openmetadata-ingestion 1.5.13.0__py3-none-any.whl → 1.6.0.0rc2__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.
Potentially problematic release.
This version of openmetadata-ingestion might be problematic. Click here for more details.
- _openmetadata_testutils/dict.py +23 -0
- _openmetadata_testutils/pydantic/test_utils.py +22 -2
- _openmetadata_testutils/pytest_openmetadata/plugin.py +86 -0
- _openmetadata_testutils/pytest_openmetadata/test_utils.py +99 -0
- metadata/cli/app.py +1 -0
- metadata/cli/classify.py +52 -0
- metadata/cli/dataquality.py +4 -1
- metadata/cli/ingest.py +2 -1
- metadata/cli/lineage.py +2 -1
- metadata/cli/profile.py +4 -1
- metadata/cli/usage.py +2 -1
- metadata/cmd.py +9 -0
- metadata/config/common.py +3 -1
- metadata/data_quality/api/models.py +2 -1
- metadata/data_quality/builders/{i_validator_builder.py → validator_builder.py} +45 -31
- metadata/data_quality/interface/pandas/pandas_test_suite_interface.py +25 -24
- metadata/data_quality/interface/sqlalchemy/sqa_test_suite_interface.py +23 -49
- metadata/data_quality/interface/test_suite_interface.py +36 -52
- metadata/data_quality/processor/test_case_runner.py +9 -9
- metadata/data_quality/runner/base_test_suite_source.py +52 -26
- metadata/data_quality/source/test_suite.py +1 -1
- metadata/data_quality/validations/base_test_handler.py +13 -8
- metadata/data_quality/validations/column/base/columnValuesToBeInSet.py +4 -2
- metadata/data_quality/validations/column/sqlalchemy/columnValueLengthsToBeBetween.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValueMaxToBeBetween.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValueMeanToBeBetween.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValueMedianToBeBetween.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValueMinToBeBetween.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValueStdDevToBeBetween.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValuesMissingCount.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValuesSumToBeBetween.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeAtExpectedLocation.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeBetween.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeInSet.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeNotInSet.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeNotNull.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToBeUnique.py +2 -8
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToMatchRegex.py +1 -1
- metadata/data_quality/validations/column/sqlalchemy/columnValuesToNotMatchRegex.py +1 -1
- metadata/data_quality/validations/models.py +6 -1
- metadata/data_quality/validations/runtime_param_setter/param_setter.py +2 -3
- metadata/data_quality/validations/runtime_param_setter/param_setter_factory.py +45 -17
- metadata/data_quality/validations/runtime_param_setter/table_diff_params_setter.py +26 -16
- metadata/data_quality/validations/table/sqlalchemy/tableDiff.py +192 -17
- metadata/data_quality/validations/table/sqlalchemy/tableRowInsertedCountToBeBetween.py +2 -2
- metadata/examples/workflows/bigquery_classifier.yaml +56 -0
- metadata/examples/workflows/bigquery_profiler.yaml +1 -2
- metadata/examples/workflows/datafactory.yaml +29 -0
- metadata/examples/workflows/db2_profiler.yaml +1 -2
- metadata/examples/workflows/dbtcloud.yaml +2 -1
- metadata/examples/workflows/matillion.yaml +31 -0
- metadata/examples/workflows/{mstr.yaml → microstrategy.yaml} +3 -2
- metadata/examples/workflows/mysql_profiler.yaml +0 -1
- metadata/examples/workflows/redshift_classifier.yaml +38 -0
- metadata/examples/workflows/redshift_profiler.yaml +2 -3
- metadata/examples/workflows/rest.yaml +20 -0
- metadata/examples/workflows/sigma.yaml +25 -0
- metadata/examples/workflows/trino_profiler.yaml +20 -0
- metadata/generated/antlr/EntityLinkLexer.py +70 -99
- 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 +1 -1
- 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 +1 -1
- metadata/generated/schema/api/addTagToAssetsRequest.py +43 -0
- metadata/generated/schema/api/analytics/__init__.py +1 -1
- metadata/generated/schema/api/analytics/createWebAnalyticEvent.py +1 -1
- metadata/generated/schema/api/automations/__init__.py +1 -1
- metadata/generated/schema/api/automations/createWorkflow.py +1 -1
- metadata/generated/schema/api/bulkAssets.py +1 -1
- metadata/generated/schema/api/classification/__init__.py +1 -1
- metadata/generated/schema/api/classification/createClassification.py +1 -1
- metadata/generated/schema/api/classification/createTag.py +1 -1
- metadata/generated/schema/api/classification/loadTags.py +1 -1
- metadata/generated/schema/api/createBot.py +1 -1
- metadata/generated/schema/api/createEventPublisherJob.py +1 -1
- metadata/generated/schema/api/createType.py +1 -1
- metadata/generated/schema/api/data/__init__.py +1 -1
- metadata/generated/schema/api/data/createAPICollection.py +1 -1
- metadata/generated/schema/api/data/createAPIEndpoint.py +1 -1
- metadata/generated/schema/api/data/createChart.py +1 -1
- metadata/generated/schema/api/data/createContainer.py +1 -1
- metadata/generated/schema/api/data/createCustomProperty.py +1 -1
- metadata/generated/schema/api/data/createDashboard.py +1 -1
- metadata/generated/schema/api/data/createDashboardDataModel.py +1 -1
- metadata/generated/schema/api/data/createDatabase.py +1 -1
- metadata/generated/schema/api/data/createDatabaseSchema.py +1 -1
- metadata/generated/schema/api/data/createGlossary.py +1 -1
- metadata/generated/schema/api/data/createGlossaryTerm.py +1 -1
- metadata/generated/schema/api/data/createMetric.py +83 -0
- metadata/generated/schema/api/data/createMlModel.py +1 -1
- metadata/generated/schema/api/data/createPipeline.py +1 -1
- metadata/generated/schema/api/data/createQuery.py +1 -1
- metadata/generated/schema/api/data/createSearchIndex.py +8 -1
- metadata/generated/schema/api/data/createStoredProcedure.py +1 -1
- metadata/generated/schema/api/data/createTable.py +1 -1
- metadata/generated/schema/api/data/createTableProfile.py +1 -1
- metadata/generated/schema/api/data/createTopic.py +1 -1
- metadata/generated/schema/api/data/loadGlossary.py +1 -1
- metadata/generated/schema/api/data/restoreEntity.py +1 -1
- metadata/generated/schema/api/dataInsight/__init__.py +1 -1
- metadata/generated/schema/api/dataInsight/createDataInsightChart.py +1 -1
- 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 +1 -1
- metadata/generated/schema/api/docStore/__init__.py +1 -1
- metadata/generated/schema/api/docStore/createDocument.py +1 -1
- metadata/generated/schema/api/domains/__init__.py +1 -1
- metadata/generated/schema/api/domains/createDataProduct.py +1 -1
- metadata/generated/schema/api/domains/createDomain.py +1 -1
- 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 +1 -1
- metadata/generated/schema/api/feed/resolveTask.py +1 -1
- metadata/generated/schema/api/feed/threadCount.py +1 -1
- metadata/generated/schema/api/governance/__init__.py +3 -0
- metadata/generated/schema/api/governance/createWorkflowDefinition.py +83 -0
- metadata/generated/schema/api/governance/createWorkflowInstanceState.py +22 -0
- metadata/generated/schema/api/lineage/__init__.py +1 -1
- metadata/generated/schema/api/lineage/addLineage.py +1 -1
- metadata/generated/schema/api/openMetadataServerVersion.py +1 -1
- metadata/generated/schema/api/policies/__init__.py +1 -1
- metadata/generated/schema/api/policies/createPolicy.py +1 -1
- metadata/generated/schema/api/services/__init__.py +1 -1
- metadata/generated/schema/api/services/createApiService.py +2 -2
- metadata/generated/schema/api/services/createDashboardService.py +1 -1
- metadata/generated/schema/api/services/createDatabaseService.py +1 -1
- metadata/generated/schema/api/services/createMessagingService.py +1 -1
- metadata/generated/schema/api/services/createMetadataService.py +1 -1
- metadata/generated/schema/api/services/createMlModelService.py +1 -1
- metadata/generated/schema/api/services/createPipelineService.py +1 -1
- metadata/generated/schema/api/services/createSearchService.py +1 -1
- metadata/generated/schema/api/services/createStorageService.py +1 -1
- metadata/generated/schema/api/services/ingestionPipelines/__init__.py +1 -1
- metadata/generated/schema/api/services/ingestionPipelines/createIngestionPipeline.py +1 -1
- metadata/generated/schema/api/setOwner.py +1 -1
- metadata/generated/schema/api/teams/__init__.py +1 -1
- metadata/generated/schema/api/teams/createPersona.py +1 -1
- metadata/generated/schema/api/teams/createRole.py +1 -1
- metadata/generated/schema/api/teams/createTeam.py +1 -1
- metadata/generated/schema/api/teams/createUser.py +1 -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 +1 -1
- metadata/generated/schema/api/tests/createTestCaseResolutionStatus.py +1 -1
- metadata/generated/schema/api/tests/createTestCaseResult.py +71 -0
- metadata/generated/schema/api/tests/createTestDefinition.py +1 -1
- metadata/generated/schema/api/tests/createTestSuite.py +1 -1
- 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 +1 -1
- 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/tokenRefreshRequest.py +1 -1
- metadata/generated/schema/configuration/__init__.py +1 -1
- metadata/generated/schema/configuration/appsPrivateConfiguration.py +1 -1
- metadata/generated/schema/configuration/assetCertificationSettings.py +22 -0
- metadata/generated/schema/configuration/authConfig.py +1 -1
- metadata/generated/schema/configuration/authenticationConfiguration.py +1 -1
- metadata/generated/schema/configuration/authorizerConfiguration.py +1 -1
- metadata/generated/schema/configuration/changeEventConfiguration.py +1 -1
- metadata/generated/schema/configuration/dataQualityConfiguration.py +1 -1
- metadata/generated/schema/configuration/elasticSearchConfiguration.py +1 -1
- 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 +1 -1
- metadata/generated/schema/configuration/loginConfiguration.py +1 -1
- metadata/generated/schema/configuration/logoConfiguration.py +1 -1
- metadata/generated/schema/configuration/pipelineServiceClientConfiguration.py +1 -1
- metadata/generated/schema/configuration/profilerConfiguration.py +1 -1
- metadata/generated/schema/configuration/searchSettings.py +1 -1
- metadata/generated/schema/configuration/slackAppConfiguration.py +1 -14
- 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/dataInsight/__init__.py +1 -1
- metadata/generated/schema/dataInsight/custom/__init__.py +1 -1
- metadata/generated/schema/dataInsight/custom/dataInsightCustomChart.py +1 -1
- metadata/generated/schema/dataInsight/custom/dataInsightCustomChartResult.py +9 -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 +23 -8
- metadata/generated/schema/dataInsight/custom/summaryCard.py +20 -9
- metadata/generated/schema/dataInsight/dataInsightChart.py +1 -1
- 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 +1 -1
- 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 +1 -1
- 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 +9 -1
- metadata/generated/schema/entity/applications/appExtension.py +1 -1
- metadata/generated/schema/entity/applications/appRunRecord.py +1 -1
- metadata/generated/schema/entity/applications/configuration/__init__.py +1 -1
- metadata/generated/schema/entity/applications/configuration/applicationConfig.py +4 -2
- 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/addDescriptionAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/addDomainAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/addOwnerAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/addTagsAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/addTierAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/lineagePropagationAction.py +54 -2
- metadata/generated/schema/entity/applications/configuration/external/automator/mlTaggingAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/removeDescriptionAction.py +1 -1
- 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 +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automator/removeTierAction.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/automatorAppConfig.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/collateAIAppConfig.py +1 -1
- metadata/generated/schema/entity/applications/configuration/external/slackAppTokenConfiguration.py +22 -0
- metadata/generated/schema/entity/applications/configuration/internal/__init__.py +1 -1
- metadata/generated/schema/entity/applications/configuration/internal/dataInsightsAppConfig.py +1 -1
- metadata/generated/schema/entity/applications/configuration/internal/dataInsightsReportAppConfig.py +1 -1
- metadata/generated/schema/entity/applications/configuration/internal/searchIndexingAppConfig.py +1 -1
- metadata/generated/schema/entity/applications/configuration/private/external/__init__.py +1 -1
- metadata/generated/schema/entity/applications/configuration/private/external/collateAIAppPrivateConfig.py +1 -1
- metadata/generated/schema/entity/applications/createAppRequest.py +8 -1
- 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 +8 -1
- metadata/generated/schema/entity/applications/marketplace/createAppMarketPlaceDefinitionReq.py +8 -1
- metadata/generated/schema/entity/applications/scheduledExecutionContext.py +1 -1
- metadata/generated/schema/entity/automations/__init__.py +1 -1
- metadata/generated/schema/entity/automations/testServiceConnection.py +3 -1
- metadata/generated/schema/entity/automations/workflow.py +1 -1
- metadata/generated/schema/entity/bot.py +1 -1
- metadata/generated/schema/entity/classification/__init__.py +1 -1
- metadata/generated/schema/entity/classification/classification.py +2 -2
- metadata/generated/schema/entity/classification/tag.py +2 -2
- metadata/generated/schema/entity/data/__init__.py +1 -1
- metadata/generated/schema/entity/data/apiCollection.py +4 -2
- metadata/generated/schema/entity/data/apiEndpoint.py +4 -2
- metadata/generated/schema/entity/data/chart.py +4 -3
- metadata/generated/schema/entity/data/container.py +4 -2
- metadata/generated/schema/entity/data/dashboard.py +4 -2
- metadata/generated/schema/entity/data/dashboardDataModel.py +5 -2
- metadata/generated/schema/entity/data/database.py +7 -7
- metadata/generated/schema/entity/data/databaseSchema.py +5 -2
- metadata/generated/schema/entity/data/glossary.py +2 -2
- metadata/generated/schema/entity/data/glossaryTerm.py +3 -2
- metadata/generated/schema/entity/data/metric.py +196 -0
- metadata/generated/schema/entity/data/mlmodel.py +4 -2
- metadata/generated/schema/entity/data/pipeline.py +4 -3
- metadata/generated/schema/entity/data/query.py +3 -5
- metadata/generated/schema/entity/data/report.py +1 -1
- metadata/generated/schema/entity/data/searchIndex.py +16 -2
- metadata/generated/schema/entity/data/storedProcedure.py +4 -2
- metadata/generated/schema/entity/data/table.py +36 -3
- metadata/generated/schema/entity/data/topic.py +4 -3
- metadata/generated/schema/entity/docStore/__init__.py +1 -1
- metadata/generated/schema/entity/docStore/document.py +1 -1
- metadata/generated/schema/entity/domains/__init__.py +1 -1
- metadata/generated/schema/entity/domains/dataProduct.py +1 -1
- metadata/generated/schema/entity/domains/domain.py +1 -1
- metadata/generated/schema/entity/events/__init__.py +1 -1
- metadata/generated/schema/entity/events/webhook.py +19 -2
- 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 +1 -1
- 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 +1 -1
- 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 +4 -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 +1 -1
- metadata/generated/schema/entity/services/__init__.py +1 -1
- metadata/generated/schema/entity/services/apiService.py +7 -7
- metadata/generated/schema/entity/services/connections/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/{apiService → api}/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/{apiService → api}/restConnection.py +15 -7
- 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 +29 -1
- metadata/generated/schema/entity/services/connections/dashboard/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/customDashboardConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/domoDashboardConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/lightdashConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/lookerConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/metabaseConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/microStrategyConnection.py +67 -0
- metadata/generated/schema/entity/services/connections/dashboard/modeConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/powerBIConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/powerBIReportServerConnection.py +1 -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 +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/qlikSenseConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/quickSightConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/redashConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/sigmaConnection.py +41 -0
- metadata/generated/schema/entity/services/connections/dashboard/supersetConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/dashboard/tableauConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/database/athenaConnection.py +5 -1
- metadata/generated/schema/entity/services/connections/database/azureSQLConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/bigQueryConnection.py +9 -1
- metadata/generated/schema/entity/services/connections/database/bigTableConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/clickhouseConnection.py +1 -1
- 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 +11 -0
- metadata/generated/schema/entity/services/connections/database/couchbaseConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/customDatabaseConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/databricksConnection.py +1 -1
- 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 +1 -1
- metadata/generated/schema/entity/services/connections/database/db2Connection.py +4 -1
- metadata/generated/schema/entity/services/connections/database/deltaLakeConnection.py +1 -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 +1 -1
- metadata/generated/schema/entity/services/connections/database/dorisConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/druidConnection.py +4 -1
- metadata/generated/schema/entity/services/connections/database/dynamoDBConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/exasolConnection.py +84 -0
- metadata/generated/schema/entity/services/connections/database/glueConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/greenplumConnection.py +4 -1
- metadata/generated/schema/entity/services/connections/database/hiveConnection.py +4 -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 +1 -1
- metadata/generated/schema/entity/services/connections/database/impalaConnection.py +4 -1
- metadata/generated/schema/entity/services/connections/database/mariaDBConnection.py +4 -1
- metadata/generated/schema/entity/services/connections/database/mongoDBConnection.py +3 -2
- metadata/generated/schema/entity/services/connections/database/mssqlConnection.py +5 -1
- metadata/generated/schema/entity/services/connections/database/mysqlConnection.py +11 -7
- metadata/generated/schema/entity/services/connections/database/oracleConnection.py +5 -1
- metadata/generated/schema/entity/services/connections/database/pinotDBConnection.py +4 -1
- metadata/generated/schema/entity/services/connections/database/postgresConnection.py +5 -1
- metadata/generated/schema/entity/services/connections/database/prestoConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/redshiftConnection.py +9 -1
- metadata/generated/schema/entity/services/connections/database/salesforceConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/sapErpConnection.py +1 -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 +5 -1
- metadata/generated/schema/entity/services/connections/database/sasConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/singleStoreConnection.py +4 -1
- metadata/generated/schema/entity/services/connections/database/snowflakeConnection.py +9 -1
- metadata/generated/schema/entity/services/connections/database/sqliteConnection.py +4 -1
- metadata/generated/schema/entity/services/connections/database/synapseConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/teradataConnection.py +4 -1
- metadata/generated/schema/entity/services/connections/database/trinoConnection.py +7 -2
- metadata/generated/schema/entity/services/connections/database/unityCatalogConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/database/verticaConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/messaging/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/messaging/customMessagingConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/messaging/kafkaConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/messaging/kinesisConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/messaging/pulsarConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/messaging/redpandaConnection.py +1 -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 +1 -1
- metadata/generated/schema/entity/services/connections/metadata/alationSinkConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/metadata/amundsenConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/metadata/atlasConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/metadata/metadataESConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/metadata/openMetadataConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/mlmodel/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/mlmodel/customMlModelConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/mlmodel/mlflowConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/mlmodel/sageMakerConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/mlmodel/sklearnConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/mlmodel/vertexaiConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/airbyteConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/airflowConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/backendConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/customPipelineConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/dagsterConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/databricksPipelineConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/datafactoryConnection.py +69 -0
- metadata/generated/schema/entity/services/connections/pipeline/dbtCloudConnection.py +17 -5
- metadata/generated/schema/entity/services/connections/pipeline/domoPipelineConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/fivetranConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/flinkConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/gluePipelineConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/kafkaConnectConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/matillionConnection.py +65 -0
- metadata/generated/schema/entity/services/connections/pipeline/nifiConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/openLineageConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/sparkConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/splineConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/pipeline/stitchConnection.py +43 -0
- metadata/generated/schema/entity/services/connections/search/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/search/customSearchConnection.py +1 -1
- 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 +1 -1
- metadata/generated/schema/entity/services/connections/search/openSearchConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/serviceConnection.py +3 -3
- metadata/generated/schema/entity/services/connections/storage/__init__.py +1 -1
- metadata/generated/schema/entity/services/connections/storage/adlsConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/storage/customStorageConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/storage/gcsConnection.py +1 -1
- metadata/generated/schema/entity/services/connections/storage/s3Connection.py +1 -1
- metadata/generated/schema/entity/services/connections/testConnectionDefinition.py +1 -1
- metadata/generated/schema/entity/services/connections/testConnectionResult.py +1 -1
- metadata/generated/schema/entity/services/dashboardService.py +8 -5
- metadata/generated/schema/entity/services/databaseService.py +5 -2
- metadata/generated/schema/entity/services/ingestionPipelines/__init__.py +1 -1
- metadata/generated/schema/entity/services/ingestionPipelines/ingestionPipeline.py +2 -1
- metadata/generated/schema/entity/services/ingestionPipelines/pipelineServiceClientResponse.py +1 -1
- metadata/generated/schema/entity/services/ingestionPipelines/status.py +1 -1
- metadata/generated/schema/entity/services/messagingService.py +2 -2
- metadata/generated/schema/entity/services/metadataService.py +2 -2
- metadata/generated/schema/entity/services/mlmodelService.py +2 -2
- metadata/generated/schema/entity/services/pipelineService.py +11 -2
- metadata/generated/schema/entity/services/searchService.py +2 -2
- metadata/generated/schema/entity/services/serviceType.py +2 -2
- metadata/generated/schema/entity/services/storageService.py +2 -2
- metadata/generated/schema/entity/teams/__init__.py +1 -1
- metadata/generated/schema/entity/teams/persona.py +1 -1
- metadata/generated/schema/entity/teams/role.py +1 -1
- metadata/generated/schema/entity/teams/team.py +1 -1
- metadata/generated/schema/entity/teams/teamHierarchy.py +1 -1
- metadata/generated/schema/entity/teams/user.py +1 -1
- metadata/generated/schema/entity/type.py +1 -1
- 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 +1 -1
- metadata/generated/schema/events/api/eventSubscriptionDiagnosticInfo.py +69 -0
- metadata/generated/schema/events/api/eventsRecord.py +37 -0
- metadata/generated/schema/events/api/testEventSubscriptionDestination.py +1 -4
- metadata/generated/schema/events/api/typedEvent.py +46 -0
- metadata/generated/schema/events/emailAlertConfig.py +1 -1
- metadata/generated/schema/events/eventFilterRule.py +1 -1
- metadata/generated/schema/events/eventSubscription.py +15 -46
- metadata/generated/schema/events/eventSubscriptionOffset.py +7 -2
- metadata/generated/schema/events/failedEvent.py +1 -1
- metadata/generated/schema/events/failedEventResponse.py +38 -0
- metadata/generated/schema/events/filterResourceDescriptor.py +1 -1
- metadata/generated/schema/events/statusContext.py +43 -0
- metadata/generated/schema/events/subscriptionResourceDescriptor.py +1 -1
- metadata/generated/schema/events/subscriptionStatus.py +78 -0
- metadata/generated/schema/events/testDestinationStatus.py +88 -0
- metadata/generated/schema/governance/workflows/__init__.py +3 -0
- metadata/generated/schema/governance/workflows/elements/__init__.py +5 -0
- metadata/generated/schema/governance/workflows/elements/edge.py +29 -0
- metadata/generated/schema/governance/workflows/elements/nodeSubType.py +16 -0
- metadata/generated/schema/governance/workflows/elements/nodeType.py +14 -0
- metadata/generated/schema/governance/workflows/elements/nodes/__init__.py +3 -0
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/__init__.py +5 -0
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/checkEntityAttributesTask.py +44 -0
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/setEntityCertificationTask.py +49 -0
- metadata/generated/schema/governance/workflows/elements/nodes/automatedTask/setGlossaryTermStatusTask.py +42 -0
- metadata/generated/schema/governance/workflows/elements/nodes/endEvent/__init__.py +3 -0
- metadata/generated/schema/governance/workflows/elements/nodes/endEvent/endEvent.py +30 -0
- metadata/generated/schema/governance/workflows/elements/nodes/startEvent/__init__.py +3 -0
- metadata/generated/schema/governance/workflows/elements/nodes/startEvent/startEvent.py +30 -0
- metadata/generated/schema/governance/workflows/elements/nodes/userTask/__init__.py +3 -0
- metadata/generated/schema/governance/workflows/elements/nodes/userTask/userApprovalTask.py +57 -0
- metadata/generated/schema/governance/workflows/elements/triggers/__init__.py +3 -0
- metadata/generated/schema/governance/workflows/elements/triggers/eventBasedEntityTrigger.py +43 -0
- metadata/generated/schema/governance/workflows/elements/triggers/periodicBatchEntityTrigger.py +44 -0
- metadata/generated/schema/{entity/data/metrics.py → governance/workflows/workflowDefinition.py} +36 -49
- metadata/generated/schema/governance/workflows/workflowInstance.py +50 -0
- metadata/generated/schema/governance/workflows/workflowInstanceState.py +73 -0
- metadata/generated/schema/metadataIngestion/__init__.py +1 -1
- metadata/generated/schema/metadataIngestion/apiServiceMetadataPipeline.py +53 -0
- metadata/generated/schema/metadataIngestion/application.py +1 -1
- metadata/generated/schema/metadataIngestion/applicationPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/dashboardServiceMetadataPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/dataInsightPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/databaseServiceAutoClassificationPipeline.py +127 -0
- metadata/generated/schema/metadataIngestion/databaseServiceMetadataPipeline.py +1 -9
- metadata/generated/schema/metadataIngestion/databaseServiceProfilerPipeline.py +28 -17
- metadata/generated/schema/metadataIngestion/databaseServiceQueryLineagePipeline.py +42 -1
- metadata/generated/schema/metadataIngestion/databaseServiceQueryUsagePipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/dbtPipeline.py +1 -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 +9 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/dbtLocalConfig.py +9 -1
- metadata/generated/schema/metadataIngestion/dbtconfig/dbtS3Config.py +1 -1
- metadata/generated/schema/metadataIngestion/messagingServiceMetadataPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/metadataToElasticSearchPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/mlmodelServiceMetadataPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/pipelineServiceMetadataPipeline.py +1 -1
- metadata/generated/schema/metadataIngestion/searchServiceMetadataPipeline.py +9 -1
- 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 +1 -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 +4 -1
- metadata/generated/schema/metadataIngestion/workflow.py +15 -1
- metadata/generated/schema/monitoring/__init__.py +1 -1
- metadata/generated/schema/monitoring/eventMonitorProvider.py +1 -1
- 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 +1 -1
- 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 +1 -1
- 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 +1 -1
- metadata/generated/schema/security/credentials/gcpCredentials.py +1 -1
- metadata/generated/schema/security/credentials/gcpExternalAccount.py +1 -1
- 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 +1 -1
- metadata/generated/schema/security/credentials/gitlabCredentials.py +1 -1
- 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 +1 -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 +7 -1
- metadata/generated/schema/system/__init__.py +1 -1
- metadata/generated/schema/system/entityError.py +1 -1
- metadata/generated/schema/system/eventPublisherJob.py +2 -1
- metadata/generated/schema/system/indexingError.py +1 -1
- metadata/generated/schema/system/limitsResponse.py +1 -1
- metadata/generated/schema/system/ui/__init__.py +1 -1
- metadata/generated/schema/system/ui/knowledgePanel.py +1 -1
- metadata/generated/schema/system/ui/page.py +1 -1
- metadata/generated/schema/system/validationResponse.py +1 -1
- metadata/generated/schema/tests/__init__.py +1 -1
- metadata/generated/schema/tests/assigned.py +1 -1
- metadata/generated/schema/tests/basic.py +19 -4
- 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 +6 -2
- metadata/generated/schema/tests/testCaseResolutionStatus.py +17 -2
- metadata/generated/schema/tests/testDefinition.py +1 -1
- metadata/generated/schema/tests/testSuite.py +9 -2
- metadata/generated/schema/type/__init__.py +1 -1
- metadata/generated/schema/type/apiSchema.py +1 -1
- metadata/generated/schema/type/assetCertification.py +26 -0
- metadata/generated/schema/type/auditLog.py +1 -1
- metadata/generated/schema/type/basic.py +1 -1
- metadata/generated/schema/type/bulkOperationResult.py +1 -1
- metadata/generated/schema/type/changeEvent.py +1 -1
- metadata/generated/schema/type/changeEventType.py +1 -1
- metadata/generated/schema/type/collectionDescriptor.py +1 -1
- 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 +8 -1
- metadata/generated/schema/type/dailyCount.py +1 -1
- metadata/generated/schema/type/databaseConnectionConfig.py +1 -1
- metadata/generated/schema/type/entityHierarchy.py +1 -1
- metadata/generated/schema/type/entityHistory.py +1 -1
- metadata/generated/schema/type/entityLineage.py +1 -1
- metadata/generated/schema/type/entityReference.py +1 -1
- metadata/generated/schema/type/entityReferenceList.py +1 -1
- metadata/generated/schema/type/entityRelationship.py +1 -1
- metadata/generated/schema/type/entityUsage.py +1 -1
- metadata/generated/schema/type/filterPattern.py +1 -1
- 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/lifeCycle.py +1 -1
- metadata/generated/schema/type/paging.py +1 -1
- metadata/generated/schema/type/profile.py +1 -1
- metadata/generated/schema/type/queryParserData.py +1 -1
- metadata/generated/schema/type/reaction.py +1 -1
- metadata/generated/schema/type/schedule.py +1 -1
- metadata/generated/schema/type/schema.py +1 -1
- metadata/generated/schema/type/tableQuery.py +1 -1
- metadata/generated/schema/type/tableUsageCount.py +1 -1
- metadata/generated/schema/type/tagLabel.py +1 -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/ingestion/api/models.py +9 -5
- metadata/ingestion/api/parser.py +14 -0
- metadata/ingestion/lineage/parser.py +4 -6
- metadata/ingestion/lineage/sql_lineage.py +29 -185
- metadata/ingestion/models/custom_basemodel_validation.py +80 -0
- metadata/ingestion/models/custom_pydantic.py +33 -1
- metadata/ingestion/models/custom_types.py +2 -0
- metadata/ingestion/models/entity_interface.py +71 -0
- metadata/ingestion/models/patch_request.py +0 -1
- metadata/ingestion/ometa/mixins/es_mixin.py +66 -8
- metadata/ingestion/ometa/models.py +1 -0
- metadata/ingestion/ometa/ometa_api.py +6 -3
- metadata/ingestion/ometa/routes.py +4 -2
- metadata/ingestion/processor/query_parser.py +2 -2
- metadata/ingestion/sink/metadata_rest.py +23 -18
- metadata/ingestion/source/api/api_service.py +212 -0
- metadata/ingestion/source/api/rest/connection.py +93 -0
- metadata/ingestion/source/api/rest/metadata.py +278 -0
- metadata/ingestion/source/api/rest/models.py +41 -0
- metadata/ingestion/source/api/rest/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/domodashboard/service_spec.py +6 -0
- metadata/ingestion/source/dashboard/lightdash/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/looker/metadata.py +33 -4
- metadata/ingestion/source/dashboard/looker/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/metabase/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/microstrategy/__init__.py +0 -0
- metadata/ingestion/source/dashboard/{mstr → microstrategy}/client.py +27 -32
- metadata/ingestion/source/dashboard/{mstr → microstrategy}/connection.py +7 -10
- metadata/ingestion/source/dashboard/{mstr → microstrategy}/metadata.py +18 -14
- metadata/ingestion/source/dashboard/{mstr → microstrategy}/models.py +1 -1
- metadata/ingestion/source/dashboard/microstrategy/service_spec.py +6 -0
- metadata/ingestion/source/dashboard/mode/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/powerbi/client.py +1 -2
- metadata/ingestion/source/dashboard/powerbi/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/qlikcloud/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/qliksense/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/quicksight/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/redash/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/sigma/__init__.py +0 -0
- metadata/ingestion/source/dashboard/sigma/client.py +186 -0
- metadata/ingestion/source/dashboard/sigma/connection.py +67 -0
- metadata/ingestion/source/dashboard/sigma/metadata.py +331 -0
- metadata/ingestion/source/dashboard/sigma/models.py +97 -0
- metadata/ingestion/source/dashboard/sigma/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/superset/service_spec.py +4 -0
- metadata/ingestion/source/dashboard/tableau/service_spec.py +4 -0
- metadata/ingestion/source/database/athena/service_spec.py +10 -0
- metadata/ingestion/source/database/azuresql/service_spec.py +12 -0
- metadata/ingestion/source/database/bigquery/connection.py +28 -2
- metadata/ingestion/source/database/bigquery/lineage.py +29 -2
- metadata/ingestion/source/database/bigquery/metadata.py +1 -25
- metadata/ingestion/source/database/bigquery/profiler/__init__.py +0 -0
- metadata/ingestion/source/database/bigquery/profiler/profiler.py +30 -0
- metadata/ingestion/source/database/bigquery/profiler/system.py +165 -0
- metadata/ingestion/source/database/bigquery/queries.py +69 -1
- metadata/ingestion/source/database/bigquery/service_spec.py +16 -0
- metadata/ingestion/source/database/bigtable/service_spec.py +4 -0
- metadata/ingestion/source/database/clickhouse/service_spec.py +12 -0
- metadata/ingestion/source/database/common_db_source.py +150 -147
- metadata/ingestion/source/database/common_nosql_source.py +0 -14
- metadata/ingestion/source/database/couchbase/metadata.py +62 -2
- metadata/{data_quality/builders/sqa_validator_builder.py → ingestion/source/database/couchbase/models.py} +15 -9
- metadata/ingestion/source/database/couchbase/queries.py +6 -1
- metadata/ingestion/source/database/couchbase/service_spec.py +4 -0
- metadata/ingestion/source/database/database_service.py +15 -22
- metadata/ingestion/source/database/databricks/metadata.py +12 -1
- metadata/ingestion/source/database/databricks/service_spec.py +20 -0
- metadata/ingestion/source/database/datalake/clients/azure_blob.py +0 -3
- metadata/ingestion/source/database/datalake/clients/base.py +10 -1
- metadata/ingestion/source/database/datalake/clients/gcs.py +10 -6
- metadata/ingestion/source/database/datalake/clients/s3.py +0 -3
- metadata/ingestion/source/database/datalake/metadata.py +2 -13
- metadata/ingestion/source/database/datalake/service_spec.py +16 -0
- metadata/ingestion/source/database/db2/lineage.py +46 -0
- metadata/ingestion/source/database/db2/service_spec.py +9 -0
- metadata/ingestion/source/database/dbt/constants.py +4 -0
- metadata/ingestion/source/database/dbt/dbt_config.py +19 -1
- metadata/ingestion/source/database/dbt/dbt_service.py +8 -1
- metadata/ingestion/source/database/dbt/dbt_utils.py +29 -0
- metadata/ingestion/source/database/dbt/metadata.py +73 -7
- metadata/ingestion/source/database/dbt/models.py +2 -0
- metadata/ingestion/source/database/dbt/service_spec.py +4 -0
- metadata/ingestion/source/database/deltalake/metadata.py +1 -12
- metadata/ingestion/source/database/deltalake/service_spec.py +4 -0
- metadata/ingestion/source/database/domodatabase/metadata.py +1 -12
- metadata/ingestion/source/database/domodatabase/service_spec.py +4 -0
- metadata/ingestion/source/database/doris/metadata.py +1 -1
- metadata/ingestion/source/database/doris/service_spec.py +4 -0
- metadata/ingestion/source/database/druid/lineage.py +46 -0
- metadata/ingestion/source/database/druid/service_spec.py +4 -0
- metadata/ingestion/source/database/dynamodb/service_spec.py +10 -0
- metadata/ingestion/source/database/exasol/__init__.py +0 -0
- metadata/ingestion/source/database/exasol/connection.py +87 -0
- metadata/ingestion/source/database/exasol/metadata.py +27 -0
- metadata/ingestion/source/database/exasol/service_spec.py +4 -0
- metadata/ingestion/source/database/glue/metadata.py +49 -19
- metadata/ingestion/source/database/glue/models.py +8 -0
- metadata/ingestion/source/database/glue/service_spec.py +4 -0
- metadata/ingestion/source/database/greenplum/lineage.py +46 -0
- metadata/ingestion/source/database/greenplum/service_spec.py +4 -0
- metadata/ingestion/source/database/hive/lineage.py +46 -0
- metadata/ingestion/source/database/hive/service_spec.py +4 -0
- metadata/ingestion/source/database/iceberg/metadata.py +1 -16
- metadata/ingestion/source/database/iceberg/service_spec.py +4 -0
- metadata/ingestion/source/database/impala/lineage.py +46 -0
- metadata/ingestion/source/database/impala/service_spec.py +4 -0
- metadata/ingestion/source/database/life_cycle_query_mixin.py +3 -5
- metadata/ingestion/source/database/lineage_source.py +124 -32
- metadata/ingestion/source/database/mariadb/lineage.py +46 -0
- metadata/ingestion/source/database/mariadb/service_spec.py +4 -0
- metadata/ingestion/source/database/mongodb/service_spec.py +10 -0
- metadata/ingestion/source/database/mssql/lineage.py +51 -2
- metadata/ingestion/source/database/mssql/metadata.py +2 -44
- metadata/ingestion/source/database/mssql/service_spec.py +10 -0
- metadata/ingestion/source/database/mysql/lineage.py +0 -2
- metadata/ingestion/source/database/mysql/queries.py +0 -2
- metadata/ingestion/source/database/mysql/service_spec.py +10 -0
- metadata/ingestion/source/database/mysql/usage.py +0 -2
- metadata/ingestion/source/database/oracle/lineage.py +28 -1
- metadata/ingestion/source/database/oracle/metadata.py +2 -23
- metadata/ingestion/source/database/oracle/service_spec.py +10 -0
- metadata/ingestion/source/database/pinotdb/lineage.py +46 -0
- metadata/ingestion/source/database/pinotdb/service_spec.py +4 -0
- metadata/ingestion/source/database/postgres/metadata.py +4 -13
- metadata/ingestion/source/database/postgres/service_spec.py +10 -0
- metadata/ingestion/source/database/postgres/utils.py +26 -0
- metadata/ingestion/source/database/presto/service_spec.py +4 -0
- metadata/ingestion/source/database/query/service_spec.py +9 -0
- metadata/ingestion/source/database/query_parser_source.py +1 -0
- metadata/ingestion/source/database/redshift/connection.py +9 -2
- metadata/ingestion/source/database/redshift/lineage.py +27 -3
- metadata/ingestion/source/database/redshift/metadata.py +2 -27
- metadata/ingestion/source/database/redshift/profiler/__init__.py +0 -0
- metadata/ingestion/source/database/redshift/profiler/profiler.py +14 -0
- metadata/ingestion/source/database/redshift/profiler/system.py +148 -0
- metadata/ingestion/source/database/redshift/queries.py +99 -8
- metadata/ingestion/source/database/redshift/service_spec.py +14 -0
- metadata/ingestion/source/database/redshift/utils.py +3 -3
- metadata/ingestion/source/database/salesforce/metadata.py +1 -12
- metadata/ingestion/source/database/salesforce/service_spec.py +4 -0
- metadata/ingestion/source/database/sample_data.py +129 -34
- metadata/ingestion/source/database/saperp/service_spec.py +4 -0
- metadata/ingestion/source/database/saphana/service_spec.py +7 -0
- metadata/ingestion/source/database/sas/metadata.py +1 -10
- metadata/ingestion/source/database/sas/service_spec.py +4 -0
- metadata/ingestion/source/database/singlestore/lineage.py +46 -0
- metadata/ingestion/source/database/singlestore/service_spec.py +9 -0
- metadata/ingestion/source/database/snowflake/connection.py +6 -1
- metadata/ingestion/source/database/snowflake/lineage.py +31 -2
- metadata/ingestion/source/database/snowflake/metadata.py +20 -62
- metadata/ingestion/source/database/snowflake/models.py +50 -2
- metadata/ingestion/source/database/snowflake/profiler/__init__.py +0 -0
- metadata/ingestion/source/database/snowflake/profiler/profiler.py +26 -0
- metadata/ingestion/source/database/snowflake/profiler/system.py +436 -0
- metadata/ingestion/source/database/snowflake/queries.py +28 -22
- metadata/ingestion/source/database/snowflake/service_spec.py +20 -0
- metadata/ingestion/source/database/sql_column_handler.py +26 -0
- metadata/ingestion/source/database/sqlite/lineage.py +46 -0
- metadata/ingestion/source/database/sqlite/service_spec.py +4 -0
- metadata/ingestion/source/database/stored_procedures_mixin.py +78 -52
- metadata/ingestion/source/database/teradata/lineage.py +46 -0
- metadata/ingestion/source/database/teradata/service_spec.py +4 -0
- metadata/ingestion/source/database/trino/connection.py +60 -23
- metadata/ingestion/source/database/trino/profiler/__init__.py +0 -0
- metadata/ingestion/source/database/trino/profiler/system_tables_profiler.py +167 -0
- metadata/ingestion/source/database/trino/service_spec.py +16 -0
- metadata/ingestion/source/database/unitycatalog/metadata.py +3 -24
- metadata/ingestion/source/database/unitycatalog/service_spec.py +22 -0
- metadata/ingestion/source/database/usage_source.py +4 -1
- metadata/ingestion/source/database/vertica/service_spec.py +10 -0
- metadata/ingestion/source/messaging/kafka/service_spec.py +4 -0
- metadata/ingestion/source/messaging/kinesis/service_spec.py +4 -0
- metadata/ingestion/source/messaging/redpanda/service_spec.py +4 -0
- metadata/ingestion/source/metadata/alationsink/service_spec.py +4 -0
- metadata/ingestion/source/metadata/amundsen/service_spec.py +4 -0
- metadata/ingestion/source/metadata/atlas/service_spec.py +4 -0
- metadata/ingestion/source/mlmodel/mlflow/service_spec.py +4 -0
- metadata/ingestion/source/mlmodel/sagemaker/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/airbyte/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/airflow/connection.py +45 -2
- metadata/ingestion/source/pipeline/airflow/metadata.py +71 -27
- metadata/ingestion/source/pipeline/airflow/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/dagster/metadata.py +3 -1
- metadata/ingestion/source/pipeline/dagster/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/databrickspipeline/service_spec.py +6 -0
- metadata/ingestion/source/pipeline/dbtcloud/client.py +67 -28
- metadata/ingestion/source/pipeline/dbtcloud/connection.py +1 -3
- metadata/ingestion/source/pipeline/dbtcloud/models.py +1 -1
- metadata/ingestion/source/pipeline/dbtcloud/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/domopipeline/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/fivetran/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/flink/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/gluepipeline/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/kafkaconnect/client.py +1 -1
- metadata/ingestion/source/pipeline/kafkaconnect/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/nifi/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/openlineage/service_spec.py +4 -0
- metadata/ingestion/source/pipeline/pipeline_service.py +3 -1
- metadata/ingestion/source/pipeline/spline/service_spec.py +4 -0
- metadata/ingestion/source/search/elasticsearch/metadata.py +53 -0
- metadata/ingestion/source/search/elasticsearch/service_spec.py +4 -0
- metadata/ingestion/source/search/search_service.py +44 -1
- metadata/ingestion/source/storage/gcs/metadata.py +2 -2
- metadata/ingestion/source/storage/gcs/service_spec.py +4 -0
- metadata/ingestion/source/storage/s3/service_spec.py +4 -0
- metadata/mixins/pandas/pandas_mixin.py +2 -31
- metadata/mixins/sqalchemy/sqa_mixin.py +18 -17
- metadata/parsers/avro_parser.py +51 -16
- metadata/parsers/json_schema_parser.py +45 -16
- metadata/pii/processor.py +10 -9
- metadata/pii/scanners/ner_scanner.py +2 -1
- metadata/profiler/api/models.py +4 -75
- metadata/profiler/config.py +39 -0
- metadata/profiler/interface/nosql/profiler_interface.py +1 -26
- metadata/profiler/interface/pandas/profiler_interface.py +37 -77
- metadata/profiler/interface/profiler_interface.py +25 -279
- metadata/profiler/interface/sqlalchemy/bigquery/profiler_interface.py +0 -19
- metadata/profiler/interface/sqlalchemy/databricks/profiler_interface.py +2 -17
- metadata/profiler/interface/sqlalchemy/db2/profiler_interface.py +1 -1
- metadata/profiler/interface/sqlalchemy/mariadb/profiler_interface.py +7 -5
- metadata/profiler/interface/sqlalchemy/profiler_interface.py +73 -125
- metadata/profiler/interface/sqlalchemy/single_store/profiler_interface.py +5 -5
- metadata/profiler/interface/sqlalchemy/snowflake/profiler_interface.py +1 -1
- metadata/profiler/interface/sqlalchemy/stored_statistics_profiler.py +145 -0
- metadata/profiler/interface/sqlalchemy/trino/profiler_interface.py +9 -4
- metadata/profiler/metrics/system/system.py +69 -374
- metadata/profiler/orm/converter/base.py +23 -13
- metadata/profiler/orm/converter/bigquery/converter.py +1 -1
- metadata/profiler/orm/functions/count.py +9 -0
- metadata/profiler/orm/functions/md5.py +39 -0
- metadata/{data_quality/builders/pandas_validator_builder.py → profiler/orm/functions/substr.py} +13 -10
- metadata/profiler/orm/functions/table_metric_computer.py +6 -5
- metadata/profiler/orm/functions/unique_count.py +6 -0
- metadata/profiler/orm/registry.py +27 -2
- metadata/profiler/processor/core.py +13 -62
- metadata/profiler/processor/handle_partition.py +0 -48
- metadata/profiler/processor/processor.py +1 -1
- metadata/profiler/processor/runner.py +111 -35
- metadata/profiler/processor/sample_data_handler.py +7 -4
- metadata/profiler/source/database/base/profiler_source.py +215 -0
- metadata/profiler/source/{bigquery → database/bigquery}/profiler_source.py +4 -4
- metadata/profiler/source/{databricks → database/databricks}/profiler_source.py +6 -7
- metadata/profiler/source/{mariadb → database/mariadb}/metrics/window/first_quartile.py +1 -1
- metadata/profiler/source/{mariadb → database/mariadb}/metrics/window/median.py +1 -1
- metadata/profiler/source/{mariadb → database/mariadb}/metrics/window/third_quartile.py +1 -1
- metadata/profiler/source/{single_store → database/single_store}/metrics/window/first_quartile.py +3 -1
- metadata/profiler/source/{single_store → database/single_store}/metrics/window/median.py +3 -1
- metadata/profiler/source/{single_store → database/single_store}/metrics/window/third_quartile.py +3 -1
- metadata/profiler/source/fetcher/config.py +44 -0
- metadata/profiler/source/fetcher/entity_fetcher.py +62 -0
- metadata/profiler/source/fetcher/fetcher_strategy.py +299 -0
- metadata/profiler/source/fetcher/profiler_source_factory.py +89 -0
- metadata/profiler/source/metadata.py +16 -201
- metadata/profiler/source/metadata_ext.py +23 -18
- metadata/profiler/source/model.py +29 -0
- metadata/sampler/config.py +237 -0
- metadata/sampler/models.py +106 -0
- metadata/{profiler/processor/sampler → sampler}/nosql/sampler.py +44 -12
- metadata/sampler/pandas/sampler.py +239 -0
- metadata/{utils → sampler}/partition.py +66 -51
- metadata/sampler/processor.py +158 -0
- metadata/sampler/sampler_interface.py +251 -0
- metadata/{profiler/processor/sampler → sampler}/sqlalchemy/azuresql/sampler.py +1 -1
- metadata/{profiler/processor/sampler → sampler}/sqlalchemy/bigquery/sampler.py +55 -26
- metadata/sampler/sqlalchemy/postgres/sampler.py +91 -0
- metadata/{profiler/processor/sampler → sampler}/sqlalchemy/sampler.py +90 -98
- metadata/sampler/sqlalchemy/snowflake/sampler.py +95 -0
- metadata/{profiler/processor/sampler → sampler}/sqlalchemy/trino/sampler.py +5 -3
- metadata/utils/class_helper.py +5 -0
- metadata/utils/collaborative_super.py +28 -0
- metadata/utils/constants.py +65 -0
- metadata/utils/constraints.py +65 -0
- metadata/utils/dict.py +8 -0
- metadata/utils/dispatch.py +1 -53
- metadata/utils/entity_link.py +8 -2
- metadata/utils/entity_utils.py +93 -0
- metadata/utils/filters.py +15 -0
- metadata/utils/fqn.py +15 -0
- metadata/utils/helpers.py +3 -2
- metadata/utils/importer.py +12 -24
- metadata/utils/logger.py +31 -1
- metadata/utils/lru_cache.py +76 -15
- metadata/utils/profiler_utils.py +43 -9
- metadata/utils/service_spec/__init__.py +5 -0
- metadata/utils/service_spec/default.py +21 -0
- metadata/utils/service_spec/service_spec.py +146 -0
- metadata/utils/ssl_manager.py +33 -0
- metadata/utils/time_utils.py +33 -5
- metadata/workflow/application.py +7 -36
- metadata/workflow/base.py +47 -17
- metadata/workflow/classification.py +53 -0
- metadata/workflow/ingestion.py +6 -38
- metadata/workflow/profiler.py +1 -17
- metadata/workflow/workflow_output_handler.py +11 -8
- metadata/workflow/workflow_status_mixin.py +0 -2
- openmetadata_ingestion-1.6.0.0rc2.dist-info/LICENSE +114 -0
- {openmetadata_ingestion-1.5.13.0.dist-info → openmetadata_ingestion-1.6.0.0rc2.dist-info}/METADATA +473 -548
- {openmetadata_ingestion-1.5.13.0.dist-info → openmetadata_ingestion-1.6.0.0rc2.dist-info}/RECORD +995 -809
- {openmetadata_ingestion-1.5.13.0.dist-info → openmetadata_ingestion-1.6.0.0rc2.dist-info}/entry_points.txt +3 -0
- metadata/data_quality/interface/test_suite_interface_factory.py +0 -128
- metadata/data_quality/runner/test_suite_source_factory.py +0 -38
- metadata/generated/schema/entity/services/connections/dashboard/mstrConnection.py +0 -54
- metadata/profiler/interface/profiler_interface_factory.py +0 -107
- metadata/profiler/metrics/system/queries/bigquery.py +0 -54
- metadata/profiler/metrics/system/queries/redshift.py +0 -103
- metadata/profiler/metrics/system/queries/snowflake.py +0 -191
- metadata/profiler/processor/sampler/pandas/sampler.py +0 -170
- metadata/profiler/processor/sampler/sampler_factory.py +0 -100
- metadata/profiler/processor/sampler/sampler_interface.py +0 -74
- metadata/profiler/processor/sampler/sqlalchemy/snowflake/sampler.py +0 -60
- metadata/profiler/source/base/profiler_source.py +0 -299
- metadata/profiler/source/profiler_source_factory.py +0 -54
- openmetadata_ingestion-1.5.13.0.dist-info/LICENSE +0 -201
- {metadata/ingestion/source/dashboard/mstr → _openmetadata_testutils/pytest_openmetadata}/__init__.py +0 -0
- /metadata/profiler/source/{bigquery → database/bigquery}/type_mapper.py +0 -0
- /metadata/profiler/source/{mariadb → database/mariadb}/functions/median.py +0 -0
- /metadata/profiler/source/{single_store → database/single_store}/functions/median.py +0 -0
- {openmetadata_ingestion-1.5.13.0.dist-info → openmetadata_ingestion-1.6.0.0rc2.dist-info}/WHEEL +0 -0
- {openmetadata_ingestion-1.5.13.0.dist-info → openmetadata_ingestion-1.6.0.0rc2.dist-info}/top_level.txt +0 -0
|
@@ -1,128 +0,0 @@
|
|
|
1
|
-
# Copyright 2021 Collate
|
|
2
|
-
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
3
|
-
# you may not use this file except in compliance with the License.
|
|
4
|
-
# You may obtain a copy of the License at
|
|
5
|
-
# http://www.apache.org/licenses/LICENSE-2.0
|
|
6
|
-
# Unless required by applicable law or agreed to in writing, software
|
|
7
|
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
8
|
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
9
|
-
# See the License for the specific language governing permissions and
|
|
10
|
-
# limitations under the License.
|
|
11
|
-
"""
|
|
12
|
-
Interface factory
|
|
13
|
-
"""
|
|
14
|
-
import traceback
|
|
15
|
-
from logging import Logger
|
|
16
|
-
|
|
17
|
-
from metadata.data_quality.interface.pandas.pandas_test_suite_interface import (
|
|
18
|
-
PandasTestSuiteInterface,
|
|
19
|
-
)
|
|
20
|
-
from metadata.data_quality.interface.sqlalchemy.databricks.test_suite_interface import (
|
|
21
|
-
DatabricksTestSuiteInterface,
|
|
22
|
-
)
|
|
23
|
-
from metadata.data_quality.interface.sqlalchemy.snowflake.test_suite_interface import (
|
|
24
|
-
SnowflakeTestSuiteInterface,
|
|
25
|
-
)
|
|
26
|
-
from metadata.data_quality.interface.sqlalchemy.sqa_test_suite_interface import (
|
|
27
|
-
SQATestSuiteInterface,
|
|
28
|
-
)
|
|
29
|
-
from metadata.data_quality.interface.sqlalchemy.unity_catalog.test_suite_interface import (
|
|
30
|
-
UnityCatalogTestSuiteInterface,
|
|
31
|
-
)
|
|
32
|
-
from metadata.data_quality.interface.test_suite_interface import TestSuiteInterface
|
|
33
|
-
from metadata.generated.schema.entity.data.table import Table
|
|
34
|
-
from metadata.generated.schema.entity.services.connections.database.databricksConnection import (
|
|
35
|
-
DatabricksConnection,
|
|
36
|
-
)
|
|
37
|
-
from metadata.generated.schema.entity.services.connections.database.datalakeConnection import (
|
|
38
|
-
DatalakeConnection,
|
|
39
|
-
)
|
|
40
|
-
from metadata.generated.schema.entity.services.connections.database.snowflakeConnection import (
|
|
41
|
-
SnowflakeConnection,
|
|
42
|
-
)
|
|
43
|
-
from metadata.generated.schema.entity.services.connections.database.unityCatalogConnection import (
|
|
44
|
-
UnityCatalogConnection,
|
|
45
|
-
)
|
|
46
|
-
from metadata.generated.schema.entity.services.databaseService import DatabaseConnection
|
|
47
|
-
from metadata.ingestion.ometa.ometa_api import OpenMetadata
|
|
48
|
-
from metadata.utils.logger import test_suite_logger
|
|
49
|
-
|
|
50
|
-
logger: Logger = test_suite_logger()
|
|
51
|
-
|
|
52
|
-
|
|
53
|
-
class TestSuiteInterfaceFactory:
|
|
54
|
-
"""Factory class for the data quality interface"""
|
|
55
|
-
|
|
56
|
-
def __init__(self):
|
|
57
|
-
"""Initialize the interface factory"""
|
|
58
|
-
self._interface_type = {
|
|
59
|
-
"base": SQATestSuiteInterface,
|
|
60
|
-
DatalakeConnection.__name__: PandasTestSuiteInterface,
|
|
61
|
-
}
|
|
62
|
-
|
|
63
|
-
def register(self, interface_type: str, interface: TestSuiteInterface):
|
|
64
|
-
"""Register the interface
|
|
65
|
-
|
|
66
|
-
Args:
|
|
67
|
-
interface_type (str): type of the interface
|
|
68
|
-
interface (TestSuiteInterface): a class that implements the TestSuiteInterface
|
|
69
|
-
"""
|
|
70
|
-
self._interface_type[interface_type] = interface
|
|
71
|
-
|
|
72
|
-
def register_many(self, interface_dict):
|
|
73
|
-
"""
|
|
74
|
-
Registers multiple profiler interfaces at once.
|
|
75
|
-
|
|
76
|
-
Args:
|
|
77
|
-
interface_dict: A dictionary mapping connection class names (strings) to their
|
|
78
|
-
corresponding profiler interface classes.
|
|
79
|
-
"""
|
|
80
|
-
for interface_type, interface_class in interface_dict.items():
|
|
81
|
-
self.register(interface_type, interface_class)
|
|
82
|
-
|
|
83
|
-
def create(
|
|
84
|
-
self,
|
|
85
|
-
service_connection_config: DatabaseConnection,
|
|
86
|
-
ometa_client: OpenMetadata,
|
|
87
|
-
table_entity: Table,
|
|
88
|
-
*args,
|
|
89
|
-
**kwargs,
|
|
90
|
-
) -> TestSuiteInterface:
|
|
91
|
-
"""Create the interface
|
|
92
|
-
|
|
93
|
-
Args:
|
|
94
|
-
service_connection_config (DatabaseService): a database service object
|
|
95
|
-
|
|
96
|
-
Raises:
|
|
97
|
-
AttributeError: if no connection is found in the database service object
|
|
98
|
-
|
|
99
|
-
Returns:
|
|
100
|
-
TestSuiteInterface:
|
|
101
|
-
"""
|
|
102
|
-
try:
|
|
103
|
-
connection_type = service_connection_config.__class__.__name__
|
|
104
|
-
except AttributeError as err:
|
|
105
|
-
logger.debug(traceback.format_exc())
|
|
106
|
-
raise AttributeError(f"Could not instantiate interface class: {err}")
|
|
107
|
-
interface = self._interface_type.get(connection_type)
|
|
108
|
-
|
|
109
|
-
if not interface:
|
|
110
|
-
interface = self._interface_type["base"]
|
|
111
|
-
|
|
112
|
-
return interface(
|
|
113
|
-
service_connection_config, ometa_client, table_entity, *args, **kwargs
|
|
114
|
-
)
|
|
115
|
-
|
|
116
|
-
|
|
117
|
-
test_suite_interface_factory = TestSuiteInterfaceFactory()
|
|
118
|
-
|
|
119
|
-
|
|
120
|
-
test_suite_interface = {
|
|
121
|
-
DatabaseConnection.__name__: SQATestSuiteInterface,
|
|
122
|
-
DatalakeConnection.__name__: PandasTestSuiteInterface,
|
|
123
|
-
SnowflakeConnection.__name__: SnowflakeTestSuiteInterface,
|
|
124
|
-
UnityCatalogConnection.__name__: UnityCatalogTestSuiteInterface,
|
|
125
|
-
DatabricksConnection.__name__: DatabricksTestSuiteInterface,
|
|
126
|
-
}
|
|
127
|
-
|
|
128
|
-
test_suite_interface_factory.register_many(test_suite_interface)
|
|
@@ -1,38 +0,0 @@
|
|
|
1
|
-
# Copyright 2021 Collate
|
|
2
|
-
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
3
|
-
# you may not use this file except in compliance with the License.
|
|
4
|
-
# You may obtain a copy of the License at
|
|
5
|
-
# http://www.apache.org/licenses/LICENSE-2.0
|
|
6
|
-
# Unless required by applicable law or agreed to in writing, software
|
|
7
|
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
8
|
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
9
|
-
# See the License for the specific language governing permissions and
|
|
10
|
-
# limitations under the License.
|
|
11
|
-
|
|
12
|
-
"""
|
|
13
|
-
Factory class for creating test suite source objects
|
|
14
|
-
"""
|
|
15
|
-
|
|
16
|
-
from metadata.data_quality.runner.base_test_suite_source import BaseTestSuiteRunner
|
|
17
|
-
|
|
18
|
-
|
|
19
|
-
class TestSuiteRunnerFactory:
|
|
20
|
-
"""Creational factory for test suite source objects"""
|
|
21
|
-
|
|
22
|
-
def __init__(self):
|
|
23
|
-
self._source_type = {"base": BaseTestSuiteRunner}
|
|
24
|
-
|
|
25
|
-
def register_source(self, source_type: str, source_class):
|
|
26
|
-
"""Register a new source type"""
|
|
27
|
-
self._source_type[source_type] = source_class
|
|
28
|
-
|
|
29
|
-
def create(self, source_type: str, *args, **kwargs) -> BaseTestSuiteRunner:
|
|
30
|
-
"""Create source object based on source type"""
|
|
31
|
-
source_class = self._source_type.get(source_type)
|
|
32
|
-
if not source_class:
|
|
33
|
-
source_class = self._source_type["base"]
|
|
34
|
-
return source_class(*args, **kwargs)
|
|
35
|
-
return source_class(*args, **kwargs)
|
|
36
|
-
|
|
37
|
-
|
|
38
|
-
test_suite_source_factory = TestSuiteRunnerFactory()
|
|
@@ -1,54 +0,0 @@
|
|
|
1
|
-
# generated by datamodel-codegen:
|
|
2
|
-
# filename: entity/services/connections/dashboard/mstrConnection.json
|
|
3
|
-
# timestamp: 2024-11-28T18:44:30+00:00
|
|
4
|
-
|
|
5
|
-
from __future__ import annotations
|
|
6
|
-
|
|
7
|
-
from enum import Enum
|
|
8
|
-
from typing import Optional
|
|
9
|
-
|
|
10
|
-
from pydantic import AnyUrl, ConfigDict, Field
|
|
11
|
-
from typing_extensions import Annotated
|
|
12
|
-
|
|
13
|
-
from metadata.ingestion.models.custom_pydantic import BaseModel, CustomSecretStr
|
|
14
|
-
|
|
15
|
-
from .. import connectionBasicType
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
class MstrType(Enum):
|
|
19
|
-
Mstr = 'Mstr'
|
|
20
|
-
|
|
21
|
-
|
|
22
|
-
class MstrConnection(BaseModel):
|
|
23
|
-
model_config = ConfigDict(
|
|
24
|
-
extra='forbid',
|
|
25
|
-
)
|
|
26
|
-
type: Annotated[
|
|
27
|
-
Optional[MstrType],
|
|
28
|
-
Field(MstrType.Mstr, description='Service Type', title='Service Type'),
|
|
29
|
-
]
|
|
30
|
-
username: Annotated[
|
|
31
|
-
str,
|
|
32
|
-
Field(
|
|
33
|
-
description='Username to connect to MSTR. This user should have privileges to read all the metadata in MSTR.',
|
|
34
|
-
title='Username',
|
|
35
|
-
),
|
|
36
|
-
]
|
|
37
|
-
password: Annotated[
|
|
38
|
-
Optional[CustomSecretStr],
|
|
39
|
-
Field(None, description='Password to connect to MSTR.', title='Password'),
|
|
40
|
-
]
|
|
41
|
-
hostPort: Annotated[
|
|
42
|
-
AnyUrl,
|
|
43
|
-
Field(
|
|
44
|
-
description='Host and Port of the Metabase instance.', title='Host and Port'
|
|
45
|
-
),
|
|
46
|
-
]
|
|
47
|
-
projectName: Annotated[
|
|
48
|
-
Optional[str],
|
|
49
|
-
Field(None, description='MSTR Project Name', title='Project Name'),
|
|
50
|
-
]
|
|
51
|
-
supportsMetadataExtraction: Annotated[
|
|
52
|
-
Optional[connectionBasicType.SupportsMetadataExtraction],
|
|
53
|
-
Field(None, title='Supports Metadata Extraction'),
|
|
54
|
-
]
|
|
@@ -1,107 +0,0 @@
|
|
|
1
|
-
# Copyright 2021 Collate
|
|
2
|
-
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
3
|
-
# you may not use this file except in compliance with the License.
|
|
4
|
-
# You may obtain a copy of the License at
|
|
5
|
-
# http://www.apache.org/licenses/LICENSE-2.0
|
|
6
|
-
# Unless required by applicable law or agreed to in writing, software
|
|
7
|
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
8
|
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
9
|
-
# See the License for the specific language governing permissions and
|
|
10
|
-
# limitations under the License.
|
|
11
|
-
|
|
12
|
-
"""
|
|
13
|
-
Factory class for creating profiler interface objects
|
|
14
|
-
"""
|
|
15
|
-
|
|
16
|
-
import importlib
|
|
17
|
-
from typing import Dict, cast
|
|
18
|
-
|
|
19
|
-
from metadata.generated.schema.entity.services.connections.database.bigQueryConnection import (
|
|
20
|
-
BigQueryConnection,
|
|
21
|
-
)
|
|
22
|
-
from metadata.generated.schema.entity.services.connections.database.databricksConnection import (
|
|
23
|
-
DatabricksConnection,
|
|
24
|
-
)
|
|
25
|
-
from metadata.generated.schema.entity.services.connections.database.datalakeConnection import (
|
|
26
|
-
DatalakeConnection,
|
|
27
|
-
)
|
|
28
|
-
from metadata.generated.schema.entity.services.connections.database.db2Connection import (
|
|
29
|
-
Db2Connection,
|
|
30
|
-
)
|
|
31
|
-
from metadata.generated.schema.entity.services.connections.database.dynamoDBConnection import (
|
|
32
|
-
DynamoDBConnection,
|
|
33
|
-
)
|
|
34
|
-
from metadata.generated.schema.entity.services.connections.database.mariaDBConnection import (
|
|
35
|
-
MariaDBConnection,
|
|
36
|
-
)
|
|
37
|
-
from metadata.generated.schema.entity.services.connections.database.mongoDBConnection import (
|
|
38
|
-
MongoDBConnection,
|
|
39
|
-
)
|
|
40
|
-
from metadata.generated.schema.entity.services.connections.database.singleStoreConnection import (
|
|
41
|
-
SingleStoreConnection,
|
|
42
|
-
)
|
|
43
|
-
from metadata.generated.schema.entity.services.connections.database.snowflakeConnection import (
|
|
44
|
-
SnowflakeConnection,
|
|
45
|
-
)
|
|
46
|
-
from metadata.generated.schema.entity.services.connections.database.trinoConnection import (
|
|
47
|
-
TrinoConnection,
|
|
48
|
-
)
|
|
49
|
-
from metadata.generated.schema.entity.services.connections.database.unityCatalogConnection import (
|
|
50
|
-
UnityCatalogConnection,
|
|
51
|
-
)
|
|
52
|
-
from metadata.generated.schema.entity.services.databaseService import DatabaseConnection
|
|
53
|
-
from metadata.profiler.factory import Factory
|
|
54
|
-
from metadata.profiler.interface.profiler_interface import ProfilerInterface
|
|
55
|
-
|
|
56
|
-
|
|
57
|
-
class ProfilerInterfaceFactory(Factory):
|
|
58
|
-
def create(self, interface_type: str, *args, **kwargs):
|
|
59
|
-
"""Create interface object based on interface type"""
|
|
60
|
-
interface_class_path = profiler_class_mapping.get(
|
|
61
|
-
interface_type, profiler_class_mapping[DatabaseConnection.__name__]
|
|
62
|
-
)
|
|
63
|
-
try:
|
|
64
|
-
module_path, class_name = interface_class_path.rsplit(".", 1)
|
|
65
|
-
module = importlib.import_module(module_path)
|
|
66
|
-
profiler_class = getattr(module, class_name)
|
|
67
|
-
except (ImportError, AttributeError) as e:
|
|
68
|
-
raise ImportError(f"Error importing {class_name} from {module_path}: {e}")
|
|
69
|
-
profiler_class = cast(ProfilerInterface, profiler_class)
|
|
70
|
-
return profiler_class.create(*args, **kwargs)
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
profiler_interface_factory = ProfilerInterfaceFactory()
|
|
74
|
-
|
|
75
|
-
BASE_PROFILER_PATH = "metadata.profiler.interface"
|
|
76
|
-
SQLALCHEMY_PROFILER_PATH = f"{BASE_PROFILER_PATH}.sqlalchemy"
|
|
77
|
-
NOSQL_PROFILER_PATH = (
|
|
78
|
-
f"{BASE_PROFILER_PATH}.nosql.profiler_interface.NoSQLProfilerInterface"
|
|
79
|
-
)
|
|
80
|
-
PANDAS_PROFILER_PATH = (
|
|
81
|
-
f"{BASE_PROFILER_PATH}.pandas.profiler_interface.PandasProfilerInterface"
|
|
82
|
-
)
|
|
83
|
-
|
|
84
|
-
# Configuration for dynamic imports
|
|
85
|
-
profiler_class_mapping: Dict[str, str] = {
|
|
86
|
-
DatabaseConnection.__name__: SQLALCHEMY_PROFILER_PATH
|
|
87
|
-
+ ".profiler_interface.SQAProfilerInterface",
|
|
88
|
-
BigQueryConnection.__name__: SQLALCHEMY_PROFILER_PATH
|
|
89
|
-
+ ".bigquery.profiler_interface.BigQueryProfilerInterface",
|
|
90
|
-
SingleStoreConnection.__name__: SQLALCHEMY_PROFILER_PATH
|
|
91
|
-
+ ".single_store.profiler_interface.SingleStoreProfilerInterface",
|
|
92
|
-
DatalakeConnection.__name__: PANDAS_PROFILER_PATH,
|
|
93
|
-
MariaDBConnection.__name__: SQLALCHEMY_PROFILER_PATH
|
|
94
|
-
+ ".mariadb.profiler_interface.MariaDBProfilerInterface",
|
|
95
|
-
SnowflakeConnection.__name__: SQLALCHEMY_PROFILER_PATH
|
|
96
|
-
+ ".snowflake.profiler_interface.SnowflakeProfilerInterface",
|
|
97
|
-
TrinoConnection.__name__: SQLALCHEMY_PROFILER_PATH
|
|
98
|
-
+ ".trino.profiler_interface.TrinoProfilerInterface",
|
|
99
|
-
UnityCatalogConnection.__name__: SQLALCHEMY_PROFILER_PATH
|
|
100
|
-
+ ".unity_catalog.profiler_interface.UnityCatalogProfilerInterface",
|
|
101
|
-
DatabricksConnection.__name__: SQLALCHEMY_PROFILER_PATH
|
|
102
|
-
+ ".databricks.profiler_interface.DatabricksProfilerInterface",
|
|
103
|
-
Db2Connection.__name__: SQLALCHEMY_PROFILER_PATH
|
|
104
|
-
+ ".db2.profiler_interface.DB2ProfilerInterface",
|
|
105
|
-
MongoDBConnection.__name__: NOSQL_PROFILER_PATH,
|
|
106
|
-
DynamoDBConnection.__name__: NOSQL_PROFILER_PATH,
|
|
107
|
-
}
|
|
@@ -1,54 +0,0 @@
|
|
|
1
|
-
# Copyright 2021 Collate
|
|
2
|
-
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
3
|
-
# you may not use this file except in compliance with the License.
|
|
4
|
-
# You may obtain a copy of the License at
|
|
5
|
-
# http://www.apache.org/licenses/LICENSE-2.0
|
|
6
|
-
# Unless required by applicable law or agreed to in writing, software
|
|
7
|
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
8
|
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
9
|
-
# See the License for the specific language governing permissions and
|
|
10
|
-
# limitations under the License.
|
|
11
|
-
|
|
12
|
-
"""
|
|
13
|
-
Bigquery System Metric Queries
|
|
14
|
-
"""
|
|
15
|
-
from datetime import datetime
|
|
16
|
-
|
|
17
|
-
from pydantic import BaseModel
|
|
18
|
-
|
|
19
|
-
from metadata.profiler.metrics.system.dml_operation import DatabaseDMLOperations
|
|
20
|
-
|
|
21
|
-
|
|
22
|
-
class BigQueryQueryResult(BaseModel):
|
|
23
|
-
table_name: dict
|
|
24
|
-
timestamp: datetime
|
|
25
|
-
query_type: str
|
|
26
|
-
dml_statistics: dict
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
DML_STAT_TO_DML_STATEMENT_MAPPING = {
|
|
30
|
-
"inserted_row_count": DatabaseDMLOperations.INSERT.value,
|
|
31
|
-
"deleted_row_count": DatabaseDMLOperations.DELETE.value,
|
|
32
|
-
"updated_row_count": DatabaseDMLOperations.UPDATE.value,
|
|
33
|
-
}
|
|
34
|
-
|
|
35
|
-
JOBS = """
|
|
36
|
-
SELECT
|
|
37
|
-
statement_type,
|
|
38
|
-
start_time,
|
|
39
|
-
destination_table,
|
|
40
|
-
dml_statistics
|
|
41
|
-
FROM
|
|
42
|
-
`region-{usage_location}`.INFORMATION_SCHEMA.JOBS
|
|
43
|
-
WHERE
|
|
44
|
-
DATE(creation_time) >= CURRENT_DATE() - 1 AND
|
|
45
|
-
destination_table.dataset_id = '{dataset_id}' AND
|
|
46
|
-
destination_table.project_id = '{project_id}' AND
|
|
47
|
-
statement_type IN (
|
|
48
|
-
'{insert}',
|
|
49
|
-
'{update}',
|
|
50
|
-
'{delete}',
|
|
51
|
-
'{merge}'
|
|
52
|
-
)
|
|
53
|
-
ORDER BY creation_time DESC;
|
|
54
|
-
"""
|
|
@@ -1,103 +0,0 @@
|
|
|
1
|
-
# Copyright 2021 Collate
|
|
2
|
-
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
3
|
-
# you may not use this file except in compliance with the License.
|
|
4
|
-
# You may obtain a copy of the License at
|
|
5
|
-
# http://www.apache.org/licenses/LICENSE-2.0
|
|
6
|
-
# Unless required by applicable law or agreed to in writing, software
|
|
7
|
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
8
|
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
9
|
-
# See the License for the specific language governing permissions and
|
|
10
|
-
# limitations under the License.
|
|
11
|
-
|
|
12
|
-
"""
|
|
13
|
-
Redshift System Metric Queries and queries operations
|
|
14
|
-
"""
|
|
15
|
-
|
|
16
|
-
from typing import List
|
|
17
|
-
|
|
18
|
-
from sqlalchemy import text
|
|
19
|
-
from sqlalchemy.orm import Session
|
|
20
|
-
|
|
21
|
-
from metadata.utils.profiler_utils import QueryResult
|
|
22
|
-
|
|
23
|
-
STL_QUERY = """
|
|
24
|
-
with data as (
|
|
25
|
-
select
|
|
26
|
-
{alias}.*
|
|
27
|
-
from
|
|
28
|
-
pg_catalog.stl_insert si
|
|
29
|
-
{join_type} join pg_catalog.stl_delete sd on si.query = sd.query
|
|
30
|
-
where
|
|
31
|
-
{condition}
|
|
32
|
-
)
|
|
33
|
-
SELECT
|
|
34
|
-
SUM(data."rows") AS "rows",
|
|
35
|
-
sti."database",
|
|
36
|
-
sti."schema",
|
|
37
|
-
sti."table",
|
|
38
|
-
DATE_TRUNC('second', data.starttime) AS starttime
|
|
39
|
-
FROM
|
|
40
|
-
data
|
|
41
|
-
INNER JOIN pg_catalog.svv_table_info sti ON data.tbl = sti.table_id
|
|
42
|
-
where
|
|
43
|
-
sti."database" = '{database}' AND
|
|
44
|
-
sti."schema" = '{schema}' AND
|
|
45
|
-
"rows" != 0 AND
|
|
46
|
-
DATE(data.starttime) >= CURRENT_DATE - 1
|
|
47
|
-
GROUP BY 2,3,4,5
|
|
48
|
-
ORDER BY 5 DESC
|
|
49
|
-
"""
|
|
50
|
-
|
|
51
|
-
|
|
52
|
-
def get_query_results(
|
|
53
|
-
session: Session,
|
|
54
|
-
query,
|
|
55
|
-
operation,
|
|
56
|
-
) -> List[QueryResult]:
|
|
57
|
-
"""get query results either from cache or from the database
|
|
58
|
-
|
|
59
|
-
Args:
|
|
60
|
-
cache (Optional[List[QueryResult]]): cache results
|
|
61
|
-
session (Session): session
|
|
62
|
-
query (_type_): query
|
|
63
|
-
operation (_type_): operation
|
|
64
|
-
|
|
65
|
-
Returns:
|
|
66
|
-
List[QueryResult]:
|
|
67
|
-
"""
|
|
68
|
-
cursor = session.execute(text(query))
|
|
69
|
-
results = [
|
|
70
|
-
QueryResult(
|
|
71
|
-
database_name=row.database,
|
|
72
|
-
schema_name=row.schema,
|
|
73
|
-
table_name=row.table,
|
|
74
|
-
query_text=None,
|
|
75
|
-
query_type=operation,
|
|
76
|
-
timestamp=row.starttime,
|
|
77
|
-
rows=row.rows,
|
|
78
|
-
)
|
|
79
|
-
for row in cursor
|
|
80
|
-
]
|
|
81
|
-
|
|
82
|
-
return results
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
def get_metric_result(ddls: List[QueryResult], table_name: str) -> List:
|
|
86
|
-
"""Given query results, retur the metric result
|
|
87
|
-
|
|
88
|
-
Args:
|
|
89
|
-
ddls (List[QueryResult]): list of query results
|
|
90
|
-
table_name (str): table name
|
|
91
|
-
|
|
92
|
-
Returns:
|
|
93
|
-
List:
|
|
94
|
-
"""
|
|
95
|
-
return [
|
|
96
|
-
{
|
|
97
|
-
"timestamp": int(ddl.timestamp.timestamp() * 1000),
|
|
98
|
-
"operation": ddl.query_type,
|
|
99
|
-
"rowsAffected": ddl.rows,
|
|
100
|
-
}
|
|
101
|
-
for ddl in ddls
|
|
102
|
-
if ddl.table_name == table_name
|
|
103
|
-
]
|
|
@@ -1,191 +0,0 @@
|
|
|
1
|
-
# Copyright 2021 Collate
|
|
2
|
-
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
3
|
-
# you may not use this file except in compliance with the License.
|
|
4
|
-
# You may obtain a copy of the License at
|
|
5
|
-
# http://www.apache.org/licenses/LICENSE-2.0
|
|
6
|
-
# Unless required by applicable law or agreed to in writing, software
|
|
7
|
-
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
8
|
-
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
9
|
-
# See the License for the specific language governing permissions and
|
|
10
|
-
# limitations under the License.
|
|
11
|
-
|
|
12
|
-
"""
|
|
13
|
-
Snowflake System Metric Queries and query operations
|
|
14
|
-
"""
|
|
15
|
-
|
|
16
|
-
import re
|
|
17
|
-
import traceback
|
|
18
|
-
from typing import Optional, Tuple
|
|
19
|
-
|
|
20
|
-
from sqlalchemy.engine.row import Row
|
|
21
|
-
|
|
22
|
-
from metadata.generated.schema.entity.services.databaseService import DatabaseService
|
|
23
|
-
from metadata.ingestion.lineage.sql_lineage import search_table_entities
|
|
24
|
-
from metadata.ingestion.ometa.ometa_api import OpenMetadata
|
|
25
|
-
from metadata.utils.logger import profiler_logger
|
|
26
|
-
from metadata.utils.profiler_utils import (
|
|
27
|
-
SnowflakeQueryResult,
|
|
28
|
-
get_identifiers_from_string,
|
|
29
|
-
)
|
|
30
|
-
|
|
31
|
-
logger = profiler_logger()
|
|
32
|
-
|
|
33
|
-
INFORMATION_SCHEMA_QUERY = """
|
|
34
|
-
SELECT
|
|
35
|
-
QUERY_ID,
|
|
36
|
-
QUERY_TEXT,
|
|
37
|
-
QUERY_TYPE,
|
|
38
|
-
START_TIME,
|
|
39
|
-
ROWS_INSERTED,
|
|
40
|
-
ROWS_UPDATED,
|
|
41
|
-
ROWS_DELETED
|
|
42
|
-
FROM "SNOWFLAKE"."ACCOUNT_USAGE"."QUERY_HISTORY"
|
|
43
|
-
WHERE
|
|
44
|
-
start_time>= DATEADD('DAY', -1, CURRENT_TIMESTAMP)
|
|
45
|
-
AND QUERY_TEXT ILIKE '%{tablename}%'
|
|
46
|
-
AND QUERY_TYPE IN (
|
|
47
|
-
'{insert}',
|
|
48
|
-
'{update}',
|
|
49
|
-
'{delete}',
|
|
50
|
-
'{merge}'
|
|
51
|
-
)
|
|
52
|
-
AND EXECUTION_STATUS = 'SUCCESS';
|
|
53
|
-
"""
|
|
54
|
-
|
|
55
|
-
RESULT_SCAN = """
|
|
56
|
-
SELECT *
|
|
57
|
-
FROM TABLE(RESULT_SCAN('{query_id}'));
|
|
58
|
-
"""
|
|
59
|
-
|
|
60
|
-
|
|
61
|
-
QUERY_PATTERN = r"(?:(INSERT\s*INTO\s*|INSERT\s*OVERWRITE\s*INTO\s*|UPDATE\s*|MERGE\s*INTO\s*|DELETE\s*FROM\s*))([\w._\"\'()]+)(?=[\s*\n])" # pylint: disable=line-too-long
|
|
62
|
-
IDENTIFIER_PATTERN = r"(IDENTIFIER\(\')([\w._\"]+)(\'\))"
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
def _parse_query(query: str) -> Optional[str]:
|
|
66
|
-
"""Parse snowflake queries to extract the identifiers"""
|
|
67
|
-
match = re.match(QUERY_PATTERN, query, re.IGNORECASE)
|
|
68
|
-
try:
|
|
69
|
-
# This will match results like `DATABASE.SCHEMA.TABLE1` or IDENTIFIER('TABLE1')
|
|
70
|
-
# If we have `IDENTIFIER` type of queries coming from Stored Procedures, we'll need to further clean it up.
|
|
71
|
-
identifier = match.group(2)
|
|
72
|
-
|
|
73
|
-
match_internal_identifier = re.match(
|
|
74
|
-
IDENTIFIER_PATTERN, identifier, re.IGNORECASE
|
|
75
|
-
)
|
|
76
|
-
internal_identifier = (
|
|
77
|
-
match_internal_identifier.group(2) if match_internal_identifier else None
|
|
78
|
-
)
|
|
79
|
-
if internal_identifier:
|
|
80
|
-
return internal_identifier
|
|
81
|
-
|
|
82
|
-
return identifier
|
|
83
|
-
except (IndexError, AttributeError):
|
|
84
|
-
logger.debug("Could not find identifier in query. Skipping row.")
|
|
85
|
-
return None
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
def get_identifiers(
|
|
89
|
-
identifier: str, ometa_client: OpenMetadata, db_service: DatabaseService
|
|
90
|
-
) -> Tuple[Optional[str], Optional[str], Optional[str]]:
|
|
91
|
-
"""Get query identifiers and if needed, fetch them from ES"""
|
|
92
|
-
database_name, schema_name, table_name = get_identifiers_from_string(identifier)
|
|
93
|
-
|
|
94
|
-
if not table_name:
|
|
95
|
-
logger.debug("Could not extract the table name. Skipping operation.")
|
|
96
|
-
return database_name, schema_name, table_name
|
|
97
|
-
|
|
98
|
-
if not all([database_name, schema_name]):
|
|
99
|
-
logger.debug(
|
|
100
|
-
"Missing database or schema info from the query. We'll look for it in ES."
|
|
101
|
-
)
|
|
102
|
-
es_tables = search_table_entities(
|
|
103
|
-
metadata=ometa_client,
|
|
104
|
-
service_name=db_service.fullyQualifiedName.root,
|
|
105
|
-
database=database_name,
|
|
106
|
-
database_schema=schema_name,
|
|
107
|
-
table=table_name,
|
|
108
|
-
)
|
|
109
|
-
|
|
110
|
-
if not es_tables:
|
|
111
|
-
logger.debug("No tables match the search criteria.")
|
|
112
|
-
return database_name, schema_name, table_name
|
|
113
|
-
|
|
114
|
-
if len(es_tables) > 1:
|
|
115
|
-
logger.debug(
|
|
116
|
-
"Found more than 1 table matching the search criteria."
|
|
117
|
-
" Skipping the computation to not mix system data."
|
|
118
|
-
)
|
|
119
|
-
return database_name, schema_name, table_name
|
|
120
|
-
|
|
121
|
-
matched_table = es_tables[0]
|
|
122
|
-
database_name = matched_table.database.name
|
|
123
|
-
schema_name = matched_table.databaseSchema.name
|
|
124
|
-
|
|
125
|
-
return database_name, schema_name, table_name
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
def get_snowflake_system_queries(
|
|
129
|
-
row: Row,
|
|
130
|
-
database: str,
|
|
131
|
-
schema: str,
|
|
132
|
-
ometa_client: OpenMetadata,
|
|
133
|
-
db_service: DatabaseService,
|
|
134
|
-
) -> Optional[SnowflakeQueryResult]:
|
|
135
|
-
"""
|
|
136
|
-
Run a regex lookup on the query to identify which operation ran against the table.
|
|
137
|
-
|
|
138
|
-
If the query does not have the complete set of `database.schema.table` when it runs,
|
|
139
|
-
we'll use ES to pick up the table, if we find it.
|
|
140
|
-
|
|
141
|
-
Args:
|
|
142
|
-
row (dict): row from the snowflake system queries table
|
|
143
|
-
database (str): database name
|
|
144
|
-
schema (str): schema name
|
|
145
|
-
ometa_client (OpenMetadata): OpenMetadata client to search against ES
|
|
146
|
-
db_service (DatabaseService): DB service where the process is running against
|
|
147
|
-
Returns:
|
|
148
|
-
QueryResult: namedtuple with the query result
|
|
149
|
-
"""
|
|
150
|
-
|
|
151
|
-
try:
|
|
152
|
-
dict_row = dict(row)
|
|
153
|
-
query_text = dict_row.get("QUERY_TEXT", dict_row.get("query_text"))
|
|
154
|
-
logger.debug(f"Trying to parse query:\n{query_text}\n")
|
|
155
|
-
|
|
156
|
-
identifier = _parse_query(query_text)
|
|
157
|
-
if not identifier:
|
|
158
|
-
return None
|
|
159
|
-
|
|
160
|
-
database_name, schema_name, table_name = get_identifiers(
|
|
161
|
-
identifier=identifier,
|
|
162
|
-
ometa_client=ometa_client,
|
|
163
|
-
db_service=db_service,
|
|
164
|
-
)
|
|
165
|
-
|
|
166
|
-
if not all([database_name, schema_name, table_name]):
|
|
167
|
-
return None
|
|
168
|
-
|
|
169
|
-
if (
|
|
170
|
-
database.lower() == database_name.lower()
|
|
171
|
-
and schema.lower() == schema_name.lower()
|
|
172
|
-
):
|
|
173
|
-
return SnowflakeQueryResult(
|
|
174
|
-
query_id=dict_row.get("QUERY_ID", dict_row.get("query_id")),
|
|
175
|
-
database_name=database_name.lower(),
|
|
176
|
-
schema_name=schema_name.lower(),
|
|
177
|
-
table_name=table_name.lower(),
|
|
178
|
-
query_text=query_text,
|
|
179
|
-
query_type=dict_row.get("QUERY_TYPE", dict_row.get("query_type")),
|
|
180
|
-
timestamp=dict_row.get("START_TIME", dict_row.get("start_time")),
|
|
181
|
-
rows_inserted=dict_row.get(
|
|
182
|
-
"ROWS_INSERTED", dict_row.get("rows_inserted")
|
|
183
|
-
),
|
|
184
|
-
rows_updated=dict_row.get("ROWS_UPDATED", dict_row.get("rows_updated")),
|
|
185
|
-
rows_deleted=dict_row.get("ROWS_DELETED", dict_row.get("rows_deleted")),
|
|
186
|
-
)
|
|
187
|
-
except Exception:
|
|
188
|
-
logger.debug(traceback.format_exc())
|
|
189
|
-
return None
|
|
190
|
-
|
|
191
|
-
return None
|