/security/credentials/basicAuth.py +1 -1
- metadata/generated/schema/security/credentials/bitbucketCredentials.py +10 -2
- metadata/generated/schema/security/credentials/gcpCredentials.py +42 -22
- metadata/generated/schema/security/credentials/gcpExternalAccount.py +2 -2
- metadata/generated/schema/security/credentials/gcpValues.py +1 -1
- metadata/generated/schema/security/credentials/gitCredentials.py +1 -1
- metadata/generated/schema/security/credentials/githubCredentials.py +10 -2
- metadata/generated/schema/security/credentials/gitlabCredentials.py +10 -2
- metadata/generated/schema/security/credentials/kubernetesCredentials.py +40 -0
- metadata/generated/schema/security/sasl/__init__.py +1 -1
- metadata/generated/schema/security/sasl/saslClientConfig.py +1 -1
- metadata/generated/schema/security/secrets/__init__.py +1 -1
- metadata/generated/schema/security/secrets/secretsManagerClientLoader.py +1 -1
- metadata/generated/schema/security/secrets/secretsManagerConfiguration.py +1 -1
- metadata/generated/schema/security/secrets/secretsManagerProvider.py +2 -1
- metadata/generated/schema/security/securityConfiguration.py +1 -1
- metadata/generated/schema/security/ssl/__init__.py +1 -1
- metadata/generated/schema/security/ssl/validateSSLClientConfig.py +1 -1
- metadata/generated/schema/security/ssl/verifySSLConfig.py +1 -1
- metadata/generated/schema/settings/__init__.py +1 -1
- metadata/generated/schema/settings/settings.py +11 -1
- metadata/generated/schema/system/__init__.py +1 -1
- metadata/generated/schema/system/entityError.py +1 -1
- metadata/generated/schema/system/eventPublisherJob.py +62 -11
- metadata/generated/schema/system/indexingError.py +2 -2
- metadata/generated/schema/system/limitsResponse.py +1 -1
- metadata/generated/schema/system/securityValidationResponse.py +62 -0
- metadata/generated/schema/system/ui/__init__.py +1 -1
- metadata/generated/schema/system/ui/knowledgePanel.py +1 -1
- metadata/generated/schema/system/ui/navigationItem.py +38 -0
- metadata/generated/schema/system/ui/page.py +34 -18
- metadata/generated/schema/system/ui/tab.py +36 -0
- metadata/generated/schema/system/ui/uiCustomization.py +52 -0
- metadata/generated/schema/system/validationResponse.py +8 -1
- metadata/generated/schema/tests/__init__.py +1 -1
- metadata/generated/schema/tests/assigned.py +1 -1
- metadata/generated/schema/tests/basic.py +1 -1
- metadata/generated/schema/tests/customMetric.py +1 -1
- metadata/generated/schema/tests/dataQualityReport.py +1 -1
- metadata/generated/schema/tests/resolved.py +1 -1
- metadata/generated/schema/tests/testCase.py +36 -5
- metadata/generated/schema/tests/testCaseResolutionStatus.py +1 -1
- metadata/generated/schema/tests/testDefinition.py +9 -5
- metadata/generated/schema/tests/testSuite.py +26 -4
- metadata/generated/schema/type/__init__.py +1 -1
- metadata/generated/schema/type/apiSchema.py +1 -1
- metadata/generated/schema/type/assetCertification.py +1 -1
- metadata/generated/schema/type/auditLog.py +1 -1
- metadata/generated/schema/type/basic.py +61 -3
- metadata/generated/schema/type/bulkOperationResult.py +1 -1
- metadata/generated/schema/type/changeEvent.py +9 -7
- metadata/generated/schema/type/changeEventType.py +1 -1
- metadata/generated/schema/type/changeSummaryMap.py +37 -0
- metadata/generated/schema/type/collectionDescriptor.py +1 -1
- metadata/generated/schema/type/contextRecognizer.py +50 -0
- metadata/generated/schema/type/contractExecutionStatus.py +16 -0
- metadata/generated/schema/type/csvDocumentation.py +1 -1
- metadata/generated/schema/type/csvErrorType.py +1 -1
- metadata/generated/schema/type/csvFile.py +1 -1
- metadata/generated/schema/type/csvImportResult.py +1 -1
- metadata/generated/schema/type/customProperties/__init__.py +1 -1
- metadata/generated/schema/type/customProperties/complexTypes.py +1 -1
- metadata/generated/schema/type/customProperties/enumConfig.py +1 -1
- metadata/generated/schema/type/customProperties/tableConfig.py +1 -1
- metadata/generated/schema/type/customProperty.py +1 -1
- metadata/generated/schema/type/customRecognizer.py +39 -0
- metadata/generated/schema/type/dailyCount.py +1 -1
- metadata/generated/schema/type/databaseConnectionConfig.py +1 -1
- metadata/generated/schema/type/denyListRecognizer.py +37 -0
- metadata/generated/schema/type/entityHierarchy.py +7 -6
- metadata/generated/schema/type/entityHistory.py +35 -1
- metadata/generated/schema/type/entityLineage.py +27 -1
- metadata/generated/schema/type/entityProfile.py +45 -0
- metadata/generated/schema/type/entityReference.py +1 -1
- metadata/generated/schema/type/entityReferenceList.py +1 -1
- metadata/generated/schema/type/{entityRelationship.py → entityRelationship/__init__.py} +2 -2
- metadata/generated/schema/type/entityRelationship/nodeInformation.py +25 -0
- metadata/generated/schema/type/entityUsage.py +1 -1
- metadata/generated/schema/type/filterPattern.py +3 -3
- metadata/generated/schema/type/function.py +1 -1
- metadata/generated/schema/type/include.py +1 -1
- metadata/generated/schema/type/jdbcConnection.py +1 -1
- metadata/generated/schema/type/layerPaging.py +24 -0
- metadata/generated/schema/type/lifeCycle.py +1 -1
- metadata/generated/schema/type/paging.py +1 -1
- metadata/generated/schema/type/patternRecognizer.py +39 -0
- metadata/generated/schema/type/personaPreferences.py +48 -0
- metadata/generated/schema/type/piiEntity.py +48 -0
- metadata/generated/schema/type/predefinedRecognizer.py +85 -0
- metadata/generated/schema/type/profile.py +1 -1
- metadata/generated/schema/type/queryParserData.py +4 -1
- metadata/generated/schema/type/reaction.py +1 -1
- metadata/generated/schema/type/recognizer.py +159 -0
- metadata/generated/schema/type/recognizerFeedback.py +120 -0
- metadata/generated/schema/type/recognizers/__init__.py +3 -0
- metadata/generated/schema/type/recognizers/patterns.py +29 -0
- metadata/generated/schema/type/recognizers/regexFlags.py +29 -0
- metadata/generated/schema/type/schedule.py +1 -1
- metadata/generated/schema/type/schema.py +1 -1
- metadata/generated/schema/type/status.py +16 -0
- metadata/generated/schema/type/tableQuery.py +4 -1
- metadata/generated/schema/type/tableUsageCount.py +18 -1
- metadata/generated/schema/type/tagLabel.py +9 -1
- metadata/generated/schema/type/usageDetails.py +1 -1
- metadata/generated/schema/type/usageRequest.py +1 -1
- metadata/generated/schema/type/votes.py +1 -1
- metadata/great_expectations/action.py +158 -21
- metadata/great_expectations/action1xx.py +471 -0
- metadata/great_expectations/utils/ometa_config_handler.py +2 -2
- metadata/ingestion/api/closeable.py +3 -3
- metadata/ingestion/api/common.py +3 -3
- metadata/ingestion/api/delete.py +3 -3
- metadata/ingestion/api/models.py +3 -3
- metadata/ingestion/api/parser.py +41 -12
- metadata/ingestion/api/status.py +9 -4
- metadata/ingestion/api/step.py +6 -4
- metadata/ingestion/api/steps.py +3 -3
- metadata/ingestion/api/topology_runner.py +54 -32
- metadata/ingestion/bulksink/metadata_usage.py +49 -11
- metadata/ingestion/connections/builders.py +22 -8
- metadata/ingestion/connections/connection.py +80 -0
- metadata/ingestion/connections/headers.py +3 -3
- metadata/ingestion/connections/query_logger.py +115 -0
- metadata/ingestion/connections/secrets.py +3 -3
- metadata/ingestion/connections/session.py +3 -3
- metadata/ingestion/connections/test_connections.py +4 -4
- metadata/ingestion/lineage/masker.py +45 -27
- metadata/ingestion/lineage/models.py +30 -3
- metadata/ingestion/lineage/parser.py +89 -36
- metadata/ingestion/lineage/sql_lineage.py +459 -112
- metadata/ingestion/models/custom_basemodel_validation.py +180 -49
- metadata/ingestion/models/custom_properties.py +6 -6
- metadata/ingestion/models/custom_pydantic.py +83 -27
- metadata/ingestion/models/custom_types.py +2 -2
- metadata/ingestion/models/data_insight.py +3 -3
- metadata/ingestion/models/delete_entity.py +3 -3
- metadata/ingestion/models/encoders.py +3 -3
- metadata/ingestion/models/entity_interface.py +3 -3
- metadata/ingestion/models/lf_tags_model.py +3 -3
- metadata/ingestion/models/life_cycle.py +3 -3
- metadata/ingestion/models/ometa_classification.py +3 -3
- metadata/ingestion/models/ometa_lineage.py +8 -4
- metadata/ingestion/models/ometa_topic_data.py +3 -3
- metadata/ingestion/models/patch_request.py +119 -11
- metadata/ingestion/models/pipeline_status.py +3 -3
- metadata/ingestion/models/profile_data.py +3 -3
- metadata/ingestion/models/search_index_data.py +3 -3
- metadata/ingestion/models/table_metadata.py +3 -3
- metadata/ingestion/models/tests_data.py +3 -3
- metadata/ingestion/models/topology.py +3 -3
- metadata/ingestion/models/user.py +3 -3
- metadata/ingestion/ometa/auth_provider.py +3 -3
- metadata/ingestion/ometa/client.py +25 -5
- metadata/ingestion/ometa/client_utils.py +3 -3
- metadata/ingestion/ometa/credentials.py +3 -3
- metadata/ingestion/ometa/mixins/csv_mixin.py +180 -0
- metadata/ingestion/ometa/mixins/custom_property_mixin.py +3 -3
- metadata/ingestion/ometa/mixins/dashboard_mixin.py +3 -3
- metadata/ingestion/ometa/mixins/data_contract_mixin.py +205 -0
- metadata/ingestion/ometa/mixins/data_insight_mixin.py +3 -3
- metadata/ingestion/ometa/mixins/domain_mixin.py +128 -0
- metadata/ingestion/ometa/mixins/es_mixin.py +134 -22
- metadata/ingestion/ometa/mixins/ingestion_pipeline_mixin.py +29 -3
- metadata/ingestion/ometa/mixins/lineage_mixin.py +39 -12
- metadata/ingestion/ometa/mixins/logs_mixin.py +316 -0
- metadata/ingestion/ometa/mixins/mlmodel_mixin.py +3 -3
- metadata/ingestion/ometa/mixins/patch_mixin.py +165 -65
- metadata/ingestion/ometa/mixins/patch_mixin_utils.py +2 -2
- metadata/ingestion/ometa/mixins/pipeline_mixin.py +21 -3
- metadata/ingestion/ometa/mixins/profile_mixin.py +77 -0
- metadata/ingestion/ometa/mixins/query_mixin.py +48 -3
- metadata/ingestion/ometa/mixins/role_policy_mixin.py +2 -2
- metadata/ingestion/ometa/mixins/search_index_mixin.py +3 -3
- metadata/ingestion/ometa/mixins/server_mixin.py +3 -3
- metadata/ingestion/ometa/mixins/service_mixin.py +3 -3
- metadata/ingestion/ometa/mixins/suggestions_mixin.py +3 -3
- metadata/ingestion/ometa/mixins/table_mixin.py +34 -4
- metadata/ingestion/ometa/mixins/tag_glossary_mixin.py +68 -0
- metadata/ingestion/ometa/mixins/tests_mixin.py +8 -8
- metadata/ingestion/ometa/mixins/topic_mixin.py +3 -3
- metadata/ingestion/ometa/mixins/user_mixin.py +60 -19
- metadata/ingestion/ometa/mixins/version_mixin.py +3 -3
- metadata/ingestion/ometa/models.py +3 -3
- metadata/ingestion/ometa/ometa_api.py +94 -10
- metadata/ingestion/ometa/routes.py +59 -3
- metadata/ingestion/ometa/ttl_cache.py +3 -3
- metadata/ingestion/ometa/utils.py +48 -4
- metadata/ingestion/processor/query_parser.py +4 -3
- metadata/ingestion/sink/file.py +3 -3
- metadata/ingestion/sink/metadata_rest.py +139 -16
- metadata/ingestion/source/api/api_service.py +4 -9
- metadata/ingestion/source/api/rest/connection.py +22 -12
- metadata/ingestion/source/api/rest/metadata.py +358 -67
- metadata/ingestion/source/api/rest/models.py +5 -4
- metadata/ingestion/source/api/rest/parser.py +103 -0
- metadata/ingestion/source/connections.py +83 -17
- metadata/ingestion/source/connections_utils.py +32 -0
- metadata/ingestion/source/dashboard/dashboard_service.py +77 -25
- metadata/ingestion/source/dashboard/domodashboard/connection.py +3 -3
- metadata/ingestion/source/dashboard/domodashboard/metadata.py +35 -19
- metadata/ingestion/source/dashboard/grafana/__init__.py +10 -0
- metadata/ingestion/source/dashboard/grafana/client.py +223 -0
- metadata/ingestion/source/dashboard/grafana/connection.py +66 -0
- metadata/ingestion/source/dashboard/grafana/metadata.py +476 -0
- metadata/ingestion/source/dashboard/grafana/models.py +159 -0
- metadata/ingestion/source/dashboard/grafana/service_spec.py +17 -0
- metadata/ingestion/source/dashboard/hex/client.py +104 -0
- metadata/ingestion/source/dashboard/hex/connection.py +60 -0
- metadata/ingestion/source/dashboard/hex/metadata.py +363 -0
- metadata/ingestion/source/dashboard/hex/models.py +143 -0
- metadata/ingestion/source/dashboard/hex/query_fetcher.py +503 -0
- metadata/ingestion/source/dashboard/hex/service_spec.py +18 -0
- metadata/ingestion/source/dashboard/hex/warehouse_queries.py +225 -0
- metadata/ingestion/source/dashboard/lightdash/client.py +86 -12
- metadata/ingestion/source/dashboard/lightdash/connection.py +3 -3
- metadata/ingestion/source/dashboard/lightdash/metadata.py +49 -28
- metadata/ingestion/source/dashboard/lightdash/models.py +11 -0
- metadata/ingestion/source/dashboard/looker/bulk_parser.py +34 -17
- metadata/ingestion/source/dashboard/looker/columns.py +3 -3
- metadata/ingestion/source/dashboard/looker/connection.py +3 -3
- metadata/ingestion/source/dashboard/looker/links.py +3 -3
- metadata/ingestion/source/dashboard/looker/metadata.py +602 -100
- metadata/ingestion/source/dashboard/looker/models.py +8 -3
- metadata/ingestion/source/dashboard/looker/parser.py +6 -3
- metadata/ingestion/source/dashboard/looker/utils.py +21 -6
- metadata/ingestion/source/dashboard/metabase/client.py +99 -18
- metadata/ingestion/source/dashboard/metabase/connection.py +3 -3
- metadata/ingestion/source/dashboard/metabase/metadata.py +181 -48
- metadata/ingestion/source/dashboard/metabase/models.py +54 -5
- metadata/ingestion/source/dashboard/microstrategy/client.py +23 -2
- metadata/ingestion/source/dashboard/microstrategy/connection.py +3 -3
- metadata/ingestion/source/dashboard/microstrategy/helpers.py +67 -0
- metadata/ingestion/source/dashboard/microstrategy/metadata.py +197 -6
- metadata/ingestion/source/dashboard/microstrategy/models.py +16 -4
- metadata/ingestion/source/dashboard/mode/client.py +5 -4
- metadata/ingestion/source/dashboard/mode/connection.py +3 -3
- metadata/ingestion/source/dashboard/mode/metadata.py +78 -25
- metadata/ingestion/source/dashboard/powerbi/client.py +322 -65
- metadata/ingestion/source/dashboard/powerbi/connection.py +3 -3
- metadata/ingestion/source/dashboard/powerbi/databricks_parser.py +108 -0
- metadata/ingestion/source/dashboard/powerbi/file_client.py +3 -3
- metadata/ingestion/source/dashboard/powerbi/metadata.py +1207 -286
- metadata/ingestion/source/dashboard/powerbi/models.py +102 -2
- metadata/ingestion/source/dashboard/qlikcloud/client.py +63 -8
- metadata/ingestion/source/dashboard/qlikcloud/connection.py +3 -3
- metadata/ingestion/source/dashboard/qlikcloud/constants.py +9 -3
- metadata/ingestion/source/dashboard/qlikcloud/metadata.py +96 -21
- metadata/ingestion/source/dashboard/qlikcloud/models.py +56 -4
- metadata/ingestion/source/dashboard/qliksense/client.py +3 -3
- metadata/ingestion/source/dashboard/qliksense/connection.py +3 -3
- metadata/ingestion/source/dashboard/qliksense/constants.py +3 -3
- metadata/ingestion/source/dashboard/qliksense/metadata.py +77 -28
- metadata/ingestion/source/dashboard/qliksense/models.py +3 -3
- metadata/ingestion/source/dashboard/quicksight/connection.py +3 -3
- metadata/ingestion/source/dashboard/quicksight/metadata.py +124 -42
- metadata/ingestion/source/dashboard/quicksight/models.py +3 -3
- metadata/ingestion/source/dashboard/redash/client.py +10 -14
- metadata/ingestion/source/dashboard/redash/connection.py +3 -3
- metadata/ingestion/source/dashboard/redash/metadata.py +80 -25
- metadata/ingestion/source/dashboard/sigma/client.py +72 -9
- metadata/ingestion/source/dashboard/sigma/connection.py +7 -4
- metadata/ingestion/source/dashboard/sigma/metadata.py +76 -26
- metadata/ingestion/source/dashboard/sigma/models.py +8 -2
- metadata/ingestion/source/dashboard/superset/api_source.py +74 -31
- metadata/ingestion/source/dashboard/superset/client.py +6 -5
- metadata/ingestion/source/dashboard/superset/connection.py +15 -17
- metadata/ingestion/source/dashboard/superset/db_source.py +71 -27
- metadata/ingestion/source/dashboard/superset/metadata.py +3 -3
- metadata/ingestion/source/dashboard/superset/mixin.py +224 -59
- metadata/ingestion/source/dashboard/superset/models.py +6 -3
- metadata/ingestion/source/dashboard/superset/queries.py +10 -7
- metadata/ingestion/source/dashboard/superset/utils.py +36 -0
- metadata/ingestion/source/dashboard/tableau/__init__.py +3 -3
- metadata/ingestion/source/dashboard/tableau/client.py +275 -109
- metadata/ingestion/source/dashboard/tableau/connection.py +72 -54
- metadata/ingestion/source/dashboard/tableau/metadata.py +472 -117
- metadata/ingestion/source/dashboard/tableau/models.py +45 -18
- metadata/ingestion/source/dashboard/tableau/queries.py +17 -3
- metadata/ingestion/source/database/athena/client.py +3 -3
- metadata/ingestion/source/database/athena/connection.py +3 -3
- metadata/ingestion/source/database/athena/lineage.py +3 -3
- metadata/ingestion/source/database/athena/metadata.py +52 -3
- metadata/ingestion/source/database/athena/models.py +3 -3
- metadata/ingestion/source/database/athena/query_parser.py +3 -3
- metadata/ingestion/source/database/athena/usage.py +3 -3
- metadata/ingestion/source/database/athena/utils.py +66 -9
- metadata/ingestion/source/database/azuresql/connection.py +3 -3
- metadata/ingestion/source/database/azuresql/lineage.py +3 -3
- metadata/ingestion/source/database/azuresql/metadata.py +3 -3
- metadata/ingestion/source/database/azuresql/queries.py +3 -3
- metadata/ingestion/source/database/azuresql/query_parser.py +3 -3
- metadata/ingestion/source/database/azuresql/usage.py +3 -3
- metadata/ingestion/source/database/bigquery/connection.py +36 -7
- metadata/ingestion/source/database/bigquery/helper.py +31 -19
- metadata/ingestion/source/database/bigquery/incremental_table_processor.py +3 -3
- metadata/ingestion/source/database/bigquery/lineage.py +8 -13
- metadata/ingestion/source/database/bigquery/metadata.py +331 -66
- metadata/ingestion/source/database/bigquery/models.py +3 -3
- metadata/ingestion/source/database/bigquery/queries.py +26 -8
- metadata/ingestion/source/database/bigquery/query_parser.py +4 -3
- metadata/ingestion/source/database/bigquery/service_spec.py +4 -4
- metadata/ingestion/source/database/bigquery/usage.py +3 -3
- metadata/ingestion/source/database/bigtable/client.py +2 -2
- metadata/ingestion/source/database/bigtable/connection.py +2 -2
- metadata/ingestion/source/database/bigtable/metadata.py +13 -5
- metadata/ingestion/source/database/bigtable/models.py +2 -2
- metadata/ingestion/source/database/cassandra/connection.py +138 -0
- metadata/ingestion/source/database/cassandra/helpers.py +100 -0
- metadata/ingestion/source/database/cassandra/metadata.py +138 -0
- metadata/ingestion/source/database/cassandra/queries.py +45 -0
- metadata/ingestion/source/database/cassandra/service_spec.py +10 -0
- metadata/ingestion/source/database/clickhouse/connection.py +3 -3
- metadata/ingestion/source/database/clickhouse/lineage.py +3 -3
- metadata/ingestion/source/database/clickhouse/metadata.py +3 -36
- metadata/ingestion/source/database/clickhouse/queries.py +3 -3
- metadata/ingestion/source/database/clickhouse/query_parser.py +3 -3
- metadata/ingestion/source/database/clickhouse/usage.py +3 -3
- metadata/ingestion/source/database/clickhouse/utils.py +3 -3
- metadata/ingestion/source/database/cockroach/connection.py +71 -0
- metadata/ingestion/source/database/cockroach/metadata.py +206 -0
- metadata/ingestion/source/database/cockroach/queries.py +90 -0
- metadata/ingestion/source/database/cockroach/service_spec.py +4 -0
- metadata/ingestion/source/database/column_helpers.py +3 -3
- metadata/ingestion/source/database/column_type_parser.py +11 -6
- metadata/ingestion/source/database/common/data_diff/databricks_base.py +42 -0
- metadata/ingestion/source/database/common_db_source.py +63 -59
- metadata/ingestion/source/database/common_nosql_source.py +105 -52
- metadata/ingestion/source/database/common_pg_mappings.py +56 -0
- metadata/ingestion/source/database/couchbase/connection.py +3 -3
- metadata/ingestion/source/database/couchbase/metadata.py +11 -5
- metadata/ingestion/source/database/couchbase/models.py +2 -2
- metadata/ingestion/source/database/couchbase/queries.py +3 -3
- metadata/ingestion/source/database/database_service.py +154 -11
- metadata/ingestion/source/database/databricks/auth.py +99 -0
- metadata/ingestion/source/database/databricks/client.py +304 -19
- metadata/ingestion/source/database/databricks/connection.py +154 -16
- metadata/ingestion/source/database/databricks/data_diff/__init__.py +1 -0
- metadata/ingestion/source/database/databricks/data_diff/data_diff.py +12 -0
- metadata/ingestion/source/database/databricks/lineage.py +15 -27
- metadata/ingestion/source/database/databricks/metadata.py +64 -17
- metadata/ingestion/source/database/databricks/queries.py +133 -9
- metadata/ingestion/source/database/databricks/query_parser.py +14 -18
- metadata/ingestion/source/database/databricks/service_spec.py +6 -0
- metadata/ingestion/source/database/databricks/usage.py +10 -40
- metadata/ingestion/source/database/datalake/clients/azure_blob.py +3 -3
- metadata/ingestion/source/database/datalake/clients/base.py +3 -3
- metadata/ingestion/source/database/datalake/clients/gcs.py +3 -3
- metadata/ingestion/source/database/datalake/clients/s3.py +3 -3
- metadata/ingestion/source/database/datalake/columns.py +3 -3
- metadata/ingestion/source/database/datalake/connection.py +54 -76
- metadata/ingestion/source/database/datalake/metadata.py +41 -34
- metadata/ingestion/source/database/datalake/service_spec.py +2 -0
- metadata/ingestion/source/database/db2/connection.py +19 -4
- metadata/ingestion/source/database/db2/lineage.py +3 -3
- metadata/ingestion/source/database/db2/metadata.py +3 -3
- metadata/ingestion/source/database/db2/service_spec.py +4 -1
- metadata/ingestion/source/database/db2/utils.py +133 -3
- metadata/ingestion/source/database/dbt/constants.py +20 -5
- metadata/ingestion/source/database/dbt/dbt_config.py +93 -31
- metadata/ingestion/source/database/dbt/dbt_service.py +34 -5
- metadata/ingestion/source/database/dbt/dbt_utils.py +4 -4
- metadata/ingestion/source/database/dbt/metadata.py +346 -116
- metadata/ingestion/source/database/dbt/models.py +4 -3
- metadata/ingestion/source/database/deltalake/clients/base.py +3 -3
- metadata/ingestion/source/database/deltalake/clients/pyspark.py +4 -4
- metadata/ingestion/source/database/deltalake/clients/s3.py +3 -3
- metadata/ingestion/source/database/deltalake/connection.py +3 -3
- metadata/ingestion/source/database/deltalake/metadata.py +21 -20
- metadata/ingestion/source/database/domodatabase/connection.py +3 -3
- metadata/ingestion/source/database/domodatabase/metadata.py +21 -20
- metadata/ingestion/source/database/domodatabase/models.py +3 -3
- metadata/ingestion/source/database/doris/connection.py +3 -3
- metadata/ingestion/source/database/doris/metadata.py +14 -6
- metadata/ingestion/source/database/doris/utils.py +3 -3
- metadata/ingestion/source/database/druid/connection.py +3 -3
- metadata/ingestion/source/database/druid/lineage.py +3 -3
- metadata/ingestion/source/database/druid/metadata.py +3 -3
- metadata/ingestion/source/database/druid/service_spec.py +5 -1
- metadata/ingestion/source/database/dynamodb/connection.py +3 -3
- metadata/ingestion/source/database/dynamodb/metadata.py +9 -6
- metadata/ingestion/source/database/dynamodb/models.py +2 -2
- metadata/ingestion/source/database/exasol/connection.py +20 -3
- metadata/ingestion/source/database/exasol/lineage.py +24 -0
- metadata/ingestion/source/database/exasol/metadata.py +6 -0
- metadata/ingestion/source/database/exasol/queries.py +35 -0
- metadata/ingestion/source/database/exasol/query_parser.py +33 -0
- metadata/ingestion/source/database/exasol/service_spec.py +5 -1
- metadata/ingestion/source/database/external_table_lineage_mixin.py +3 -3
- metadata/ingestion/source/database/glue/connection.py +3 -3
- metadata/ingestion/source/database/glue/metadata.py +100 -25
- metadata/ingestion/source/database/glue/models.py +3 -3
- metadata/ingestion/source/database/greenplum/connection.py +3 -3
- metadata/ingestion/source/database/greenplum/lineage.py +3 -3
- metadata/ingestion/source/database/greenplum/metadata.py +9 -40
- metadata/ingestion/source/database/greenplum/queries.py +3 -3
- metadata/ingestion/source/database/greenplum/service_spec.py +5 -1
- metadata/ingestion/source/database/greenplum/utils.py +3 -3
- metadata/ingestion/source/database/hive/connection.py +44 -5
- metadata/ingestion/source/database/hive/custom_hive_connection.py +195 -0
- metadata/ingestion/source/database/hive/lineage.py +3 -3
- metadata/ingestion/source/database/hive/metadata.py +11 -4
- metadata/ingestion/source/database/hive/metastore_dialects/mixin.py +3 -3
- metadata/ingestion/source/database/hive/metastore_dialects/mysql/__init__.py +3 -3
- metadata/ingestion/source/database/hive/metastore_dialects/mysql/dialect.py +38 -16
- metadata/ingestion/source/database/hive/metastore_dialects/postgres/__init__.py +3 -3
- metadata/ingestion/source/database/hive/metastore_dialects/postgres/dialect.py +41 -17
- metadata/ingestion/source/database/hive/queries.py +3 -3
- metadata/ingestion/source/database/hive/service_spec.py +5 -1
- metadata/ingestion/source/database/hive/utils.py +3 -3
- metadata/ingestion/source/database/iceberg/catalog/__init__.py +3 -3
- metadata/ingestion/source/database/iceberg/catalog/base.py +3 -3
- metadata/ingestion/source/database/iceberg/catalog/dynamodb.py +3 -3
- metadata/ingestion/source/database/iceberg/catalog/glue.py +3 -3
- metadata/ingestion/source/database/iceberg/catalog/hive.py +3 -3
- metadata/ingestion/source/database/iceberg/catalog/rest.py +3 -3
- metadata/ingestion/source/database/iceberg/connection.py +3 -3
- metadata/ingestion/source/database/iceberg/fs/__init__.py +3 -3
- metadata/ingestion/source/database/iceberg/fs/azure.py +3 -3
- metadata/ingestion/source/database/iceberg/fs/base.py +3 -3
- metadata/ingestion/source/database/iceberg/fs/s3.py +3 -3
- metadata/ingestion/source/database/iceberg/helper.py +3 -3
- metadata/ingestion/source/database/iceberg/metadata.py +64 -21
- metadata/ingestion/source/database/iceberg/models.py +25 -16
- metadata/ingestion/source/database/impala/connection.py +3 -3
- metadata/ingestion/source/database/impala/lineage.py +3 -3
- metadata/ingestion/source/database/impala/metadata.py +3 -3
- metadata/ingestion/source/database/impala/queries.py +3 -3
- metadata/ingestion/source/database/impala/service_spec.py +5 -1
- metadata/ingestion/source/database/incremental_metadata_extraction.py +3 -3
- metadata/ingestion/source/database/life_cycle_query_mixin.py +23 -10
- metadata/ingestion/source/database/lineage_processors.py +386 -0
- metadata/ingestion/source/database/lineage_source.py +417 -111
- metadata/ingestion/source/database/mariadb/connection.py +3 -3
- metadata/ingestion/source/database/mariadb/lineage.py +6 -6
- metadata/ingestion/source/database/mariadb/metadata.py +89 -4
- metadata/ingestion/source/database/mariadb/models.py +36 -0
- metadata/ingestion/source/database/mariadb/queries.py +43 -0
- metadata/ingestion/source/database/mariadb/service_spec.py +9 -1
- metadata/ingestion/source/database/mongodb/connection.py +3 -3
- metadata/ingestion/source/database/mongodb/metadata.py +11 -6
- metadata/ingestion/source/database/mssql/connection.py +3 -3
- metadata/ingestion/source/database/mssql/constants.py +3 -3
- metadata/ingestion/source/database/mssql/lineage.py +6 -30
- metadata/ingestion/source/database/mssql/metadata.py +71 -5
- metadata/ingestion/source/database/mssql/models.py +3 -3
- metadata/ingestion/source/database/mssql/queries.py +51 -8
- metadata/ingestion/source/database/mssql/query_parser.py +3 -3
- metadata/ingestion/source/database/mssql/usage.py +3 -3
- metadata/ingestion/source/database/mssql/utils.py +3 -3
- metadata/ingestion/source/database/multi_db_source.py +3 -3
- metadata/ingestion/source/database/mysql/connection.py +55 -46
- metadata/ingestion/source/database/mysql/lineage.py +7 -7
- metadata/ingestion/source/database/mysql/metadata.py +3 -3
- metadata/ingestion/source/database/mysql/queries.py +32 -3
- metadata/ingestion/source/database/mysql/query_parser.py +34 -3
- metadata/ingestion/source/database/mysql/service_spec.py +2 -0
- metadata/ingestion/source/database/mysql/usage.py +3 -3
- metadata/ingestion/source/database/mysql/utils.py +8 -6
- metadata/ingestion/source/database/oracle/connection.py +165 -99
- metadata/ingestion/source/database/oracle/lineage.py +6 -13
- metadata/ingestion/source/database/oracle/metadata.py +40 -62
- metadata/ingestion/source/database/oracle/models.py +5 -4
- metadata/ingestion/source/database/oracle/queries.py +64 -6
- metadata/ingestion/source/database/oracle/query_parser.py +3 -3
- metadata/ingestion/source/database/oracle/service_spec.py +2 -0
- metadata/ingestion/source/database/oracle/usage.py +3 -3
- metadata/ingestion/source/database/oracle/utils.py +22 -17
- metadata/ingestion/source/database/pinotdb/connection.py +3 -3
- metadata/ingestion/source/database/pinotdb/lineage.py +3 -3
- metadata/ingestion/source/database/pinotdb/metadata.py +3 -3
- metadata/ingestion/source/database/pinotdb/service_spec.py +5 -1
- metadata/ingestion/source/database/postgres/connection.py +60 -51
- metadata/ingestion/source/database/postgres/converter_orm.py +3 -3
- metadata/ingestion/source/database/postgres/lineage.py +5 -4
- metadata/ingestion/source/database/postgres/metadata.py +51 -63
- metadata/ingestion/source/database/postgres/models.py +4 -2
- metadata/ingestion/source/database/postgres/pgspider/lineage.py +2 -2
- metadata/ingestion/source/database/postgres/pgspider/queries.py +2 -2
- metadata/ingestion/source/database/postgres/queries.py +55 -91
- metadata/ingestion/source/database/postgres/query_parser.py +3 -3
- metadata/ingestion/source/database/postgres/service_spec.py +2 -0
- metadata/ingestion/source/database/postgres/types/money.py +3 -3
- metadata/ingestion/source/database/postgres/usage.py +18 -3
- metadata/ingestion/source/database/postgres/utils.py +58 -59
- metadata/ingestion/source/database/presto/connection.py +3 -3
- metadata/ingestion/source/database/presto/metadata.py +3 -3
- metadata/ingestion/source/database/presto/queries.py +3 -3
- metadata/ingestion/source/database/query/lineage.py +3 -3
- metadata/ingestion/source/database/query/usage.py +3 -3
- metadata/ingestion/source/database/query_parser_source.py +17 -15
- metadata/ingestion/source/database/redshift/connection.py +4 -4
- metadata/ingestion/source/database/redshift/incremental_table_processor.py +3 -3
- metadata/ingestion/source/database/redshift/lineage.py +7 -13
- metadata/ingestion/source/database/redshift/metadata.py +40 -22
- metadata/ingestion/source/database/redshift/models.py +3 -3
- metadata/ingestion/source/database/redshift/queries.py +8 -38
- metadata/ingestion/source/database/redshift/query_parser.py +3 -3
- metadata/ingestion/source/database/redshift/service_spec.py +4 -4
- metadata/ingestion/source/database/redshift/usage.py +3 -3
- metadata/ingestion/source/database/redshift/utils.py +31 -4
- metadata/ingestion/source/database/salesforce/connection.py +17 -9
- metadata/ingestion/source/database/salesforce/metadata.py +73 -34
- metadata/ingestion/source/database/sample_data.py +692 -6
- metadata/ingestion/source/database/sample_usage.py +5 -3
- metadata/ingestion/source/database/saperp/client.py +3 -3
- metadata/ingestion/source/database/saperp/connection.py +3 -3
- metadata/ingestion/source/database/saperp/constants.py +2 -2
- metadata/ingestion/source/database/saperp/metadata.py +3 -3
- metadata/ingestion/source/database/saperp/models.py +3 -3
- metadata/ingestion/source/database/saphana/cdata_parser.py +330 -120
- metadata/ingestion/source/database/saphana/connection.py +3 -3
- metadata/ingestion/source/database/saphana/lineage.py +20 -12
- metadata/ingestion/source/database/saphana/metadata.py +3 -3
- metadata/ingestion/source/database/saphana/models.py +3 -3
- metadata/ingestion/source/database/saphana/queries.py +10 -3
- metadata/ingestion/source/database/sas/client.py +5 -4
- metadata/ingestion/source/database/sas/connection.py +3 -3
- metadata/ingestion/source/database/sas/extension_attr.py +3 -3
- metadata/ingestion/source/database/sas/metadata.py +22 -25
- metadata/ingestion/source/database/singlestore/connection.py +3 -3
- metadata/ingestion/source/database/singlestore/lineage.py +3 -3
- metadata/ingestion/source/database/singlestore/metadata.py +3 -3
- metadata/ingestion/source/database/singlestore/service_spec.py +6 -1
- metadata/ingestion/source/database/snowflake/connection.py +207 -152
- metadata/ingestion/source/database/snowflake/constants.py +69 -0
- metadata/ingestion/source/database/snowflake/data_diff/data_diff.py +37 -0
- metadata/ingestion/source/database/snowflake/lineage.py +55 -32
- metadata/ingestion/source/database/snowflake/metadata.py +416 -63
- metadata/ingestion/source/database/snowflake/models.py +6 -6
- metadata/ingestion/source/database/snowflake/queries.py +76 -20
- metadata/ingestion/source/database/snowflake/query_parser.py +18 -5
- metadata/ingestion/source/database/snowflake/service_spec.py +10 -4
- metadata/ingestion/source/database/snowflake/usage.py +100 -3
- metadata/ingestion/source/database/snowflake/utils.py +143 -39
- metadata/ingestion/source/database/sql_column_handler.py +112 -90
- metadata/ingestion/source/database/sqlalchemy_source.py +3 -3
- metadata/ingestion/source/database/sqlite/connection.py +3 -3
- metadata/ingestion/source/database/sqlite/lineage.py +3 -3
- metadata/ingestion/source/database/sqlite/metadata.py +3 -3
- metadata/ingestion/source/database/sqlite/service_spec.py +5 -1
- metadata/ingestion/source/database/stored_procedures_mixin.py +113 -176
- metadata/ingestion/source/database/teradata/connection.py +3 -3
- metadata/ingestion/source/database/teradata/lineage.py +3 -3
- metadata/ingestion/source/database/teradata/metadata.py +3 -3
- metadata/ingestion/source/database/teradata/queries.py +3 -3
- metadata/ingestion/source/database/teradata/service_spec.py +5 -1
- metadata/ingestion/source/database/teradata/utils.py +3 -3
- metadata/ingestion/source/database/timescale/__init__.py +3 -0
- metadata/ingestion/source/database/timescale/connection.py +103 -0
- metadata/ingestion/source/database/timescale/lineage.py +124 -0
- metadata/ingestion/source/database/timescale/metadata.py +228 -0
- metadata/ingestion/source/database/timescale/models.py +60 -0
- metadata/ingestion/source/database/timescale/queries.py +96 -0
- metadata/ingestion/source/database/timescale/query_parser.py +106 -0
- metadata/ingestion/source/database/timescale/service_spec.py +12 -0
- metadata/ingestion/source/database/timescale/usage.py +43 -0
- metadata/ingestion/source/database/timescale/utils.py +545 -0
- metadata/ingestion/source/database/trino/connection.py +290 -123
- metadata/ingestion/source/database/trino/lineage.py +98 -3
- metadata/ingestion/source/database/trino/metadata.py +46 -7
- metadata/ingestion/source/database/trino/profiler/system_tables_profiler.py +29 -20
- metadata/ingestion/source/database/trino/queries.py +8 -4
- metadata/ingestion/source/database/trino/query_parser.py +18 -3
- metadata/ingestion/source/database/trino/service_spec.py +2 -0
- metadata/ingestion/source/database/trino/usage.py +3 -3
- metadata/ingestion/source/database/unitycatalog/client.py +38 -4
- metadata/ingestion/source/database/unitycatalog/connection.py +121 -17
- metadata/ingestion/source/database/unitycatalog/data_diff/__init__.py +1 -0
- metadata/ingestion/source/database/unitycatalog/data_diff/data_diff.py +12 -0
- metadata/ingestion/source/database/unitycatalog/lineage.py +306 -63
- metadata/ingestion/source/database/unitycatalog/metadata.py +160 -48
- metadata/ingestion/source/database/unitycatalog/models.py +22 -6
- metadata/ingestion/source/database/unitycatalog/queries.py +55 -0
- metadata/ingestion/source/database/unitycatalog/query_parser.py +14 -3
- metadata/ingestion/source/database/unitycatalog/service_spec.py +8 -0
- metadata/ingestion/source/database/unitycatalog/usage.py +30 -5
- metadata/ingestion/source/database/usage_source.py +11 -8
- metadata/ingestion/source/database/vertica/connection.py +3 -3
- metadata/ingestion/source/database/vertica/lineage.py +3 -3
- metadata/ingestion/source/database/vertica/metadata.py +3 -3
- metadata/ingestion/source/database/vertica/queries.py +8 -23
- metadata/ingestion/source/database/vertica/query_parser.py +3 -3
- metadata/ingestion/source/database/vertica/usage.py +3 -3
- metadata/ingestion/source/drive/drive_service.py +648 -0
- metadata/ingestion/source/drives/__init__.py +13 -0
- metadata/ingestion/source/messaging/common_broker_source.py +3 -3
- metadata/ingestion/source/messaging/kafka/connection.py +16 -7
- metadata/ingestion/source/messaging/kafka/metadata.py +11 -13
- metadata/ingestion/source/messaging/kinesis/connection.py +3 -3
- metadata/ingestion/source/messaging/kinesis/metadata.py +7 -5
- metadata/ingestion/source/messaging/kinesis/models.py +2 -2
- metadata/ingestion/source/messaging/messaging_service.py +7 -10
- metadata/ingestion/source/messaging/redpanda/connection.py +3 -3
- metadata/ingestion/source/messaging/redpanda/metadata.py +3 -3
- metadata/ingestion/source/metadata/alationsink/client.py +2 -2
- metadata/ingestion/source/metadata/alationsink/connection.py +2 -2
- metadata/ingestion/source/metadata/alationsink/constants.py +2 -2
- metadata/ingestion/source/metadata/alationsink/metadata.py +36 -27
- metadata/ingestion/source/metadata/alationsink/models.py +2 -2
- metadata/ingestion/source/metadata/amundsen/client.py +9 -9
- metadata/ingestion/source/metadata/amundsen/connection.py +3 -3
- metadata/ingestion/source/metadata/amundsen/metadata.py +18 -20
- metadata/ingestion/source/metadata/amundsen/queries.py +3 -3
- metadata/ingestion/source/metadata/atlas/client.py +5 -4
- metadata/ingestion/source/metadata/atlas/connection.py +3 -3
- metadata/ingestion/source/metadata/atlas/metadata.py +7 -11
- metadata/ingestion/source/mlmodel/mlflow/connection.py +3 -3
- metadata/ingestion/source/mlmodel/mlflow/metadata.py +3 -3
- metadata/ingestion/source/mlmodel/mlmodel_service.py +82 -11
- metadata/ingestion/source/mlmodel/sagemaker/connection.py +3 -3
- metadata/ingestion/source/mlmodel/sagemaker/metadata.py +3 -3
- metadata/ingestion/source/models.py +3 -3
- metadata/ingestion/source/pipeline/airbyte/client.py +145 -7
- metadata/ingestion/source/pipeline/airbyte/connection.py +21 -8
- metadata/ingestion/source/pipeline/airbyte/constants.py +29 -0
- metadata/ingestion/source/pipeline/airbyte/metadata.py +172 -31
- metadata/ingestion/source/pipeline/airbyte/utils.py +99 -0
- metadata/ingestion/source/pipeline/airflow/connection.py +14 -14
- metadata/ingestion/source/pipeline/airflow/lineage_parser.py +3 -3
- metadata/ingestion/source/pipeline/airflow/metadata.py +112 -41
- metadata/ingestion/source/pipeline/airflow/models.py +4 -3
- metadata/ingestion/source/pipeline/airflow/utils.py +49 -7
- metadata/ingestion/source/pipeline/dagster/client.py +3 -3
- metadata/ingestion/source/pipeline/dagster/connection.py +3 -3
- metadata/ingestion/source/pipeline/dagster/metadata.py +3 -3
- metadata/ingestion/source/pipeline/dagster/models.py +3 -3
- metadata/ingestion/source/pipeline/dagster/queries.py +3 -3
- metadata/ingestion/source/pipeline/databrickspipeline/connection.py +30 -5
- metadata/ingestion/source/pipeline/databrickspipeline/kafka_parser.py +603 -0
- metadata/ingestion/source/pipeline/databrickspipeline/metadata.py +1138 -47
- metadata/ingestion/source/pipeline/databrickspipeline/models.py +22 -6
- metadata/ingestion/source/pipeline/dbtcloud/client.py +126 -68
- metadata/ingestion/source/pipeline/dbtcloud/connection.py +3 -3
- metadata/ingestion/source/pipeline/dbtcloud/metadata.py +138 -87
- metadata/ingestion/source/pipeline/dbtcloud/models.py +20 -3
- metadata/ingestion/source/pipeline/dbtcloud/queries.py +10 -14
- metadata/ingestion/source/pipeline/domopipeline/connection.py +3 -3
- metadata/ingestion/source/pipeline/domopipeline/metadata.py +3 -3
- metadata/ingestion/source/pipeline/fivetran/client.py +10 -6
- metadata/ingestion/source/pipeline/fivetran/connection.py +3 -3
- metadata/ingestion/source/pipeline/fivetran/metadata.py +127 -95
- metadata/ingestion/source/pipeline/flink/client.py +3 -3
- metadata/ingestion/source/pipeline/flink/connection.py +3 -3
- metadata/ingestion/source/pipeline/flink/metadata.py +3 -3
- metadata/ingestion/source/pipeline/flink/models.py +3 -3
- metadata/ingestion/source/pipeline/gluepipeline/connection.py +3 -3
- metadata/ingestion/source/pipeline/gluepipeline/metadata.py +3 -3
- metadata/ingestion/source/pipeline/gluepipeline/models.py +3 -3
- metadata/ingestion/source/pipeline/kafkaconnect/client.py +391 -110
- metadata/ingestion/source/pipeline/kafkaconnect/connection.py +3 -3
- metadata/ingestion/source/pipeline/kafkaconnect/metadata.py +1683 -77
- metadata/ingestion/source/pipeline/kafkaconnect/models.py +58 -21
- metadata/ingestion/source/pipeline/nifi/client.py +71 -77
- metadata/ingestion/source/pipeline/nifi/connection.py +4 -19
- metadata/ingestion/source/pipeline/nifi/metadata.py +197 -10
- metadata/ingestion/source/pipeline/openlineage/connection.py +3 -3
- metadata/ingestion/source/pipeline/openlineage/metadata.py +10 -5
- metadata/ingestion/source/pipeline/openlineage/models.py +6 -5
- metadata/ingestion/source/pipeline/pipeline_service.py +146 -15
- metadata/ingestion/source/pipeline/spline/client.py +3 -3
- metadata/ingestion/source/pipeline/spline/connection.py +3 -3
- metadata/ingestion/source/pipeline/spline/metadata.py +3 -3
- metadata/ingestion/source/pipeline/spline/models.py +3 -3
- metadata/ingestion/source/pipeline/spline/utils.py +3 -3
- metadata/ingestion/source/search/elasticsearch/connection.py +4 -4
- metadata/ingestion/source/search/elasticsearch/metadata.py +3 -3
- metadata/ingestion/source/search/elasticsearch/parser.py +3 -3
- metadata/ingestion/source/search/opensearch/connection.py +211 -0
- metadata/ingestion/source/search/opensearch/metadata.py +270 -0
- metadata/ingestion/source/search/opensearch/parser.py +71 -0
- metadata/ingestion/source/search/opensearch/service_spec.py +4 -0
- metadata/ingestion/source/search/search_service.py +7 -10
- metadata/ingestion/source/security/security_service.py +128 -0
- metadata/ingestion/source/sqa_types.py +3 -3
- metadata/ingestion/source/storage/gcs/client.py +2 -2
- metadata/ingestion/source/storage/gcs/connection.py +17 -7
- metadata/ingestion/source/storage/gcs/metadata.py +287 -28
- metadata/ingestion/source/storage/gcs/models.py +14 -8
- metadata/ingestion/source/storage/s3/connection.py +3 -3
- metadata/ingestion/source/storage/s3/metadata.py +75 -31
- metadata/ingestion/source/storage/s3/models.py +3 -3
- metadata/ingestion/source/storage/storage_service.py +9 -10
- metadata/ingestion/stage/table_usage.py +56 -4
- metadata/mixins/pandas/pandas_mixin.py +3 -3
- metadata/mixins/sqalchemy/sqa_mixin.py +22 -11
- metadata/parsers/avro_parser.py +3 -3
- metadata/parsers/json_schema_parser.py +44 -4
- metadata/parsers/protobuf_parser.py +3 -3
- metadata/parsers/schema_parsers.py +3 -3
- metadata/pii/algorithms/classifiers.py +242 -0
- metadata/pii/algorithms/column_patterns.py +61 -0
- metadata/pii/algorithms/feature_extraction.py +165 -0
- metadata/pii/algorithms/patterns.py +171 -0
- metadata/pii/algorithms/preprocessing.py +68 -0
- metadata/pii/algorithms/presidio_patches.py +90 -0
- metadata/pii/algorithms/presidio_recognizer_factory.py +302 -0
- metadata/pii/algorithms/presidio_utils.py +292 -0
- metadata/pii/algorithms/tags.py +111 -0
- metadata/pii/algorithms/utils.py +42 -0
- metadata/pii/base_processor.py +126 -0
- metadata/pii/constants.py +11 -3
- metadata/pii/models.py +3 -3
- metadata/pii/ner.py +3 -3
- metadata/pii/processor.py +52 -139
- metadata/pii/processor_factory.py +28 -0
- metadata/pii/scanners/base.py +3 -3
- metadata/pii/scanners/column_name_scanner.py +5 -7
- metadata/pii/scanners/custom_ner_scanner.py +290 -0
- metadata/pii/scanners/ner_scanner.py +5 -12
- metadata/pii/tag_analyzer.py +124 -0
- metadata/pii/tag_processor.py +100 -0
- metadata/profiler/adaptors/adaptor_factory.py +2 -2
- metadata/profiler/adaptors/dynamodb.py +2 -2
- metadata/profiler/adaptors/factory.py +2 -2
- metadata/profiler/adaptors/mongodb.py +2 -2
- metadata/profiler/adaptors/nosql_adaptor.py +2 -2
- metadata/profiler/api/models.py +3 -3
- metadata/profiler/config.py +3 -3
- metadata/profiler/factory.py +3 -3
- metadata/profiler/interface/nosql/profiler_interface.py +3 -3
- metadata/profiler/interface/pandas/profiler_interface.py +3 -3
- metadata/profiler/interface/profiler_interface.py +3 -3
- metadata/profiler/interface/sqlalchemy/bigquery/profiler_interface.py +52 -6
- metadata/profiler/interface/sqlalchemy/databricks/profiler_interface.py +59 -5
- metadata/profiler/interface/sqlalchemy/db2/profiler_interface.py +3 -3
- metadata/profiler/interface/sqlalchemy/mariadb/profiler_interface.py +3 -3
- metadata/profiler/interface/sqlalchemy/profiler_interface.py +86 -52
- metadata/profiler/interface/sqlalchemy/redshift/profiler_interface.py +50 -0
- metadata/profiler/interface/sqlalchemy/single_store/profiler_interface.py +3 -3
- metadata/profiler/interface/sqlalchemy/snowflake/profiler_interface.py +23 -3
- metadata/profiler/interface/sqlalchemy/stored_statistics_profiler.py +3 -3
- metadata/profiler/interface/sqlalchemy/trino/profiler_interface.py +3 -3
- metadata/profiler/interface/sqlalchemy/unity_catalog/profiler_interface.py +19 -5
- metadata/profiler/metrics/composed/distinct_ratio.py +3 -3
- metadata/profiler/metrics/composed/duplicate_count.py +3 -3
- metadata/profiler/metrics/composed/ilike_ratio.py +3 -3
- metadata/profiler/metrics/composed/iqr.py +3 -3
- metadata/profiler/metrics/composed/like_ratio.py +3 -3
- metadata/profiler/metrics/composed/non_parametric_skew.py +3 -3
- metadata/profiler/metrics/composed/null_ratio.py +3 -3
- metadata/profiler/metrics/composed/unique_ratio.py +3 -3
- metadata/profiler/metrics/core.py +3 -3
- metadata/profiler/metrics/hybrid/cardinality_distribution.py +232 -0
- metadata/profiler/metrics/hybrid/histogram.py +4 -4
- metadata/profiler/metrics/registry.py +7 -3
- metadata/profiler/metrics/static/column_count.py +3 -3
- metadata/profiler/metrics/static/column_names.py +3 -3
- metadata/profiler/metrics/static/count.py +3 -3
- metadata/profiler/metrics/static/count_in_set.py +3 -3
- metadata/profiler/metrics/static/distinct_count.py +3 -3
- metadata/profiler/metrics/static/ilike_count.py +3 -3
- metadata/profiler/metrics/static/like_count.py +3 -3
- metadata/profiler/metrics/static/max.py +3 -3
- metadata/profiler/metrics/static/max_length.py +3 -3
- metadata/profiler/metrics/static/mean.py +3 -3
- metadata/profiler/metrics/static/min.py +3 -3
- metadata/profiler/metrics/static/min_length.py +3 -3
- metadata/profiler/metrics/static/not_like_count.py +3 -3
- metadata/profiler/metrics/static/not_regexp_match_count.py +7 -6
- metadata/profiler/metrics/static/null_count.py +3 -3
- metadata/profiler/metrics/static/null_missing_count.py +3 -3
- metadata/profiler/metrics/static/regexp_match_count.py +6 -5
- metadata/profiler/metrics/static/row_count.py +3 -3
- metadata/profiler/metrics/static/stddev.py +3 -3
- metadata/profiler/metrics/static/sum.py +3 -3
- metadata/profiler/metrics/static/unique_count.py +3 -3
- metadata/{ingestion/source/database/bigquery/profiler → profiler/metrics/system/bigquery}/system.py +44 -59
- metadata/profiler/metrics/system/databricks/system.py +132 -0
- metadata/profiler/metrics/system/dml_operation.py +5 -3
- metadata/{ingestion/source/database/redshift/profiler → profiler/metrics/system/redshift}/system.py +39 -57
- metadata/{ingestion/source/database/snowflake/profiler → profiler/metrics/system/snowflake}/system.py +29 -52
- metadata/profiler/metrics/system/system.py +110 -41
- metadata/profiler/metrics/window/first_quartile.py +6 -6
- metadata/profiler/metrics/window/median.py +6 -6
- metadata/profiler/metrics/window/third_quartile.py +6 -6
- metadata/profiler/orm/converter/azuresql/converter.py +39 -0
- metadata/profiler/orm/converter/base.py +3 -3
- metadata/profiler/orm/converter/bigquery/converter.py +3 -3
- metadata/profiler/orm/converter/common.py +3 -3
- metadata/profiler/orm/converter/converter_registry.py +9 -3
- metadata/profiler/orm/converter/mariadb/converter.py +28 -0
- metadata/profiler/orm/converter/mssql/converter.py +3 -3
- metadata/profiler/orm/converter/redshift/converter.py +3 -3
- metadata/profiler/orm/converter/snowflake/converter.py +3 -3
- metadata/profiler/orm/converter/trino/__init__.py +14 -0
- metadata/profiler/orm/converter/trino/converter.py +39 -0
- metadata/profiler/orm/functions/concat.py +3 -3
- metadata/profiler/orm/functions/conn_test.py +8 -3
- metadata/profiler/orm/functions/count.py +3 -3
- metadata/profiler/orm/functions/datetime.py +3 -3
- metadata/profiler/orm/functions/length.py +6 -3
- metadata/profiler/orm/functions/md5.py +12 -3
- metadata/profiler/orm/functions/median.py +22 -3
- metadata/profiler/orm/functions/modulo.py +5 -3
- metadata/profiler/orm/functions/random_num.py +8 -3
- metadata/profiler/orm/functions/regexp.py +40 -0
- metadata/profiler/orm/functions/substr.py +3 -3
- metadata/profiler/orm/functions/sum.py +4 -3
- metadata/profiler/orm/functions/table_metric_computer.py +28 -15
- metadata/profiler/orm/functions/unique_count.py +3 -3
- metadata/profiler/orm/registry.py +15 -3
- metadata/profiler/orm/types/__init__.py +10 -0
- metadata/profiler/orm/types/bytea_to_string.py +3 -3
- metadata/profiler/orm/types/custom_array.py +3 -3
- metadata/profiler/orm/types/custom_datetimerange.py +3 -3
- metadata/profiler/orm/types/custom_hex_byte_string.py +3 -3
- metadata/profiler/orm/types/custom_image.py +3 -3
- metadata/profiler/orm/types/custom_ip.py +3 -3
- metadata/profiler/orm/types/custom_time.py +54 -0
- metadata/profiler/orm/types/custom_timestamp.py +3 -3
- metadata/profiler/orm/types/trino.py +70 -0
- metadata/profiler/orm/types/undetermined_type.py +3 -3
- metadata/profiler/orm/types/uuid.py +3 -3
- metadata/profiler/processor/core.py +9 -8
- metadata/profiler/processor/default.py +33 -31
- metadata/profiler/processor/handle_partition.py +3 -3
- metadata/profiler/processor/metric_filter.py +20 -7
- metadata/profiler/processor/models.py +17 -7
- metadata/profiler/processor/processor.py +21 -6
- metadata/profiler/processor/runner.py +32 -9
- metadata/profiler/processor/sample_data_handler.py +3 -3
- metadata/profiler/registry.py +3 -3
- metadata/profiler/source/database/base/profiler_resolver.py +38 -0
- metadata/profiler/source/database/base/profiler_source.py +60 -37
- metadata/profiler/source/database/bigquery/profiler_source.py +5 -23
- metadata/profiler/source/database/bigquery/type_mapper.py +3 -3
- metadata/profiler/source/database/mssql/profiler_source.py +86 -0
- metadata/profiler/source/fetcher/config.py +7 -3
- metadata/profiler/source/fetcher/entity_fetcher.py +3 -3
- metadata/profiler/source/fetcher/fetcher_strategy.py +26 -5
- metadata/profiler/source/fetcher/profiler_source_factory.py +16 -3
- metadata/profiler/source/metadata.py +3 -3
- metadata/profiler/source/metadata_ext.py +3 -3
- metadata/profiler/source/model.py +3 -3
- metadata/profiler/source/profiler_source_interface.py +20 -16
- metadata/readers/dataframe/avro.py +3 -3
- metadata/readers/dataframe/base.py +5 -3
- metadata/readers/dataframe/common.py +3 -3
- metadata/readers/dataframe/dsv.py +53 -8
- metadata/readers/dataframe/json.py +8 -4
- metadata/readers/dataframe/mf4.py +75 -0
- metadata/readers/dataframe/models.py +3 -3
- metadata/readers/dataframe/parquet.py +226 -39
- metadata/readers/dataframe/reader_factory.py +20 -4
- metadata/readers/file/adls.py +3 -3
- metadata/readers/file/api_reader.py +3 -3
- metadata/readers/file/base.py +3 -3
- metadata/readers/file/bitbucket.py +3 -3
- metadata/readers/file/config_source_factory.py +3 -3
- metadata/readers/file/credentials.py +3 -3
- metadata/readers/file/gcs.py +3 -3
- metadata/readers/file/github.py +3 -3
- metadata/readers/file/gitlab.py +3 -3
- metadata/readers/file/local.py +3 -3
- metadata/readers/file/s3.py +3 -3
- metadata/readers/models.py +3 -3
- metadata/sampler/config.py +7 -6
- metadata/sampler/models.py +9 -6
- metadata/sampler/nosql/sampler.py +11 -7
- metadata/sampler/pandas/sampler.py +12 -11
- metadata/sampler/partition.py +14 -6
- metadata/sampler/processor.py +30 -7
- metadata/sampler/sampler_interface.py +28 -17
- metadata/sampler/sqlalchemy/azuresql/sampler.py +32 -6
- metadata/sampler/sqlalchemy/bigquery/sampler.py +30 -9
- metadata/sampler/sqlalchemy/databricks/sampler.py +69 -0
- metadata/sampler/sqlalchemy/mssql/sampler.py +16 -14
- metadata/sampler/sqlalchemy/postgres/sampler.py +8 -8
- metadata/sampler/sqlalchemy/sampler.py +155 -61
- metadata/sampler/sqlalchemy/snowflake/sampler.py +12 -11
- metadata/sampler/sqlalchemy/trino/sampler.py +12 -11
- metadata/sampler/sqlalchemy/unitycatalog/sampler.py +27 -0
- metadata/sdk/__init__.py +188 -0
- metadata/sdk/api/__init__.py +7 -0
- metadata/sdk/api/lineage.py +421 -0
- metadata/sdk/api/search.py +382 -0
- metadata/sdk/client.py +91 -0
- metadata/sdk/config.py +121 -0
- metadata/sdk/entities/__init__.py +64 -0
- metadata/sdk/entities/apicollections.py +19 -0
- metadata/sdk/entities/apiendpoints.py +19 -0
- metadata/sdk/entities/base.py +487 -0
- metadata/sdk/entities/charts.py +17 -0
- metadata/sdk/entities/classifications.py +21 -0
- metadata/sdk/entities/containers.py +17 -0
- metadata/sdk/entities/custom_properties.py +144 -0
- metadata/sdk/entities/dashboard_services.py +18 -0
- metadata/sdk/entities/dashboarddatamodels.py +21 -0
- metadata/sdk/entities/dashboards.py +17 -0
- metadata/sdk/entities/database_services.py +18 -0
- metadata/sdk/entities/databases.py +17 -0
- metadata/sdk/entities/databaseschemas.py +19 -0
- metadata/sdk/entities/datacontracts.py +19 -0
- metadata/sdk/entities/dataproducts.py +19 -0
- metadata/sdk/entities/domains.py +17 -0
- metadata/sdk/entities/glossaries.py +17 -0
- metadata/sdk/entities/glossary_terms.py +18 -0
- metadata/sdk/entities/glossaryterms.py +19 -0
- metadata/sdk/entities/metrics.py +49 -0
- metadata/sdk/entities/mlmodels.py +17 -0
- metadata/sdk/entities/pipelines.py +17 -0
- metadata/sdk/entities/queries.py +17 -0
- metadata/sdk/entities/searchindexes.py +19 -0
- metadata/sdk/entities/storedprocedures.py +19 -0
- metadata/sdk/entities/tables.py +68 -0
- metadata/sdk/entities/tags.py +17 -0
- metadata/sdk/entities/teams.py +17 -0
- metadata/sdk/entities/testcases.py +17 -0
- metadata/sdk/entities/testdefinitions.py +19 -0
- metadata/sdk/entities/testsuites.py +17 -0
- metadata/sdk/entities/users.py +17 -0
- metadata/sdk/examples/builder_end_to_end.py +387 -0
- metadata/sdk/types.py +15 -0
- metadata/timer/repeated_timer.py +3 -3
- metadata/utils/bigquery_utils.py +45 -7
- metadata/utils/class_helper.py +4 -3
- metadata/utils/client_version.py +3 -3
- metadata/utils/constants.py +6 -3
- metadata/utils/constraints.py +2 -2
- metadata/utils/credentials.py +63 -7
- metadata/utils/custom_thread_pool.py +3 -3
- metadata/utils/datalake/datalake_utils.py +134 -6
- metadata/utils/db_utils.py +42 -10
- metadata/utils/dependency_injector/dependency_injector.py +403 -0
- metadata/utils/deprecation.py +12 -9
- metadata/utils/dispatch.py +9 -6
- metadata/utils/elasticsearch.py +10 -3
- metadata/utils/entity_link.py +3 -3
- metadata/utils/entity_utils.py +9 -3
- metadata/utils/execution_time_tracker.py +3 -3
- metadata/utils/filters.py +106 -3
- metadata/utils/fqn.py +190 -14
- metadata/utils/helpers.py +121 -4
- metadata/utils/importer.py +6 -5
- metadata/utils/life_cycle_utils.py +3 -3
- metadata/utils/logger.py +34 -11
- metadata/utils/lru_cache.py +3 -3
- metadata/utils/messaging_utils.py +2 -2
- metadata/utils/metadata_service_helper.py +1 -1
- metadata/utils/profiler_utils.py +3 -3
- metadata/utils/s3_utils.py +3 -3
- metadata/utils/secrets/aws_based_secrets_manager.py +2 -2
- metadata/utils/secrets/aws_secrets_manager.py +2 -2
- metadata/utils/secrets/aws_ssm_secrets_manager.py +2 -2
- metadata/utils/secrets/azure_kv_secrets_manager.py +2 -2
- metadata/utils/secrets/db_secrets_manager.py +2 -2
- metadata/utils/secrets/external_secrets_manager.py +2 -2
- metadata/utils/secrets/kubernetes_secrets_manager.py +176 -0
- metadata/utils/secrets/secrets_manager.py +2 -2
- metadata/utils/secrets/secrets_manager_factory.py +5 -2
- metadata/utils/service_spec/default.py +4 -0
- metadata/utils/service_spec/service_spec.py +67 -10
- metadata/utils/singleton.py +3 -3
- metadata/utils/source_hash.py +3 -3
- metadata/utils/sqa_like_column.py +3 -3
- metadata/utils/sqa_utils.py +21 -3
- metadata/utils/sqlalchemy_utils.py +29 -5
- metadata/utils/ssl_manager.py +186 -12
- metadata/utils/ssl_registry.py +3 -3
- metadata/utils/storage_metadata_config.py +3 -3
- metadata/utils/stored_procedures.py +3 -3
- metadata/utils/streamable_logger.py +569 -0
- metadata/utils/tag_utils.py +75 -30
- metadata/utils/test_utils.py +2 -2
- metadata/utils/time_utils.py +3 -3
- metadata/utils/timeout.py +3 -3
- metadata/utils/uuid_encoder.py +3 -3
- metadata/workflow/application.py +3 -3
- metadata/workflow/application_output_handler.py +3 -3
- metadata/workflow/base.py +60 -6
- metadata/workflow/classification.py +8 -5
- metadata/workflow/context/base.py +24 -0
- metadata/workflow/context/context_manager.py +124 -0
- metadata/workflow/context/workflow_context.py +29 -0
- metadata/workflow/data_quality.py +3 -3
- metadata/workflow/ingestion.py +18 -5
- metadata/workflow/metadata.py +3 -3
- metadata/workflow/output_handler.py +3 -3
- metadata/workflow/profiler.py +24 -21
- metadata/workflow/usage.py +3 -3
- metadata/workflow/workflow_init_error_handler.py +3 -3
- metadata/workflow/workflow_output_handler.py +3 -3
- metadata/workflow/workflow_resource_metrics.py +111 -0
- metadata/workflow/workflow_status_mixin.py +40 -19
- {openmetadata_ingestion-1.6.2.0.dist-info → openmetadata_ingestion-1.10.14.4.dist-info}/METADATA +607 -376
- openmetadata_ingestion-1.10.14.4.dist-info/RECORD +2162 -0
- metadata/ingestion/source/database/bigquery/profiler/profiler.py +0 -30
- metadata/ingestion/source/database/extended_sample_data.py +0 -534
- metadata/ingestion/source/database/redshift/profiler/profiler.py +0 -14
- metadata/ingestion/source/database/snowflake/profiler/profiler.py +0 -26
- openmetadata_ingestion-1.6.2.0.dist-info/RECORD +0 -1814
- /metadata/ingestion/source/database/{bigquery/profiler → cassandra}/__init__.py +0 -0
- /metadata/ingestion/source/database/{redshift/profiler → cockroach}/__init__.py +0 -0
- /metadata/ingestion/source/database/snowflake/{profiler → data_diff}/__init__.py +0 -0
- {openmetadata_ingestion-1.6.2.0.dist-info → openmetadata_ingestion-1.10.14.4.dist-info}/LICENSE +0 -0
- {openmetadata_ingestion-1.6.2.0.dist-info → openmetadata_ingestion-1.10.14.4.dist-info}/WHEEL +0 -0
- {openmetadata_ingestion-1.6.2.0.dist-info → openmetadata_ingestion-1.10.14.4.dist-info}/entry_points.txt +0 -0
- {openmetadata_ingestion-1.6.2.0.dist-info → openmetadata_ingestion-1.10.14.4.dist-info}/top_level.txt +0 -0
|
@@ -1,8 +1,8 @@
|
|
|
1
|
-
# Copyright
|
|
2
|
-
# Licensed under the
|
|
1
|
+
# Copyright 2025 Collate
|
|
2
|
+
# Licensed under the Collate Community License, Version 1.0 (the "License");
|
|
3
3
|
# you may not use this file except in compliance with the License.
|
|
4
4
|
# You may obtain a copy of the License at
|
|
5
|
-
#
|
|
5
|
+
# https://github.com/open-metadata/OpenMetadata/blob/main/ingestion/LICENSE
|
|
6
6
|
# Unless required by applicable law or agreed to in writing, software
|
|
7
7
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
8
8
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
@@ -10,9 +10,13 @@
|
|
|
10
10
|
# limitations under the License.
|
|
11
11
|
"""PowerBI source module"""
|
|
12
12
|
|
|
13
|
+
import re
|
|
13
14
|
import traceback
|
|
14
15
|
from typing import Any, Iterable, List, Optional, Union
|
|
15
16
|
|
|
17
|
+
from pydantic import EmailStr
|
|
18
|
+
from pydantic_core import PydanticCustomError
|
|
19
|
+
|
|
16
20
|
from metadata.generated.schema.api.data.createChart import CreateChartRequest
|
|
17
21
|
from metadata.generated.schema.api.data.createDashboard import CreateDashboardRequest
|
|
18
22
|
from metadata.generated.schema.api.data.createDashboardDataModel import (
|
|
@@ -47,17 +51,27 @@ from metadata.generated.schema.type.basic import (
|
|
|
47
51
|
Markdown,
|
|
48
52
|
SourceUrl,
|
|
49
53
|
)
|
|
54
|
+
from metadata.generated.schema.type.entityLineage import ColumnLineage
|
|
55
|
+
from metadata.generated.schema.type.entityReferenceList import EntityReferenceList
|
|
50
56
|
from metadata.ingestion.api.models import Either
|
|
51
57
|
from metadata.ingestion.api.steps import InvalidSourceException
|
|
58
|
+
from metadata.ingestion.lineage.models import Dialect
|
|
59
|
+
from metadata.ingestion.lineage.parser import LineageParser
|
|
52
60
|
from metadata.ingestion.ometa.ometa_api import OpenMetadata
|
|
53
61
|
from metadata.ingestion.ometa.utils import model_str
|
|
54
62
|
from metadata.ingestion.source.dashboard.dashboard_service import DashboardServiceSource
|
|
63
|
+
from metadata.ingestion.source.dashboard.powerbi.databricks_parser import (
|
|
64
|
+
parse_databricks_native_query_source,
|
|
65
|
+
)
|
|
55
66
|
from metadata.ingestion.source.dashboard.powerbi.models import (
|
|
67
|
+
Dataflow,
|
|
56
68
|
Dataset,
|
|
57
69
|
Group,
|
|
58
70
|
PowerBIDashboard,
|
|
71
|
+
PowerBiMeasureModel,
|
|
59
72
|
PowerBIReport,
|
|
60
73
|
PowerBiTable,
|
|
74
|
+
ReportPage,
|
|
61
75
|
)
|
|
62
76
|
from metadata.ingestion.source.database.column_type_parser import ColumnTypeParser
|
|
63
77
|
from metadata.utils import fqn
|
|
@@ -65,13 +79,17 @@ from metadata.utils.filters import (
|
|
|
65
79
|
filter_by_chart,
|
|
66
80
|
filter_by_dashboard,
|
|
67
81
|
filter_by_datamodel,
|
|
68
|
-
filter_by_project,
|
|
69
82
|
)
|
|
83
|
+
from metadata.utils.fqn import build_es_fqn_search_string
|
|
70
84
|
from metadata.utils.helpers import clean_uri
|
|
71
85
|
from metadata.utils.logger import ingestion_logger
|
|
72
86
|
|
|
73
87
|
logger = ingestion_logger()
|
|
74
88
|
|
|
89
|
+
OWNER_ACCESS_RIGHTS_KEYWORDS = ["owner", "write", "admin"]
|
|
90
|
+
SNOWFLAKE_QUERY_EXPRESSION_KW = "Value.NativeQuery(Snowflake.Databases("
|
|
91
|
+
DATABRICKS_QUERY_EXPRESSION_KW = "Value.NativeQuery(Databricks.Catalogs("
|
|
92
|
+
|
|
75
93
|
|
|
76
94
|
class PowerbiSource(DashboardServiceSource):
|
|
77
95
|
"""PowerBi Source Class"""
|
|
@@ -91,31 +109,111 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
91
109
|
self.workspace_data = []
|
|
92
110
|
self.datamodel_file_mappings = []
|
|
93
111
|
|
|
94
|
-
def prepare(self):
|
|
95
|
-
return super().prepare()
|
|
96
|
-
|
|
97
112
|
def close(self):
|
|
98
113
|
self.metadata.close()
|
|
99
114
|
if self.client.file_client:
|
|
100
115
|
self.client.file_client.delete_tmp_files()
|
|
101
116
|
|
|
102
|
-
def
|
|
117
|
+
def get_org_workspace_data(self) -> Iterable[Optional[Group]]:
|
|
103
118
|
"""
|
|
104
|
-
|
|
119
|
+
fetch all the workspace data for non-admin users
|
|
105
120
|
"""
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
|
|
121
|
+
filter_pattern = self.source_config.projectFilterPattern
|
|
122
|
+
workspaces = self.client.api_client.fetch_all_workspaces(filter_pattern)
|
|
123
|
+
for workspace in workspaces:
|
|
124
|
+
# add the dashboards to the workspace
|
|
125
|
+
workspace.dashboards.extend(
|
|
126
|
+
self.client.api_client.fetch_all_org_dashboards(group_id=workspace.id)
|
|
127
|
+
or []
|
|
128
|
+
)
|
|
129
|
+
for dashboard in workspace.dashboards:
|
|
130
|
+
# add the tiles to the dashboards
|
|
131
|
+
dashboard.tiles.extend(
|
|
132
|
+
self.client.api_client.fetch_all_org_tiles(
|
|
133
|
+
group_id=workspace.id, dashboard_id=dashboard.id
|
|
134
|
+
)
|
|
135
|
+
or []
|
|
136
|
+
)
|
|
137
|
+
|
|
138
|
+
# add the reports to the workspaces
|
|
139
|
+
workspace.reports.extend(
|
|
140
|
+
self.client.api_client.fetch_all_org_reports(group_id=workspace.id)
|
|
141
|
+
or []
|
|
142
|
+
)
|
|
143
|
+
|
|
144
|
+
# add the datasets to the workspaces
|
|
145
|
+
workspace.datasets.extend(
|
|
146
|
+
self.client.api_client.fetch_all_org_datasets(group_id=workspace.id)
|
|
147
|
+
or []
|
|
148
|
+
)
|
|
149
|
+
for dataset in workspace.datasets:
|
|
150
|
+
# add the tables to the datasets
|
|
151
|
+
dataset.tables.extend(
|
|
152
|
+
self.client.api_client.fetch_dataset_tables(
|
|
153
|
+
group_id=workspace.id, dataset_id=dataset.id
|
|
154
|
+
)
|
|
155
|
+
or []
|
|
115
156
|
)
|
|
116
|
-
|
|
117
|
-
|
|
118
|
-
|
|
157
|
+
yield workspace
|
|
158
|
+
|
|
159
|
+
def get_admin_workspace_data(self) -> Iterable[Optional[Group]]:
|
|
160
|
+
"""
|
|
161
|
+
fetch all the workspace data
|
|
162
|
+
"""
|
|
163
|
+
filter_pattern = self.source_config.projectFilterPattern
|
|
164
|
+
workspaces = self.client.api_client.fetch_all_workspaces(filter_pattern)
|
|
165
|
+
if workspaces:
|
|
166
|
+
workspace_id_list = [workspace.id for workspace in workspaces]
|
|
167
|
+
|
|
168
|
+
# Start the scan of the available workspaces for dashboard metadata
|
|
169
|
+
workspace_paginated_list = [
|
|
170
|
+
workspace_id_list[i : i + self.pagination_entity_per_page]
|
|
171
|
+
for i in range(
|
|
172
|
+
0, len(workspace_id_list), self.pagination_entity_per_page
|
|
173
|
+
)
|
|
174
|
+
]
|
|
175
|
+
count = 1
|
|
176
|
+
for workspace_ids_chunk in workspace_paginated_list:
|
|
177
|
+
logger.info(
|
|
178
|
+
f"Scanning {count}/{len(workspace_paginated_list)} set of workspaces"
|
|
179
|
+
)
|
|
180
|
+
workspace_scan = self.client.api_client.initiate_workspace_scan(
|
|
181
|
+
workspace_ids_chunk
|
|
182
|
+
)
|
|
183
|
+
if not workspace_scan:
|
|
184
|
+
logger.error(
|
|
185
|
+
f"Error initiating workspace scan for ids:{str(workspace_ids_chunk)}\n moving to next set of workspaces"
|
|
186
|
+
)
|
|
187
|
+
count += 1
|
|
188
|
+
continue
|
|
189
|
+
|
|
190
|
+
# Keep polling the scan status endpoint to check if scan is succeeded
|
|
191
|
+
workspace_scan_status = self.client.api_client.wait_for_scan_complete(
|
|
192
|
+
scan_id=workspace_scan.id
|
|
193
|
+
)
|
|
194
|
+
if not workspace_scan_status:
|
|
195
|
+
logger.error(
|
|
196
|
+
f"Max poll hit to scan status for scan_id: {workspace_scan.id}, moving to next set of workspaces"
|
|
197
|
+
)
|
|
198
|
+
count += 1
|
|
199
|
+
continue
|
|
200
|
+
|
|
201
|
+
# Get scan result for successfull scan
|
|
202
|
+
response = self.client.api_client.fetch_workspace_scan_result(
|
|
203
|
+
scan_id=workspace_scan.id
|
|
204
|
+
)
|
|
205
|
+
if not response:
|
|
206
|
+
logger.error(
|
|
207
|
+
f"Error getting workspace scan result for scan_id: {workspace_scan.id}"
|
|
208
|
+
)
|
|
209
|
+
count += 1
|
|
210
|
+
continue
|
|
211
|
+
for active_workspace in response.workspaces:
|
|
212
|
+
if active_workspace.state == "Active":
|
|
213
|
+
yield active_workspace
|
|
214
|
+
count += 1
|
|
215
|
+
else:
|
|
216
|
+
logger.error("Unable to fetch any PowerBI workspaces")
|
|
119
217
|
|
|
120
218
|
@classmethod
|
|
121
219
|
def create(
|
|
@@ -129,62 +227,31 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
129
227
|
)
|
|
130
228
|
return cls(config, metadata)
|
|
131
229
|
|
|
132
|
-
def
|
|
133
|
-
"""
|
|
134
|
-
|
|
135
|
-
|
|
136
|
-
|
|
137
|
-
|
|
138
|
-
|
|
139
|
-
|
|
140
|
-
|
|
141
|
-
|
|
142
|
-
|
|
143
|
-
|
|
144
|
-
or []
|
|
145
|
-
)
|
|
146
|
-
|
|
147
|
-
# add the reports to the groups
|
|
148
|
-
workspace.reports.extend(
|
|
149
|
-
self.client.api_client.fetch_all_org_reports(group_id=workspace.id) or []
|
|
150
|
-
)
|
|
151
|
-
|
|
152
|
-
# add the datasets to the groups
|
|
153
|
-
workspace.datasets.extend(
|
|
154
|
-
self.client.api_client.fetch_all_org_datasets(group_id=workspace.id) or []
|
|
155
|
-
)
|
|
156
|
-
for dataset in workspace.datasets:
|
|
157
|
-
# add the tables to the datasets
|
|
158
|
-
dataset.tables.extend(
|
|
159
|
-
self.client.api_client.fetch_dataset_tables(
|
|
160
|
-
group_id=workspace.id, dataset_id=dataset.id
|
|
161
|
-
)
|
|
162
|
-
or []
|
|
163
|
-
)
|
|
230
|
+
def _prepare_workspace_data(self):
|
|
231
|
+
"""
|
|
232
|
+
- Since we get all the required info i.e. reports, dashboards, charts, datasets
|
|
233
|
+
with workflow scan approach, we are populating bulk data for workspace.
|
|
234
|
+
- Some individual APIs are not able to yield data with details.
|
|
235
|
+
"""
|
|
236
|
+
if self.service_connection.useAdminApis:
|
|
237
|
+
for workspace in self.get_admin_workspace_data():
|
|
238
|
+
yield workspace
|
|
239
|
+
else:
|
|
240
|
+
for workspace in self.get_org_workspace_data():
|
|
241
|
+
yield workspace
|
|
164
242
|
|
|
165
243
|
def get_dashboard(self) -> Any:
|
|
166
244
|
"""
|
|
167
245
|
Method to iterate through dashboard lists filter dashboards & yield dashboard details
|
|
168
246
|
"""
|
|
169
|
-
|
|
170
|
-
all_workspaces = self.client.api_client.fetch_all_workspaces() or []
|
|
171
|
-
all_workspaces = self.get_filtered_workspaces(all_workspaces)
|
|
172
|
-
for workspace in all_workspaces:
|
|
173
|
-
# prepare additional data for specific workspace (datasets, reports, dashboards)
|
|
174
|
-
self.prepare_workspace_data(workspace)
|
|
175
|
-
self.context.get().workspace = workspace
|
|
247
|
+
for workspace in self._prepare_workspace_data():
|
|
176
248
|
self.workspace_data.append(workspace)
|
|
177
|
-
|
|
178
|
-
|
|
179
|
-
|
|
180
|
-
|
|
181
|
-
|
|
182
|
-
logger.warning(
|
|
183
|
-
f"Cannot extract dashboard details from {dashboard}: {exc}"
|
|
184
|
-
)
|
|
185
|
-
continue
|
|
249
|
+
self.context.get().workspace = workspace
|
|
250
|
+
self.filtered_dashboards = []
|
|
251
|
+
self.filtered_datamodels = []
|
|
252
|
+
for dashboard in self.get_dashboards_list() or []:
|
|
253
|
+
dashboard_details = self.get_dashboard_details(dashboard)
|
|
186
254
|
dashboard_name = self.get_dashboard_name(dashboard_details)
|
|
187
|
-
|
|
188
255
|
if filter_by_dashboard(
|
|
189
256
|
self.source_config.dashboardFilterPattern,
|
|
190
257
|
dashboard_name,
|
|
@@ -194,7 +261,8 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
194
261
|
"Dashboard Filtered Out",
|
|
195
262
|
)
|
|
196
263
|
continue
|
|
197
|
-
|
|
264
|
+
self.filtered_dashboards.append(dashboard_details)
|
|
265
|
+
yield workspace
|
|
198
266
|
|
|
199
267
|
def get_dashboards_list(
|
|
200
268
|
self,
|
|
@@ -231,16 +299,48 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
231
299
|
"""
|
|
232
300
|
return (
|
|
233
301
|
f"{clean_uri(self.service_connection.hostPort)}/groups/"
|
|
234
|
-
f"{workspace_id}/dashboards/{dashboard_id}"
|
|
302
|
+
f"{workspace_id}/dashboards/{dashboard_id}?experience=power-bi"
|
|
235
303
|
)
|
|
236
304
|
|
|
237
305
|
def _get_report_url(self, workspace_id: str, dashboard_id: str) -> str:
|
|
238
306
|
"""
|
|
239
307
|
Method to build the dashboard url
|
|
240
308
|
"""
|
|
309
|
+
page_id = ""
|
|
310
|
+
try:
|
|
311
|
+
pages: Optional[
|
|
312
|
+
List[ReportPage]
|
|
313
|
+
] = self.client.api_client.fetch_report_pages(workspace_id, dashboard_id)
|
|
314
|
+
if len(pages) >= 1:
|
|
315
|
+
# get first page out of multiple pages otherwise
|
|
316
|
+
# get page if of single page
|
|
317
|
+
page_id = pages[0].name
|
|
318
|
+
page_id = f"/{page_id}" if page_id else ""
|
|
319
|
+
except Exception as exc:
|
|
320
|
+
logger.debug(traceback.format_exc())
|
|
321
|
+
logger.warning(f"Error building report page url: {exc}")
|
|
322
|
+
# https://app.powerbi.com/groups/4e57dcbb-***/reports/a2902011-***/098b***?experience=power-bi
|
|
323
|
+
return (
|
|
324
|
+
f"{clean_uri(self.service_connection.hostPort)}/groups/"
|
|
325
|
+
f"{workspace_id}/reports/{dashboard_id}{page_id}?experience=power-bi"
|
|
326
|
+
)
|
|
327
|
+
|
|
328
|
+
def _get_dataset_url(self, workspace_id: str, dataset_id: str) -> str:
|
|
329
|
+
"""
|
|
330
|
+
Method to build the dataset url
|
|
331
|
+
"""
|
|
241
332
|
return (
|
|
242
333
|
f"{clean_uri(self.service_connection.hostPort)}/groups/"
|
|
243
|
-
f"{workspace_id}/
|
|
334
|
+
f"{workspace_id}/datasets/{dataset_id}/details?experience=power-bi"
|
|
335
|
+
)
|
|
336
|
+
|
|
337
|
+
def _get_dataflow_url(self, workspace_id: str, dataflow_id: str) -> str:
|
|
338
|
+
"""
|
|
339
|
+
Method to build the dataset url
|
|
340
|
+
"""
|
|
341
|
+
return (
|
|
342
|
+
f"{clean_uri(self.service_connection.hostPort)}/groups/"
|
|
343
|
+
f"{workspace_id}/dataflows/{dataflow_id}?experience=power-bi"
|
|
244
344
|
)
|
|
245
345
|
|
|
246
346
|
def _get_chart_url(
|
|
@@ -257,45 +357,147 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
257
357
|
f"{workspace_id}/{chart_url_postfix}"
|
|
258
358
|
)
|
|
259
359
|
|
|
260
|
-
def
|
|
261
|
-
self, dashboard_details:
|
|
360
|
+
def yield_dashboard(
|
|
361
|
+
self, dashboard_details: Group
|
|
262
362
|
) -> Iterable[Either[CreateDashboardRequest]]:
|
|
263
363
|
"""
|
|
264
|
-
Method to
|
|
364
|
+
Method to Get Dashboard Entity, Dashboard Charts & Lineage
|
|
265
365
|
"""
|
|
266
|
-
|
|
267
|
-
|
|
268
|
-
|
|
269
|
-
|
|
270
|
-
|
|
271
|
-
|
|
272
|
-
|
|
273
|
-
|
|
274
|
-
|
|
275
|
-
|
|
276
|
-
|
|
277
|
-
|
|
278
|
-
|
|
279
|
-
|
|
280
|
-
|
|
281
|
-
|
|
282
|
-
|
|
283
|
-
|
|
284
|
-
|
|
285
|
-
|
|
286
|
-
|
|
366
|
+
try:
|
|
367
|
+
for dashboard in self.filtered_dashboards or []:
|
|
368
|
+
dashboard_details = self.get_dashboard_details(dashboard)
|
|
369
|
+
if isinstance(dashboard_details, PowerBIDashboard):
|
|
370
|
+
dashboard_request = CreateDashboardRequest(
|
|
371
|
+
name=EntityName(dashboard_details.id),
|
|
372
|
+
sourceUrl=SourceUrl(
|
|
373
|
+
self._get_dashboard_url(
|
|
374
|
+
workspace_id=self.context.get().workspace.id,
|
|
375
|
+
dashboard_id=dashboard_details.id,
|
|
376
|
+
)
|
|
377
|
+
),
|
|
378
|
+
project=self.get_project_name(dashboard_details),
|
|
379
|
+
displayName=dashboard_details.displayName,
|
|
380
|
+
dashboardType=DashboardType.Dashboard,
|
|
381
|
+
charts=[
|
|
382
|
+
FullyQualifiedEntityName(
|
|
383
|
+
fqn.build(
|
|
384
|
+
self.metadata,
|
|
385
|
+
entity_type=Chart,
|
|
386
|
+
service_name=self.context.get().dashboard_service,
|
|
387
|
+
chart_name=chart,
|
|
388
|
+
)
|
|
389
|
+
)
|
|
390
|
+
for chart in self.context.get().charts or []
|
|
391
|
+
],
|
|
392
|
+
service=FullyQualifiedEntityName(
|
|
393
|
+
self.context.get().dashboard_service
|
|
394
|
+
),
|
|
395
|
+
owners=self.get_owner_ref(dashboard_details=dashboard_details),
|
|
396
|
+
)
|
|
397
|
+
else:
|
|
398
|
+
dashboard_request = CreateDashboardRequest(
|
|
399
|
+
name=EntityName(dashboard_details.id),
|
|
400
|
+
dashboardType=DashboardType.Report,
|
|
401
|
+
sourceUrl=SourceUrl(
|
|
402
|
+
self._get_report_url(
|
|
403
|
+
workspace_id=self.context.get().workspace.id,
|
|
404
|
+
dashboard_id=dashboard_details.id,
|
|
405
|
+
)
|
|
406
|
+
),
|
|
407
|
+
project=self.get_project_name(dashboard_details),
|
|
408
|
+
displayName=dashboard_details.name,
|
|
409
|
+
service=self.context.get().dashboard_service,
|
|
410
|
+
owners=self.get_owner_ref(dashboard_details=dashboard_details),
|
|
411
|
+
)
|
|
412
|
+
yield Either(right=dashboard_request)
|
|
413
|
+
self.register_record(dashboard_request=dashboard_request)
|
|
414
|
+
except Exception as exc: # pylint: disable=broad-except
|
|
415
|
+
yield Either(
|
|
416
|
+
left=StackTraceError(
|
|
417
|
+
name=dashboard_details.name,
|
|
418
|
+
error=f"Error creating dashboard [{dashboard_details}]: {exc}",
|
|
419
|
+
stackTrace=traceback.format_exc(),
|
|
287
420
|
)
|
|
288
|
-
|
|
289
|
-
self.register_record_datamodel(datamodel_request=data_model_request)
|
|
421
|
+
)
|
|
290
422
|
|
|
291
|
-
|
|
292
|
-
|
|
293
|
-
|
|
294
|
-
|
|
295
|
-
|
|
296
|
-
|
|
297
|
-
|
|
423
|
+
def yield_dashboard_chart(
|
|
424
|
+
self, dashboard_details: Group
|
|
425
|
+
) -> Iterable[Either[CreateChartRequest]]:
|
|
426
|
+
"""Get chart method
|
|
427
|
+
Args:
|
|
428
|
+
dashboard_details:
|
|
429
|
+
Returns:
|
|
430
|
+
Iterable[Chart]
|
|
431
|
+
"""
|
|
432
|
+
for dashboard in self.filtered_dashboards or []:
|
|
433
|
+
dashboard_details = self.get_dashboard_details(dashboard)
|
|
434
|
+
if isinstance(dashboard_details, PowerBIDashboard):
|
|
435
|
+
charts = dashboard_details.tiles
|
|
436
|
+
for chart in charts or []:
|
|
437
|
+
try:
|
|
438
|
+
chart_title = chart.title
|
|
439
|
+
chart_display_name = chart_title if chart_title else chart.id
|
|
440
|
+
if filter_by_chart(
|
|
441
|
+
self.source_config.chartFilterPattern, chart_display_name
|
|
442
|
+
):
|
|
443
|
+
self.status.filter(
|
|
444
|
+
chart_display_name, "Chart Pattern not Allowed"
|
|
445
|
+
)
|
|
446
|
+
continue
|
|
447
|
+
yield Either(
|
|
448
|
+
right=CreateChartRequest(
|
|
449
|
+
name=EntityName(chart.id),
|
|
450
|
+
displayName=chart_display_name,
|
|
451
|
+
chartType=ChartType.Other.value,
|
|
452
|
+
sourceUrl=SourceUrl(
|
|
453
|
+
self._get_chart_url(
|
|
454
|
+
report_id=chart.reportId,
|
|
455
|
+
workspace_id=self.context.get().workspace.id,
|
|
456
|
+
dashboard_id=dashboard_details.id,
|
|
457
|
+
)
|
|
458
|
+
),
|
|
459
|
+
service=FullyQualifiedEntityName(
|
|
460
|
+
self.context.get().dashboard_service
|
|
461
|
+
),
|
|
462
|
+
)
|
|
463
|
+
)
|
|
464
|
+
except Exception as exc:
|
|
465
|
+
yield Either(
|
|
466
|
+
left=StackTraceError(
|
|
467
|
+
name=chart.title,
|
|
468
|
+
error=f"Error creating chart [{chart.title}]: {exc}",
|
|
469
|
+
stackTrace=traceback.format_exc(),
|
|
470
|
+
)
|
|
471
|
+
)
|
|
472
|
+
|
|
473
|
+
def _get_child_measures(self, table: PowerBiTable) -> List[Column]:
|
|
474
|
+
"""
|
|
475
|
+
Extract the measures of the table
|
|
476
|
+
"""
|
|
477
|
+
measures = []
|
|
478
|
+
for measure in table.measures or []:
|
|
479
|
+
try:
|
|
480
|
+
measure_type = DataType.MEASURE_VISIBLE
|
|
481
|
+
if measure.isHidden:
|
|
482
|
+
measure_type = DataType.MEASURE_HIDDEN
|
|
483
|
+
expression_text = f"Expression : {measure.expression}"
|
|
484
|
+
description_text = (
|
|
485
|
+
f"Description : {measure.description}"
|
|
486
|
+
if measure.description
|
|
487
|
+
else ""
|
|
488
|
+
)
|
|
489
|
+
description_field_text = f"{expression_text}\n\n{description_text}"
|
|
490
|
+
parsed_measure = PowerBiMeasureModel(
|
|
491
|
+
dataType=measure_type,
|
|
492
|
+
dataTypeDisplay=measure_type,
|
|
493
|
+
name=measure.name,
|
|
494
|
+
description=description_field_text,
|
|
298
495
|
)
|
|
496
|
+
measures.append(Column(**parsed_measure.model_dump()))
|
|
497
|
+
except Exception as err:
|
|
498
|
+
logger.debug(traceback.format_exc())
|
|
499
|
+
logger.debug(f"Error processing datamodel nested measure: {err}")
|
|
500
|
+
return measures
|
|
299
501
|
|
|
300
502
|
def _get_child_columns(self, table: PowerBiTable) -> List[Column]:
|
|
301
503
|
"""
|
|
@@ -305,14 +507,15 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
305
507
|
for column in table.columns or []:
|
|
306
508
|
try:
|
|
307
509
|
parsed_column = {
|
|
308
|
-
"dataTypeDisplay":
|
|
309
|
-
|
|
310
|
-
|
|
510
|
+
"dataTypeDisplay": (
|
|
511
|
+
column.dataType if column.dataType else DataType.UNKNOWN.value
|
|
512
|
+
),
|
|
311
513
|
"dataType": ColumnTypeParser.get_column_type(
|
|
312
514
|
column.dataType if column.dataType else None
|
|
313
515
|
),
|
|
314
516
|
"name": column.name,
|
|
315
517
|
"displayName": column.name,
|
|
518
|
+
"description": column.description,
|
|
316
519
|
}
|
|
317
520
|
if column.dataType and column.dataType == DataType.ARRAY.value:
|
|
318
521
|
parsed_column["arrayDataType"] = DataType.UNKNOWN
|
|
@@ -327,79 +530,103 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
327
530
|
datasource_columns = []
|
|
328
531
|
for table in dataset.tables or []:
|
|
329
532
|
try:
|
|
533
|
+
table_display_name = None
|
|
534
|
+
if self.service_connection.displayTableNameFromSource:
|
|
535
|
+
table_display_name = self.parse_table_name_from_source(table=table)
|
|
536
|
+
if table_display_name:
|
|
537
|
+
logger.debug(
|
|
538
|
+
f"Parsed Table display name: {table_display_name} for table: {table.name}"
|
|
539
|
+
)
|
|
540
|
+
if not table_display_name:
|
|
541
|
+
table_display_name = table.name
|
|
330
542
|
parsed_table = {
|
|
331
543
|
"dataTypeDisplay": "PowerBI Table",
|
|
332
544
|
"dataType": DataType.TABLE,
|
|
333
545
|
"name": table.name,
|
|
334
|
-
"displayName":
|
|
546
|
+
"displayName": table_display_name,
|
|
335
547
|
"description": table.description,
|
|
548
|
+
"children": [],
|
|
336
549
|
}
|
|
337
550
|
child_columns = self._get_child_columns(table=table)
|
|
551
|
+
child_measures = self._get_child_measures(table=table)
|
|
338
552
|
if child_columns:
|
|
339
553
|
parsed_table["children"] = child_columns
|
|
554
|
+
if child_measures:
|
|
555
|
+
parsed_table["children"].extend(child_measures)
|
|
340
556
|
datasource_columns.append(Column(**parsed_table))
|
|
341
557
|
except Exception as exc:
|
|
342
558
|
logger.debug(traceback.format_exc())
|
|
343
559
|
logger.warning(f"Error to yield datamodel column: {exc}")
|
|
344
560
|
return datasource_columns
|
|
345
561
|
|
|
346
|
-
def
|
|
347
|
-
self, dashboard_details: Union[PowerBIDashboard, PowerBIReport]
|
|
348
|
-
) -> Iterable[Either[CreateDashboardRequest]]:
|
|
562
|
+
def _get_datamodels_list(self) -> List[Union[Dataset, Dataflow]]:
|
|
349
563
|
"""
|
|
350
|
-
|
|
564
|
+
Get All the Powerbi Datasets
|
|
565
|
+
"""
|
|
566
|
+
return (
|
|
567
|
+
self.context.get().workspace.datasets
|
|
568
|
+
+ self.context.get().workspace.dataflows
|
|
569
|
+
)
|
|
570
|
+
|
|
571
|
+
def yield_datamodel(
|
|
572
|
+
self, dashboard_details: Group
|
|
573
|
+
) -> Iterable[Either[CreateDashboardDataModelRequest]]:
|
|
574
|
+
"""
|
|
575
|
+
Get All the Powerbi Datasets
|
|
351
576
|
"""
|
|
352
577
|
try:
|
|
353
|
-
if
|
|
354
|
-
|
|
355
|
-
|
|
356
|
-
|
|
357
|
-
|
|
578
|
+
if self.source_config.includeDataModels:
|
|
579
|
+
for dataset in self._get_datamodels_list() or []:
|
|
580
|
+
if filter_by_datamodel(
|
|
581
|
+
self.source_config.dataModelFilterPattern, dataset.name
|
|
582
|
+
):
|
|
583
|
+
self.status.filter(dataset.name, "Data model filtered out.")
|
|
584
|
+
continue
|
|
585
|
+
self.filtered_datamodels.append(dataset)
|
|
586
|
+
if isinstance(dataset, Dataset):
|
|
587
|
+
data_model_type = DataModelType.PowerBIDataModel.value
|
|
588
|
+
datamodel_columns = self._get_column_info(dataset)
|
|
589
|
+
source_url = self._get_dataset_url(
|
|
358
590
|
workspace_id=self.context.get().workspace.id,
|
|
359
|
-
|
|
591
|
+
dataset_id=dataset.id,
|
|
360
592
|
)
|
|
361
|
-
)
|
|
362
|
-
|
|
363
|
-
|
|
364
|
-
|
|
365
|
-
charts=[
|
|
366
|
-
FullyQualifiedEntityName(
|
|
367
|
-
fqn.build(
|
|
368
|
-
self.metadata,
|
|
369
|
-
entity_type=Chart,
|
|
370
|
-
service_name=self.context.get().dashboard_service,
|
|
371
|
-
chart_name=chart,
|
|
372
|
-
)
|
|
373
|
-
)
|
|
374
|
-
for chart in self.context.get().charts or []
|
|
375
|
-
],
|
|
376
|
-
service=FullyQualifiedEntityName(
|
|
377
|
-
self.context.get().dashboard_service
|
|
378
|
-
),
|
|
379
|
-
owners=self.get_owner_ref(dashboard_details=dashboard_details),
|
|
380
|
-
)
|
|
381
|
-
else:
|
|
382
|
-
dashboard_request = CreateDashboardRequest(
|
|
383
|
-
name=EntityName(dashboard_details.id),
|
|
384
|
-
dashboardType=DashboardType.Report,
|
|
385
|
-
sourceUrl=SourceUrl(
|
|
386
|
-
self._get_report_url(
|
|
593
|
+
elif isinstance(dataset, Dataflow):
|
|
594
|
+
data_model_type = DataModelType.PowerBIDataFlow.value
|
|
595
|
+
datamodel_columns = []
|
|
596
|
+
source_url = self._get_dataflow_url(
|
|
387
597
|
workspace_id=self.context.get().workspace.id,
|
|
388
|
-
|
|
598
|
+
dataflow_id=dataset.id,
|
|
389
599
|
)
|
|
390
|
-
|
|
391
|
-
|
|
392
|
-
|
|
393
|
-
|
|
394
|
-
|
|
395
|
-
|
|
396
|
-
|
|
397
|
-
|
|
398
|
-
|
|
600
|
+
else:
|
|
601
|
+
logger.warning(
|
|
602
|
+
f"Unknown dataset type: {type(dataset)}, name: {dataset.name}"
|
|
603
|
+
)
|
|
604
|
+
continue
|
|
605
|
+
data_model_request = CreateDashboardDataModelRequest(
|
|
606
|
+
name=EntityName(dataset.id),
|
|
607
|
+
displayName=dataset.name,
|
|
608
|
+
description=(
|
|
609
|
+
Markdown(dataset.description)
|
|
610
|
+
if dataset.description
|
|
611
|
+
else None
|
|
612
|
+
),
|
|
613
|
+
service=FullyQualifiedEntityName(
|
|
614
|
+
self.context.get().dashboard_service
|
|
615
|
+
),
|
|
616
|
+
dataModelType=data_model_type,
|
|
617
|
+
serviceType=DashboardServiceType.PowerBI.value,
|
|
618
|
+
columns=datamodel_columns,
|
|
619
|
+
project=self.get_project_name(dashboard_details=dataset),
|
|
620
|
+
owners=self.get_owner_ref(dashboard_details=dataset),
|
|
621
|
+
sourceUrl=SourceUrl(source_url),
|
|
622
|
+
)
|
|
623
|
+
yield Either(right=data_model_request)
|
|
624
|
+
self.register_record_datamodel(datamodel_request=data_model_request)
|
|
625
|
+
except Exception as exc:
|
|
399
626
|
yield Either(
|
|
400
627
|
left=StackTraceError(
|
|
401
|
-
name=
|
|
402
|
-
error=f"Error
|
|
628
|
+
name=dataset.name,
|
|
629
|
+
error=f"Error yielding Data Model [{dataset.name}]: {exc}",
|
|
403
630
|
stackTrace=traceback.format_exc(),
|
|
404
631
|
)
|
|
405
632
|
)
|
|
@@ -441,18 +668,21 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
441
668
|
except Exception as exc: # pylint: disable=broad-except
|
|
442
669
|
yield Either(
|
|
443
670
|
left=StackTraceError(
|
|
444
|
-
name="Lineage",
|
|
671
|
+
name="Report and Dashboard Lineage",
|
|
445
672
|
error=f"Error to yield report and dashboard lineage details: {exc}",
|
|
446
673
|
stackTrace=traceback.format_exc(),
|
|
447
674
|
)
|
|
448
675
|
)
|
|
449
676
|
|
|
450
677
|
def create_datamodel_report_lineage(
|
|
451
|
-
self,
|
|
678
|
+
self,
|
|
679
|
+
db_service_prefix: Optional[str],
|
|
680
|
+
dashboard_details: PowerBIReport,
|
|
452
681
|
) -> Iterable[Either[CreateDashboardRequest]]:
|
|
453
682
|
"""
|
|
454
683
|
create the lineage between datamodel and report
|
|
455
684
|
"""
|
|
685
|
+
|
|
456
686
|
try:
|
|
457
687
|
report_fqn = fqn.build(
|
|
458
688
|
self.metadata,
|
|
@@ -464,14 +694,12 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
464
694
|
entity=Dashboard,
|
|
465
695
|
fqn=report_fqn,
|
|
466
696
|
)
|
|
467
|
-
|
|
468
|
-
dataset = self._fetch_dataset_from_workspace(dashboard_details.datasetId)
|
|
469
|
-
if dataset:
|
|
697
|
+
if dashboard_details.datasetId:
|
|
470
698
|
datamodel_fqn = fqn.build(
|
|
471
699
|
self.metadata,
|
|
472
700
|
entity_type=DashboardDataModel,
|
|
473
701
|
service_name=self.context.get().dashboard_service,
|
|
474
|
-
data_model_name=
|
|
702
|
+
data_model_name=dashboard_details.datasetId,
|
|
475
703
|
)
|
|
476
704
|
datamodel_entity = self.metadata.get_by_name(
|
|
477
705
|
entity=DashboardDataModel,
|
|
@@ -482,27 +710,18 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
482
710
|
yield self._get_add_lineage_request(
|
|
483
711
|
to_entity=report_entity, from_entity=datamodel_entity
|
|
484
712
|
)
|
|
713
|
+
else:
|
|
714
|
+
logger.debug(
|
|
715
|
+
f"Skipping datamodel and report lineage for"
|
|
716
|
+
f" {dashboard_details.id} as datasetId is not found"
|
|
717
|
+
)
|
|
485
718
|
|
|
486
|
-
for table in dataset.tables or []:
|
|
487
|
-
yield self._get_table_and_datamodel_lineage(
|
|
488
|
-
db_service_name=db_service_name,
|
|
489
|
-
table=table,
|
|
490
|
-
datamodel_entity=datamodel_entity,
|
|
491
|
-
)
|
|
492
|
-
|
|
493
|
-
# create the lineage between table and datamodel using the pbit files
|
|
494
|
-
if self.client.file_client:
|
|
495
|
-
yield from self.create_table_datamodel_lineage_from_files(
|
|
496
|
-
db_service_name=db_service_name,
|
|
497
|
-
datamodel_entity=datamodel_entity,
|
|
498
|
-
)
|
|
499
719
|
except Exception as exc: # pylint: disable=broad-except
|
|
500
720
|
yield Either(
|
|
501
721
|
left=StackTraceError(
|
|
502
|
-
name=f"
|
|
722
|
+
name=f"Datamodel and Report Lineage",
|
|
503
723
|
error=(
|
|
504
|
-
"Error to yield datamodel and report lineage details
|
|
505
|
-
f"service name [{db_service_name}]: {exc}"
|
|
724
|
+
f"Error to yield datamodel and report lineage details: {exc}"
|
|
506
725
|
),
|
|
507
726
|
stackTrace=traceback.format_exc(),
|
|
508
727
|
)
|
|
@@ -527,41 +746,421 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
527
746
|
logger.debug(f"Error to get data_model_column_fqn {exc}")
|
|
528
747
|
logger.debug(traceback.format_exc())
|
|
529
748
|
|
|
749
|
+
def parse_table_name_from_source(self, table: PowerBiTable) -> Optional[str]:
|
|
750
|
+
"""
|
|
751
|
+
Parse the snowflake table name
|
|
752
|
+
"""
|
|
753
|
+
try:
|
|
754
|
+
if not isinstance(table.source, list):
|
|
755
|
+
return None
|
|
756
|
+
source_expression = table.source[0].expression
|
|
757
|
+
if not source_expression:
|
|
758
|
+
logger.debug(f"No source expression found for table: {table.name}")
|
|
759
|
+
return None
|
|
760
|
+
|
|
761
|
+
if "Snowflake.Databases" in source_expression:
|
|
762
|
+
# snowflake expression
|
|
763
|
+
table_match = re.search(
|
|
764
|
+
r'\[Name=(?:"([^"]+)"|([^,]+)),Kind="Table"\]', source_expression
|
|
765
|
+
)
|
|
766
|
+
view_match = re.search(
|
|
767
|
+
r'\[Name=(?:"([^"]+)"|([^,]+)),Kind="View"\]', source_expression
|
|
768
|
+
)
|
|
769
|
+
table = table_match.group(1) if table_match else None
|
|
770
|
+
view = view_match.group(1) if view_match else None
|
|
771
|
+
return table if table else view
|
|
772
|
+
|
|
773
|
+
# other general expressions
|
|
774
|
+
table_match = re.findall(r'\[Name="([^"]+)"\]', source_expression)
|
|
775
|
+
table = None
|
|
776
|
+
if isinstance(table_match, list):
|
|
777
|
+
table = table_match[1] if len(table_match) > 1 else None
|
|
778
|
+
return table
|
|
779
|
+
except Exception as exc:
|
|
780
|
+
logger.debug(f"Error to parse display table name: {exc}")
|
|
781
|
+
logger.debug(traceback.format_exc())
|
|
782
|
+
return None
|
|
783
|
+
|
|
784
|
+
def _parse_expression_regex_exp(
|
|
785
|
+
self, match: re.Match, datamodel_entity: DashboardDataModel
|
|
786
|
+
) -> Optional[str]:
|
|
787
|
+
"""parse snowflake regex expression"""
|
|
788
|
+
try:
|
|
789
|
+
if not match:
|
|
790
|
+
return None
|
|
791
|
+
elif match.group(1):
|
|
792
|
+
return match.group(1)
|
|
793
|
+
elif match.group(2):
|
|
794
|
+
dataset = self._fetch_dataset_from_workspace(datamodel_entity.name.root)
|
|
795
|
+
if dataset and dataset.expressions:
|
|
796
|
+
# find keyword from dataset expressions
|
|
797
|
+
for dexpression in dataset.expressions:
|
|
798
|
+
if not dexpression.expression:
|
|
799
|
+
logger.debug(
|
|
800
|
+
f"No expression value found inside dataset"
|
|
801
|
+
f"({dataset.name}) expressions' name={dexpression.name}"
|
|
802
|
+
)
|
|
803
|
+
continue
|
|
804
|
+
if dexpression.name == match.group(2):
|
|
805
|
+
pattern = r'^"([^"]+)"\s+meta'
|
|
806
|
+
kw_match = re.search(pattern, dexpression.expression)
|
|
807
|
+
if kw_match:
|
|
808
|
+
return kw_match.group(1)
|
|
809
|
+
except Exception as exc:
|
|
810
|
+
logger.debug(f"Error to parse snowflake regex expression: {exc}")
|
|
811
|
+
logger.debug(traceback.format_exc())
|
|
812
|
+
return None
|
|
813
|
+
|
|
814
|
+
def _parse_redshift_source(self, source_expression: str) -> Optional[List[dict]]:
|
|
815
|
+
try:
|
|
816
|
+
db_match = re.search(
|
|
817
|
+
r'AmazonRedshift\.Database\("[^"]+","([^"]+)"\)', source_expression
|
|
818
|
+
)
|
|
819
|
+
if not db_match:
|
|
820
|
+
# not valid redshift source
|
|
821
|
+
return None
|
|
822
|
+
schema_table_match = re.findall(r'\[Name="([^"]+)"\]', source_expression)
|
|
823
|
+
|
|
824
|
+
database = db_match.group(1) if db_match else None
|
|
825
|
+
schema = table = None
|
|
826
|
+
if isinstance(schema_table_match, list):
|
|
827
|
+
schema = schema_table_match[0] if len(schema_table_match) > 0 else None
|
|
828
|
+
table = schema_table_match[1] if len(schema_table_match) > 1 else None
|
|
829
|
+
|
|
830
|
+
if table: # atlease table should be fetched
|
|
831
|
+
return [{"database": database, "schema": schema, "table": table}]
|
|
832
|
+
return None
|
|
833
|
+
except Exception as exc:
|
|
834
|
+
logger.debug(f"Error to parse redshift table source: {exc}")
|
|
835
|
+
logger.debug(traceback.format_exc())
|
|
836
|
+
return None
|
|
837
|
+
|
|
838
|
+
def _parse_snowflake_query_source(
|
|
839
|
+
self, source_expression: str
|
|
840
|
+
) -> Optional[List[dict]]:
|
|
841
|
+
"""
|
|
842
|
+
Parse snowflake query source
|
|
843
|
+
source expressions like `Value.NativeQuery(Snowflake.Databases())`
|
|
844
|
+
"""
|
|
845
|
+
try:
|
|
846
|
+
logger.debug(
|
|
847
|
+
f"parsing source expression through query parser: {source_expression[:100]}"
|
|
848
|
+
)
|
|
849
|
+
|
|
850
|
+
# Look for SQL query after [Data],
|
|
851
|
+
# The pattern needs to handle the concatenated strings with & operators
|
|
852
|
+
m = re.search(
|
|
853
|
+
r'\[Data\],\s*"(.+?)"(?:,\s*null|\s*\))',
|
|
854
|
+
source_expression,
|
|
855
|
+
re.IGNORECASE | re.DOTALL,
|
|
856
|
+
)
|
|
857
|
+
|
|
858
|
+
if not m:
|
|
859
|
+
logger.debug("sql query not found in source expression")
|
|
860
|
+
return None
|
|
861
|
+
|
|
862
|
+
# Extract and clean the SQL query
|
|
863
|
+
sql_query = m.group(1).replace('""', '"')
|
|
864
|
+
|
|
865
|
+
# Handle PowerBI parameter concatenation (e.g., "& Database &")
|
|
866
|
+
# For now, replace parameter references with wildcards for parsing
|
|
867
|
+
sql_query_cleaned = re.sub(r'"?\s*&\s*\w+\s*&\s*"?\.?', "%.", sql_query)
|
|
868
|
+
|
|
869
|
+
logger.debug(f"Extracted SQL query: {sql_query}")
|
|
870
|
+
logger.debug(f"Cleaned SQL query: {sql_query_cleaned}")
|
|
871
|
+
|
|
872
|
+
if not sql_query_cleaned:
|
|
873
|
+
logger.debug("Empty SQL query after extraction")
|
|
874
|
+
return None
|
|
875
|
+
|
|
876
|
+
# Clean the query for parser
|
|
877
|
+
# 1. Replace % with a placeholder database name
|
|
878
|
+
parser_query = sql_query_cleaned.replace("%", "PLACEHOLDER_DB")
|
|
879
|
+
|
|
880
|
+
# 2. Remove PowerBI line feed markers #(lf) and clean up the query
|
|
881
|
+
parser_query = parser_query.replace("#(lf)", "\n")
|
|
882
|
+
|
|
883
|
+
# 3. Remove SQL comments that might cause issues (// style comments)
|
|
884
|
+
parser_query = re.sub(r"//[^\n]*", "", parser_query)
|
|
885
|
+
|
|
886
|
+
# 4. Clean up excessive whitespace
|
|
887
|
+
parser_query = re.sub(r"\s+", " ", parser_query).strip()
|
|
888
|
+
|
|
889
|
+
logger.debug(
|
|
890
|
+
f"Attempting LineageParser with cleaned query: {parser_query[:200]}"
|
|
891
|
+
)
|
|
892
|
+
|
|
893
|
+
try:
|
|
894
|
+
parser = LineageParser(
|
|
895
|
+
parser_query, dialect=Dialect.SNOWFLAKE, timeout_seconds=30
|
|
896
|
+
)
|
|
897
|
+
except Exception as parser_exc:
|
|
898
|
+
logger.debug(f"LineageParser failed with error: {parser_exc}")
|
|
899
|
+
logger.debug(f"Failed query was: {parser_query[:200]}...")
|
|
900
|
+
return None
|
|
901
|
+
|
|
902
|
+
if parser.source_tables:
|
|
903
|
+
logger.debug(
|
|
904
|
+
f"LineageParser found {len(parser.source_tables)} source table(s)"
|
|
905
|
+
)
|
|
906
|
+
for table in parser.source_tables:
|
|
907
|
+
logger.debug(
|
|
908
|
+
f"source table: {table.raw_name}, schema: {table.schema if hasattr(table, 'schema') else 'N/A'}"
|
|
909
|
+
)
|
|
910
|
+
lineage_tables_list = []
|
|
911
|
+
for source_table in parser.source_tables:
|
|
912
|
+
# source_table = parser.source_tables[0]
|
|
913
|
+
|
|
914
|
+
# Extract database from schema's parent if it exists
|
|
915
|
+
database = None
|
|
916
|
+
schema = None
|
|
917
|
+
|
|
918
|
+
if hasattr(source_table, "schema") and source_table.schema:
|
|
919
|
+
# Log what we have in the schema object
|
|
920
|
+
logger.debug(
|
|
921
|
+
f"Schema object type: {type(source_table.schema)}, value: {source_table.schema}"
|
|
922
|
+
)
|
|
923
|
+
|
|
924
|
+
# Get schema as string first
|
|
925
|
+
schema_str = (
|
|
926
|
+
source_table.schema.raw_name
|
|
927
|
+
if hasattr(source_table.schema, "raw_name")
|
|
928
|
+
else str(source_table.schema)
|
|
929
|
+
)
|
|
930
|
+
|
|
931
|
+
# If schema contains dots, it might be database.schema format
|
|
932
|
+
if "." in schema_str:
|
|
933
|
+
parts = schema_str.split(".")
|
|
934
|
+
if len(parts) == 2:
|
|
935
|
+
# Format: database.schema
|
|
936
|
+
# Check for placeholder (case insensitive)
|
|
937
|
+
database = (
|
|
938
|
+
parts[0]
|
|
939
|
+
if parts[0].upper() != "PLACEHOLDER_DB"
|
|
940
|
+
else None
|
|
941
|
+
)
|
|
942
|
+
schema = parts[1]
|
|
943
|
+
else:
|
|
944
|
+
# Just use as is
|
|
945
|
+
schema = schema_str
|
|
946
|
+
else:
|
|
947
|
+
schema = schema_str
|
|
948
|
+
# Check if schema has a parent (database)
|
|
949
|
+
if (
|
|
950
|
+
hasattr(source_table.schema, "parent")
|
|
951
|
+
and source_table.schema.parent
|
|
952
|
+
):
|
|
953
|
+
database = (
|
|
954
|
+
source_table.schema.parent.raw_name
|
|
955
|
+
if hasattr(source_table.schema.parent, "raw_name")
|
|
956
|
+
else str(source_table.schema.parent)
|
|
957
|
+
)
|
|
958
|
+
|
|
959
|
+
# Filter out placeholder values (case insensitive)
|
|
960
|
+
if database and database.upper() == "PLACEHOLDER_DB":
|
|
961
|
+
database = None
|
|
962
|
+
|
|
963
|
+
table = source_table.raw_name
|
|
964
|
+
|
|
965
|
+
if table:
|
|
966
|
+
logger.debug(f"tables found = {database}.{schema}.{table}")
|
|
967
|
+
lineage_tables_list.append(
|
|
968
|
+
{
|
|
969
|
+
"database": database,
|
|
970
|
+
"schema": schema,
|
|
971
|
+
"table": table,
|
|
972
|
+
}
|
|
973
|
+
)
|
|
974
|
+
return lineage_tables_list
|
|
975
|
+
logger.debug("tables in query not found through parser")
|
|
976
|
+
return None
|
|
977
|
+
except Exception as exc:
|
|
978
|
+
logger.debug(f"Error parsing snowflake query source: {exc}")
|
|
979
|
+
logger.debug(traceback.format_exc())
|
|
980
|
+
return None
|
|
981
|
+
|
|
982
|
+
def _parse_catalog_table_definition(
|
|
983
|
+
self, source_expression: str, datamodel_entity: DashboardDataModel
|
|
984
|
+
) -> Optional[List[dict]]:
|
|
985
|
+
"""parse catalog table definition"""
|
|
986
|
+
db_match = re.search(
|
|
987
|
+
r'\[Name=(?:"([^"]+)"|([^,]+)),Kind="Database"\]', source_expression
|
|
988
|
+
)
|
|
989
|
+
schema_match = re.search(
|
|
990
|
+
r'\[Name=(?:"([^"]+)"|([^,]+)),Kind="Schema"\]', source_expression
|
|
991
|
+
)
|
|
992
|
+
table_match = re.search(
|
|
993
|
+
r'\[Name=(?:"([^"]+)"|([^,]+)),Kind="Table"\]', source_expression
|
|
994
|
+
)
|
|
995
|
+
view_match = re.search(
|
|
996
|
+
r'\[Name=(?:"([^"]+)"|([^,]+)),Kind="View"\]', source_expression
|
|
997
|
+
)
|
|
998
|
+
try:
|
|
999
|
+
database = self._parse_expression_regex_exp(db_match, datamodel_entity)
|
|
1000
|
+
schema = self._parse_expression_regex_exp(schema_match, datamodel_entity)
|
|
1001
|
+
table = self._parse_expression_regex_exp(table_match, datamodel_entity)
|
|
1002
|
+
view = self._parse_expression_regex_exp(view_match, datamodel_entity)
|
|
1003
|
+
if table or view: # at least table or view should be present
|
|
1004
|
+
return [
|
|
1005
|
+
{
|
|
1006
|
+
"database": database,
|
|
1007
|
+
"schema": schema,
|
|
1008
|
+
"table": table if table else view,
|
|
1009
|
+
}
|
|
1010
|
+
]
|
|
1011
|
+
except Exception as exc:
|
|
1012
|
+
logger.debug(f"Error to parse databricks table source: {exc}")
|
|
1013
|
+
logger.debug(traceback.format_exc())
|
|
1014
|
+
return None
|
|
1015
|
+
|
|
1016
|
+
def _parse_databricks_source(
|
|
1017
|
+
self, source_expression: str, datamodel_entity: DashboardDataModel
|
|
1018
|
+
) -> Optional[List[dict]]:
|
|
1019
|
+
if "Databricks.Catalogs" not in source_expression:
|
|
1020
|
+
return None
|
|
1021
|
+
dataset = self._fetch_dataset_from_workspace(datamodel_entity.name.root)
|
|
1022
|
+
if dataset and dataset.expressions:
|
|
1023
|
+
try:
|
|
1024
|
+
if DATABRICKS_QUERY_EXPRESSION_KW in source_expression:
|
|
1025
|
+
return parse_databricks_native_query_source(source_expression)
|
|
1026
|
+
else:
|
|
1027
|
+
return self._parse_catalog_table_definition(
|
|
1028
|
+
source_expression, datamodel_entity
|
|
1029
|
+
)
|
|
1030
|
+
except Exception as exc:
|
|
1031
|
+
logger.debug(f"Error to parse databricks table source: {exc}")
|
|
1032
|
+
logger.debug(traceback.format_exc())
|
|
1033
|
+
return None
|
|
1034
|
+
|
|
1035
|
+
def _parse_snowflake_source(
|
|
1036
|
+
self, source_expression: str, datamodel_entity: DashboardDataModel
|
|
1037
|
+
) -> Optional[List[dict]]:
|
|
1038
|
+
try:
|
|
1039
|
+
if "Snowflake.Databases" not in source_expression:
|
|
1040
|
+
# Not a snowflake valid expression
|
|
1041
|
+
return None
|
|
1042
|
+
if SNOWFLAKE_QUERY_EXPRESSION_KW in source_expression:
|
|
1043
|
+
# snowflake query source identified
|
|
1044
|
+
return self._parse_snowflake_query_source(source_expression)
|
|
1045
|
+
return self._parse_catalog_table_definition(
|
|
1046
|
+
source_expression, datamodel_entity
|
|
1047
|
+
)
|
|
1048
|
+
except Exception as exc:
|
|
1049
|
+
logger.debug(f"Error to parse snowflake table source: {exc}")
|
|
1050
|
+
logger.debug(traceback.format_exc())
|
|
1051
|
+
return None
|
|
1052
|
+
|
|
1053
|
+
def _parse_table_info_from_source_exp(
|
|
1054
|
+
self, table: PowerBiTable, datamodel_entity: DashboardDataModel
|
|
1055
|
+
) -> Optional[List[dict]]:
|
|
1056
|
+
try:
|
|
1057
|
+
if not isinstance(table.source, list):
|
|
1058
|
+
return None
|
|
1059
|
+
source_expression = table.source[0].expression
|
|
1060
|
+
if not source_expression:
|
|
1061
|
+
logger.debug(f"No source expression found for table: {table.name}")
|
|
1062
|
+
return None
|
|
1063
|
+
# parse snowflake source
|
|
1064
|
+
table_info_list = self._parse_snowflake_source(
|
|
1065
|
+
source_expression, datamodel_entity
|
|
1066
|
+
)
|
|
1067
|
+
if isinstance(table_info_list, List):
|
|
1068
|
+
return table_info_list
|
|
1069
|
+
# parse redshift source
|
|
1070
|
+
table_info_list = self._parse_redshift_source(source_expression)
|
|
1071
|
+
if isinstance(table_info_list, List):
|
|
1072
|
+
return table_info_list
|
|
1073
|
+
|
|
1074
|
+
# parse databricks source
|
|
1075
|
+
table_info_list = self._parse_databricks_source(
|
|
1076
|
+
source_expression, datamodel_entity
|
|
1077
|
+
)
|
|
1078
|
+
if isinstance(table_info_list, List):
|
|
1079
|
+
return table_info_list
|
|
1080
|
+
|
|
1081
|
+
return None
|
|
1082
|
+
except Exception as exc:
|
|
1083
|
+
logger.debug(f"Error to parse table source: {exc}")
|
|
1084
|
+
logger.debug(traceback.format_exc())
|
|
1085
|
+
return None
|
|
1086
|
+
|
|
530
1087
|
def _get_table_and_datamodel_lineage(
|
|
531
1088
|
self,
|
|
532
|
-
|
|
1089
|
+
db_service_prefix: Optional[str],
|
|
533
1090
|
table: PowerBiTable,
|
|
534
1091
|
datamodel_entity: DashboardDataModel,
|
|
535
|
-
) ->
|
|
1092
|
+
) -> Iterable[Either[AddLineageRequest]]:
|
|
536
1093
|
"""
|
|
537
1094
|
Method to create lineage between table and datamodels
|
|
538
1095
|
"""
|
|
1096
|
+
(
|
|
1097
|
+
prefix_service_name,
|
|
1098
|
+
prefix_database_name,
|
|
1099
|
+
prefix_schema_name,
|
|
1100
|
+
prefix_table_name,
|
|
1101
|
+
) = self.parse_db_service_prefix(db_service_prefix)
|
|
1102
|
+
|
|
539
1103
|
try:
|
|
540
|
-
|
|
541
|
-
|
|
542
|
-
entity_type=Table,
|
|
543
|
-
service_name=db_service_name,
|
|
544
|
-
database_name=None,
|
|
545
|
-
schema_name=None,
|
|
546
|
-
table_name=table.name,
|
|
547
|
-
)
|
|
548
|
-
table_entity = self.metadata.get_by_name(
|
|
549
|
-
entity=Table,
|
|
550
|
-
fqn=table_fqn,
|
|
1104
|
+
table_info_list = self._parse_table_info_from_source_exp(
|
|
1105
|
+
table, datamodel_entity
|
|
551
1106
|
)
|
|
1107
|
+
if isinstance(table_info_list, List):
|
|
1108
|
+
for table_info in table_info_list:
|
|
1109
|
+
table_name = table_info.get("table") or table.name
|
|
1110
|
+
schema_name = table_info.get("schema")
|
|
1111
|
+
database_name = table_info.get("database")
|
|
1112
|
+
if (
|
|
1113
|
+
prefix_table_name
|
|
1114
|
+
and table_name
|
|
1115
|
+
and prefix_table_name.lower() != table_name.lower()
|
|
1116
|
+
):
|
|
1117
|
+
logger.debug(
|
|
1118
|
+
f"Table {table_name} does not match prefix {prefix_table_name}"
|
|
1119
|
+
)
|
|
1120
|
+
return
|
|
552
1121
|
|
|
553
|
-
|
|
554
|
-
|
|
555
|
-
|
|
556
|
-
|
|
557
|
-
|
|
558
|
-
|
|
559
|
-
|
|
560
|
-
|
|
561
|
-
|
|
562
|
-
|
|
1122
|
+
if (
|
|
1123
|
+
prefix_schema_name
|
|
1124
|
+
and schema_name
|
|
1125
|
+
and prefix_schema_name.lower() != schema_name.lower()
|
|
1126
|
+
):
|
|
1127
|
+
logger.debug(
|
|
1128
|
+
f"Schema {table_info.get('schema')} does not match prefix {prefix_schema_name}"
|
|
1129
|
+
)
|
|
1130
|
+
return
|
|
1131
|
+
|
|
1132
|
+
if (
|
|
1133
|
+
prefix_database_name
|
|
1134
|
+
and database_name
|
|
1135
|
+
and prefix_database_name.lower() != database_name.lower()
|
|
1136
|
+
):
|
|
1137
|
+
logger.debug(
|
|
1138
|
+
f"Database {table_info.get('database')} does not match prefix {prefix_database_name}"
|
|
1139
|
+
)
|
|
1140
|
+
return
|
|
1141
|
+
|
|
1142
|
+
fqn_search_string = build_es_fqn_search_string(
|
|
1143
|
+
service_name=prefix_service_name or "*",
|
|
1144
|
+
table_name=(prefix_table_name or table_name),
|
|
1145
|
+
schema_name=(prefix_schema_name or schema_name),
|
|
1146
|
+
database_name=(prefix_database_name or database_name),
|
|
1147
|
+
)
|
|
1148
|
+
table_entity = self.metadata.search_in_any_service(
|
|
1149
|
+
entity_type=Table,
|
|
1150
|
+
fqn_search_string=fqn_search_string,
|
|
1151
|
+
)
|
|
1152
|
+
if table_entity and datamodel_entity:
|
|
1153
|
+
columns_list = [column.name for column in table.columns]
|
|
1154
|
+
column_lineage = self._get_column_lineage(
|
|
1155
|
+
table_entity, datamodel_entity, columns_list
|
|
1156
|
+
)
|
|
1157
|
+
yield self._get_add_lineage_request(
|
|
1158
|
+
to_entity=datamodel_entity,
|
|
1159
|
+
from_entity=table_entity,
|
|
1160
|
+
column_lineage=column_lineage,
|
|
1161
|
+
)
|
|
563
1162
|
except Exception as exc: # pylint: disable=broad-except
|
|
564
|
-
|
|
1163
|
+
yield Either(
|
|
565
1164
|
left=StackTraceError(
|
|
566
1165
|
name="DataModel Lineage for pbit files",
|
|
567
1166
|
error=(
|
|
@@ -571,16 +1170,17 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
571
1170
|
stackTrace=traceback.format_exc(),
|
|
572
1171
|
)
|
|
573
1172
|
)
|
|
574
|
-
return None
|
|
575
1173
|
|
|
576
1174
|
def create_table_datamodel_lineage_from_files(
|
|
577
1175
|
self,
|
|
578
|
-
|
|
1176
|
+
db_service_prefix: Optional[str],
|
|
579
1177
|
datamodel_entity: Optional[DashboardDataModel],
|
|
580
1178
|
) -> Iterable[Either[AddLineageRequest]]:
|
|
581
1179
|
"""
|
|
582
1180
|
Method to create lineage between table and datamodels using pbit files
|
|
583
1181
|
"""
|
|
1182
|
+
(prefix_service_name, *_) = self.parse_db_service_prefix(db_service_prefix)
|
|
1183
|
+
|
|
584
1184
|
try:
|
|
585
1185
|
# check if the datamodel_file_mappings is populated or not
|
|
586
1186
|
# if not, then populate the datamodel_file_mappings and process the lineage
|
|
@@ -600,8 +1200,8 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
600
1200
|
|
|
601
1201
|
for datamodel_schema_file in datamodel_file_list:
|
|
602
1202
|
for table in datamodel_schema_file.tables or []:
|
|
603
|
-
yield self._get_table_and_datamodel_lineage(
|
|
604
|
-
|
|
1203
|
+
yield from self._get_table_and_datamodel_lineage(
|
|
1204
|
+
db_service_prefix=db_service_prefix,
|
|
605
1205
|
table=table,
|
|
606
1206
|
datamodel_entity=datamodel_entity,
|
|
607
1207
|
)
|
|
@@ -611,88 +1211,329 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
611
1211
|
name="DataModel Lineage",
|
|
612
1212
|
error=(
|
|
613
1213
|
"Error to yield datamodel lineage details for DB "
|
|
614
|
-
f"service name [{
|
|
1214
|
+
f"service name [{prefix_service_name}]: {exc}"
|
|
615
1215
|
),
|
|
616
1216
|
stackTrace=traceback.format_exc(),
|
|
617
1217
|
)
|
|
618
1218
|
)
|
|
619
1219
|
|
|
620
|
-
def
|
|
621
|
-
self,
|
|
622
|
-
dashboard_details: Union[PowerBIDashboard, PowerBIReport],
|
|
623
|
-
db_service_name: str,
|
|
1220
|
+
def create_dataset_upstream_dataflow_lineage(
|
|
1221
|
+
self, datamodel: Dataset, datamodel_entity: DashboardDataModel
|
|
624
1222
|
) -> Iterable[Either[AddLineageRequest]]:
|
|
625
1223
|
"""
|
|
626
|
-
|
|
627
|
-
tables - datamodel - report - dashboard
|
|
1224
|
+
Create lineage between dataset and upstreamDataflow
|
|
628
1225
|
"""
|
|
629
|
-
|
|
630
|
-
|
|
631
|
-
|
|
632
|
-
|
|
1226
|
+
for upstream_dataflow in datamodel.upstreamDataflows or []:
|
|
1227
|
+
try:
|
|
1228
|
+
if not upstream_dataflow.targetDataflowId:
|
|
1229
|
+
logger.debug(
|
|
1230
|
+
f"No targetDataflowId found for upstreamDataflow in "
|
|
1231
|
+
f"datamodel [{datamodel_entity.name.root}], "
|
|
1232
|
+
f"Moving to next upstreamDataflow"
|
|
1233
|
+
)
|
|
1234
|
+
continue
|
|
1235
|
+
upstream_dataflow_fqn = fqn.build(
|
|
1236
|
+
self.metadata,
|
|
1237
|
+
entity_type=DashboardDataModel,
|
|
1238
|
+
service_name=self.context.get().dashboard_service,
|
|
1239
|
+
data_model_name=upstream_dataflow.targetDataflowId,
|
|
633
1240
|
)
|
|
634
|
-
|
|
635
|
-
|
|
636
|
-
|
|
637
|
-
dashboard_details=dashboard_details
|
|
1241
|
+
upstream_dataflow_entity = self.metadata.get_by_name(
|
|
1242
|
+
entity=DashboardDataModel,
|
|
1243
|
+
fqn=upstream_dataflow_fqn,
|
|
638
1244
|
)
|
|
639
|
-
|
|
640
|
-
|
|
641
|
-
|
|
642
|
-
|
|
643
|
-
|
|
644
|
-
|
|
645
|
-
|
|
1245
|
+
if upstream_dataflow_entity and datamodel_entity:
|
|
1246
|
+
yield self._get_add_lineage_request(
|
|
1247
|
+
from_entity=upstream_dataflow_entity,
|
|
1248
|
+
to_entity=datamodel_entity,
|
|
1249
|
+
)
|
|
1250
|
+
else:
|
|
1251
|
+
logger.debug(
|
|
1252
|
+
f"No upstreamDataflow entity with id={str(upstream_dataflow.targetDataflowId)} "
|
|
1253
|
+
f"found for datamodel [{datamodel_entity.name.root}]"
|
|
1254
|
+
)
|
|
1255
|
+
except Exception as exc: # pylint: disable=broad-except
|
|
1256
|
+
yield Either(
|
|
1257
|
+
left=StackTraceError(
|
|
1258
|
+
name="Dataset and UpstreamDataflow Lineage",
|
|
1259
|
+
error=(
|
|
1260
|
+
"Error to yield dataset and upstreamDataflow lineage "
|
|
1261
|
+
f"between [{datamodel_entity.name.root}, {str(upstream_dataflow.targetDataflowId)}]: {exc}"
|
|
1262
|
+
),
|
|
1263
|
+
stackTrace=traceback.format_exc(),
|
|
1264
|
+
)
|
|
646
1265
|
)
|
|
1266
|
+
|
|
1267
|
+
def _get_downstream_data_model_column_fqn(
|
|
1268
|
+
self, data_model_entity: DashboardDataModel, table_name: str, column: str
|
|
1269
|
+
) -> Optional[str]:
|
|
1270
|
+
"""
|
|
1271
|
+
Get the FQN of the column if it exists in the
|
|
1272
|
+
downstream data model entity's table and column.
|
|
1273
|
+
"""
|
|
1274
|
+
try:
|
|
1275
|
+
if not data_model_entity:
|
|
1276
|
+
return None
|
|
1277
|
+
for table in data_model_entity.columns:
|
|
1278
|
+
if table.name.root != table_name:
|
|
1279
|
+
continue
|
|
1280
|
+
for child_column in table.children or []:
|
|
1281
|
+
if column.lower() == child_column.name.root.lower():
|
|
1282
|
+
return child_column.fullyQualifiedName.root
|
|
1283
|
+
except Exception as exc:
|
|
1284
|
+
logger.error(
|
|
1285
|
+
f"Error to get downstream data_model_column_fqn for data_model_entity="
|
|
1286
|
+
f"{data_model_entity.name.root}, table_name={table_name}, column={column}: {exc}"
|
|
647
1287
|
)
|
|
1288
|
+
logger.debug(traceback.format_exc())
|
|
1289
|
+
return None
|
|
648
1290
|
|
|
649
|
-
def
|
|
650
|
-
self,
|
|
651
|
-
|
|
652
|
-
|
|
653
|
-
|
|
654
|
-
dashboard_details:
|
|
655
|
-
Returns:
|
|
656
|
-
Iterable[Chart]
|
|
1291
|
+
def _create_dataset_upstream_dataset_column_lineage(
|
|
1292
|
+
self,
|
|
1293
|
+
datamodel_entity: DashboardDataModel,
|
|
1294
|
+
upstream_dataset_entity: DashboardDataModel,
|
|
1295
|
+
) -> Optional[List[ColumnLineage]]:
|
|
657
1296
|
"""
|
|
658
|
-
|
|
659
|
-
|
|
660
|
-
|
|
661
|
-
|
|
662
|
-
|
|
663
|
-
|
|
664
|
-
|
|
665
|
-
|
|
666
|
-
|
|
667
|
-
|
|
668
|
-
|
|
1297
|
+
Create column lineage between powerbi dataset/datamodel and
|
|
1298
|
+
its upstream dataset/datamodel
|
|
1299
|
+
"""
|
|
1300
|
+
try:
|
|
1301
|
+
target_tables = [table.name.root for table in datamodel_entity.columns]
|
|
1302
|
+
if not target_tables:
|
|
1303
|
+
return []
|
|
1304
|
+
column_lineage = []
|
|
1305
|
+
for table in upstream_dataset_entity.columns:
|
|
1306
|
+
if table.name.root not in target_tables:
|
|
1307
|
+
continue
|
|
1308
|
+
for column in table.children or []:
|
|
1309
|
+
source_column = column.fullyQualifiedName.root
|
|
1310
|
+
target_column = self._get_downstream_data_model_column_fqn(
|
|
1311
|
+
data_model_entity=datamodel_entity,
|
|
1312
|
+
table_name=table.name.root,
|
|
1313
|
+
column=column.name.root,
|
|
1314
|
+
)
|
|
1315
|
+
if source_column and target_column:
|
|
1316
|
+
column_lineage.append(
|
|
1317
|
+
ColumnLineage(
|
|
1318
|
+
fromColumns=[source_column], toColumn=target_column
|
|
1319
|
+
)
|
|
669
1320
|
)
|
|
670
|
-
|
|
671
|
-
|
|
672
|
-
|
|
673
|
-
|
|
674
|
-
|
|
675
|
-
|
|
676
|
-
|
|
677
|
-
|
|
678
|
-
|
|
679
|
-
|
|
680
|
-
|
|
681
|
-
|
|
682
|
-
|
|
683
|
-
|
|
684
|
-
|
|
685
|
-
|
|
1321
|
+
return column_lineage
|
|
1322
|
+
except Exception as exc:
|
|
1323
|
+
logger.debug(traceback.format_exc())
|
|
1324
|
+
logger.error(
|
|
1325
|
+
"Error while creating column lineage between dataset = "
|
|
1326
|
+
f"{datamodel_entity.name.root} and upstream dataset = "
|
|
1327
|
+
f"{upstream_dataset_entity.name.root}: {exc}"
|
|
1328
|
+
)
|
|
1329
|
+
return []
|
|
1330
|
+
|
|
1331
|
+
def create_dataset_upstream_dataset_lineage(
|
|
1332
|
+
self, datamodel: Dataset, datamodel_entity: DashboardDataModel
|
|
1333
|
+
) -> Iterable[Either[AddLineageRequest]]:
|
|
1334
|
+
"""
|
|
1335
|
+
Create lineage between dataset and upstreamDataset
|
|
1336
|
+
"""
|
|
1337
|
+
for upstream_dataset in datamodel.upstreamDatasets or []:
|
|
1338
|
+
try:
|
|
1339
|
+
if not upstream_dataset.targetDatasetId:
|
|
1340
|
+
logger.debug(
|
|
1341
|
+
f"No targetDatasetId found for upstreamDataset in "
|
|
1342
|
+
f"datamodel [{datamodel_entity.name.root}], "
|
|
1343
|
+
f"Moving to next upstreamDataset"
|
|
1344
|
+
)
|
|
1345
|
+
continue
|
|
1346
|
+
upstream_dataset_fqn = fqn.build(
|
|
1347
|
+
self.metadata,
|
|
1348
|
+
entity_type=DashboardDataModel,
|
|
1349
|
+
service_name=self.context.get().dashboard_service,
|
|
1350
|
+
data_model_name=upstream_dataset.targetDatasetId,
|
|
1351
|
+
)
|
|
1352
|
+
upstream_dataset_entity = self.metadata.get_by_name(
|
|
1353
|
+
entity=DashboardDataModel,
|
|
1354
|
+
fqn=upstream_dataset_fqn,
|
|
1355
|
+
)
|
|
1356
|
+
if upstream_dataset_entity and datamodel_entity:
|
|
1357
|
+
# create column lineage between current dataset/datamodel
|
|
1358
|
+
# and its upstream dataset.
|
|
1359
|
+
column_lineage = (
|
|
1360
|
+
self._create_dataset_upstream_dataset_column_lineage(
|
|
1361
|
+
datamodel_entity, upstream_dataset_entity
|
|
686
1362
|
)
|
|
687
1363
|
)
|
|
688
|
-
|
|
689
|
-
|
|
690
|
-
|
|
691
|
-
|
|
692
|
-
|
|
693
|
-
|
|
1364
|
+
yield self._get_add_lineage_request(
|
|
1365
|
+
from_entity=upstream_dataset_entity,
|
|
1366
|
+
to_entity=datamodel_entity,
|
|
1367
|
+
column_lineage=column_lineage,
|
|
1368
|
+
)
|
|
1369
|
+
else:
|
|
1370
|
+
logger.debug(
|
|
1371
|
+
f"No upstreamDataset entity with id={str(upstream_dataset.targetDatasetId)} "
|
|
1372
|
+
f"found for datamodel [{datamodel_entity.name.root}]"
|
|
1373
|
+
)
|
|
1374
|
+
except Exception as exc: # pylint: disable=broad-except
|
|
1375
|
+
yield Either(
|
|
1376
|
+
left=StackTraceError(
|
|
1377
|
+
name="Dataset and UpstreamDataset Lineage",
|
|
1378
|
+
error=(
|
|
1379
|
+
"Error to yield dataset and upstreamDataset lineage "
|
|
1380
|
+
f"between [{datamodel_entity.name.root}, {str(upstream_dataset.targetDatasetId)}]: {exc}"
|
|
1381
|
+
),
|
|
1382
|
+
stackTrace=traceback.format_exc(),
|
|
1383
|
+
)
|
|
1384
|
+
)
|
|
1385
|
+
|
|
1386
|
+
def create_dataflow_upstream_dataflow_lineage(
|
|
1387
|
+
self, datamodel: Dataflow, datamodel_entity: DashboardDataModel
|
|
1388
|
+
) -> Iterable[Either[AddLineageRequest]]:
|
|
1389
|
+
"""
|
|
1390
|
+
Create lineage between dataflow and upstreamDataflow
|
|
1391
|
+
"""
|
|
1392
|
+
for upstream_dataflow in datamodel.upstreamDataflows or []:
|
|
1393
|
+
try:
|
|
1394
|
+
if not upstream_dataflow.targetDataflowId:
|
|
1395
|
+
logger.debug(
|
|
1396
|
+
f"No targetDataflowId found for upstreamDataflow in "
|
|
1397
|
+
f"datamodel [{datamodel_entity.name.root}], "
|
|
1398
|
+
f"Moving to next upstreamDataflow"
|
|
1399
|
+
)
|
|
1400
|
+
continue
|
|
1401
|
+
upstream_dataflow_fqn = fqn.build(
|
|
1402
|
+
self.metadata,
|
|
1403
|
+
entity_type=DashboardDataModel,
|
|
1404
|
+
service_name=self.context.get().dashboard_service,
|
|
1405
|
+
data_model_name=upstream_dataflow.targetDataflowId,
|
|
1406
|
+
)
|
|
1407
|
+
upstream_dataflow_entity = self.metadata.get_by_name(
|
|
1408
|
+
entity=DashboardDataModel,
|
|
1409
|
+
fqn=upstream_dataflow_fqn,
|
|
1410
|
+
)
|
|
1411
|
+
if upstream_dataflow_entity and datamodel_entity:
|
|
1412
|
+
yield self._get_add_lineage_request(
|
|
1413
|
+
from_entity=upstream_dataflow_entity,
|
|
1414
|
+
to_entity=datamodel_entity,
|
|
1415
|
+
)
|
|
1416
|
+
else:
|
|
1417
|
+
logger.debug(
|
|
1418
|
+
f"No upstreamDataflow entity with id={str(upstream_dataflow.targetDataflowId)} "
|
|
1419
|
+
f"found for datamodel [{datamodel_entity.name.root}]"
|
|
1420
|
+
)
|
|
1421
|
+
except Exception as exc: # pylint: disable=broad-except
|
|
1422
|
+
yield Either(
|
|
1423
|
+
left=StackTraceError(
|
|
1424
|
+
name="Dataflow and UpstreamDataflow Lineage",
|
|
1425
|
+
error=(
|
|
1426
|
+
f"Error to yield dataflow and upstreamDataflow lineage "
|
|
1427
|
+
f"between [{datamodel_entity.name.root}, {str(upstream_dataflow.targetDataflowId)}]: {exc}"
|
|
1428
|
+
),
|
|
1429
|
+
stackTrace=traceback.format_exc(),
|
|
1430
|
+
)
|
|
1431
|
+
)
|
|
1432
|
+
|
|
1433
|
+
def yield_dashboard_lineage_details(
|
|
1434
|
+
self,
|
|
1435
|
+
dashboard_details: Group,
|
|
1436
|
+
db_service_prefix: Optional[str] = None,
|
|
1437
|
+
) -> Iterable[Either[AddLineageRequest]]:
|
|
1438
|
+
"""
|
|
1439
|
+
We will build the logic to build the logic as below
|
|
1440
|
+
tables - datamodel - report - dashboard
|
|
1441
|
+
"""
|
|
1442
|
+
(prefix_service_name, *_) = self.parse_db_service_prefix(db_service_prefix)
|
|
1443
|
+
|
|
1444
|
+
for dashboard in self.filtered_dashboards or []:
|
|
1445
|
+
dashboard_details = self.get_dashboard_details(dashboard)
|
|
1446
|
+
try:
|
|
1447
|
+
if isinstance(dashboard_details, PowerBIReport):
|
|
1448
|
+
yield from self.create_datamodel_report_lineage(
|
|
1449
|
+
db_service_prefix=db_service_prefix,
|
|
1450
|
+
dashboard_details=dashboard_details,
|
|
1451
|
+
)
|
|
1452
|
+
if isinstance(dashboard_details, PowerBIDashboard):
|
|
1453
|
+
yield from self.create_report_dashboard_lineage(
|
|
1454
|
+
dashboard_details=dashboard_details
|
|
1455
|
+
)
|
|
1456
|
+
except Exception as exc: # pylint: disable=broad-except
|
|
1457
|
+
yield Either(
|
|
1458
|
+
left=StackTraceError(
|
|
1459
|
+
name="Dashboard Lineage",
|
|
1460
|
+
error=f"Error to yield dashboard lineage details for DB service name [{str(prefix_service_name)}]: {exc}",
|
|
1461
|
+
stackTrace=traceback.format_exc(),
|
|
1462
|
+
)
|
|
1463
|
+
)
|
|
1464
|
+
"""
|
|
1465
|
+
Iterate loop for filtered datamodels so datamodels which are not connected to
|
|
1466
|
+
any report but have tables would be eligible for a dataset-db_table lineage.
|
|
1467
|
+
Also create below lineages:
|
|
1468
|
+
1. dataset-db_table
|
|
1469
|
+
2. dataset-upstreamDataflow
|
|
1470
|
+
3. dataset-upstreamDataset
|
|
1471
|
+
4. dataflow-upstreamDataflow
|
|
1472
|
+
"""
|
|
1473
|
+
for datamodel in self.filtered_datamodels or []:
|
|
1474
|
+
try:
|
|
1475
|
+
datamodel_fqn = fqn.build(
|
|
1476
|
+
self.metadata,
|
|
1477
|
+
entity_type=DashboardDataModel,
|
|
1478
|
+
service_name=self.context.get().dashboard_service,
|
|
1479
|
+
data_model_name=datamodel.id,
|
|
1480
|
+
)
|
|
1481
|
+
datamodel_entity = self.metadata.get_by_name(
|
|
1482
|
+
entity=DashboardDataModel,
|
|
1483
|
+
fqn=datamodel_fqn,
|
|
1484
|
+
)
|
|
1485
|
+
if datamodel_entity:
|
|
1486
|
+
if isinstance(datamodel, Dataset):
|
|
1487
|
+
# 1. datamodel-db_table lineage
|
|
1488
|
+
for table in datamodel.tables or []:
|
|
1489
|
+
yield from self._get_table_and_datamodel_lineage(
|
|
1490
|
+
db_service_prefix=db_service_prefix,
|
|
1491
|
+
table=table,
|
|
1492
|
+
datamodel_entity=datamodel_entity,
|
|
1493
|
+
)
|
|
1494
|
+
# 2. dataset-upstreamDataflow lineage
|
|
1495
|
+
yield from self.create_dataset_upstream_dataflow_lineage(
|
|
1496
|
+
datamodel, datamodel_entity
|
|
1497
|
+
)
|
|
1498
|
+
# 3. dataset-upstreamDataset lineage
|
|
1499
|
+
yield from self.create_dataset_upstream_dataset_lineage(
|
|
1500
|
+
datamodel, datamodel_entity
|
|
694
1501
|
)
|
|
1502
|
+
# create the lineage between table and datamodel using the pbit files
|
|
1503
|
+
if self.client.file_client:
|
|
1504
|
+
yield from self.create_table_datamodel_lineage_from_files(
|
|
1505
|
+
db_service_prefix=db_service_prefix,
|
|
1506
|
+
datamodel_entity=datamodel_entity,
|
|
1507
|
+
)
|
|
1508
|
+
elif isinstance(datamodel, Dataflow):
|
|
1509
|
+
# create dataflow-upstreamDataflow lineage
|
|
1510
|
+
yield from self.create_dataflow_upstream_dataflow_lineage(
|
|
1511
|
+
datamodel, datamodel_entity
|
|
1512
|
+
)
|
|
1513
|
+
else:
|
|
1514
|
+
logger.warning(
|
|
1515
|
+
f"Unknown datamodel type: {type(datamodel)}, name: {datamodel.name}"
|
|
1516
|
+
)
|
|
1517
|
+
except Exception as exc: # pylint: disable=broad-except
|
|
1518
|
+
yield Either(
|
|
1519
|
+
left=StackTraceError(
|
|
1520
|
+
name="Datamodel Lineage",
|
|
1521
|
+
error=f"Error to yield datamodel lineage details for DB service name [{str(prefix_service_name)}]: {exc}",
|
|
1522
|
+
stackTrace=traceback.format_exc(),
|
|
695
1523
|
)
|
|
1524
|
+
)
|
|
1525
|
+
|
|
1526
|
+
def yield_datamodel_dashboard_lineage(
|
|
1527
|
+
self,
|
|
1528
|
+
) -> Iterable[Either[AddLineageRequest]]:
|
|
1529
|
+
"""
|
|
1530
|
+
Returns:
|
|
1531
|
+
Lineage request between Data Models and Dashboards
|
|
1532
|
+
"""
|
|
1533
|
+
"""
|
|
1534
|
+
We're implementing this differently inside `yield_dashboard_lineage_details`
|
|
1535
|
+
since we have report and dashboard both as dashboard.
|
|
1536
|
+
"""
|
|
696
1537
|
|
|
697
1538
|
def _fetch_dataset_from_workspace(
|
|
698
1539
|
self, dataset_id: Optional[str]
|
|
@@ -734,21 +1575,6 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
734
1575
|
return report_data
|
|
735
1576
|
return None
|
|
736
1577
|
|
|
737
|
-
def _fetch_dataset_workspace(self, dataset_id: Optional[str]) -> Optional[str]:
|
|
738
|
-
"""
|
|
739
|
-
Method to search the workspace name in which the dataset in contained
|
|
740
|
-
"""
|
|
741
|
-
if dataset_id:
|
|
742
|
-
workspace_names = (
|
|
743
|
-
workspace.name
|
|
744
|
-
for workspace in self.workspace_data
|
|
745
|
-
for dataset in workspace.datasets
|
|
746
|
-
if dataset.id == dataset_id
|
|
747
|
-
)
|
|
748
|
-
return next(iter(workspace_names), None)
|
|
749
|
-
|
|
750
|
-
return None
|
|
751
|
-
|
|
752
1578
|
def get_project_name(self, dashboard_details: Any) -> Optional[str]:
|
|
753
1579
|
"""
|
|
754
1580
|
Get the project / workspace / folder / collection name of the dashboard
|
|
@@ -761,3 +1587,98 @@ class PowerbiSource(DashboardServiceSource):
|
|
|
761
1587
|
f"Error fetching project name for {dashboard_details.id}: {exc}"
|
|
762
1588
|
)
|
|
763
1589
|
return None
|
|
1590
|
+
|
|
1591
|
+
def get_owner_ref( # pylint: disable=unused-argument, useless-return
|
|
1592
|
+
self, dashboard_details: Any
|
|
1593
|
+
) -> Optional[EntityReferenceList]:
|
|
1594
|
+
"""
|
|
1595
|
+
Method to process the dashboard owners
|
|
1596
|
+
"""
|
|
1597
|
+
try:
|
|
1598
|
+
if not self.source_config.includeOwners:
|
|
1599
|
+
logger.debug(
|
|
1600
|
+
f"Skipping owner processing for {dashboard_details.id} "
|
|
1601
|
+
f"as includeOwners is False"
|
|
1602
|
+
)
|
|
1603
|
+
return None
|
|
1604
|
+
owner_ref_list = [] # to assign multiple owners to entity if they exist
|
|
1605
|
+
for owner in dashboard_details.users or []:
|
|
1606
|
+
owner_ref = None
|
|
1607
|
+
# put filtering conditions
|
|
1608
|
+
if isinstance(dashboard_details, Dataset):
|
|
1609
|
+
access_right = owner.datasetUserAccessRight
|
|
1610
|
+
elif isinstance(dashboard_details, Dataflow):
|
|
1611
|
+
access_right = owner.dataflowUserAccessRight
|
|
1612
|
+
elif isinstance(dashboard_details, PowerBIReport):
|
|
1613
|
+
access_right = owner.reportUserAccessRight
|
|
1614
|
+
elif isinstance(dashboard_details, PowerBIDashboard):
|
|
1615
|
+
access_right = owner.dashboardUserAccessRight
|
|
1616
|
+
|
|
1617
|
+
if owner.userType != "Member":
|
|
1618
|
+
logger.debug(
|
|
1619
|
+
f"User is not a member of {dashboard_details.id}:"
|
|
1620
|
+
f" ({owner.displayName}, {owner.email})"
|
|
1621
|
+
)
|
|
1622
|
+
continue
|
|
1623
|
+
if access_right and any(
|
|
1624
|
+
keyword in access_right.lower()
|
|
1625
|
+
for keyword in OWNER_ACCESS_RIGHTS_KEYWORDS
|
|
1626
|
+
):
|
|
1627
|
+
if owner.email:
|
|
1628
|
+
try:
|
|
1629
|
+
owner_email = EmailStr._validate(owner.email)
|
|
1630
|
+
except PydanticCustomError:
|
|
1631
|
+
logger.debug(f"Invalid email for owner: {owner.email}")
|
|
1632
|
+
owner_email = None
|
|
1633
|
+
if owner_email:
|
|
1634
|
+
try:
|
|
1635
|
+
owner_ref = self.metadata.get_reference_by_email(
|
|
1636
|
+
owner_email.lower()
|
|
1637
|
+
)
|
|
1638
|
+
except Exception as err:
|
|
1639
|
+
logger.debug(
|
|
1640
|
+
f"Could not process owner data with email"
|
|
1641
|
+
f" {owner.email} in {dashboard_details.id}: {err}"
|
|
1642
|
+
)
|
|
1643
|
+
elif owner.displayName:
|
|
1644
|
+
try:
|
|
1645
|
+
owner_ref = self.metadata.get_reference_by_name(
|
|
1646
|
+
name=owner.displayName
|
|
1647
|
+
)
|
|
1648
|
+
except Exception as err:
|
|
1649
|
+
logger.debug(
|
|
1650
|
+
f"Could not process owner data with name"
|
|
1651
|
+
f" {owner.displayName} in {dashboard_details.id}: {err}"
|
|
1652
|
+
)
|
|
1653
|
+
if owner_ref:
|
|
1654
|
+
owner_ref_list.append(owner_ref.root[0])
|
|
1655
|
+
else:
|
|
1656
|
+
logger.debug(
|
|
1657
|
+
f"User does not have owner, admin or write access to"
|
|
1658
|
+
f" {dashboard_details.id}: ({owner.displayName}, {owner.email})"
|
|
1659
|
+
)
|
|
1660
|
+
# check for last modified, configuredBy user
|
|
1661
|
+
current_active_user = None
|
|
1662
|
+
if isinstance(dashboard_details, Dataset):
|
|
1663
|
+
current_active_user = dashboard_details.configuredBy
|
|
1664
|
+
elif isinstance(dashboard_details, (Dataflow, PowerBIReport)):
|
|
1665
|
+
current_active_user = dashboard_details.modifiedBy
|
|
1666
|
+
if current_active_user:
|
|
1667
|
+
try:
|
|
1668
|
+
owner_ref = self.metadata.get_reference_by_email(
|
|
1669
|
+
current_active_user.lower()
|
|
1670
|
+
)
|
|
1671
|
+
if owner_ref and owner_ref.root[0] not in owner_ref_list:
|
|
1672
|
+
owner_ref_list.append(owner_ref.root[0])
|
|
1673
|
+
except Exception as err:
|
|
1674
|
+
logger.debug(f"Could not fetch current active user due to {err}")
|
|
1675
|
+
if len(owner_ref_list) > 0:
|
|
1676
|
+
logger.debug(
|
|
1677
|
+
f"Successfully fetched owners data for {dashboard_details.id}"
|
|
1678
|
+
)
|
|
1679
|
+
return EntityReferenceList(root=owner_ref_list)
|
|
1680
|
+
return None
|
|
1681
|
+
except Exception as err:
|
|
1682
|
+
logger.debug(traceback.format_exc())
|
|
1683
|
+
logger.warning(f"Could not fetch owner data due to {err}")
|
|
1684
|
+
return None
|