acryl-datahub 1.0.0rc18__py3-none-any.whl → 1.3.0.1rc9__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 acryl-datahub might be problematic. Click here for more details.

Files changed (503) hide show
  1. {acryl_datahub-1.0.0rc18.dist-info → acryl_datahub-1.3.0.1rc9.dist-info}/METADATA +2686 -2563
  2. {acryl_datahub-1.0.0rc18.dist-info → acryl_datahub-1.3.0.1rc9.dist-info}/RECORD +499 -392
  3. {acryl_datahub-1.0.0rc18.dist-info → acryl_datahub-1.3.0.1rc9.dist-info}/WHEEL +1 -1
  4. {acryl_datahub-1.0.0rc18.dist-info → acryl_datahub-1.3.0.1rc9.dist-info}/entry_points.txt +7 -1
  5. datahub/_version.py +1 -1
  6. datahub/api/circuit_breaker/operation_circuit_breaker.py +2 -2
  7. datahub/api/entities/assertion/assertion.py +1 -1
  8. datahub/api/entities/common/serialized_value.py +1 -1
  9. datahub/api/entities/corpgroup/corpgroup.py +1 -1
  10. datahub/api/entities/datacontract/datacontract.py +35 -3
  11. datahub/api/entities/datajob/dataflow.py +18 -3
  12. datahub/api/entities/datajob/datajob.py +24 -4
  13. datahub/api/entities/dataprocess/dataprocess_instance.py +4 -0
  14. datahub/api/entities/dataproduct/dataproduct.py +32 -3
  15. datahub/api/entities/dataset/dataset.py +47 -72
  16. datahub/api/entities/external/__init__.py +0 -0
  17. datahub/api/entities/external/external_entities.py +724 -0
  18. datahub/api/entities/external/external_tag.py +147 -0
  19. datahub/api/entities/external/lake_formation_external_entites.py +162 -0
  20. datahub/api/entities/external/restricted_text.py +172 -0
  21. datahub/api/entities/external/unity_catalog_external_entites.py +172 -0
  22. datahub/api/entities/forms/forms.py +37 -37
  23. datahub/api/entities/structuredproperties/structuredproperties.py +6 -6
  24. datahub/api/graphql/assertion.py +1 -1
  25. datahub/api/graphql/base.py +8 -6
  26. datahub/api/graphql/operation.py +14 -10
  27. datahub/cli/check_cli.py +91 -9
  28. datahub/cli/cli_utils.py +63 -0
  29. datahub/cli/config_utils.py +20 -12
  30. datahub/cli/container_cli.py +5 -0
  31. datahub/cli/delete_cli.py +133 -34
  32. datahub/cli/docker_check.py +110 -14
  33. datahub/cli/docker_cli.py +155 -231
  34. datahub/cli/exists_cli.py +2 -3
  35. datahub/cli/get_cli.py +2 -3
  36. datahub/cli/graphql_cli.py +1422 -0
  37. datahub/cli/iceberg_cli.py +11 -5
  38. datahub/cli/ingest_cli.py +25 -26
  39. datahub/cli/migrate.py +12 -9
  40. datahub/cli/migration_utils.py +4 -3
  41. datahub/cli/put_cli.py +4 -6
  42. datahub/cli/quickstart_versioning.py +53 -10
  43. datahub/cli/specific/assertions_cli.py +39 -7
  44. datahub/cli/specific/datacontract_cli.py +57 -9
  45. datahub/cli/specific/dataproduct_cli.py +12 -24
  46. datahub/cli/specific/dataset_cli.py +31 -21
  47. datahub/cli/specific/forms_cli.py +2 -5
  48. datahub/cli/specific/group_cli.py +2 -3
  49. datahub/cli/specific/structuredproperties_cli.py +5 -7
  50. datahub/cli/specific/user_cli.py +174 -4
  51. datahub/cli/state_cli.py +2 -3
  52. datahub/cli/timeline_cli.py +2 -3
  53. datahub/configuration/common.py +46 -2
  54. datahub/configuration/connection_resolver.py +5 -2
  55. datahub/configuration/env_vars.py +331 -0
  56. datahub/configuration/import_resolver.py +7 -4
  57. datahub/configuration/kafka.py +21 -1
  58. datahub/configuration/pydantic_migration_helpers.py +6 -13
  59. datahub/configuration/source_common.py +4 -3
  60. datahub/configuration/validate_field_deprecation.py +5 -2
  61. datahub/configuration/validate_field_removal.py +8 -2
  62. datahub/configuration/validate_field_rename.py +6 -5
  63. datahub/configuration/validate_multiline_string.py +5 -2
  64. datahub/emitter/mce_builder.py +12 -8
  65. datahub/emitter/mcp.py +20 -5
  66. datahub/emitter/mcp_builder.py +12 -0
  67. datahub/emitter/request_helper.py +138 -15
  68. datahub/emitter/response_helper.py +111 -19
  69. datahub/emitter/rest_emitter.py +399 -163
  70. datahub/entrypoints.py +10 -5
  71. datahub/errors.py +12 -0
  72. datahub/ingestion/api/auto_work_units/auto_ensure_aspect_size.py +299 -2
  73. datahub/ingestion/api/auto_work_units/auto_validate_input_fields.py +87 -0
  74. datahub/ingestion/api/common.py +9 -0
  75. datahub/ingestion/api/decorators.py +15 -3
  76. datahub/ingestion/api/report.py +381 -3
  77. datahub/ingestion/api/sink.py +27 -2
  78. datahub/ingestion/api/source.py +174 -62
  79. datahub/ingestion/api/source_helpers.py +41 -3
  80. datahub/ingestion/api/source_protocols.py +23 -0
  81. datahub/ingestion/autogenerated/__init__.py +0 -0
  82. datahub/ingestion/autogenerated/capability_summary.json +3652 -0
  83. datahub/ingestion/autogenerated/lineage.json +402 -0
  84. datahub/ingestion/autogenerated/lineage_helper.py +177 -0
  85. datahub/ingestion/extractor/schema_util.py +31 -5
  86. datahub/ingestion/glossary/classification_mixin.py +9 -2
  87. datahub/ingestion/graph/client.py +492 -55
  88. datahub/ingestion/graph/config.py +18 -2
  89. datahub/ingestion/graph/filters.py +96 -32
  90. datahub/ingestion/graph/links.py +55 -0
  91. datahub/ingestion/reporting/datahub_ingestion_run_summary_provider.py +21 -11
  92. datahub/ingestion/run/pipeline.py +90 -23
  93. datahub/ingestion/run/pipeline_config.py +3 -3
  94. datahub/ingestion/sink/datahub_kafka.py +1 -0
  95. datahub/ingestion/sink/datahub_rest.py +31 -23
  96. datahub/ingestion/sink/file.py +1 -0
  97. datahub/ingestion/source/abs/config.py +1 -1
  98. datahub/ingestion/source/abs/datalake_profiler_config.py +1 -1
  99. datahub/ingestion/source/abs/source.py +15 -30
  100. datahub/ingestion/source/apply/datahub_apply.py +6 -5
  101. datahub/ingestion/source/aws/aws_common.py +185 -13
  102. datahub/ingestion/source/aws/glue.py +517 -244
  103. datahub/ingestion/source/aws/platform_resource_repository.py +30 -0
  104. datahub/ingestion/source/aws/s3_boto_utils.py +100 -5
  105. datahub/ingestion/source/aws/sagemaker_processors/feature_groups.py +1 -1
  106. datahub/ingestion/source/aws/sagemaker_processors/models.py +4 -4
  107. datahub/ingestion/source/aws/tag_entities.py +270 -0
  108. datahub/ingestion/source/azure/azure_common.py +3 -3
  109. datahub/ingestion/source/bigquery_v2/bigquery.py +51 -7
  110. datahub/ingestion/source/bigquery_v2/bigquery_config.py +51 -81
  111. datahub/ingestion/source/bigquery_v2/bigquery_connection.py +81 -0
  112. datahub/ingestion/source/bigquery_v2/bigquery_queries.py +6 -1
  113. datahub/ingestion/source/bigquery_v2/bigquery_report.py +0 -2
  114. datahub/ingestion/source/bigquery_v2/bigquery_schema.py +23 -16
  115. datahub/ingestion/source/bigquery_v2/bigquery_schema_gen.py +20 -5
  116. datahub/ingestion/source/bigquery_v2/common.py +1 -1
  117. datahub/ingestion/source/bigquery_v2/lineage.py +1 -1
  118. datahub/ingestion/source/bigquery_v2/profiler.py +4 -2
  119. datahub/ingestion/source/bigquery_v2/queries.py +3 -3
  120. datahub/ingestion/source/bigquery_v2/queries_extractor.py +45 -9
  121. datahub/ingestion/source/cassandra/cassandra.py +7 -18
  122. datahub/ingestion/source/cassandra/cassandra_api.py +36 -0
  123. datahub/ingestion/source/cassandra/cassandra_config.py +20 -0
  124. datahub/ingestion/source/cassandra/cassandra_profiling.py +26 -24
  125. datahub/ingestion/source/cassandra/cassandra_utils.py +1 -2
  126. datahub/ingestion/source/common/data_platforms.py +23 -0
  127. datahub/ingestion/source/common/gcp_credentials_config.py +9 -1
  128. datahub/ingestion/source/common/subtypes.py +73 -1
  129. datahub/ingestion/source/data_lake_common/data_lake_utils.py +59 -10
  130. datahub/ingestion/source/data_lake_common/object_store.py +732 -0
  131. datahub/ingestion/source/data_lake_common/path_spec.py +87 -38
  132. datahub/ingestion/source/datahub/config.py +19 -5
  133. datahub/ingestion/source/datahub/datahub_database_reader.py +205 -36
  134. datahub/ingestion/source/datahub/datahub_source.py +11 -1
  135. datahub/ingestion/source/dbt/dbt_cloud.py +17 -10
  136. datahub/ingestion/source/dbt/dbt_common.py +270 -26
  137. datahub/ingestion/source/dbt/dbt_core.py +88 -47
  138. datahub/ingestion/source/dbt/dbt_tests.py +8 -6
  139. datahub/ingestion/source/debug/__init__.py +0 -0
  140. datahub/ingestion/source/debug/datahub_debug.py +300 -0
  141. datahub/ingestion/source/delta_lake/config.py +9 -5
  142. datahub/ingestion/source/delta_lake/source.py +8 -0
  143. datahub/ingestion/source/dremio/dremio_api.py +114 -73
  144. datahub/ingestion/source/dremio/dremio_aspects.py +3 -2
  145. datahub/ingestion/source/dremio/dremio_config.py +5 -4
  146. datahub/ingestion/source/dremio/dremio_datahub_source_mapping.py +1 -1
  147. datahub/ingestion/source/dremio/dremio_entities.py +6 -5
  148. datahub/ingestion/source/dremio/dremio_reporting.py +22 -3
  149. datahub/ingestion/source/dremio/dremio_source.py +228 -215
  150. datahub/ingestion/source/dremio/dremio_sql_queries.py +82 -21
  151. datahub/ingestion/source/dynamodb/dynamodb.py +19 -13
  152. datahub/ingestion/source/excel/__init__.py +0 -0
  153. datahub/ingestion/source/excel/config.py +92 -0
  154. datahub/ingestion/source/excel/excel_file.py +539 -0
  155. datahub/ingestion/source/excel/profiling.py +308 -0
  156. datahub/ingestion/source/excel/report.py +49 -0
  157. datahub/ingestion/source/excel/source.py +662 -0
  158. datahub/ingestion/source/excel/util.py +18 -0
  159. datahub/ingestion/source/feast.py +12 -14
  160. datahub/ingestion/source/file.py +3 -0
  161. datahub/ingestion/source/fivetran/config.py +67 -8
  162. datahub/ingestion/source/fivetran/fivetran.py +228 -43
  163. datahub/ingestion/source/fivetran/fivetran_log_api.py +42 -9
  164. datahub/ingestion/source/fivetran/fivetran_query.py +58 -36
  165. datahub/ingestion/source/fivetran/fivetran_rest_api.py +65 -0
  166. datahub/ingestion/source/fivetran/response_models.py +97 -0
  167. datahub/ingestion/source/gc/datahub_gc.py +0 -2
  168. datahub/ingestion/source/gc/soft_deleted_entity_cleanup.py +101 -104
  169. datahub/ingestion/source/gcs/gcs_source.py +53 -10
  170. datahub/ingestion/source/gcs/gcs_utils.py +36 -9
  171. datahub/ingestion/source/ge_data_profiler.py +146 -33
  172. datahub/ingestion/source/ge_profiling_config.py +26 -11
  173. datahub/ingestion/source/grafana/entity_mcp_builder.py +272 -0
  174. datahub/ingestion/source/grafana/field_utils.py +307 -0
  175. datahub/ingestion/source/grafana/grafana_api.py +142 -0
  176. datahub/ingestion/source/grafana/grafana_config.py +104 -0
  177. datahub/ingestion/source/grafana/grafana_source.py +522 -84
  178. datahub/ingestion/source/grafana/lineage.py +202 -0
  179. datahub/ingestion/source/grafana/models.py +137 -0
  180. datahub/ingestion/source/grafana/report.py +90 -0
  181. datahub/ingestion/source/grafana/types.py +16 -0
  182. datahub/ingestion/source/hex/__init__.py +0 -0
  183. datahub/ingestion/source/hex/api.py +402 -0
  184. datahub/ingestion/source/hex/constants.py +8 -0
  185. datahub/ingestion/source/hex/hex.py +311 -0
  186. datahub/ingestion/source/hex/mapper.py +412 -0
  187. datahub/ingestion/source/hex/model.py +78 -0
  188. datahub/ingestion/source/hex/query_fetcher.py +307 -0
  189. datahub/ingestion/source/iceberg/iceberg.py +385 -164
  190. datahub/ingestion/source/iceberg/iceberg_common.py +2 -2
  191. datahub/ingestion/source/iceberg/iceberg_profiler.py +25 -20
  192. datahub/ingestion/source/identity/azure_ad.py +1 -1
  193. datahub/ingestion/source/identity/okta.py +1 -14
  194. datahub/ingestion/source/kafka/kafka.py +28 -71
  195. datahub/ingestion/source/kafka/kafka_config.py +78 -0
  196. datahub/ingestion/source/kafka_connect/common.py +2 -2
  197. datahub/ingestion/source/kafka_connect/sink_connectors.py +157 -48
  198. datahub/ingestion/source/kafka_connect/source_connectors.py +63 -5
  199. datahub/ingestion/source/ldap.py +1 -1
  200. datahub/ingestion/source/looker/looker_common.py +216 -86
  201. datahub/ingestion/source/looker/looker_config.py +15 -4
  202. datahub/ingestion/source/looker/looker_constant.py +4 -0
  203. datahub/ingestion/source/looker/looker_lib_wrapper.py +37 -4
  204. datahub/ingestion/source/looker/looker_liquid_tag.py +56 -5
  205. datahub/ingestion/source/looker/looker_source.py +539 -555
  206. datahub/ingestion/source/looker/looker_view_id_cache.py +1 -1
  207. datahub/ingestion/source/looker/lookml_concept_context.py +1 -1
  208. datahub/ingestion/source/looker/lookml_config.py +31 -3
  209. datahub/ingestion/source/looker/lookml_refinement.py +1 -1
  210. datahub/ingestion/source/looker/lookml_source.py +103 -118
  211. datahub/ingestion/source/looker/view_upstream.py +494 -1
  212. datahub/ingestion/source/metabase.py +32 -6
  213. datahub/ingestion/source/metadata/business_glossary.py +7 -7
  214. datahub/ingestion/source/metadata/lineage.py +11 -10
  215. datahub/ingestion/source/mlflow.py +254 -23
  216. datahub/ingestion/source/mock_data/__init__.py +0 -0
  217. datahub/ingestion/source/mock_data/datahub_mock_data.py +533 -0
  218. datahub/ingestion/source/mock_data/datahub_mock_data_report.py +12 -0
  219. datahub/ingestion/source/mock_data/table_naming_helper.py +97 -0
  220. datahub/ingestion/source/mode.py +359 -181
  221. datahub/ingestion/source/mongodb.py +11 -1
  222. datahub/ingestion/source/neo4j/neo4j_source.py +122 -153
  223. datahub/ingestion/source/nifi.py +5 -5
  224. datahub/ingestion/source/openapi.py +85 -38
  225. datahub/ingestion/source/openapi_parser.py +59 -40
  226. datahub/ingestion/source/powerbi/config.py +92 -27
  227. datahub/ingestion/source/powerbi/m_query/data_classes.py +3 -0
  228. datahub/ingestion/source/powerbi/m_query/odbc.py +185 -0
  229. datahub/ingestion/source/powerbi/m_query/parser.py +2 -2
  230. datahub/ingestion/source/powerbi/m_query/pattern_handler.py +358 -14
  231. datahub/ingestion/source/powerbi/m_query/resolver.py +10 -0
  232. datahub/ingestion/source/powerbi/powerbi.py +66 -32
  233. datahub/ingestion/source/powerbi/rest_api_wrapper/data_resolver.py +2 -2
  234. datahub/ingestion/source/powerbi/rest_api_wrapper/powerbi_api.py +11 -12
  235. datahub/ingestion/source/powerbi_report_server/report_server.py +0 -23
  236. datahub/ingestion/source/powerbi_report_server/report_server_domain.py +2 -4
  237. datahub/ingestion/source/preset.py +3 -3
  238. datahub/ingestion/source/qlik_sense/data_classes.py +28 -8
  239. datahub/ingestion/source/qlik_sense/qlik_sense.py +2 -1
  240. datahub/ingestion/source/redash.py +1 -1
  241. datahub/ingestion/source/redshift/config.py +15 -9
  242. datahub/ingestion/source/redshift/datashares.py +1 -1
  243. datahub/ingestion/source/redshift/lineage.py +386 -687
  244. datahub/ingestion/source/redshift/profile.py +2 -2
  245. datahub/ingestion/source/redshift/query.py +24 -20
  246. datahub/ingestion/source/redshift/redshift.py +52 -111
  247. datahub/ingestion/source/redshift/redshift_schema.py +17 -12
  248. datahub/ingestion/source/redshift/report.py +0 -2
  249. datahub/ingestion/source/redshift/usage.py +13 -11
  250. datahub/ingestion/source/s3/report.py +4 -2
  251. datahub/ingestion/source/s3/source.py +515 -244
  252. datahub/ingestion/source/sac/sac.py +3 -1
  253. datahub/ingestion/source/salesforce.py +28 -13
  254. datahub/ingestion/source/schema/json_schema.py +14 -14
  255. datahub/ingestion/source/schema_inference/object.py +22 -6
  256. datahub/ingestion/source/sigma/config.py +75 -8
  257. datahub/ingestion/source/sigma/data_classes.py +3 -0
  258. datahub/ingestion/source/sigma/sigma.py +36 -7
  259. datahub/ingestion/source/sigma/sigma_api.py +99 -58
  260. datahub/ingestion/source/slack/slack.py +403 -140
  261. datahub/ingestion/source/snaplogic/__init__.py +0 -0
  262. datahub/ingestion/source/snaplogic/snaplogic.py +355 -0
  263. datahub/ingestion/source/snaplogic/snaplogic_config.py +37 -0
  264. datahub/ingestion/source/snaplogic/snaplogic_lineage_extractor.py +107 -0
  265. datahub/ingestion/source/snaplogic/snaplogic_parser.py +168 -0
  266. datahub/ingestion/source/snaplogic/snaplogic_utils.py +31 -0
  267. datahub/ingestion/source/snowflake/constants.py +4 -0
  268. datahub/ingestion/source/snowflake/snowflake_config.py +103 -34
  269. datahub/ingestion/source/snowflake/snowflake_connection.py +47 -25
  270. datahub/ingestion/source/snowflake/snowflake_lineage_v2.py +25 -6
  271. datahub/ingestion/source/snowflake/snowflake_profiler.py +1 -6
  272. datahub/ingestion/source/snowflake/snowflake_queries.py +511 -107
  273. datahub/ingestion/source/snowflake/snowflake_query.py +100 -72
  274. datahub/ingestion/source/snowflake/snowflake_report.py +4 -2
  275. datahub/ingestion/source/snowflake/snowflake_schema.py +381 -16
  276. datahub/ingestion/source/snowflake/snowflake_schema_gen.py +163 -52
  277. datahub/ingestion/source/snowflake/snowflake_summary.py +7 -1
  278. datahub/ingestion/source/snowflake/snowflake_tag.py +4 -1
  279. datahub/ingestion/source/snowflake/snowflake_usage_v2.py +8 -2
  280. datahub/ingestion/source/snowflake/snowflake_utils.py +62 -17
  281. datahub/ingestion/source/snowflake/snowflake_v2.py +56 -10
  282. datahub/ingestion/source/snowflake/stored_proc_lineage.py +143 -0
  283. datahub/ingestion/source/sql/athena.py +219 -26
  284. datahub/ingestion/source/sql/athena_properties_extractor.py +795 -0
  285. datahub/ingestion/source/sql/clickhouse.py +29 -9
  286. datahub/ingestion/source/sql/cockroachdb.py +5 -4
  287. datahub/ingestion/source/sql/druid.py +9 -4
  288. datahub/ingestion/source/sql/hana.py +3 -1
  289. datahub/ingestion/source/sql/hive.py +28 -8
  290. datahub/ingestion/source/sql/hive_metastore.py +24 -25
  291. datahub/ingestion/source/sql/mariadb.py +0 -1
  292. datahub/ingestion/source/sql/mssql/job_models.py +18 -2
  293. datahub/ingestion/source/sql/mssql/source.py +376 -62
  294. datahub/ingestion/source/sql/mysql.py +154 -4
  295. datahub/ingestion/source/sql/oracle.py +62 -11
  296. datahub/ingestion/source/sql/postgres.py +142 -6
  297. datahub/ingestion/source/sql/presto.py +20 -2
  298. datahub/ingestion/source/sql/sql_common.py +281 -49
  299. datahub/ingestion/source/sql/sql_config.py +1 -34
  300. datahub/ingestion/source/sql/sql_generic_profiler.py +2 -1
  301. datahub/ingestion/source/sql/sql_types.py +27 -2
  302. datahub/ingestion/source/sql/sqlalchemy_uri.py +68 -0
  303. datahub/ingestion/source/sql/stored_procedures/__init__.py +0 -0
  304. datahub/ingestion/source/sql/stored_procedures/base.py +253 -0
  305. datahub/ingestion/source/sql/{mssql/stored_procedure_lineage.py → stored_procedures/lineage.py} +2 -29
  306. datahub/ingestion/source/sql/teradata.py +1028 -245
  307. datahub/ingestion/source/sql/trino.py +43 -10
  308. datahub/ingestion/source/sql/two_tier_sql_source.py +3 -4
  309. datahub/ingestion/source/sql/vertica.py +14 -7
  310. datahub/ingestion/source/sql_queries.py +219 -121
  311. datahub/ingestion/source/state/checkpoint.py +8 -29
  312. datahub/ingestion/source/state/entity_removal_state.py +5 -2
  313. datahub/ingestion/source/state/redundant_run_skip_handler.py +21 -0
  314. datahub/ingestion/source/state/stale_entity_removal_handler.py +0 -1
  315. datahub/ingestion/source/state/stateful_ingestion_base.py +36 -11
  316. datahub/ingestion/source/state_provider/datahub_ingestion_checkpointing_provider.py +2 -1
  317. datahub/ingestion/source/superset.py +810 -126
  318. datahub/ingestion/source/tableau/tableau.py +172 -69
  319. datahub/ingestion/source/tableau/tableau_common.py +11 -4
  320. datahub/ingestion/source/tableau/tableau_constant.py +1 -4
  321. datahub/ingestion/source/tableau/tableau_server_wrapper.py +3 -0
  322. datahub/ingestion/source/tableau/tableau_validation.py +1 -1
  323. datahub/ingestion/source/unity/config.py +161 -40
  324. datahub/ingestion/source/unity/connection.py +61 -0
  325. datahub/ingestion/source/unity/connection_test.py +1 -0
  326. datahub/ingestion/source/unity/platform_resource_repository.py +19 -0
  327. datahub/ingestion/source/unity/proxy.py +794 -51
  328. datahub/ingestion/source/unity/proxy_patch.py +321 -0
  329. datahub/ingestion/source/unity/proxy_types.py +36 -2
  330. datahub/ingestion/source/unity/report.py +15 -3
  331. datahub/ingestion/source/unity/source.py +465 -131
  332. datahub/ingestion/source/unity/tag_entities.py +197 -0
  333. datahub/ingestion/source/unity/usage.py +46 -4
  334. datahub/ingestion/source/usage/clickhouse_usage.py +11 -4
  335. datahub/ingestion/source/usage/starburst_trino_usage.py +10 -5
  336. datahub/ingestion/source/usage/usage_common.py +4 -68
  337. datahub/ingestion/source/vertexai/__init__.py +0 -0
  338. datahub/ingestion/source/vertexai/vertexai.py +1367 -0
  339. datahub/ingestion/source/vertexai/vertexai_config.py +29 -0
  340. datahub/ingestion/source/vertexai/vertexai_result_type_utils.py +89 -0
  341. datahub/ingestion/source_config/pulsar.py +3 -1
  342. datahub/ingestion/source_report/ingestion_stage.py +50 -11
  343. datahub/ingestion/transformer/add_dataset_dataproduct.py +1 -1
  344. datahub/ingestion/transformer/add_dataset_ownership.py +19 -3
  345. datahub/ingestion/transformer/base_transformer.py +8 -5
  346. datahub/ingestion/transformer/dataset_domain.py +1 -1
  347. datahub/ingestion/transformer/set_browse_path.py +112 -0
  348. datahub/integrations/assertion/common.py +3 -2
  349. datahub/integrations/assertion/snowflake/compiler.py +4 -3
  350. datahub/lite/lite_util.py +2 -2
  351. datahub/metadata/{_schema_classes.py → _internal_schema_classes.py} +3095 -631
  352. datahub/metadata/_urns/urn_defs.py +1866 -1582
  353. datahub/metadata/com/linkedin/pegasus2avro/application/__init__.py +19 -0
  354. datahub/metadata/com/linkedin/pegasus2avro/common/__init__.py +2 -0
  355. datahub/metadata/com/linkedin/pegasus2avro/dataplatform/slack/__init__.py +15 -0
  356. datahub/metadata/com/linkedin/pegasus2avro/event/__init__.py +11 -0
  357. datahub/metadata/com/linkedin/pegasus2avro/event/notification/__init__.py +15 -0
  358. datahub/metadata/com/linkedin/pegasus2avro/event/notification/settings/__init__.py +19 -0
  359. datahub/metadata/com/linkedin/pegasus2avro/file/__init__.py +19 -0
  360. datahub/metadata/com/linkedin/pegasus2avro/identity/__init__.py +2 -0
  361. datahub/metadata/com/linkedin/pegasus2avro/logical/__init__.py +15 -0
  362. datahub/metadata/com/linkedin/pegasus2avro/metadata/key/__init__.py +8 -0
  363. datahub/metadata/com/linkedin/pegasus2avro/module/__init__.py +31 -0
  364. datahub/metadata/com/linkedin/pegasus2avro/platform/event/v1/__init__.py +4 -0
  365. datahub/metadata/com/linkedin/pegasus2avro/role/__init__.py +2 -0
  366. datahub/metadata/com/linkedin/pegasus2avro/settings/asset/__init__.py +19 -0
  367. datahub/metadata/com/linkedin/pegasus2avro/settings/global/__init__.py +8 -0
  368. datahub/metadata/com/linkedin/pegasus2avro/template/__init__.py +31 -0
  369. datahub/metadata/schema.avsc +18404 -16617
  370. datahub/metadata/schema_classes.py +3 -3
  371. datahub/metadata/schemas/Actors.avsc +38 -1
  372. datahub/metadata/schemas/ApplicationKey.avsc +31 -0
  373. datahub/metadata/schemas/ApplicationProperties.avsc +72 -0
  374. datahub/metadata/schemas/Applications.avsc +38 -0
  375. datahub/metadata/schemas/AssetSettings.avsc +63 -0
  376. datahub/metadata/schemas/ChartInfo.avsc +2 -1
  377. datahub/metadata/schemas/ChartKey.avsc +1 -0
  378. datahub/metadata/schemas/ContainerKey.avsc +1 -0
  379. datahub/metadata/schemas/ContainerProperties.avsc +8 -0
  380. datahub/metadata/schemas/CorpUserEditableInfo.avsc +15 -1
  381. datahub/metadata/schemas/CorpUserKey.avsc +2 -1
  382. datahub/metadata/schemas/CorpUserSettings.avsc +145 -0
  383. datahub/metadata/schemas/DashboardKey.avsc +1 -0
  384. datahub/metadata/schemas/DataContractKey.avsc +2 -1
  385. datahub/metadata/schemas/DataFlowInfo.avsc +8 -0
  386. datahub/metadata/schemas/DataFlowKey.avsc +1 -0
  387. datahub/metadata/schemas/DataHubFileInfo.avsc +230 -0
  388. datahub/metadata/schemas/DataHubFileKey.avsc +21 -0
  389. datahub/metadata/schemas/DataHubIngestionSourceKey.avsc +2 -1
  390. datahub/metadata/schemas/DataHubOpenAPISchemaKey.avsc +22 -0
  391. datahub/metadata/schemas/DataHubPageModuleKey.avsc +21 -0
  392. datahub/metadata/schemas/DataHubPageModuleProperties.avsc +298 -0
  393. datahub/metadata/schemas/DataHubPageTemplateKey.avsc +21 -0
  394. datahub/metadata/schemas/DataHubPageTemplateProperties.avsc +251 -0
  395. datahub/metadata/schemas/DataHubPolicyInfo.avsc +12 -1
  396. datahub/metadata/schemas/DataJobInfo.avsc +8 -0
  397. datahub/metadata/schemas/DataJobInputOutput.avsc +8 -0
  398. datahub/metadata/schemas/DataJobKey.avsc +1 -0
  399. datahub/metadata/schemas/DataProcessInstanceInput.avsc +2 -1
  400. datahub/metadata/schemas/DataProcessInstanceOutput.avsc +2 -1
  401. datahub/metadata/schemas/DataProcessKey.avsc +8 -0
  402. datahub/metadata/schemas/DataProductKey.avsc +3 -1
  403. datahub/metadata/schemas/DataProductProperties.avsc +1 -1
  404. datahub/metadata/schemas/DataTransformLogic.avsc +4 -2
  405. datahub/metadata/schemas/DatasetKey.avsc +11 -1
  406. datahub/metadata/schemas/DatasetUsageStatistics.avsc +8 -0
  407. datahub/metadata/schemas/Deprecation.avsc +2 -0
  408. datahub/metadata/schemas/DomainKey.avsc +2 -1
  409. datahub/metadata/schemas/ExecutionRequestInput.avsc +5 -0
  410. datahub/metadata/schemas/FormInfo.avsc +5 -0
  411. datahub/metadata/schemas/GlobalSettingsInfo.avsc +134 -0
  412. datahub/metadata/schemas/GlossaryNodeKey.avsc +2 -1
  413. datahub/metadata/schemas/GlossaryTermKey.avsc +3 -1
  414. datahub/metadata/schemas/IcebergWarehouseInfo.avsc +8 -0
  415. datahub/metadata/schemas/IncidentInfo.avsc +3 -3
  416. datahub/metadata/schemas/InstitutionalMemory.avsc +31 -0
  417. datahub/metadata/schemas/LogicalParent.avsc +145 -0
  418. datahub/metadata/schemas/MLFeatureKey.avsc +1 -0
  419. datahub/metadata/schemas/MLFeatureTableKey.avsc +1 -0
  420. datahub/metadata/schemas/MLModelDeploymentKey.avsc +8 -0
  421. datahub/metadata/schemas/MLModelDeploymentProperties.avsc +3 -0
  422. datahub/metadata/schemas/MLModelGroupKey.avsc +11 -1
  423. datahub/metadata/schemas/MLModelGroupProperties.avsc +16 -0
  424. datahub/metadata/schemas/MLModelKey.avsc +9 -0
  425. datahub/metadata/schemas/MLPrimaryKeyKey.avsc +1 -0
  426. datahub/metadata/schemas/MetadataChangeEvent.avsc +189 -47
  427. datahub/metadata/schemas/MetadataChangeLog.avsc +65 -44
  428. datahub/metadata/schemas/MetadataChangeProposal.avsc +64 -0
  429. datahub/metadata/schemas/NotebookKey.avsc +1 -0
  430. datahub/metadata/schemas/Operation.avsc +21 -2
  431. datahub/metadata/schemas/Ownership.avsc +69 -0
  432. datahub/metadata/schemas/QueryProperties.avsc +24 -2
  433. datahub/metadata/schemas/QuerySubjects.avsc +1 -12
  434. datahub/metadata/schemas/RelationshipChangeEvent.avsc +215 -0
  435. datahub/metadata/schemas/SchemaFieldKey.avsc +4 -1
  436. datahub/metadata/schemas/Siblings.avsc +2 -0
  437. datahub/metadata/schemas/SlackUserInfo.avsc +160 -0
  438. datahub/metadata/schemas/StructuredProperties.avsc +69 -0
  439. datahub/metadata/schemas/StructuredPropertySettings.avsc +9 -0
  440. datahub/metadata/schemas/SystemMetadata.avsc +147 -0
  441. datahub/metadata/schemas/UpstreamLineage.avsc +9 -0
  442. datahub/metadata/schemas/__init__.py +3 -3
  443. datahub/sdk/__init__.py +7 -0
  444. datahub/sdk/_all_entities.py +15 -0
  445. datahub/sdk/_shared.py +393 -10
  446. datahub/sdk/_utils.py +4 -0
  447. datahub/sdk/chart.py +386 -0
  448. datahub/sdk/container.py +7 -0
  449. datahub/sdk/dashboard.py +453 -0
  450. datahub/sdk/dataflow.py +309 -0
  451. datahub/sdk/datajob.py +367 -0
  452. datahub/sdk/dataset.py +180 -4
  453. datahub/sdk/entity.py +99 -3
  454. datahub/sdk/entity_client.py +154 -12
  455. datahub/sdk/lineage_client.py +943 -0
  456. datahub/sdk/main_client.py +83 -8
  457. datahub/sdk/mlmodel.py +383 -0
  458. datahub/sdk/mlmodelgroup.py +240 -0
  459. datahub/sdk/search_client.py +85 -8
  460. datahub/sdk/search_filters.py +393 -68
  461. datahub/secret/datahub_secret_store.py +5 -1
  462. datahub/secret/environment_secret_store.py +29 -0
  463. datahub/secret/file_secret_store.py +49 -0
  464. datahub/specific/aspect_helpers/fine_grained_lineage.py +76 -0
  465. datahub/specific/aspect_helpers/siblings.py +73 -0
  466. datahub/specific/aspect_helpers/structured_properties.py +27 -0
  467. datahub/specific/chart.py +1 -1
  468. datahub/specific/datajob.py +15 -1
  469. datahub/specific/dataproduct.py +4 -0
  470. datahub/specific/dataset.py +51 -59
  471. datahub/sql_parsing/_sqlglot_patch.py +1 -2
  472. datahub/sql_parsing/fingerprint_utils.py +6 -0
  473. datahub/sql_parsing/split_statements.py +30 -3
  474. datahub/sql_parsing/sql_parsing_aggregator.py +144 -63
  475. datahub/sql_parsing/sqlglot_lineage.py +517 -44
  476. datahub/sql_parsing/sqlglot_utils.py +30 -18
  477. datahub/sql_parsing/tool_meta_extractor.py +25 -2
  478. datahub/telemetry/telemetry.py +30 -16
  479. datahub/testing/check_imports.py +1 -1
  480. datahub/testing/docker_utils.py +8 -2
  481. datahub/testing/mce_helpers.py +421 -0
  482. datahub/testing/mcp_diff.py +17 -21
  483. datahub/testing/sdk_v2_helpers.py +18 -0
  484. datahub/upgrade/upgrade.py +86 -30
  485. datahub/utilities/file_backed_collections.py +14 -15
  486. datahub/utilities/hive_schema_to_avro.py +2 -2
  487. datahub/utilities/ingest_utils.py +2 -2
  488. datahub/utilities/is_pytest.py +3 -2
  489. datahub/utilities/logging_manager.py +30 -7
  490. datahub/utilities/mapping.py +29 -2
  491. datahub/utilities/sample_data.py +5 -4
  492. datahub/utilities/server_config_util.py +298 -10
  493. datahub/utilities/sqlalchemy_query_combiner.py +6 -4
  494. datahub/utilities/stats_collections.py +4 -0
  495. datahub/utilities/threaded_iterator_executor.py +16 -3
  496. datahub/utilities/urn_encoder.py +1 -1
  497. datahub/utilities/urns/urn.py +41 -2
  498. datahub/emitter/sql_parsing_builder.py +0 -306
  499. datahub/ingestion/source/redshift/lineage_v2.py +0 -458
  500. datahub/ingestion/source/vertexai.py +0 -697
  501. datahub/ingestion/transformer/system_metadata_transformer.py +0 -45
  502. {acryl_datahub-1.0.0rc18.dist-info → acryl_datahub-1.3.0.1rc9.dist-info/licenses}/LICENSE +0 -0
  503. {acryl_datahub-1.0.0rc18.dist-info → acryl_datahub-1.3.0.1rc9.dist-info}/top_level.txt +0 -0
@@ -0,0 +1,412 @@
1
+ import logging
2
+ from datetime import datetime
3
+ from typing import Iterable, List, Optional, Tuple, Union
4
+
5
+ from datahub._codegen.aspect import (
6
+ _Aspect, # TODO: is there a better import than this one?
7
+ )
8
+ from datahub.emitter.mce_builder import (
9
+ make_container_urn,
10
+ make_dashboard_urn,
11
+ make_data_platform_urn,
12
+ make_dataplatform_instance_urn,
13
+ make_tag_urn,
14
+ make_ts_millis,
15
+ make_user_urn,
16
+ )
17
+ from datahub.emitter.mcp import MetadataChangeProposalWrapper
18
+ from datahub.emitter.mcp_builder import ContainerKey
19
+ from datahub.ingestion.api.incremental_lineage_helper import (
20
+ convert_dashboard_info_to_patch,
21
+ )
22
+ from datahub.ingestion.api.workunit import MetadataWorkUnit
23
+ from datahub.ingestion.source.common.subtypes import BIAssetSubTypes
24
+ from datahub.ingestion.source.hex.constants import (
25
+ HEX_API_BASE_URL_DEFAULT,
26
+ HEX_PLATFORM_NAME,
27
+ )
28
+ from datahub.ingestion.source.hex.model import (
29
+ Analytics,
30
+ Category,
31
+ Collection,
32
+ Component,
33
+ Owner,
34
+ Project,
35
+ Status,
36
+ )
37
+ from datahub.metadata.com.linkedin.pegasus2avro.common import (
38
+ AuditStampClass,
39
+ ChangeAuditStampsClass,
40
+ OwnershipType,
41
+ )
42
+ from datahub.metadata.schema_classes import (
43
+ CalendarIntervalClass,
44
+ ContainerClass,
45
+ ContainerPropertiesClass,
46
+ DashboardInfoClass,
47
+ DashboardUsageStatisticsClass,
48
+ DataPlatformInstanceClass,
49
+ EdgeClass,
50
+ GlobalTagsClass,
51
+ OwnerClass,
52
+ OwnershipClass,
53
+ SubTypesClass,
54
+ TagAssociationClass,
55
+ TimeWindowSizeClass,
56
+ )
57
+ from datahub.metadata.urns import (
58
+ ContainerUrn,
59
+ CorpUserUrn,
60
+ DashboardUrn,
61
+ DatasetUrn,
62
+ SchemaFieldUrn,
63
+ Urn,
64
+ )
65
+
66
+ logger = logging.getLogger(__name__)
67
+
68
+
69
+ class WorkspaceKey(ContainerKey):
70
+ workspace_name: str
71
+
72
+
73
+ DEFAULT_INGESTION_USER_URN = CorpUserUrn("_ingestion")
74
+ DEFAULT_OWNERSHIP_TYPE = OwnershipType.TECHNICAL_OWNER
75
+
76
+
77
+ class Mapper:
78
+ def __init__(
79
+ self,
80
+ workspace_name: str,
81
+ platform_instance: Optional[str] = None,
82
+ env: Optional[str] = None,
83
+ base_url: str = HEX_API_BASE_URL_DEFAULT,
84
+ patch_metadata: bool = True,
85
+ collections_as_tags: bool = True,
86
+ status_as_tag: bool = True,
87
+ categories_as_tags: bool = True,
88
+ set_ownership_from_email: bool = True,
89
+ ):
90
+ self._workspace_name = workspace_name
91
+ self._env = env
92
+ self._platform_instance = platform_instance
93
+ self._workspace_urn = Mapper._get_workspace_urn(
94
+ workspace_name=workspace_name,
95
+ platform=HEX_PLATFORM_NAME,
96
+ env=env,
97
+ platform_instance=platform_instance,
98
+ )
99
+ self._base_url = base_url.strip("/").replace("/api/v1", "")
100
+ self._patch_metadata = patch_metadata
101
+ self._collections_as_tags = collections_as_tags
102
+ self._status_as_tag = status_as_tag
103
+ self._categories_as_tags = categories_as_tags
104
+ self._set_ownership_from_email = set_ownership_from_email
105
+
106
+ def map_workspace(self) -> Iterable[MetadataWorkUnit]:
107
+ container_properties = ContainerPropertiesClass(
108
+ name=self._workspace_name,
109
+ env=self._env,
110
+ )
111
+ yield from self._yield_mcps(
112
+ entity_urn=self._workspace_urn,
113
+ aspects=[container_properties],
114
+ )
115
+
116
+ def map_project(self, project: Project) -> Iterable[MetadataWorkUnit]:
117
+ dashboard_urn = self._get_dashboard_urn(name=project.id)
118
+
119
+ dashboard_info = DashboardInfoClass(
120
+ title=project.title,
121
+ description=project.description or "",
122
+ lastModified=self._change_audit_stamps(
123
+ created_at=project.created_at, last_edited_at=project.last_edited_at
124
+ ),
125
+ externalUrl=self._get_project_or_component_external_url(project),
126
+ customProperties=dict(id=project.id),
127
+ datasetEdges=self._dataset_edges(project.upstream_datasets),
128
+ # TODO: support schema field upstream, maybe InputFields?
129
+ )
130
+
131
+ subtypes = SubTypesClass(
132
+ typeNames=[BIAssetSubTypes.HEX_PROJECT],
133
+ )
134
+
135
+ platform_instance = self._platform_instance_aspect()
136
+
137
+ container = ContainerClass(
138
+ container=self._workspace_urn.urn(),
139
+ )
140
+
141
+ tags = self._global_tags(
142
+ status=project.status,
143
+ categories=project.categories,
144
+ collections=project.collections,
145
+ )
146
+
147
+ ownership = self._ownership(creator=project.creator, owner=project.owner)
148
+
149
+ usage_stats_all_time, usage_stats_last_7_days = (
150
+ self._dashboard_usage_statistics(analytics=project.analytics)
151
+ )
152
+
153
+ yield from self._yield_mcps(
154
+ entity_urn=dashboard_urn,
155
+ aspects=[
156
+ dashboard_info,
157
+ subtypes,
158
+ platform_instance,
159
+ container,
160
+ tags,
161
+ ownership,
162
+ usage_stats_all_time,
163
+ usage_stats_last_7_days,
164
+ ],
165
+ )
166
+
167
+ def map_component(self, component: Component) -> Iterable[MetadataWorkUnit]:
168
+ dashboard_urn = self._get_dashboard_urn(name=component.id)
169
+
170
+ dashboard_info = DashboardInfoClass(
171
+ title=component.title,
172
+ description=component.description or "",
173
+ lastModified=self._change_audit_stamps(
174
+ created_at=component.created_at, last_edited_at=component.last_edited_at
175
+ ),
176
+ externalUrl=self._get_project_or_component_external_url(component),
177
+ customProperties=dict(id=component.id),
178
+ )
179
+
180
+ subtypes = SubTypesClass(
181
+ typeNames=[BIAssetSubTypes.HEX_COMPONENT],
182
+ )
183
+
184
+ platform_instance = self._platform_instance_aspect()
185
+
186
+ container = ContainerClass(
187
+ container=self._workspace_urn.urn(),
188
+ )
189
+
190
+ tags = self._global_tags(
191
+ status=component.status,
192
+ categories=component.categories,
193
+ collections=component.collections,
194
+ )
195
+
196
+ ownership = self._ownership(creator=component.creator, owner=component.owner)
197
+
198
+ usage_stats_all_time, usage_stats_last_7_days = (
199
+ self._dashboard_usage_statistics(analytics=component.analytics)
200
+ )
201
+
202
+ yield from self._yield_mcps(
203
+ entity_urn=dashboard_urn,
204
+ aspects=[
205
+ dashboard_info,
206
+ subtypes,
207
+ platform_instance,
208
+ container,
209
+ tags,
210
+ ownership,
211
+ usage_stats_all_time,
212
+ usage_stats_last_7_days,
213
+ ],
214
+ )
215
+
216
+ @classmethod
217
+ def _get_workspace_urn(
218
+ cls,
219
+ workspace_name: str,
220
+ platform: str = HEX_PLATFORM_NAME,
221
+ env: Optional[str] = None,
222
+ platform_instance: Optional[str] = None,
223
+ ) -> ContainerUrn:
224
+ workspace_key = WorkspaceKey(
225
+ platform=platform,
226
+ env=env,
227
+ platform_instance=platform_instance,
228
+ workspace_name=workspace_name,
229
+ )
230
+ container_urn_str = make_container_urn(guid=workspace_key)
231
+ container_urn = Urn.from_string(container_urn_str)
232
+ assert isinstance(container_urn, ContainerUrn)
233
+ return container_urn
234
+
235
+ def _get_dashboard_urn(self, name: str) -> DashboardUrn:
236
+ dashboard_urn_str = make_dashboard_urn(
237
+ platform=HEX_PLATFORM_NAME,
238
+ name=name,
239
+ platform_instance=self._platform_instance,
240
+ )
241
+ dashboard_urn = Urn.from_string(dashboard_urn_str)
242
+ assert isinstance(dashboard_urn, DashboardUrn)
243
+ return dashboard_urn
244
+
245
+ def _get_project_or_component_external_url(
246
+ self,
247
+ project_or_component: Union[Project, Component],
248
+ ) -> Optional[str]:
249
+ if project_or_component.last_published_at is None:
250
+ return (
251
+ f"{self._base_url}/{self._workspace_name}/hex/{project_or_component.id}"
252
+ )
253
+ else:
254
+ # published Projects/Components have a different URL that everybody, not just editors, can access
255
+ return (
256
+ f"{self._base_url}/{self._workspace_name}/app/{project_or_component.id}"
257
+ )
258
+
259
+ def _change_audit_stamps(
260
+ self, created_at: Optional[datetime], last_edited_at: Optional[datetime]
261
+ ) -> ChangeAuditStampsClass:
262
+ return ChangeAuditStampsClass(
263
+ created=AuditStampClass(
264
+ time=make_ts_millis(created_at),
265
+ actor=DEFAULT_INGESTION_USER_URN.urn(),
266
+ )
267
+ if created_at
268
+ else None,
269
+ lastModified=AuditStampClass(
270
+ time=make_ts_millis(last_edited_at),
271
+ actor=DEFAULT_INGESTION_USER_URN.urn(),
272
+ )
273
+ if last_edited_at
274
+ else None,
275
+ )
276
+
277
+ def _global_tags(
278
+ self,
279
+ status: Optional[Status],
280
+ categories: Optional[List[Category]],
281
+ collections: Optional[List[Collection]],
282
+ ) -> Optional[GlobalTagsClass]:
283
+ tag_associations: List[TagAssociationClass] = []
284
+ if status and self._status_as_tag:
285
+ tag_associations.append(
286
+ TagAssociationClass(tag=make_tag_urn(tag=f"hex:status:{status.name}"))
287
+ )
288
+ if categories and self._categories_as_tags:
289
+ tag_associations.extend(
290
+ [
291
+ TagAssociationClass(
292
+ tag=make_tag_urn(tag=f"hex:category:{cat.name}")
293
+ )
294
+ for cat in categories
295
+ ]
296
+ )
297
+ if collections and self._collections_as_tags:
298
+ tag_associations.extend(
299
+ [
300
+ TagAssociationClass(
301
+ tag=make_tag_urn(tag=f"hex:collection:{col.name}")
302
+ )
303
+ for col in collections
304
+ ]
305
+ )
306
+
307
+ return GlobalTagsClass(tags=tag_associations) if tag_associations else None
308
+
309
+ def _ownership(
310
+ self, creator: Optional[Owner], owner: Optional[Owner]
311
+ ) -> Optional[OwnershipClass]:
312
+ if self._set_ownership_from_email:
313
+ # since we are not making any diff of creator/owner, we usually have duplicates
314
+ # TODO: set or ownership types to properly differentiate them, maybe by config?
315
+ unique_owners = set(o for o in [creator, owner] if o)
316
+ owners: List[OwnerClass] = [
317
+ OwnerClass(owner=make_user_urn(o.email), type=DEFAULT_OWNERSHIP_TYPE)
318
+ for o in unique_owners
319
+ ]
320
+
321
+ return OwnershipClass(owners=owners) if owners else None
322
+ return None
323
+
324
+ def _dashboard_usage_statistics(
325
+ self, analytics: Optional[Analytics]
326
+ ) -> Tuple[
327
+ Optional[DashboardUsageStatisticsClass], Optional[DashboardUsageStatisticsClass]
328
+ ]:
329
+ tm_millis = make_ts_millis(datetime.now())
330
+ last_viewed_at = (
331
+ make_ts_millis(analytics.last_viewed_at)
332
+ if analytics and analytics.last_viewed_at
333
+ else None
334
+ )
335
+
336
+ usage_all_time: Optional[DashboardUsageStatisticsClass] = (
337
+ DashboardUsageStatisticsClass(
338
+ timestampMillis=tm_millis,
339
+ viewsCount=analytics.appviews_all_time,
340
+ lastViewedAt=last_viewed_at,
341
+ )
342
+ if analytics and analytics.appviews_all_time
343
+ else None
344
+ )
345
+
346
+ usage_last_7_days: Optional[DashboardUsageStatisticsClass] = (
347
+ DashboardUsageStatisticsClass(
348
+ timestampMillis=tm_millis,
349
+ viewsCount=analytics.appviews_last_7_days,
350
+ eventGranularity=TimeWindowSizeClass(
351
+ unit=CalendarIntervalClass.WEEK, multiple=1
352
+ ),
353
+ lastViewedAt=last_viewed_at,
354
+ )
355
+ if analytics and analytics.appviews_last_7_days
356
+ else None
357
+ )
358
+ return (usage_all_time, usage_last_7_days)
359
+
360
+ def _platform_instance_aspect(self) -> DataPlatformInstanceClass:
361
+ return DataPlatformInstanceClass(
362
+ platform=make_data_platform_urn(HEX_PLATFORM_NAME),
363
+ instance=make_dataplatform_instance_urn(
364
+ platform=HEX_PLATFORM_NAME, instance=self._platform_instance
365
+ )
366
+ if self._platform_instance
367
+ else None,
368
+ )
369
+
370
+ def _dataset_edges(
371
+ self, upstream: List[Union[DatasetUrn, SchemaFieldUrn]]
372
+ ) -> Optional[List[EdgeClass]]:
373
+ # TBC: is there support for CLL in Dashboards? for the moment, skip SchemaFieldUrns
374
+ return (
375
+ [
376
+ EdgeClass(
377
+ destinationUrn=upstream_urn.urn(),
378
+ )
379
+ for upstream_urn in upstream
380
+ if isinstance(upstream_urn, DatasetUrn)
381
+ ]
382
+ if upstream
383
+ else None
384
+ )
385
+
386
+ def _yield_mcps(
387
+ self, entity_urn: Urn, aspects: List[Optional[_Aspect]]
388
+ ) -> Iterable[MetadataWorkUnit]:
389
+ for mcpw in MetadataChangeProposalWrapper.construct_many(
390
+ entityUrn=entity_urn.urn(),
391
+ aspects=aspects,
392
+ ):
393
+ wu = MetadataWorkUnit.from_metadata(metadata=mcpw)
394
+ maybe_wu = self._maybe_patch_wu(wu)
395
+ if maybe_wu:
396
+ yield maybe_wu
397
+
398
+ def _maybe_patch_wu(self, wu: MetadataWorkUnit) -> Optional[MetadataWorkUnit]:
399
+ # So far we only have support for DashboardInfo aspect
400
+
401
+ dashboard_info_aspect: Optional[DashboardInfoClass] = wu.get_aspect_of_type(
402
+ DashboardInfoClass
403
+ )
404
+
405
+ if dashboard_info_aspect and self._patch_metadata:
406
+ return convert_dashboard_info_to_patch(
407
+ wu.get_urn(),
408
+ dashboard_info_aspect,
409
+ wu.metadata.systemMetadata,
410
+ )
411
+ else:
412
+ return wu
@@ -0,0 +1,78 @@
1
+ from dataclasses import dataclass, field
2
+ from datetime import datetime
3
+ from typing import List, Optional, Union
4
+
5
+ from datahub.metadata.urns import DatasetUrn, SchemaFieldUrn
6
+
7
+
8
+ @dataclass
9
+ class Workspace:
10
+ name: str
11
+
12
+
13
+ @dataclass
14
+ class Status:
15
+ name: str
16
+
17
+
18
+ @dataclass
19
+ class Category:
20
+ name: str
21
+ description: Optional[str] = None
22
+
23
+
24
+ @dataclass
25
+ class Collection:
26
+ name: str
27
+
28
+
29
+ @dataclass(frozen=True)
30
+ class Owner:
31
+ email: str
32
+
33
+
34
+ @dataclass
35
+ class Analytics:
36
+ appviews_all_time: Optional[int]
37
+ appviews_last_7_days: Optional[int]
38
+ appviews_last_14_days: Optional[int]
39
+ appviews_last_30_days: Optional[int]
40
+ last_viewed_at: Optional[datetime]
41
+
42
+
43
+ @dataclass
44
+ class Project:
45
+ id: str
46
+ title: str
47
+ description: Optional[str]
48
+ last_edited_at: Optional[datetime] = None
49
+ last_published_at: Optional[datetime] = None
50
+ created_at: Optional[datetime] = None
51
+ status: Optional[Status] = None
52
+ categories: Optional[List[Category]] = None # TODO: emit category description!
53
+ collections: Optional[List[Collection]] = None
54
+ creator: Optional[Owner] = None
55
+ owner: Optional[Owner] = None
56
+ analytics: Optional[Analytics] = None
57
+ upstream_datasets: List[Union[DatasetUrn, SchemaFieldUrn]] = field(
58
+ default_factory=list
59
+ )
60
+ upstream_schema_fields: List[Union[DatasetUrn, SchemaFieldUrn]] = field(
61
+ default_factory=list
62
+ )
63
+
64
+
65
+ @dataclass
66
+ class Component:
67
+ id: str
68
+ title: str
69
+ description: Optional[str]
70
+ last_edited_at: Optional[datetime] = None
71
+ last_published_at: Optional[datetime] = None
72
+ created_at: Optional[datetime] = None
73
+ status: Optional[Status] = None
74
+ categories: Optional[List[Category]] = None
75
+ collections: Optional[List[Collection]] = None
76
+ creator: Optional[Owner] = None
77
+ owner: Optional[Owner] = None
78
+ analytics: Optional[Analytics] = None