acryl-datahub 1.1.0.5rc1__py3-none-any.whl → 1.1.0.5rc3__py3-none-any.whl

{acryl_datahub-1.1.0.5rc1.dist-info → acryl_datahub-1.1.0.5rc3.dist-info}/RECORD

@@ -1,7 +1,7 @@
-acryl_datahub-1.1.0.5rc1.dist-info/licenses/LICENSE,sha256=9xNHpsD0uYF5ONzXsKDCuHHB-xbiCrSbueWXqrTNsxk,11365
+acryl_datahub-1.1.0.5rc3.dist-info/licenses/LICENSE,sha256=9xNHpsD0uYF5ONzXsKDCuHHB-xbiCrSbueWXqrTNsxk,11365
 datahub/__init__.py,sha256=aq_i5lVREmoLfYIqcx_pEQicO855YlhD19tWc1eZZNI,59
 datahub/__main__.py,sha256=pegIvQ9hzK7IhqVeUi1MeADSZ2QlP-D3K0OQdEg55RU,106
-datahub/_version.py,sha256=TZdOK9cFifb1yYky21qrv8R4D8Q8htQJWiKxMGKJpS0,323
+datahub/_version.py,sha256=81T1DSxdYWctKOc3Yt4DKvZkQAVEO5Rw4fHuqucBno4,323
 datahub/entrypoints.py,sha256=H-YFTvxTJOgpWsFBVlxyb1opjkq-hjTzNmjy5Fq3RHg,8992
 datahub/errors.py,sha256=p5rFAdAGVCk4Lqolol1YvthceadUSwpaCxLXRcyCCFQ,676
 datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -48,7 +48,7 @@ datahub/api/entities/dataprocess/dataprocess_instance.py,sha256=IhY-rcXs-r8EatwW
 datahub/api/entities/dataproduct/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/api/entities/dataproduct/dataproduct.py,sha256=148TmItxDDyGNzfZdL8aDreSEtyAw79IN8N8oSmNOPE,21461
 datahub/api/entities/dataset/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/api/entities/dataset/dataset.py,sha256=BLOn7o-A3OLD4D2JTq8vF_dwTTW9HHDnDwd6lhI1TMI,49512
+datahub/api/entities/dataset/dataset.py,sha256=ed3-0moqef0Zgs_rwYxbskTa5BqmM2fH2_Z8U12wVRs,49526
 datahub/api/entities/external/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/api/entities/external/external_entities.py,sha256=7HRt3Oz2sw_7BDxEVc6W7jK4UqHXLZuo_rGPJ5NdAoE,7727
 datahub/api/entities/external/external_tag.py,sha256=LKz4j2KnBvo8xhGLxA9g1DWtHfHUA7WqO2vOIsqwy44,4490
@@ -141,7 +141,7 @@ datahub/ingestion/api/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3h
 datahub/ingestion/api/closeable.py,sha256=k12AT--s4GDtZ-po_rVm5QKgvGIDteeRPByZPIOfecA,599
 datahub/ingestion/api/committable.py,sha256=4S6GuBzvX2vb1A8P506NbspOKfZ1621sBG8t0lvRb8o,886
 datahub/ingestion/api/common.py,sha256=26lXJiM4YfdnVH1xfe2bpZNp2VKCdJcJ8ynK7rhh0FY,3029
-datahub/ingestion/api/decorators.py,sha256=-0Kl9G_8DhmSP9Q0ryrWB3a-Odce8WTpvGETJOqRdgw,4015
+datahub/ingestion/api/decorators.py,sha256=qsI2HkyruhVR7VU-K0sCZmpAR9kg49CDdL9Tu24QpsA,4016
 datahub/ingestion/api/global_context.py,sha256=OdSJg4a_RKE52nu8MSiEkK2UqRRDhDTyOleHEAzPKho,575
 datahub/ingestion/api/incremental_lineage_helper.py,sha256=7a6FTJ_uz4EEJS1vPtbYB2KvNlcZB3py28_FKxmRiSk,5993
 datahub/ingestion/api/incremental_properties_helper.py,sha256=KzdxdrQtaMV2XMHfPsCtRa7ffDGPA1w1hgPUjeenZBU,2514
@@ -158,6 +158,7 @@ datahub/ingestion/api/workunit.py,sha256=e8n8RfSjHZZm2R4ShNH0UuMtUkMjyqqM2j2t7oL
 datahub/ingestion/api/auto_work_units/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/api/auto_work_units/auto_dataset_properties_aspect.py,sha256=ID_6N3nWl2qohsSGizUCqo3d2MNyDeVbyWroQpSOSsc,5059
 datahub/ingestion/api/auto_work_units/auto_ensure_aspect_size.py,sha256=fMjPnyWEofIZV52E2AFYU3IgBJwyZvbygXxCJyEtcWI,4442
+datahub/ingestion/autogenerated/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/extractor/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/extractor/extractor_registry.py,sha256=f7CLfW3pr29QZkXSHbp7HjUrsdw7ejQJmot-tiSPcqc,342
 datahub/ingestion/extractor/json_ref_patch.py,sha256=4g3ZWHn7rwS74jUvSXJiGpi-UKHhiSYKKgBeU4E5ukE,1448
@@ -218,13 +219,13 @@ datahub/ingestion/source/mongodb.py,sha256=2C2Cxn8DXL53IbNiywIuKt8UT_EMcPg9f8su-
 datahub/ingestion/source/nifi.py,sha256=2jxSzuHIRQFF7DLoceGbUd_10lkhHRlqA9hhApSt3Yw,56882
 datahub/ingestion/source/openapi.py,sha256=VaR2xYaH1IhvRixpTBC7-168F74eIIyKiEKb5EqTO64,19253
 datahub/ingestion/source/openapi_parser.py,sha256=T87e2r-oPGgQl_FDMHnSGFZzApvWDCyKWnzIrVI5Alo,15420
-datahub/ingestion/source/preset.py,sha256=1goxuFoLw50dokr2gp1MhUrDJ8CFNX-wBZIvv7laEXA,3966
+datahub/ingestion/source/preset.py,sha256=fncn-fgYcITsYEHVsvV6cGTQ9_xc_R06ejrw6ZbY3QA,3966
 datahub/ingestion/source/pulsar.py,sha256=u5F8QnCLJsht5-7XCiUTsnfhCPIpKVB_l32CgMCU-As,20187
 datahub/ingestion/source/redash.py,sha256=YxjSad-X_wPmxYH8dJmFz_VCFhiLTCTSlK99WdvcYiA,30653
 datahub/ingestion/source/salesforce.py,sha256=CQtDFv1OsbC1vyzNbKOc6GxhFQ5GdYj45hgAF0-oIcw,40487
 datahub/ingestion/source/source_registry.py,sha256=a2mLjJPLkSI-gYCTb_7U7Jo4D8jGknNQ_yScPIihXFk,1208
 datahub/ingestion/source/sql_queries.py,sha256=Ip7UZub7fgMh7P5jL_zJPY7lSkc9GGTy8GJ8lqZrcsE,9502
-datahub/ingestion/source/superset.py,sha256=dSXbsPj4_BY9O6esrJRt5WYcHj7QWoBk7PTfIFxS_Zw,48387
+datahub/ingestion/source/superset.py,sha256=4wEjhBj_Zf__1EhXDE9ltbaR9NTMZVdXFviHLf5VFL4,48387
 datahub/ingestion/source/abs/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/abs/config.py,sha256=mBQe0JTaP-Rcv4HnMUUySoYbSr4r3jDEMioxaXHnxXU,6709
 datahub/ingestion/source/abs/datalake_profiler_config.py,sha256=Rkf64evufyVGPiE4VK8QAjzBiJFu85tOGMmJ0lJZ2Og,3600
@@ -272,7 +273,7 @@ datahub/ingestion/source/bigquery_v2/queries.py,sha256=c1BpeQP8p8y-FOhmiQkkY2IqG
 datahub/ingestion/source/bigquery_v2/queries_extractor.py,sha256=_5cAXVU8b8T_nAPDsvN2JRd2dmM1t1J1mRylfKiPen4,19530
 datahub/ingestion/source/bigquery_v2/usage.py,sha256=A9c-ofclaRk0NSnc4IRaqJYqMPv6ecCld_TPy3V2qFs,40748
 datahub/ingestion/source/cassandra/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/source/cassandra/cassandra.py,sha256=lKvPP0Uahi9xw_yh9cArPPtwvAauXolaEk-6f-jhpz4,14558
+datahub/ingestion/source/cassandra/cassandra.py,sha256=k2EKpeUUZ0O_RwQ6NXM15vnTsY3OgST4e5LG8Eg6Coo,14534
 datahub/ingestion/source/cassandra/cassandra_api.py,sha256=b7MApc3_tEfHoj-6ub6snkcv_DweL1wi_TGJjAA1-yU,13516
 datahub/ingestion/source/cassandra/cassandra_config.py,sha256=Ga9915cDZukR5-u2tMNx5Jkf8eza2oAE5YS_sQIVEVQ,4222
 datahub/ingestion/source/cassandra/cassandra_profiling.py,sha256=ZqsAY8NFsrrLqOduV7Aem2eJLtc2_OU9tW4tc_dh0V8,10984
@@ -350,8 +351,8 @@ datahub/ingestion/source/iceberg/iceberg.py,sha256=BNDGooK9cmqpOjzkV1u4rpsduVPNW
 datahub/ingestion/source/iceberg/iceberg_common.py,sha256=CD_yHQ_wEgivyLQUTRO9BZJB29S7j5fUVllki-BPwUU,12292
 datahub/ingestion/source/iceberg/iceberg_profiler.py,sha256=9iwp2vpQTi4OMbIKoDZV5lAdvjMR0ls6Llpck9grJIE,9875
 datahub/ingestion/source/identity/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/source/identity/azure_ad.py,sha256=9Hrvm4CSfc02yjnPUsCYSY4Qw9fXPnDFWLexab0mcpc,28559
-datahub/ingestion/source/identity/okta.py,sha256=jC21myJuMRTaPgj0OD9heaC-mz8ECjqpy2hSJwlUSwM,31943
+datahub/ingestion/source/identity/azure_ad.py,sha256=V3z8PWvNHdb5rv_HTyx-mezX0pRUw3O3Z27MwvIhTpU,28559
+datahub/ingestion/source/identity/okta.py,sha256=4t1F1xdPjWuPfy1esbh4FtfU9pl2rGdi7GosIRc7Dfc,31943
 datahub/ingestion/source/kafka/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/kafka/kafka.py,sha256=CzuW6CNbBMlB9ILD2GvsNIbm8MbHd0_bRI-J3_CFyRs,23322
 datahub/ingestion/source/kafka/kafka_config.py,sha256=ijUB8PS5p-o3uLCHkAxAJAIM88s47rVaAUYXmi_lR4M,4406
@@ -387,7 +388,8 @@ datahub/ingestion/source/metadata/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeR
 datahub/ingestion/source/metadata/business_glossary.py,sha256=T_RJHst6iQRghJNmLLPeSBMEDsbEKf3yBldOAgMcGuo,19666
 datahub/ingestion/source/metadata/lineage.py,sha256=PA4JwSeQ-30XFMN4O5tPwIu-hZF1e-xMZ_CnEUE2c-Q,9595
 datahub/ingestion/source/mock_data/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/source/mock_data/datahub_mock_data.py,sha256=i19UFii3wOPG5CVQfV_20MwIRAJjr0TxsIiVjKcBND4,14314
+datahub/ingestion/source/mock_data/datahub_mock_data.py,sha256=YQ0QUhzn1PyUPscS4sJOpo0pxRwmzXjcs2W0ZiPqsqI,14543
+datahub/ingestion/source/mock_data/datahub_mock_data_report.py,sha256=sV_H7JgcuVbrpIBqtGse_BBigMdqP32ZXuanpeXmwVI,331
 datahub/ingestion/source/mock_data/table_naming_helper.py,sha256=oIC1vcOx76Vl63O9kcjP_iInBHyS-ATdN3Y932TCCZg,3283
 datahub/ingestion/source/neo4j/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/neo4j/neo4j_source.py,sha256=JqKCwxBJfOrC8SF7CmDG0cseWxHk_7E2v4Diw3Q0-WM,14181
@@ -467,12 +469,12 @@ datahub/ingestion/source/snowflake/constants.py,sha256=XCW3vw4JfLn_s8-oXBX6WFNMP
 datahub/ingestion/source/snowflake/oauth_config.py,sha256=ol9D3RmruGStJAeL8PYSQguSqcD2HfkjPkMF2AB_eZs,1277
 datahub/ingestion/source/snowflake/oauth_generator.py,sha256=fu2VnREGuJXeTqIV2jx4TwieVnznf83HQkrE0h2DGGM,3423
 datahub/ingestion/source/snowflake/snowflake_assertion.py,sha256=_l3k4aI9wvioE81xxdeizJn9nJCZ_nMIXgk9N6pEk5o,4803
-datahub/ingestion/source/snowflake/snowflake_config.py,sha256=x73EnJHOrLWNjdiQ65HUgaToC0iAaFWmDtJSyufOl9A,20918
+datahub/ingestion/source/snowflake/snowflake_config.py,sha256=LbDUzqKCXSMV6j0ClLGxAAMbxQFqsmmXjy0xgCVd7UE,20965
 datahub/ingestion/source/snowflake/snowflake_connection.py,sha256=3-nP3HHCblUnUHYo_fvFp5VOAteCtR4GNjaUEvyNTNQ,18175
 datahub/ingestion/source/snowflake/snowflake_data_reader.py,sha256=ffR5E2uhD71FUMXd3XOg2rHwrp1rbbGEFTAbqKcmI2s,2195
 datahub/ingestion/source/snowflake/snowflake_lineage_v2.py,sha256=eWUlWMc5X2X_Y1I2peworFWLLsXQjryEHxPDuSqrowg,21683
 datahub/ingestion/source/snowflake/snowflake_profiler.py,sha256=PmQi-qDlRhdJ-PsJ7x-EScIiswWRAxDDOKHydvN3mTY,7404
-datahub/ingestion/source/snowflake/snowflake_queries.py,sha256=PyuooQO6_OoUno3kzi7m_n1tv5V3GKSmJPXBk29pgWw,30480
+datahub/ingestion/source/snowflake/snowflake_queries.py,sha256=4ZNyBsFPGKatxT6B9CW-gEQt6rUMy2FkRX1seh6Ppog,31165
 datahub/ingestion/source/snowflake/snowflake_query.py,sha256=kqOxibplHyDhdioue8MeXBzeDS9d274-dspb-wyWMMI,38165
 datahub/ingestion/source/snowflake/snowflake_report.py,sha256=O-465aBA8uaYZ6WepP7i6cgK6Q1jXJPjDA1j9C8klus,6762
 datahub/ingestion/source/snowflake/snowflake_schema.py,sha256=1yGBbs2aWIdHnrwgeTR7J2lqxbbBsIt8ejCLumIpLEA,27274
@@ -482,7 +484,7 @@ datahub/ingestion/source/snowflake/snowflake_summary.py,sha256=WJfsP8w3HceUkM6GK
 datahub/ingestion/source/snowflake/snowflake_tag.py,sha256=eA9xh-G1Ydr1OwUUtrbXUWp26hE1jF0zvyKNky_i_nQ,8887
 datahub/ingestion/source/snowflake/snowflake_usage_v2.py,sha256=ySFm7WDk8FW9KjCnX4HQfTqObIrlUS-V8WIHl3j0CTI,24848
 datahub/ingestion/source/snowflake/snowflake_utils.py,sha256=w-SftpjHSOf_6S2T2srHJyJta3MHS0usmn4Z9jgx4QE,13858
-datahub/ingestion/source/snowflake/snowflake_v2.py,sha256=BU9vSXgnaPyKqCHfrHUZxRZOyOHUYMlf0CBUm0OZXn4,34730
+datahub/ingestion/source/snowflake/snowflake_v2.py,sha256=eqOVRYmFiWQPAIDWUcOo5QZkG05xanlEbxz4aDTvSNM,34706
 datahub/ingestion/source/sql/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/sql/athena.py,sha256=U-btrqnbOgrBptl6b--QJrio_AN-c6pF0yL-wbUt4P0,23986
 datahub/ingestion/source/sql/clickhouse.py,sha256=_uf-6fQYxI1-kMw0dNUvJf1f75XX0Qps4HMVlgzdpmo,25670
@@ -509,10 +511,10 @@ datahub/ingestion/source/sql/sqlalchemy_uri_mapper.py,sha256=KOpbmDIE2h1hyYEsbVH
 datahub/ingestion/source/sql/teradata.py,sha256=9WdrxDy02lRJi9IZgsAATFsmxcQnIw5Gr6yCqHJQy5k,33507
 datahub/ingestion/source/sql/trino.py,sha256=zIfQ6GvW8Sbw4sxqsTcnibT51STka_nzNYvmld6HfHw,18947
 datahub/ingestion/source/sql/two_tier_sql_source.py,sha256=AB3Gtx4omAy_08zadHQpmUGmIGufkZ6o_ihWNnfvzYc,5783
-datahub/ingestion/source/sql/vertica.py,sha256=_9OgSgIgqBml0av063rb8nACiT3SAmzpw0ouyF91wv8,33382
+datahub/ingestion/source/sql/vertica.py,sha256=kGfL6exvfKOY5PeC97f7ukdY0Id_pe8Wn9gL0bQN8dE,33358
 datahub/ingestion/source/sql/mssql/__init__.py,sha256=1agpl8S_uDW40olkhCX_W19dbr5GO9qgjS3R7pLRZSk,87
 datahub/ingestion/source/sql/mssql/job_models.py,sha256=nAo3rciu-w2-dXCz6_ekDEbGMEjCMEfh8WvSfXoF2l0,9359
-datahub/ingestion/source/sql/mssql/source.py,sha256=2kCZZyiAOwAVKpr8NkiA-sM6ivMb8UAqQV004h6LVsw,42889
+datahub/ingestion/source/sql/mssql/source.py,sha256=krV80XcNFu-N73xoUA-W2zLiop3pJP7zkb3Dvj4PDKU,42783
 datahub/ingestion/source/sql/stored_procedures/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/sql/stored_procedures/base.py,sha256=n0l5OaTuW-m3TRvkxs3TqvgMeWF6BagzW3tjyWUcC1A,8631
 datahub/ingestion/source/sql/stored_procedures/lineage.py,sha256=fryLhuAlsjr9SHIjHJ-PmtCMx89bjzWVnJZ3f1bwQVU,1905
@@ -524,7 +526,7 @@ datahub/ingestion/source/state/profiling_state_handler.py,sha256=jDMiIrAq8k4GrYo
 datahub/ingestion/source/state/redundant_run_skip_handler.py,sha256=h28twxcsMNvI74bUjAKleRYid8kfIyWS7Y11aBldDlY,9435
 datahub/ingestion/source/state/sql_common_state.py,sha256=OtJpJfMTBSgyR37dn3w-nnZwlc0nFNb2GoUzIWhnyAc,143
 datahub/ingestion/source/state/stale_entity_removal_handler.py,sha256=Lr2HYGx_b2FQ8A36s7s11tl-4-mGIM13bfy5JbQ3LtM,14890
-datahub/ingestion/source/state/stateful_ingestion_base.py,sha256=N0Qzp2t2qAf27WndhsvGbCYCd29dSrLY3TSfbO0hoKA,17369
+datahub/ingestion/source/state/stateful_ingestion_base.py,sha256=M9J2-JMjNR1KYSDPOTIa8zSCu6fFEufmGBVh4XrcNy0,17345
 datahub/ingestion/source/state/usage_common_state.py,sha256=TJyb0CpwibsduJYI854EFdtrwWnz7JC-IkzKUXVGDx0,983
 datahub/ingestion/source/state/use_case_handler.py,sha256=3g8ddTvGXHe0dCiyTkyFeNmR8a3bhwywtIt8EpK5oQs,1271
 datahub/ingestion/source/state_provider/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -547,7 +549,7 @@ datahub/ingestion/source/unity/proxy.py,sha256=E8ZhWIY3j7gjEz8ttWOeHlom0jCMJXkWH
 datahub/ingestion/source/unity/proxy_profiling.py,sha256=WLqvYP6MziaisA4LYL4T_GA-kPt6Xdde7bfaYsjYw40,9663
 datahub/ingestion/source/unity/proxy_types.py,sha256=qrvHiwPzl5cPX-KRvcIGGeJVdr0I8XUQmoAI6ErZ-v8,9371
 datahub/ingestion/source/unity/report.py,sha256=XFT9oQfvEB4RkTvWGgFOoQuLPUN_AIoPXZ79xeDhGHQ,2831
-datahub/ingestion/source/unity/source.py,sha256=uJBjgZ7qhJpn25t0ZOcLuZ0vn2Uz4n9AiideRGCFADw,49027
+datahub/ingestion/source/unity/source.py,sha256=79a7qzFdO66sx6q0E_9Sd3smyQXS8OiRzJFG8Cbcrlk,49003
 datahub/ingestion/source/unity/tag_entities.py,sha256=iWl6nRAWSye1hoFDx_Xh4aT53PN0sGzlX7n1-oTVUv8,11568
 datahub/ingestion/source/unity/usage.py,sha256=0wETBAaZvHI_EGgBlxX3bKsVHEAdnUV8_bKI_lbyWjY,11500
 datahub/ingestion/source/usage/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -967,7 +969,7 @@ datahub/sql_parsing/fingerprint_utils.py,sha256=3hGiexaQXnE7eZLxo-t7hlTyVQz7womb
 datahub/sql_parsing/query_types.py,sha256=FKjDzszZzsrCfYfm7dgD6T_8865qxWl767fdGyHWBh4,2720
 datahub/sql_parsing/schema_resolver.py,sha256=ISuingLcQnOJZkNXBkc73uPwYUbbOtERAjgGhJajDiQ,10782
 datahub/sql_parsing/split_statements.py,sha256=OIQXA9e4k3G9Z1y7rbgdtZhMWt4FPnq41cE8Jkm9cBY,9542
-datahub/sql_parsing/sql_parsing_aggregator.py,sha256=tqFZsE-7owUiU0q49nmkTt50CU4vn8ffUbNcTv9nRbc,71431
+datahub/sql_parsing/sql_parsing_aggregator.py,sha256=Y70tMjCZTHf86g7REwi0A3jSvwQzv5qsfEfaDyiqnhw,71595
 datahub/sql_parsing/sql_parsing_common.py,sha256=cZ4WvVyHZuXDGjnBvKMX2_fz2EMextB5WQWcK0_saBo,3155
 datahub/sql_parsing/sql_parsing_result_utils.py,sha256=prwWTj1EB2fRPv1eMB4EkpFNafIYAt-X8TIK0NWqank,796
 datahub/sql_parsing/sqlglot_lineage.py,sha256=jchSPPYkFtHpyTRTWR5K0YQM6LIgWR5MtyVNQ6zA2Ig,59915
@@ -1078,8 +1080,8 @@ datahub_provider/operators/datahub_assertion_operator.py,sha256=uvTQ-jk2F0sbqqxp
 datahub_provider/operators/datahub_assertion_sensor.py,sha256=lCBj_3x1cf5GMNpHdfkpHuyHfVxsm6ff5x2Z5iizcAo,140
 datahub_provider/operators/datahub_operation_operator.py,sha256=aevDp2FzX7FxGlXrR0khoHNbxbhKR2qPEX5e8O2Jyzw,174
 datahub_provider/operators/datahub_operation_sensor.py,sha256=8fcdVBCEPgqy1etTXgLoiHoJrRt_nzFZQMdSzHqSG7M,168
-acryl_datahub-1.1.0.5rc1.dist-info/METADATA,sha256=sKbnEOEUAOv5hgUfoytH-EIKqUc7QxK96ENS7SZUIPs,182347
-acryl_datahub-1.1.0.5rc1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-acryl_datahub-1.1.0.5rc1.dist-info/entry_points.txt,sha256=bnGf6eX9UhiW8yVHtt6MJCVcmLErvrVQxTJAayA-PKc,9885
-acryl_datahub-1.1.0.5rc1.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
-acryl_datahub-1.1.0.5rc1.dist-info/RECORD,,
+acryl_datahub-1.1.0.5rc3.dist-info/METADATA,sha256=8c04d-AKh-9Af0x9FeL24ah-XHUXxMCyGaklXQqhHLA,182347
+acryl_datahub-1.1.0.5rc3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+acryl_datahub-1.1.0.5rc3.dist-info/entry_points.txt,sha256=bnGf6eX9UhiW8yVHtt6MJCVcmLErvrVQxTJAayA-PKc,9885
+acryl_datahub-1.1.0.5rc3.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
+acryl_datahub-1.1.0.5rc3.dist-info/RECORD,,

datahub/_version.py

@@ -1,6 +1,6 @@
 # Published at https://pypi.org/project/acryl-datahub/.
 __package_name__ = "acryl-datahub"
-__version__ = "1.1.0.5rc1"
+__version__ = "1.1.0.5rc3"
 
 
 def is_dev_mode() -> bool:

datahub/api/entities/dataset/dataset.py

@@ -383,7 +383,7 @@ class Dataset(StrictModel):
     urn: Optional[str] = None
     description: Optional[str] = None
     name: Optional[str] = None
-    schema_metadata: Optional[SchemaSpecification] = Field(alias="schema")
+    schema_metadata: Optional[SchemaSpecification] = Field(default=None, alias="schema")
     downstreams: Optional[List[str]] = None
     properties: Optional[Dict[str, str]] = None
     subtype: Optional[str] = None

datahub/ingestion/api/decorators.py

@@ -104,6 +104,7 @@ def capability(
             for base in cls.__bases__
         ):
             cls.__capabilities = {}
+
             cls.get_capabilities = lambda: cls.__capabilities.values()
 
             # If the superclasses have capability annotations, copy those over.

datahub/ingestion/source/cassandra/cassandra.py

@@ -80,7 +80,7 @@ class KeyspaceKey(ContainerKey):
 @capability(SourceCapability.PLATFORM_INSTANCE, "Enabled by default")
 @capability(
     SourceCapability.DELETION_DETECTION,
-    "Optionally enabled via `stateful_ingestion.remove_stale_metadata`",
+    "Enabled by default via stateful ingestion",
     supported=True,
 )
 class CassandraSource(StatefulIngestionSourceBase):

datahub/ingestion/source/identity/azure_ad.py

@@ -167,7 +167,7 @@ class AzureADSourceReport(StaleEntityRemovalSourceReport):
 @config_class(AzureADConfig)
 @support_status(SupportStatus.CERTIFIED)
 @capability(
-    SourceCapability.DELETION_DETECTION, "Optionally enabled via stateful_ingestion"
+    SourceCapability.DELETION_DETECTION, "Enabled by default via stateful ingestion"
 )
 class AzureADSource(StatefulIngestionSourceBase):
     """

datahub/ingestion/source/identity/okta.py

@@ -202,7 +202,7 @@ class OktaSourceReport(StaleEntityRemovalSourceReport):
 @support_status(SupportStatus.CERTIFIED)
 @capability(SourceCapability.DESCRIPTIONS, "Optionally enabled via configuration")
 @capability(
-    SourceCapability.DELETION_DETECTION, "Optionally enabled via stateful_ingestion"
+    SourceCapability.DELETION_DETECTION, "Enabled by default via stateful ingestion"
 )
 class OktaSource(StatefulIngestionSourceBase):
     """

datahub/ingestion/source/mock_data/datahub_mock_data.py

@@ -15,6 +15,9 @@ from datahub.ingestion.api.decorators import (
 )
 from datahub.ingestion.api.source import Source, SourceReport
 from datahub.ingestion.api.workunit import MetadataWorkUnit
+from datahub.ingestion.source.mock_data.datahub_mock_data_report import (
+    DataHubMockDataReport,
+)
 from datahub.ingestion.source.mock_data.table_naming_helper import TableNamingHelper
 from datahub.metadata.schema_classes import (
     DatasetLineageTypeClass,
@@ -117,13 +120,15 @@ class DataHubMockDataSource(Source):
     def __init__(self, ctx: PipelineContext, config: DataHubMockDataConfig):
         self.ctx = ctx
         self.config = config
-        self.report = SourceReport()
+        self.report = DataHubMockDataReport()
 
     def get_workunits(self) -> Iterable[MetadataWorkUnit]:
         # We don't want any implicit aspects to be produced
         # so we are not using get_workunits_internal
         if self.config.gen_1.emit_lineage:
             for wu in self._data_gen_1():
+                if self.report.first_urn_seen is None:
+                    self.report.first_urn_seen = wu.get_urn()
                 self.report.report_workunit(wu)
                 yield wu
 

datahub/ingestion/source/mock_data/datahub_mock_data_report.py

@@ -0,0 +1,12 @@
+from dataclasses import dataclass, field
+from typing import Optional
+
+from datahub.ingestion.api.source import SourceReport
+
+
+@dataclass
+class DataHubMockDataReport(SourceReport):
+    first_urn_seen: Optional[str] = field(
+        default=None,
+        metadata={"description": "The first URN encountered during ingestion"},
+    )

datahub/ingestion/source/preset.py

@@ -71,7 +71,7 @@ class PresetConfig(SupersetConfig):
 @config_class(PresetConfig)
 @support_status(SupportStatus.CERTIFIED)
 @capability(
-    SourceCapability.DELETION_DETECTION, "Optionally enabled via stateful_ingestion"
+    SourceCapability.DELETION_DETECTION, "Enabled by default via stateful ingestion"
 )
 class PresetSource(SupersetSource):
     """

datahub/ingestion/source/snowflake/snowflake_config.py

@@ -49,6 +49,7 @@ DEFAULT_TEMP_TABLES_PATTERNS = [
     rf".*\.SEGMENT_{UUID_REGEX}",  # segment
     rf".*\.STAGING_.*_{UUID_REGEX}",  # stitch
     r".*\.(GE_TMP_|GE_TEMP_|GX_TEMP_)[0-9A-F]{8}",  # great expectations
+    r".*\.SNOWPARK_TEMP_TABLE_.+",  # snowpark
 ]
 
 

datahub/ingestion/source/snowflake/snowflake_queries.py

@@ -63,7 +63,10 @@ from datahub.sql_parsing.sqlglot_lineage import (
     DownstreamColumnRef,
 )
 from datahub.sql_parsing.sqlglot_utils import get_query_fingerprint
-from datahub.utilities.file_backed_collections import ConnectionWrapper, FileBackedList
+from datahub.utilities.file_backed_collections import (
+    ConnectionWrapper,
+    FileBackedList,
+)
 from datahub.utilities.perf_timer import PerfTimer
 
 logger = logging.getLogger(__name__)
@@ -124,6 +127,8 @@ class SnowflakeQueriesExtractorReport(Report):
     users_fetch_timer: PerfTimer = dataclasses.field(default_factory=PerfTimer)
 
     audit_log_load_timer: PerfTimer = dataclasses.field(default_factory=PerfTimer)
+    aggregator_generate_timer: PerfTimer = dataclasses.field(default_factory=PerfTimer)
+
     sql_aggregator: Optional[SqlAggregatorReport] = None
 
     num_ddl_queries_dropped: int = 0
@@ -243,6 +248,12 @@ class SnowflakeQueriesExtractor(SnowflakeStructuredReportMixin, Closeable):
         audit_log_file = self.local_temp_path / "audit_log.sqlite"
         use_cached_audit_log = audit_log_file.exists()
 
+        if self.config.local_temp_path is None:
+            self._exit_stack.callback(lambda: audit_log_file.unlink(missing_ok=True))
+
+        shared_connection = self._exit_stack.enter_context(
+            ConnectionWrapper(audit_log_file)
+        )
         queries: FileBackedList[
             Union[
                 KnownLineageMapping,
@@ -251,27 +262,16 @@ class SnowflakeQueriesExtractor(SnowflakeStructuredReportMixin, Closeable):
                 TableSwap,
                 ObservedQuery,
             ]
-        ]
+        ] = self._exit_stack.enter_context(FileBackedList(shared_connection))
+
         if use_cached_audit_log:
-            logger.info("Using cached audit log")
-            shared_connection = ConnectionWrapper(audit_log_file)
-            queries = FileBackedList(shared_connection)
+            logger.info(f"Using cached audit log at {audit_log_file}")
         else:
-            audit_log_file.unlink(missing_ok=True)
-
-            shared_connection = ConnectionWrapper(audit_log_file)
-            queries = FileBackedList(shared_connection)
-            entry: Union[
-                KnownLineageMapping,
-                PreparsedQuery,
-                TableRename,
-                TableSwap,
-                ObservedQuery,
-            ]
+            logger.info(f"Fetching audit log into {audit_log_file}")
 
             with self.report.copy_history_fetch_timer:
-                for entry in self.fetch_copy_history():
-                    queries.append(entry)
+                for copy_entry in self.fetch_copy_history():
+                    queries.append(copy_entry)
 
             with self.report.query_log_fetch_timer:
                 for entry in self.fetch_query_log(users):
@@ -281,13 +281,11 @@ class SnowflakeQueriesExtractor(SnowflakeStructuredReportMixin, Closeable):
             for i, query in enumerate(queries):
                 if i % 1000 == 0:
                     logger.info(f"Added {i} query log entries to SQL aggregator")
+
                 self.aggregator.add(query)
 
-        yield from auto_workunit(self.aggregator.gen_metadata())
-        if not use_cached_audit_log:
-            queries.close()
-            shared_connection.close()
-            audit_log_file.unlink(missing_ok=True)
+        with self.report.aggregator_generate_timer:
+            yield from auto_workunit(self.aggregator.gen_metadata())
 
     def fetch_users(self) -> UsersMapping:
         users: UsersMapping = dict()
@@ -403,8 +401,9 @@ class SnowflakeQueriesExtractor(SnowflakeStructuredReportMixin, Closeable):
 
         # TODO need to map snowflake query types to ours
         query_text: str = res["query_text"]
+        snowflake_query_type: str = res["query_type"]
         query_type: QueryType = SNOWFLAKE_QUERY_TYPE_MAPPING.get(
-            res["query_type"], QueryType.UNKNOWN
+            snowflake_query_type, QueryType.UNKNOWN
         )
 
         direct_objects_accessed = res["direct_objects_accessed"]
@@ -421,7 +420,7 @@ class SnowflakeQueriesExtractor(SnowflakeStructuredReportMixin, Closeable):
                     res["session_id"],
                     timestamp,
                     object_modified_by_ddl,
-                    res["query_type"],
+                    snowflake_query_type,
                 )
             if known_ddl_entry:
                 return known_ddl_entry
@@ -436,6 +435,16 @@ class SnowflakeQueriesExtractor(SnowflakeStructuredReportMixin, Closeable):
                 res["user_name"], users.get(res["user_name"])
             )
         )
+        extra_info = {
+            "snowflake_query_id": res["query_id"],
+            "snowflake_root_query_id": res["root_query_id"],
+            "snowflake_query_type": res["query_type"],
+            "snowflake_role_name": res["role_name"],
+            "query_duration": res["query_duration"],
+            "rows_inserted": res["rows_inserted"],
+            "rows_updated": res["rows_updated"],
+            "rows_deleted": res["rows_deleted"],
+        }
 
         # There are a couple cases when we'd want to prefer our own SQL parsing
         # over Snowflake's metadata.
@@ -470,6 +479,7 @@ class SnowflakeQueriesExtractor(SnowflakeStructuredReportMixin, Closeable):
                 query_hash=get_query_fingerprint(
                     query_text, self.identifiers.platform, fast=True
                 ),
+                extra_info=extra_info,
             )
 
         upstreams = []
@@ -556,6 +566,7 @@ class SnowflakeQueriesExtractor(SnowflakeStructuredReportMixin, Closeable):
             timestamp=timestamp,
             session_id=res["session_id"],
             query_type=query_type,
+            extra_info=extra_info,
         )
         return entry
 
@@ -667,7 +678,7 @@ def _build_enriched_query_log_query(
     start_time_millis = int(start_time.timestamp() * 1000)
     end_time_millis = int(end_time.timestamp() * 1000)
 
-    users_filter = ""
+    users_filter = "TRUE"
     if deny_usernames:
         user_not_in = ",".join(f"'{user.upper()}'" for user in deny_usernames)
         users_filter = f"user_name NOT IN ({user_not_in})"
@@ -694,10 +705,10 @@ fingerprinted_queries as (
     FROM
         snowflake.account_usage.query_history
     WHERE
-        query_history.start_time >= to_timestamp_ltz({start_time_millis}, 3)
-        AND query_history.start_time < to_timestamp_ltz({end_time_millis}, 3)
+        query_history.start_time >= to_timestamp_ltz({start_time_millis}, 3) -- {start_time.isoformat()}
+        AND query_history.start_time < to_timestamp_ltz({end_time_millis}, 3) -- {end_time.isoformat()}
         AND execution_status = 'SUCCESS'
-        AND {users_filter or "TRUE"}
+        AND {users_filter}
 )
 , deduplicated_queries as (
     SELECT
@@ -715,6 +726,7 @@ fingerprinted_queries as (
 , raw_access_history AS (
     SELECT
         query_id,
+        root_query_id,
         query_start_time,
         user_name,
         direct_objects_accessed,
@@ -723,9 +735,9 @@ fingerprinted_queries as (
     FROM
         snowflake.account_usage.access_history
     WHERE
-        query_start_time >= to_timestamp_ltz({start_time_millis}, 3)
-        AND query_start_time < to_timestamp_ltz({end_time_millis}, 3)
-        AND {users_filter or "TRUE"}
+        query_start_time >= to_timestamp_ltz({start_time_millis}, 3) -- {start_time.isoformat()}
+        AND query_start_time < to_timestamp_ltz({end_time_millis}, 3) -- {end_time.isoformat()}
+        AND {users_filter}
         AND query_id IN (
             SELECT query_id FROM deduplicated_queries
         )
@@ -734,6 +746,7 @@ fingerprinted_queries as (
     -- TODO: Add table filter clause.
     SELECT
         query_id,
+        root_query_id,
         query_start_time,
         ARRAY_SLICE(
             FILTER(direct_objects_accessed, o -> o:objectDomain IN {SnowflakeQuery.ACCESS_HISTORY_TABLE_VIEW_DOMAINS_FILTER}),
@@ -764,6 +777,7 @@ fingerprinted_queries as (
         q.rows_deleted AS "ROWS_DELETED",
         q.user_name AS "USER_NAME",
         q.role_name AS "ROLE_NAME",
+        a.root_query_id,
         a.direct_objects_accessed,
         a.objects_modified,
         a.object_modified_by_ddl

datahub/ingestion/source/snowflake/snowflake_v2.py

@@ -118,7 +118,7 @@ logger: logging.Logger = logging.getLogger(__name__)
 )
 @capability(
     SourceCapability.DELETION_DETECTION,
-    "Optionally enabled via `stateful_ingestion.remove_stale_metadata`",
+    "Enabled by default via stateful ingestion",
     supported=True,
 )
 @capability(

datahub/ingestion/source/sql/mssql/source.py

@@ -936,25 +936,25 @@ class SQLServerSource(SQLAlchemySource):
         url = self.config.get_sql_alchemy_url()
         logger.debug(f"sql_alchemy_url={url}")
         engine = create_engine(url, **self.config.options)
-        with engine.connect() as conn:
-            if self.config.database and self.config.database != "":
-                inspector = inspect(conn)
-                yield inspector
-            else:
+
+        if self.config.database and self.config.database != "":
+            inspector = inspect(engine)
+            yield inspector
+        else:
+            with engine.begin() as conn:
                 databases = conn.execute(
                     "SELECT name FROM master.sys.databases WHERE name NOT IN \
                   ('master', 'model', 'msdb', 'tempdb', 'Resource', \
                        'distribution' , 'reportserver', 'reportservertempdb'); "
-                )
-                for db in databases:
-                    if self.config.database_pattern.allowed(db["name"]):
-                        url = self.config.get_sql_alchemy_url(current_db=db["name"])
-                        with create_engine(
-                            url, **self.config.options
-                        ).connect() as conn:
-                            inspector = inspect(conn)
-                            self.current_database = db["name"]
-                            yield inspector
+                ).fetchall()
+
+            for db in databases:
+                if self.config.database_pattern.allowed(db["name"]):
+                    url = self.config.get_sql_alchemy_url(current_db=db["name"])
+                    engine = create_engine(url, **self.config.options)
+                    inspector = inspect(engine)
+                    self.current_database = db["name"]
+                    yield inspector
 
     def get_identifier(
         self, *, schema: str, entity: str, inspector: Inspector, **kwargs: Any

datahub/ingestion/source/sql/vertica.py

@@ -116,7 +116,7 @@ class VerticaConfig(BasicSQLAlchemyConfig):
 )
 @capability(
     SourceCapability.DELETION_DETECTION,
-    "Optionally enabled via `stateful_ingestion.remove_stale_metadata`",
+    "Enabled by default via stateful ingestion",
     supported=True,
 )
 class VerticaSource(SQLAlchemySource):

datahub/ingestion/source/state/stateful_ingestion_base.py

@@ -179,7 +179,7 @@ class StatefulIngestionReport(SourceReport):
 
 @capability(
     SourceCapability.DELETION_DETECTION,
-    "Optionally enabled via `stateful_ingestion.remove_stale_metadata`",
+    "Enabled by default via stateful ingestion",
     supported=True,
 )
 class StatefulIngestionSourceBase(Source):

datahub/ingestion/source/superset.py

@@ -272,7 +272,7 @@ def get_filter_name(filter_obj):
 @config_class(SupersetConfig)
 @support_status(SupportStatus.CERTIFIED)
 @capability(
-    SourceCapability.DELETION_DETECTION, "Optionally enabled via stateful_ingestion"
+    SourceCapability.DELETION_DETECTION, "Enabled by default via stateful ingestion"
 )
 @capability(SourceCapability.DOMAINS, "Enabled by `domain` config to assign domain_key")
 @capability(SourceCapability.LINEAGE_COARSE, "Supported by default")

datahub/ingestion/source/unity/source.py

@@ -159,7 +159,7 @@ logger: logging.Logger = logging.getLogger(__name__)
 )
 @capability(
     SourceCapability.DELETION_DETECTION,
-    "Optionally enabled via `stateful_ingestion.remove_stale_metadata`",
+    "Enabled by default via stateful ingestion",
     supported=True,
 )
 @capability(SourceCapability.TEST_CONNECTION, "Enabled by default")

datahub/sql_parsing/sql_parsing_aggregator.py

@@ -58,6 +58,7 @@ from datahub.sql_parsing.tool_meta_extractor import (
     ToolMetaExtractorReport,
 )
 from datahub.utilities.cooperative_timeout import CooperativeTimeoutError
+from datahub.utilities.dedup_list import deduplicate_list
 from datahub.utilities.file_backed_collections import (
     ConnectionWrapper,
     FileBackedDict,
@@ -140,6 +141,7 @@ class QueryMetadata:
 
     used_temp_tables: bool = True
 
+    extra_info: Optional[dict] = None
     origin: Optional[Urn] = None
 
     def make_created_audit_stamp(self) -> models.AuditStampClass:
@@ -263,7 +265,7 @@ class PreparsedQuery:
     query_type_props: QueryTypeProps = dataclasses.field(
         default_factory=lambda: QueryTypeProps()
     )
-    # Use this to store addtitional key-value information about query for debugging
+    # Use this to store additional key-value information about the query for debugging.
     extra_info: Optional[dict] = None
     origin: Optional[Urn] = None
 
@@ -948,6 +950,7 @@ class SqlParsingAggregator(Closeable):
                 column_usage=parsed.column_usage or {},
                 confidence_score=parsed.confidence_score,
                 used_temp_tables=session_has_temp_tables,
+                extra_info=parsed.extra_info,
                 origin=parsed.origin,
             )
         )
@@ -1706,7 +1709,7 @@ class SqlParsingAggregator(Closeable):
         )
 
         merged_query_text = ";\n\n".join(
-            [q.formatted_query_string for q in ordered_queries]
+            deduplicate_list([q.formatted_query_string for q in ordered_queries])
         )
 
         resolved_query = dataclasses.replace(