acryl-datahub 1.2.0.7rc4__py3-none-any.whl → 1.2.0.8rc1__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.
- {acryl_datahub-1.2.0.7rc4.dist-info → acryl_datahub-1.2.0.8rc1.dist-info}/METADATA +2674 -2674
- {acryl_datahub-1.2.0.7rc4.dist-info → acryl_datahub-1.2.0.8rc1.dist-info}/RECORD +34 -32
- datahub/_version.py +1 -1
- datahub/cli/delete_cli.py +1 -0
- datahub/ingestion/api/report.py +4 -0
- datahub/ingestion/autogenerated/capability_summary.json +1 -1
- datahub/ingestion/graph/client.py +8 -1
- datahub/ingestion/source/datahub/config.py +4 -0
- datahub/ingestion/source/datahub/datahub_database_reader.py +6 -1
- datahub/ingestion/source/metadata/lineage.py +8 -8
- datahub/ingestion/source/redshift/redshift.py +1 -1
- datahub/ingestion/source/sql/athena.py +95 -18
- datahub/ingestion/source/sql/athena_properties_extractor.py +43 -25
- datahub/ingestion/source/superset.py +3 -2
- datahub/ingestion/source/tableau/tableau.py +8 -5
- datahub/metadata/_internal_schema_classes.py +207 -12
- datahub/metadata/com/linkedin/pegasus2avro/settings/asset/__init__.py +19 -0
- datahub/metadata/com/linkedin/pegasus2avro/template/__init__.py +6 -0
- datahub/metadata/schema.avsc +160 -12
- datahub/metadata/schemas/AssetSettings.avsc +63 -0
- datahub/metadata/schemas/DataHubPageModuleProperties.avsc +9 -1
- datahub/metadata/schemas/DataHubPageTemplateProperties.avsc +77 -1
- datahub/metadata/schemas/DataProductKey.avsc +2 -1
- datahub/metadata/schemas/DomainKey.avsc +2 -1
- datahub/metadata/schemas/GlossaryNodeKey.avsc +2 -1
- datahub/metadata/schemas/GlossaryTermKey.avsc +2 -1
- datahub/metadata/schemas/IncidentInfo.avsc +3 -3
- datahub/metadata/schemas/StructuredPropertyDefinition.avsc +0 -3
- datahub/sql_parsing/sqlglot_lineage.py +121 -28
- datahub/sql_parsing/sqlglot_utils.py +12 -1
- {acryl_datahub-1.2.0.7rc4.dist-info → acryl_datahub-1.2.0.8rc1.dist-info}/WHEEL +0 -0
- {acryl_datahub-1.2.0.7rc4.dist-info → acryl_datahub-1.2.0.8rc1.dist-info}/entry_points.txt +0 -0
- {acryl_datahub-1.2.0.7rc4.dist-info → acryl_datahub-1.2.0.8rc1.dist-info}/licenses/LICENSE +0 -0
- {acryl_datahub-1.2.0.7rc4.dist-info → acryl_datahub-1.2.0.8rc1.dist-info}/top_level.txt +0 -0
|
@@ -1,7 +1,7 @@
|
|
|
1
|
-
acryl_datahub-1.2.0.
|
|
1
|
+
acryl_datahub-1.2.0.8rc1.dist-info/licenses/LICENSE,sha256=9xNHpsD0uYF5ONzXsKDCuHHB-xbiCrSbueWXqrTNsxk,11365
|
|
2
2
|
datahub/__init__.py,sha256=aq_i5lVREmoLfYIqcx_pEQicO855YlhD19tWc1eZZNI,59
|
|
3
3
|
datahub/__main__.py,sha256=pegIvQ9hzK7IhqVeUi1MeADSZ2QlP-D3K0OQdEg55RU,106
|
|
4
|
-
datahub/_version.py,sha256=
|
|
4
|
+
datahub/_version.py,sha256=Rdij3ffZjrkKXarGFXcv2MZfNld3LEFCYwjv7W_kgqg,323
|
|
5
5
|
datahub/entrypoints.py,sha256=9Qf-37rNnTzbGlx8S75OCDazIclFp6zWNcCEL1zCZto,9015
|
|
6
6
|
datahub/errors.py,sha256=p5rFAdAGVCk4Lqolol1YvthceadUSwpaCxLXRcyCCFQ,676
|
|
7
7
|
datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
@@ -71,7 +71,7 @@ datahub/cli/check_cli.py,sha256=PVd3DT9TOK7Ejd0ODKnbRrBnPxsZLfUWgubFRUKwKAA,1697
|
|
|
71
71
|
datahub/cli/cli_utils.py,sha256=0jTTAKuDZ8GzZwGHYytcT_MPR3Rb2DAcbr9n1H2T2sE,16170
|
|
72
72
|
datahub/cli/config_utils.py,sha256=EeBGfhmf4AxYoTfnZ4GSiGIgpzJFkduNjN_FwmxZGhA,4889
|
|
73
73
|
datahub/cli/container_cli.py,sha256=D0zWP3_3aww8_RTkMugOoOlILz3dPJ0TE9asQDLCm6E,1697
|
|
74
|
-
datahub/cli/delete_cli.py,sha256=
|
|
74
|
+
datahub/cli/delete_cli.py,sha256=0YJeWuXPGY0kbSn1AXK1-8SfCGBxb78ZbO53RAgyjQg,26515
|
|
75
75
|
datahub/cli/docker_check.py,sha256=NdMmIG8H15QbJqzuHchrs99Mva7kVE-t7FTd6t3qT_c,12978
|
|
76
76
|
datahub/cli/docker_cli.py,sha256=aAZIF3oblTSXzQVSanbphd3dCvdypHCyuMt1u3_QZ9M,33067
|
|
77
77
|
datahub/cli/env_utils.py,sha256=RQzjg4JE29hjPt4v7p-RuqoOr99w8E3DBHWiN2Sm7T4,252
|
|
@@ -148,7 +148,7 @@ datahub/ingestion/api/incremental_properties_helper.py,sha256=KzdxdrQtaMV2XMHfPs
|
|
|
148
148
|
datahub/ingestion/api/ingestion_job_checkpointing_provider_base.py,sha256=3lLdkkxVqE9MVc26cdXImPeWy16az5BwgcorWxeBV50,1759
|
|
149
149
|
datahub/ingestion/api/pipeline_run_listener.py,sha256=5uBP__LbMQxJ2utlf07cIzQINqPbUOKiZyOJta6a0og,713
|
|
150
150
|
datahub/ingestion/api/registry.py,sha256=LbdZr89465Lj7ptQRVB4vI1JR1igWABvQFj9-WX63bI,7454
|
|
151
|
-
datahub/ingestion/api/report.py,sha256
|
|
151
|
+
datahub/ingestion/api/report.py,sha256=OuVZAgNkzSGkKhpOhpqebd9_bEsBCTeoWR1VcanPeD4,18509
|
|
152
152
|
datahub/ingestion/api/report_helpers.py,sha256=WbUC1kQeaKqIagGV3XzfPmPs7slAT1mfNY4og2BH2A8,994
|
|
153
153
|
datahub/ingestion/api/sink.py,sha256=GZt48PV56FAhNoma-V5EwwRZvezhb40YH_zprm8_Yo0,4961
|
|
154
154
|
datahub/ingestion/api/source.py,sha256=uf0fNbiOy0bS_aKFOcNv6NvuZe0LSDIDdNza9hraP7s,21857
|
|
@@ -159,7 +159,7 @@ datahub/ingestion/api/auto_work_units/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCe
|
|
|
159
159
|
datahub/ingestion/api/auto_work_units/auto_dataset_properties_aspect.py,sha256=ID_6N3nWl2qohsSGizUCqo3d2MNyDeVbyWroQpSOSsc,5059
|
|
160
160
|
datahub/ingestion/api/auto_work_units/auto_ensure_aspect_size.py,sha256=0BwkpLhORbsiTHq0g_N_1cVVoZYdLR3qz02mNmsV9-M,4444
|
|
161
161
|
datahub/ingestion/autogenerated/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
162
|
-
datahub/ingestion/autogenerated/capability_summary.json,sha256=
|
|
162
|
+
datahub/ingestion/autogenerated/capability_summary.json,sha256=78zS8mOo54IkG9l5KJ_YBXkZmERSSo3pDIrefQx-PWA,110199
|
|
163
163
|
datahub/ingestion/autogenerated/lineage.json,sha256=8BdZF-5V5kJbX4mfFav8Zg-jHjzfkAEGk-pu1atLN4I,10029
|
|
164
164
|
datahub/ingestion/autogenerated/lineage_helper.py,sha256=I_k1pZSCCCjDbUVifPTfy6fkmV8jqdVhbirE8EkpmxI,4748
|
|
165
165
|
datahub/ingestion/extractor/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
@@ -181,7 +181,7 @@ datahub/ingestion/glossary/classifier.py,sha256=daLxnVv_JlfB_jBOxH5LrU_xQRndrsGo
|
|
|
181
181
|
datahub/ingestion/glossary/classifier_registry.py,sha256=yFOYLQhDgCLqXYMG3L1BquXafeLcZDcmp8meyw6k9ts,307
|
|
182
182
|
datahub/ingestion/glossary/datahub_classifier.py,sha256=O7wm6gQT1Jf2QSKdWjJQbS5oSzJwplXzfza26Gdq5Mg,7555
|
|
183
183
|
datahub/ingestion/graph/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
184
|
-
datahub/ingestion/graph/client.py,sha256=
|
|
184
|
+
datahub/ingestion/graph/client.py,sha256=Orf9BHyTeHufB_LRIC-LyJj1kii73tyWTN_I1HJ6-_k,74955
|
|
185
185
|
datahub/ingestion/graph/config.py,sha256=rmkcqAL8fJoY9QyAeS0Xm8HvwHzV3pCjY-Om-50JJTI,1015
|
|
186
186
|
datahub/ingestion/graph/connections.py,sha256=9462L0ZWGKURyypAln25eMPhK3pcufBar9tNDoqspXs,741
|
|
187
187
|
datahub/ingestion/graph/entity_versioning.py,sha256=nrcNz0Qm6kpE6oTu_mrYUQDx14KPspBTc6R9SyFUY6c,6901
|
|
@@ -228,7 +228,7 @@ datahub/ingestion/source/redash.py,sha256=YxjSad-X_wPmxYH8dJmFz_VCFhiLTCTSlK99Wd
|
|
|
228
228
|
datahub/ingestion/source/salesforce.py,sha256=UttN3y4Ylbx_yCFCr-33wUEZFR48nTiYeUfOjGIFj2E,40872
|
|
229
229
|
datahub/ingestion/source/source_registry.py,sha256=a2mLjJPLkSI-gYCTb_7U7Jo4D8jGknNQ_yScPIihXFk,1208
|
|
230
230
|
datahub/ingestion/source/sql_queries.py,sha256=Zf6Y84WFCD-j0v4HdtcXshAPrtGq5InncXCjnLaAuS4,14657
|
|
231
|
-
datahub/ingestion/source/superset.py,sha256=
|
|
231
|
+
datahub/ingestion/source/superset.py,sha256=SLv6DqglPlFJwEskLOqEr-rfgCAb_2AXqWYupk8FqH4,53756
|
|
232
232
|
datahub/ingestion/source/abs/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
233
233
|
datahub/ingestion/source/abs/config.py,sha256=WW9JWbzqAJDblAcJKtNeuBHqOeJsB57lW2PqSD65-BU,6729
|
|
234
234
|
datahub/ingestion/source/abs/datalake_profiler_config.py,sha256=FfrcgK-JEF94vw-l3q6pN6FENXb-wZzW2w1VUZVkwW8,3620
|
|
@@ -293,9 +293,9 @@ datahub/ingestion/source/data_lake_common/data_lake_utils.py,sha256=IYr5y8vy_6Ct
|
|
|
293
293
|
datahub/ingestion/source/data_lake_common/object_store.py,sha256=i9Hgb8Ww23QD_jEjzj_2qxA8Nr56krnZfo1qyOWmH9M,23608
|
|
294
294
|
datahub/ingestion/source/data_lake_common/path_spec.py,sha256=idEhxKXPCbkp7NzLmaTcS2ebAUwGZIBvn8zAUHcD1QM,23801
|
|
295
295
|
datahub/ingestion/source/datahub/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
296
|
-
datahub/ingestion/source/datahub/config.py,sha256=
|
|
296
|
+
datahub/ingestion/source/datahub/config.py,sha256=ElMoJDuaByhR7ZgXmGTwHJDZDzfGydMkHlSLDnq2Ws0,5257
|
|
297
297
|
datahub/ingestion/source/datahub/datahub_api_reader.py,sha256=hlKADVEPoTFiRGKqRsMF5mL4fSu_IrIW8Nx7LpEzvkM,2134
|
|
298
|
-
datahub/ingestion/source/datahub/datahub_database_reader.py,sha256=
|
|
298
|
+
datahub/ingestion/source/datahub/datahub_database_reader.py,sha256=KjVa7aoKALbqXvTo1iJpUfyJdW0_sxRe_VFh-9-FgrI,15599
|
|
299
299
|
datahub/ingestion/source/datahub/datahub_kafka_reader.py,sha256=gnxhhlK-jrfnHqD_4eVmfcdtBNW6pi1N_qkDZ7uSb3o,4187
|
|
300
300
|
datahub/ingestion/source/datahub/datahub_source.py,sha256=mpBrHxf3sitdPZni5B5atlEheQs8cMuN1i2q72eDelQ,8993
|
|
301
301
|
datahub/ingestion/source/datahub/report.py,sha256=VHBfCbwFRzdLdB7hQG9ST4EiZxl_vBCU0XxGcZR6Xxs,940
|
|
@@ -405,7 +405,7 @@ datahub/ingestion/source/looker/urn_functions.py,sha256=4VvqEfGvIMq3rNHHps0-HlPu
|
|
|
405
405
|
datahub/ingestion/source/looker/view_upstream.py,sha256=4FCjZaU6p2G7npB2RJpP4Gv2yLjbvbsYWEbAg55IvjY,26110
|
|
406
406
|
datahub/ingestion/source/metadata/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
407
407
|
datahub/ingestion/source/metadata/business_glossary.py,sha256=T_RJHst6iQRghJNmLLPeSBMEDsbEKf3yBldOAgMcGuo,19666
|
|
408
|
-
datahub/ingestion/source/metadata/lineage.py,sha256=
|
|
408
|
+
datahub/ingestion/source/metadata/lineage.py,sha256=YgerAUptUVMwrg-s_H8CwTKbbI91Ta74JE-rK3oGUSE,9611
|
|
409
409
|
datahub/ingestion/source/mock_data/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
410
410
|
datahub/ingestion/source/mock_data/datahub_mock_data.py,sha256=3i3SdBp267cZRszhmD_JWJLTGIot2FI8REFpjJQ4jD8,19822
|
|
411
411
|
datahub/ingestion/source/mock_data/datahub_mock_data_report.py,sha256=sV_H7JgcuVbrpIBqtGse_BBigMdqP32ZXuanpeXmwVI,331
|
|
@@ -451,7 +451,7 @@ datahub/ingestion/source/redshift/exception.py,sha256=dxzYUIv5B_FAWhOuzG2u5We7FX
|
|
|
451
451
|
datahub/ingestion/source/redshift/lineage.py,sha256=nqrvWJqaI493i1hIZ_7patrdOb16sZrgSSGapdMcEiU,31710
|
|
452
452
|
datahub/ingestion/source/redshift/profile.py,sha256=H1Xtc2rXScUv4w0b2BbM7POjYEwqIql_rpWvlumY_EM,4309
|
|
453
453
|
datahub/ingestion/source/redshift/query.py,sha256=vVIuNUaU4a7AfMFJZlgLuqi0cGVl0gVz8xZUSnPhWvs,47845
|
|
454
|
-
datahub/ingestion/source/redshift/redshift.py,sha256=
|
|
454
|
+
datahub/ingestion/source/redshift/redshift.py,sha256=zalndYg_LK5aJ8cX_ZuXLcTYajtlavmV-dmQIsjGxjg,41260
|
|
455
455
|
datahub/ingestion/source/redshift/redshift_data_reader.py,sha256=zc69jwXHdF-w8J4Hq-ZQ6BjHQ75Ij2iNDMpoRJlcmlU,1724
|
|
456
456
|
datahub/ingestion/source/redshift/redshift_schema.py,sha256=7F-l_omOuKMuGE_rBWXVPG_GWXFKnCMzC4frNxZB9cs,24800
|
|
457
457
|
datahub/ingestion/source/redshift/report.py,sha256=O3QFozHlmMbH9b7KxbqhgTgr_0tCryj6FIzMiN6kRxw,3044
|
|
@@ -505,8 +505,8 @@ datahub/ingestion/source/snowflake/snowflake_utils.py,sha256=dmgpwApayUIevyn6l55
|
|
|
505
505
|
datahub/ingestion/source/snowflake/snowflake_v2.py,sha256=c6wg_s97Hrckqi0BgAbmnnRQRDDda1-BHFLlnRx0xuw,35753
|
|
506
506
|
datahub/ingestion/source/snowflake/stored_proc_lineage.py,sha256=rOb78iHiWiK8v8WdVs1xDwVut4Y0OHmszej6IopQfCo,5341
|
|
507
507
|
datahub/ingestion/source/sql/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
508
|
-
datahub/ingestion/source/sql/athena.py,sha256=
|
|
509
|
-
datahub/ingestion/source/sql/athena_properties_extractor.py,sha256=
|
|
508
|
+
datahub/ingestion/source/sql/athena.py,sha256=S5R-3HL7nN2FcryEUlwi2fFWzWQb55iycYFSlTRLcoc,31460
|
|
509
|
+
datahub/ingestion/source/sql/athena_properties_extractor.py,sha256=mKu0ZGyt8qvpWoP6CUf0vLUAz5k7GO2keof5KJd4Wak,29469
|
|
510
510
|
datahub/ingestion/source/sql/clickhouse.py,sha256=zd5qE6XPw0AXtY_71-n0yz4ua69xP3oxMuIoabAuT3Q,25987
|
|
511
511
|
datahub/ingestion/source/sql/cockroachdb.py,sha256=WoOKCq7YjsuzSPm1SmKIYZ9CrvlSF8zWmP1fNHn4G3Q,1360
|
|
512
512
|
datahub/ingestion/source/sql/druid.py,sha256=_tzgTa5jhPUXk6WCmS7p10feCwJm6yUFcOgMZA-OcE8,2922
|
|
@@ -554,7 +554,7 @@ datahub/ingestion/source/state_provider/datahub_ingestion_checkpointing_provider
|
|
|
554
554
|
datahub/ingestion/source/state_provider/file_ingestion_checkpointing_provider.py,sha256=DziD57PbHn2Tcy51tYXCG-GQgyTGMUxnkuzVS_xihFY,4079
|
|
555
555
|
datahub/ingestion/source/state_provider/state_provider_registry.py,sha256=SVq4mIyGNmLXE9OZx1taOiNPqDoQp03-Ot9rYnB5F3k,401
|
|
556
556
|
datahub/ingestion/source/tableau/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
557
|
-
datahub/ingestion/source/tableau/tableau.py,sha256=
|
|
557
|
+
datahub/ingestion/source/tableau/tableau.py,sha256=OCdEHTYhJllnvQKalaEdiRATt04syAs2-O1oBYhCu74,156162
|
|
558
558
|
datahub/ingestion/source/tableau/tableau_common.py,sha256=4cUm3E8wLfjLSjcAXXWDWVUXAEho_hbsQa2BzAF-vtM,27012
|
|
559
559
|
datahub/ingestion/source/tableau/tableau_constant.py,sha256=4ngrOwqxf4cgbLR3i0OKI4pUxmHMABKyywfhXQ0GazA,2592
|
|
560
560
|
datahub/ingestion/source/tableau/tableau_server_wrapper.py,sha256=nSyx9RzC6TCQDm-cTVJ657qT8iDwzk_8JMKpohhmOc4,1046
|
|
@@ -635,8 +635,8 @@ datahub/lite/lite_registry.py,sha256=bpH0kasP-LtwwUFNA2QsOIehfekAYfJtN-AkQLmSWnw
|
|
|
635
635
|
datahub/lite/lite_server.py,sha256=p9Oa2nNs65mqcssSIVOr7VOzWqfVstz6ZQEdT4f82S0,1949
|
|
636
636
|
datahub/lite/lite_util.py,sha256=G0LQHKkyEb1pc_q183g6hflShclGx7kikgMaOxtVVcs,4545
|
|
637
637
|
datahub/metadata/__init__.py,sha256=AjhXPjI6cnpdcrBRrE5gOWo15vv2TTl2ctU4UAnUN7A,238
|
|
638
|
-
datahub/metadata/_internal_schema_classes.py,sha256=
|
|
639
|
-
datahub/metadata/schema.avsc,sha256=
|
|
638
|
+
datahub/metadata/_internal_schema_classes.py,sha256=AExBM83VHb4vsnAWF2nCR2x6hGg8NIsGXtAIAmYMs04,1061410
|
|
639
|
+
datahub/metadata/schema.avsc,sha256=HluHCVmYg7RpOaw9xUMigEJBxlHF5WLdNcqVBKPeoOU,704514
|
|
640
640
|
datahub/metadata/schema_classes.py,sha256=tPT8iHCak4IsZi_oL0nirbPpI8ETTPTZzapqLRpeKU4,1326
|
|
641
641
|
datahub/metadata/urns.py,sha256=nfrCTExR-k2P9w272WVtWSN3xW1VUJngPwP3xnvULjU,1217
|
|
642
642
|
datahub/metadata/_urns/__init__.py,sha256=cOF3GHMDgPhmbLKbN02NPpuLGHSu0qNgQyBRv08eqF0,243
|
|
@@ -708,12 +708,13 @@ datahub/metadata/com/linkedin/pegasus2avro/schema/__init__.py,sha256=7JpzLs6S_Ey
|
|
|
708
708
|
datahub/metadata/com/linkedin/pegasus2avro/schemafield/__init__.py,sha256=HTWeznycKnHBfPEGcCHXPEz83Iq9ypjNaoSfeQeDU9g,397
|
|
709
709
|
datahub/metadata/com/linkedin/pegasus2avro/secret/__init__.py,sha256=qk61EqqVZF6k1Ct6t4Uo-pLb0WtM1EwJKn1XjVy9LHE,305
|
|
710
710
|
datahub/metadata/com/linkedin/pegasus2avro/settings/__init__.py,sha256=gsAIuTxzfJdI7a9ybZlgMIHMAYksM1SxGxXjtySgKSc,202
|
|
711
|
+
datahub/metadata/com/linkedin/pegasus2avro/settings/asset/__init__.py,sha256=ASD1mi7q19HVN9QKzbu2T3GxzdDQ_LqZNTehV3SI12c,531
|
|
711
712
|
datahub/metadata/com/linkedin/pegasus2avro/settings/global/__init__.py,sha256=hKpTDXj2YEBja4hk2fZx3G9yzJ7eCFxcGCbbN0NydTk,933
|
|
712
713
|
datahub/metadata/com/linkedin/pegasus2avro/step/__init__.py,sha256=HLNNbqBlyhcg09eXWx_AMD_JoOtBPYEi2kv12PE0R9E,329
|
|
713
714
|
datahub/metadata/com/linkedin/pegasus2avro/structured/__init__.py,sha256=Cry61gPw6m5MQuJpPxADRm3jhI0XVqzznyD3fVKMkvc,1013
|
|
714
715
|
datahub/metadata/com/linkedin/pegasus2avro/tag/__init__.py,sha256=Odb4mzloKJIlpoFHODEIxt_OIgFNrZExcyQtvXxjOFQ,290
|
|
715
716
|
datahub/metadata/com/linkedin/pegasus2avro/telemetry/__init__.py,sha256=N4CJwzAqTrRoCQ2Aoa_e8cUZI_fzn9Zdo2okvO-_nWE,302
|
|
716
|
-
datahub/metadata/com/linkedin/pegasus2avro/template/__init__.py,sha256
|
|
717
|
+
datahub/metadata/com/linkedin/pegasus2avro/template/__init__.py,sha256=-TWG4__SDxZfD9X6VDalU1ErZRCtyRPwoI8eN4wSta4,1254
|
|
717
718
|
datahub/metadata/com/linkedin/pegasus2avro/test/__init__.py,sha256=Z4DlDtf-NELFpx44Pk4RL1JlGuxtgEAMa6Sko8QBsGw,711
|
|
718
719
|
datahub/metadata/com/linkedin/pegasus2avro/timeseries/__init__.py,sha256=6Pbit2drar8n99RFNQiXfYj7PhIzrO1SIpsGELZR4oA,637
|
|
719
720
|
datahub/metadata/com/linkedin/pegasus2avro/upgrade/__init__.py,sha256=o3U2TuzRSU1uPL-4AOMCPDqEwngqRb6g4-CBFY7eSvQ,525
|
|
@@ -729,6 +730,7 @@ datahub/metadata/schemas/AssertionActions.avsc,sha256=zrvXzX2Nv_dmK6I3ZXCVWQ1bVs
|
|
|
729
730
|
datahub/metadata/schemas/AssertionInfo.avsc,sha256=djiUVdw0pGd_Ex7uJspasTx2xwwCnd9cpItd76VJqYw,125296
|
|
730
731
|
datahub/metadata/schemas/AssertionKey.avsc,sha256=EjNaTyzGxtWzTsZd87P2ZSaGX5dn8Y7HGBdqvlQVrFI,638
|
|
731
732
|
datahub/metadata/schemas/AssertionRunEvent.avsc,sha256=FUyV73bUliBC-a_XFUlfgh75o99-lu1fl36b2q8Pqx4,12886
|
|
733
|
+
datahub/metadata/schemas/AssetSettings.avsc,sha256=LBmSybFUBkQPVmIEF1GwQl0ePmTrXlm0kHbKRIm--lo,2004
|
|
732
734
|
datahub/metadata/schemas/BrowsePaths.avsc,sha256=NR_4dKuJMk1X2RB4DLkHVSqyMQc4PvT7eR0n6lM5aOM,654
|
|
733
735
|
datahub/metadata/schemas/BrowsePathsV2.avsc,sha256=rpcXUONOmxCzTzUJZ4UxR5rk3AotiaHkCfqM5uJELjY,1876
|
|
734
736
|
datahub/metadata/schemas/BusinessAttributeInfo.avsc,sha256=5HpEN9ZP6qdkcIib01F9n54dHoNDrGrml_9o_ZO5JTc,22062
|
|
@@ -769,9 +771,9 @@ datahub/metadata/schemas/DataHubIngestionSourceInfo.avsc,sha256=4wac7sluRIq-0ZjO
|
|
|
769
771
|
datahub/metadata/schemas/DataHubIngestionSourceKey.avsc,sha256=TGmm9WEGTaABs7kt5Uc-N-kbc5Sd-2sQwx-JpfAptvw,545
|
|
770
772
|
datahub/metadata/schemas/DataHubOpenAPISchemaKey.avsc,sha256=q6ZyMoxInwmrkrXkUgMe-i-WZzAxbjcvJ-EI99SnEp8,599
|
|
771
773
|
datahub/metadata/schemas/DataHubPageModuleKey.avsc,sha256=NyFN8cVO6s6rtgoLGJJGfcPfpGr5PfmZlIhM6ajldfQ,460
|
|
772
|
-
datahub/metadata/schemas/DataHubPageModuleProperties.avsc,sha256=
|
|
774
|
+
datahub/metadata/schemas/DataHubPageModuleProperties.avsc,sha256=53Fj4ztBJqo9QMWuza2Kdtfpr2nTOTW0XuuXW77ugB8,10347
|
|
773
775
|
datahub/metadata/schemas/DataHubPageTemplateKey.avsc,sha256=0sVqwL97Rp8YHPytp2RqUP5hIW048hmT2hPNP5k6arc,472
|
|
774
|
-
datahub/metadata/schemas/DataHubPageTemplateProperties.avsc,sha256=
|
|
776
|
+
datahub/metadata/schemas/DataHubPageTemplateProperties.avsc,sha256=FyNcZIniQy9m6yN9DT4XsPkDrxUsU7tRTqmfdGoEtMU,8565
|
|
775
777
|
datahub/metadata/schemas/DataHubPersonaInfo.avsc,sha256=OUvbTgPQsBtzkDDb9pxHXpQ6A7dkL77ZnCXZ-MLEG14,227
|
|
776
778
|
datahub/metadata/schemas/DataHubPersonaKey.avsc,sha256=ddj-DhXa0_YMdLaGkKLLSklfIeDRvSwPXu8o__YEXUE,448
|
|
777
779
|
datahub/metadata/schemas/DataHubPolicyInfo.avsc,sha256=yBQe7pAuTMg9aovhugF4EkCRSHO_AN2TP_NM-0-Jg3A,10037
|
|
@@ -805,7 +807,7 @@ datahub/metadata/schemas/DataProcessInstanceProperties.avsc,sha256=2qsDFeSA2-ag5
|
|
|
805
807
|
datahub/metadata/schemas/DataProcessInstanceRelationships.avsc,sha256=VhBpnyGGvO06WEnM6zy4PmjiT0nivRQfkSdJCUgIavw,2358
|
|
806
808
|
datahub/metadata/schemas/DataProcessInstanceRunEvent.avsc,sha256=zwTYULEnpMbqwkLN8NbXW9PQWFG4X6TZkZwTQ1Wb53Y,6713
|
|
807
809
|
datahub/metadata/schemas/DataProcessKey.avsc,sha256=ZZE2HN4mwZtm_TJNcdohFS97WXytFq9HAs_-shor6sY,2518
|
|
808
|
-
datahub/metadata/schemas/DataProductKey.avsc,sha256=
|
|
810
|
+
datahub/metadata/schemas/DataProductKey.avsc,sha256=ECDUbxMwvmgI3tTRbO7UXncbNJUrqLkHTSDoMtBpYYY,684
|
|
809
811
|
datahub/metadata/schemas/DataProductProperties.avsc,sha256=Lc3duV7YMJLvo_RwckLbW4bbmPrhSS1D-bxVVboNX2c,6930
|
|
810
812
|
datahub/metadata/schemas/DataTransformLogic.avsc,sha256=nHTH6UzJ2Zz88N2aWa96hawLUR20HP7eSynfPtI1kzg,2111
|
|
811
813
|
datahub/metadata/schemas/DataTypeInfo.avsc,sha256=MCjzal71P8uIXZg161LrU8rZTJocZeizK-YxYA0Det0,704
|
|
@@ -821,7 +823,7 @@ datahub/metadata/schemas/DatasetUsageStatistics.avsc,sha256=jgF1u31kP1XAnnV2B0X8
|
|
|
821
823
|
datahub/metadata/schemas/Deprecation.avsc,sha256=p8SBIuKP3XVGeaBI7rROpLNACuoX8eMLRlZz8lGOYV8,1354
|
|
822
824
|
datahub/metadata/schemas/DisplayProperties.avsc,sha256=MTa_g2s0roxNFFggWU8rslUH3UFe3xe11uUXyh0Go_I,1732
|
|
823
825
|
datahub/metadata/schemas/Documentation.avsc,sha256=9vIJG9B08FFrC3y5c1XVaT5U3c-b5sOAc5foUxMnyCs,4836
|
|
824
|
-
datahub/metadata/schemas/DomainKey.avsc,sha256=
|
|
826
|
+
datahub/metadata/schemas/DomainKey.avsc,sha256=hDlX4jJTeAXW_VpKEhF8w5_tMZi7JRaVGqbvJL7GojE,699
|
|
825
827
|
datahub/metadata/schemas/DomainProperties.avsc,sha256=6do6wZ9G6gyt1QowQyi1xldqgdTXspb05FaqWpKJ6eM,3843
|
|
826
828
|
datahub/metadata/schemas/Domains.avsc,sha256=5mRQcba6Zmp6Y1srbxhOjETutg0I_ZG4ikuS2r9fkR0,804
|
|
827
829
|
datahub/metadata/schemas/DynamicFormAssignment.avsc,sha256=SXRL5D6kIYWdGl3zLQYxPnkQX71JXQOKrjQNavFqVp0,7339
|
|
@@ -859,15 +861,15 @@ datahub/metadata/schemas/GlobalSettingsInfo.avsc,sha256=HKZisxW2HNm6kwJUBaAVol7s
|
|
|
859
861
|
datahub/metadata/schemas/GlobalSettingsKey.avsc,sha256=Yj8s5IdM9yF7xrhJcLGCPCXBWqSsrPbufBaQjlZ3JlU,563
|
|
860
862
|
datahub/metadata/schemas/GlobalTags.avsc,sha256=-SurkodMqTDnPpkRV6qYqmpNWjQNvynUiPZX7EhL5uc,4624
|
|
861
863
|
datahub/metadata/schemas/GlossaryNodeInfo.avsc,sha256=G1Cb-w9VxIAEhNqyiEsDL_ABRO9QxyTpUANKU6DQrFw,1888
|
|
862
|
-
datahub/metadata/schemas/GlossaryNodeKey.avsc,sha256=
|
|
864
|
+
datahub/metadata/schemas/GlossaryNodeKey.avsc,sha256=KjdsiVm93SWXQdxllwE0GQei-NmvSatzrl9sQIA6TAE,687
|
|
863
865
|
datahub/metadata/schemas/GlossaryRelatedTerms.avsc,sha256=ZTP0mrFD4y-C6JekRy8IVuHvICUkJib-ZAYD93Gv1tA,2763
|
|
864
866
|
datahub/metadata/schemas/GlossaryTermInfo.avsc,sha256=j4s9NCyMOIF03HfaXoQEIkiMTRaCy_-euhenptfu7IA,2935
|
|
865
|
-
datahub/metadata/schemas/GlossaryTermKey.avsc,sha256=
|
|
867
|
+
datahub/metadata/schemas/GlossaryTermKey.avsc,sha256=_5Nf0kljjFVz0wNsE-BiIZvn6yL1iymaBQIptRKVwoU,881
|
|
866
868
|
datahub/metadata/schemas/GlossaryTerms.avsc,sha256=ogOFO6Hr2Xb7s1JHqxsCPY8r_qY_9kwu69k5-E3j2BM,7123
|
|
867
869
|
datahub/metadata/schemas/GroupMembership.avsc,sha256=wT3Hbpv2Z7V4X_-rIoed0cukAOMyYEL93udK8mMCjn0,557
|
|
868
870
|
datahub/metadata/schemas/IcebergCatalogInfo.avsc,sha256=X9Ejqzn1DyxNIth7vDhtPjGG4xMPQMhl7f-S7fBFxek,691
|
|
869
871
|
datahub/metadata/schemas/IcebergWarehouseInfo.avsc,sha256=0m7cQm8cCnBWNI5jGGgr5ZdOg66RQGWSf3gf8ay53So,2705
|
|
870
|
-
datahub/metadata/schemas/IncidentInfo.avsc,sha256=
|
|
872
|
+
datahub/metadata/schemas/IncidentInfo.avsc,sha256=bxw_OwxhTarm1TxY4R-Q12RFQBIjeoUGgvgoeRiNT04,12428
|
|
871
873
|
datahub/metadata/schemas/IncidentKey.avsc,sha256=Pip__DyNNTal7NxryM3kFi9qHlwntp1rIA8Al8Zz264,542
|
|
872
874
|
datahub/metadata/schemas/IncidentSource.avsc,sha256=lY_SarA3cM55KNENcB5z1Gu2MygxEl9l7R8LdMak9AQ,1199
|
|
873
875
|
datahub/metadata/schemas/IncidentsSummary.avsc,sha256=NTYp-6Oe92ALApbM3759TJ5pLXRArsSriIPq-f7w9vI,4514
|
|
@@ -929,7 +931,7 @@ datahub/metadata/schemas/SlackUserInfo.avsc,sha256=IY7InWaiDzJa3hJ9J4W3Eg8EUKuh0
|
|
|
929
931
|
datahub/metadata/schemas/SourceCode.avsc,sha256=tUgo2rczO5x1fxw3fYNWQj-51vRNmNIj38b1wayA0aQ,1370
|
|
930
932
|
datahub/metadata/schemas/Status.avsc,sha256=rPZSXSJdwnNywqNx2qll8cdt54aYgI-YUbRr3GK7h78,522
|
|
931
933
|
datahub/metadata/schemas/StructuredProperties.avsc,sha256=qe45sKZ9XrLcf15Gt03Ttzt2J_kJYHvN-DAOSErSYuY,7028
|
|
932
|
-
datahub/metadata/schemas/StructuredPropertyDefinition.avsc,sha256=
|
|
934
|
+
datahub/metadata/schemas/StructuredPropertyDefinition.avsc,sha256=a-6TaOQ4A7LDFLshmaFRBcXjz11p4vM0Q3X35GN4Zo0,11737
|
|
933
935
|
datahub/metadata/schemas/StructuredPropertyKey.avsc,sha256=lp7tQBgeriEU1YMQ6a4-6aUGSWDqNl00lLDym97j1yI,618
|
|
934
936
|
datahub/metadata/schemas/StructuredPropertySettings.avsc,sha256=EDNlXfT1TqogfulCanIc-nuYO9ZxRFOGzD9tl3ZJdB8,3732
|
|
935
937
|
datahub/metadata/schemas/SubTypes.avsc,sha256=bhXbzK020zDyQno97Xp05vmoMeZ82IGu2jz7pWDo3RQ,655
|
|
@@ -1004,8 +1006,8 @@ datahub/sql_parsing/split_statements.py,sha256=OIQXA9e4k3G9Z1y7rbgdtZhMWt4FPnq41
|
|
|
1004
1006
|
datahub/sql_parsing/sql_parsing_aggregator.py,sha256=kxxSVe3YNoz_T2OG6-F30ZuXNSXuBZ-E54RqObo6qTI,72323
|
|
1005
1007
|
datahub/sql_parsing/sql_parsing_common.py,sha256=cZ4WvVyHZuXDGjnBvKMX2_fz2EMextB5WQWcK0_saBo,3155
|
|
1006
1008
|
datahub/sql_parsing/sql_parsing_result_utils.py,sha256=prwWTj1EB2fRPv1eMB4EkpFNafIYAt-X8TIK0NWqank,796
|
|
1007
|
-
datahub/sql_parsing/sqlglot_lineage.py,sha256=
|
|
1008
|
-
datahub/sql_parsing/sqlglot_utils.py,sha256=
|
|
1009
|
+
datahub/sql_parsing/sqlglot_lineage.py,sha256=05ryqMkFK5_KDvr-ytXb1FyiC-SzaFB9_USnbL36LEg,66316
|
|
1010
|
+
datahub/sql_parsing/sqlglot_utils.py,sha256=zH8V9tAcSVO7Y8I3sIKPhs0D_9HzdNBlranBDmk1NB4,15454
|
|
1009
1011
|
datahub/sql_parsing/tool_meta_extractor.py,sha256=5JsLPcKjuXSrPGxNIhRvX72dFPmlV33-hyvhJwlWxCY,7543
|
|
1010
1012
|
datahub/telemetry/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
1011
1013
|
datahub/telemetry/stats.py,sha256=TwaQisQlD2Bk0uw__pP6u3Ovz9r-Ip4pCwpnto4r5e0,959
|
|
@@ -1112,8 +1114,8 @@ datahub_provider/operators/datahub_assertion_operator.py,sha256=uvTQ-jk2F0sbqqxp
|
|
|
1112
1114
|
datahub_provider/operators/datahub_assertion_sensor.py,sha256=lCBj_3x1cf5GMNpHdfkpHuyHfVxsm6ff5x2Z5iizcAo,140
|
|
1113
1115
|
datahub_provider/operators/datahub_operation_operator.py,sha256=aevDp2FzX7FxGlXrR0khoHNbxbhKR2qPEX5e8O2Jyzw,174
|
|
1114
1116
|
datahub_provider/operators/datahub_operation_sensor.py,sha256=8fcdVBCEPgqy1etTXgLoiHoJrRt_nzFZQMdSzHqSG7M,168
|
|
1115
|
-
acryl_datahub-1.2.0.
|
|
1116
|
-
acryl_datahub-1.2.0.
|
|
1117
|
-
acryl_datahub-1.2.0.
|
|
1118
|
-
acryl_datahub-1.2.0.
|
|
1119
|
-
acryl_datahub-1.2.0.
|
|
1117
|
+
acryl_datahub-1.2.0.8rc1.dist-info/METADATA,sha256=_TXC2AAKI66LHx6fTnBdVxsJBwedMdTKiIhnrBVSTQk,186651
|
|
1118
|
+
acryl_datahub-1.2.0.8rc1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
|
|
1119
|
+
acryl_datahub-1.2.0.8rc1.dist-info/entry_points.txt,sha256=qopCAD6qrsijaZ9mTw3UlPCKsE00C3t9MbkkWow7pi4,9943
|
|
1120
|
+
acryl_datahub-1.2.0.8rc1.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
|
|
1121
|
+
acryl_datahub-1.2.0.8rc1.dist-info/RECORD,,
|
datahub/_version.py
CHANGED
datahub/cli/delete_cli.py
CHANGED
datahub/ingestion/api/report.py
CHANGED
|
@@ -204,6 +204,7 @@ class ExamplesReport(Report, Closeable):
|
|
|
204
204
|
samples: Dict[str, Dict[str, List[str]]] = field(
|
|
205
205
|
default_factory=lambda: defaultdict(lambda: defaultdict(list))
|
|
206
206
|
)
|
|
207
|
+
compute_stats_time_seconds: float = 0.0
|
|
207
208
|
_file_based_dict: Optional[FileBackedDict[SourceReportSubtypes]] = None
|
|
208
209
|
|
|
209
210
|
# We are adding this to make querying easier for fine-grained lineage
|
|
@@ -405,6 +406,7 @@ class ExamplesReport(Report, Closeable):
|
|
|
405
406
|
self._update_file_based_dict(urn, entityType, aspectName, mcp)
|
|
406
407
|
|
|
407
408
|
def compute_stats(self) -> None:
|
|
409
|
+
start_time = datetime.now()
|
|
408
410
|
if self._file_based_dict is None:
|
|
409
411
|
return
|
|
410
412
|
|
|
@@ -466,6 +468,8 @@ class ExamplesReport(Report, Closeable):
|
|
|
466
468
|
list(self._lineage_aspects_seen), "lineage"
|
|
467
469
|
)
|
|
468
470
|
self._collect_samples_with_all_conditions("all_3")
|
|
471
|
+
end_time = datetime.now()
|
|
472
|
+
self.compute_stats_time_seconds += (end_time - start_time).total_seconds()
|
|
469
473
|
|
|
470
474
|
|
|
471
475
|
class EntityFilterReport(ReportAttribute):
|
|
@@ -2678,7 +2678,7 @@
|
|
|
2678
2678
|
},
|
|
2679
2679
|
{
|
|
2680
2680
|
"capability": "USAGE_STATS",
|
|
2681
|
-
"description": "
|
|
2681
|
+
"description": "Optionally enabled via `include_usage_statistics`",
|
|
2682
2682
|
"subtype_modifier": null,
|
|
2683
2683
|
"supported": true
|
|
2684
2684
|
},
|
|
@@ -971,7 +971,8 @@ class DataHubGraph(DatahubRestEmitter, EntityVersioningAPI):
|
|
|
971
971
|
$orFilters: [AndFilterInput!],
|
|
972
972
|
$batchSize: Int!,
|
|
973
973
|
$scrollId: String,
|
|
974
|
-
$skipCache: Boolean
|
|
974
|
+
$skipCache: Boolean!,
|
|
975
|
+
$includeSoftDeleted: Boolean) {
|
|
975
976
|
|
|
976
977
|
scrollAcrossEntities(input: {
|
|
977
978
|
query: $query,
|
|
@@ -983,6 +984,7 @@ class DataHubGraph(DatahubRestEmitter, EntityVersioningAPI):
|
|
|
983
984
|
skipHighlighting: true
|
|
984
985
|
skipAggregates: true
|
|
985
986
|
skipCache: $skipCache
|
|
987
|
+
includeSoftDeleted: $includeSoftDeleted
|
|
986
988
|
}
|
|
987
989
|
}) {
|
|
988
990
|
nextScrollId
|
|
@@ -1002,6 +1004,11 @@ class DataHubGraph(DatahubRestEmitter, EntityVersioningAPI):
|
|
|
1002
1004
|
"orFilters": orFilters,
|
|
1003
1005
|
"batchSize": batch_size,
|
|
1004
1006
|
"skipCache": skip_cache,
|
|
1007
|
+
"includeSoftDeleted": (
|
|
1008
|
+
None
|
|
1009
|
+
if status is None
|
|
1010
|
+
else status != RemovedStatusFilter.NOT_SOFT_DELETED
|
|
1011
|
+
),
|
|
1005
1012
|
}
|
|
1006
1013
|
|
|
1007
1014
|
for entity in self._scroll_across_entities(graphql_query, variables):
|
|
@@ -129,6 +129,10 @@ class DataHubSourceConfig(StatefulIngestionConfigBase):
|
|
|
129
129
|
description="Timeout for each query in seconds. ",
|
|
130
130
|
)
|
|
131
131
|
|
|
132
|
+
preserve_system_metadata: bool = Field(
|
|
133
|
+
default=True, description="Copy system metadata from the source system"
|
|
134
|
+
)
|
|
135
|
+
|
|
132
136
|
@root_validator(skip_on_failure=True)
|
|
133
137
|
def check_ingesting_data(cls, values):
|
|
134
138
|
if (
|
|
@@ -380,7 +380,12 @@ class DataHubDatabaseReader:
|
|
|
380
380
|
json_metadata = post_json_transform(
|
|
381
381
|
json.loads(row["systemmetadata"] or "{}")
|
|
382
382
|
)
|
|
383
|
-
system_metadata =
|
|
383
|
+
system_metadata = None
|
|
384
|
+
if self.config.preserve_system_metadata:
|
|
385
|
+
system_metadata = SystemMetadataClass.from_obj(json_metadata)
|
|
386
|
+
if system_metadata.properties:
|
|
387
|
+
is_no_op = system_metadata.properties.pop("isNoOp", None)
|
|
388
|
+
logger.debug(f"Removed potential value for is_no_op={is_no_op}")
|
|
384
389
|
return MetadataChangeProposalWrapper(
|
|
385
390
|
entityUrn=row["urn"],
|
|
386
391
|
aspect=ASPECT_MAP[row["aspect"]].from_obj(json_aspect),
|
|
@@ -37,9 +37,9 @@ from datahub.ingestion.api.source_helpers import (
|
|
|
37
37
|
from datahub.ingestion.api.workunit import MetadataWorkUnit
|
|
38
38
|
from datahub.ingestion.graph.client import get_default_graph
|
|
39
39
|
from datahub.ingestion.graph.config import ClientMode
|
|
40
|
-
from datahub.metadata.
|
|
41
|
-
|
|
42
|
-
|
|
40
|
+
from datahub.metadata.schema_classes import (
|
|
41
|
+
FineGrainedLineageDownstreamTypeClass,
|
|
42
|
+
FineGrainedLineageUpstreamTypeClass,
|
|
43
43
|
)
|
|
44
44
|
|
|
45
45
|
logger = logging.getLogger(__name__)
|
|
@@ -80,9 +80,9 @@ class FineGrainedLineageConfig(ConfigModel):
|
|
|
80
80
|
@validator("upstreamType")
|
|
81
81
|
def upstream_type_must_be_supported(cls, v: str) -> str:
|
|
82
82
|
allowed_types = [
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
|
|
83
|
+
FineGrainedLineageUpstreamTypeClass.FIELD_SET,
|
|
84
|
+
FineGrainedLineageUpstreamTypeClass.DATASET,
|
|
85
|
+
FineGrainedLineageUpstreamTypeClass.NONE,
|
|
86
86
|
]
|
|
87
87
|
if v not in allowed_types:
|
|
88
88
|
raise ValueError(
|
|
@@ -93,8 +93,8 @@ class FineGrainedLineageConfig(ConfigModel):
|
|
|
93
93
|
@validator("downstreamType")
|
|
94
94
|
def downstream_type_must_be_supported(cls, v: str) -> str:
|
|
95
95
|
allowed_types = [
|
|
96
|
-
|
|
97
|
-
|
|
96
|
+
FineGrainedLineageDownstreamTypeClass.FIELD_SET,
|
|
97
|
+
FineGrainedLineageDownstreamTypeClass.FIELD,
|
|
98
98
|
]
|
|
99
99
|
if v not in allowed_types:
|
|
100
100
|
raise ValueError(
|
|
@@ -143,7 +143,7 @@ logger: logging.Logger = logging.getLogger(__name__)
|
|
|
143
143
|
@capability(SourceCapability.SCHEMA_METADATA, "Enabled by default")
|
|
144
144
|
@capability(
|
|
145
145
|
SourceCapability.USAGE_STATS,
|
|
146
|
-
"
|
|
146
|
+
"Optionally enabled via `include_usage_statistics`",
|
|
147
147
|
)
|
|
148
148
|
@capability(
|
|
149
149
|
SourceCapability.DELETION_DETECTION, "Enabled by default via stateful ingestion"
|
|
@@ -73,6 +73,11 @@ except ImportError:
|
|
|
73
73
|
|
|
74
74
|
logger = logging.getLogger(__name__)
|
|
75
75
|
|
|
76
|
+
# Precompiled regex for SQL identifier validation
|
|
77
|
+
# Athena identifiers can only contain lowercase letters, numbers, underscore, and period (for complex types)
|
|
78
|
+
# Note: Athena automatically converts uppercase to lowercase, but we're being strict for security
|
|
79
|
+
_IDENTIFIER_PATTERN = re.compile(r"^[a-zA-Z0-9_.]+$")
|
|
80
|
+
|
|
76
81
|
assert STRUCT, "required type modules are not available"
|
|
77
82
|
register_custom_type(STRUCT, RecordTypeClass)
|
|
78
83
|
register_custom_type(MapType, MapTypeClass)
|
|
@@ -510,20 +515,76 @@ class AthenaSource(SQLAlchemySource):
|
|
|
510
515
|
return [schema for schema in schemas if schema == athena_config.database]
|
|
511
516
|
return schemas
|
|
512
517
|
|
|
518
|
+
@classmethod
|
|
519
|
+
def _sanitize_identifier(cls, identifier: str) -> str:
|
|
520
|
+
"""Sanitize SQL identifiers to prevent injection attacks.
|
|
521
|
+
|
|
522
|
+
Args:
|
|
523
|
+
identifier: The SQL identifier to sanitize
|
|
524
|
+
|
|
525
|
+
Returns:
|
|
526
|
+
Sanitized identifier safe for SQL queries
|
|
527
|
+
|
|
528
|
+
Raises:
|
|
529
|
+
ValueError: If identifier contains unsafe characters
|
|
530
|
+
"""
|
|
531
|
+
if not identifier:
|
|
532
|
+
raise ValueError("Identifier cannot be empty")
|
|
533
|
+
|
|
534
|
+
# Allow only alphanumeric characters, underscores, and periods for identifiers
|
|
535
|
+
# This matches Athena's identifier naming rules
|
|
536
|
+
if not _IDENTIFIER_PATTERN.match(identifier):
|
|
537
|
+
raise ValueError(
|
|
538
|
+
f"Identifier '{identifier}' contains unsafe characters. Only alphanumeric characters, underscores, and periods are allowed."
|
|
539
|
+
)
|
|
540
|
+
|
|
541
|
+
return identifier
|
|
542
|
+
|
|
513
543
|
@classmethod
|
|
514
544
|
def _casted_partition_key(cls, key: str) -> str:
|
|
515
545
|
# We need to cast the partition keys to a VARCHAR, since otherwise
|
|
516
546
|
# Athena may throw an error during concatenation / comparison.
|
|
517
|
-
|
|
547
|
+
sanitized_key = cls._sanitize_identifier(key)
|
|
548
|
+
return f"CAST({sanitized_key} as VARCHAR)"
|
|
549
|
+
|
|
550
|
+
@classmethod
|
|
551
|
+
def _build_max_partition_query(
|
|
552
|
+
cls, schema: str, table: str, partitions: List[str]
|
|
553
|
+
) -> str:
|
|
554
|
+
"""Build SQL query to find the row with maximum partition values.
|
|
555
|
+
|
|
556
|
+
Args:
|
|
557
|
+
schema: Database schema name
|
|
558
|
+
table: Table name
|
|
559
|
+
partitions: List of partition column names
|
|
560
|
+
|
|
561
|
+
Returns:
|
|
562
|
+
SQL query string to find the maximum partition
|
|
563
|
+
|
|
564
|
+
Raises:
|
|
565
|
+
ValueError: If any identifier contains unsafe characters
|
|
566
|
+
"""
|
|
567
|
+
# Sanitize all identifiers to prevent SQL injection
|
|
568
|
+
sanitized_schema = cls._sanitize_identifier(schema)
|
|
569
|
+
sanitized_table = cls._sanitize_identifier(table)
|
|
570
|
+
sanitized_partitions = [
|
|
571
|
+
cls._sanitize_identifier(partition) for partition in partitions
|
|
572
|
+
]
|
|
573
|
+
|
|
574
|
+
casted_keys = [cls._casted_partition_key(key) for key in partitions]
|
|
575
|
+
if len(casted_keys) == 1:
|
|
576
|
+
part_concat = casted_keys[0]
|
|
577
|
+
else:
|
|
578
|
+
separator = "CAST('-' AS VARCHAR)"
|
|
579
|
+
part_concat = f"CONCAT({f', {separator}, '.join(casted_keys)})"
|
|
580
|
+
|
|
581
|
+
return f'select {",".join(sanitized_partitions)} from "{sanitized_schema}"."{sanitized_table}$partitions" where {part_concat} = (select max({part_concat}) from "{sanitized_schema}"."{sanitized_table}$partitions")'
|
|
518
582
|
|
|
519
583
|
@override
|
|
520
584
|
def get_partitions(
|
|
521
585
|
self, inspector: Inspector, schema: str, table: str
|
|
522
586
|
) -> Optional[List[str]]:
|
|
523
|
-
if
|
|
524
|
-
not self.config.extract_partitions
|
|
525
|
-
and not self.config.extract_partitions_using_create_statements
|
|
526
|
-
):
|
|
587
|
+
if not self.config.extract_partitions:
|
|
527
588
|
return None
|
|
528
589
|
|
|
529
590
|
if not self.cursor:
|
|
@@ -557,11 +618,9 @@ class AthenaSource(SQLAlchemySource):
|
|
|
557
618
|
context=f"{schema}.{table}",
|
|
558
619
|
level=StructuredLogLevel.WARN,
|
|
559
620
|
):
|
|
560
|
-
|
|
561
|
-
|
|
562
|
-
self._casted_partition_key(key) for key in partitions
|
|
621
|
+
max_partition_query = self._build_max_partition_query(
|
|
622
|
+
schema, table, partitions
|
|
563
623
|
)
|
|
564
|
-
max_partition_query = f'select {",".join(partitions)} from "{schema}"."{table}$partitions" where {part_concat} = (select max({part_concat}) from "{schema}"."{table}$partitions")'
|
|
565
624
|
ret = self.cursor.execute(max_partition_query)
|
|
566
625
|
max_partition: Dict[str, str] = {}
|
|
567
626
|
if ret:
|
|
@@ -678,16 +737,34 @@ class AthenaSource(SQLAlchemySource):
|
|
|
678
737
|
).get(table, None)
|
|
679
738
|
|
|
680
739
|
if partition and partition.max_partition:
|
|
681
|
-
|
|
682
|
-
|
|
683
|
-
|
|
684
|
-
|
|
740
|
+
try:
|
|
741
|
+
# Sanitize identifiers to prevent SQL injection
|
|
742
|
+
sanitized_schema = self._sanitize_identifier(schema)
|
|
743
|
+
sanitized_table = self._sanitize_identifier(table)
|
|
744
|
+
|
|
745
|
+
max_partition_filters = []
|
|
746
|
+
for key, value in partition.max_partition.items():
|
|
747
|
+
# Sanitize partition key and properly escape the value
|
|
748
|
+
sanitized_key = self._sanitize_identifier(key)
|
|
749
|
+
# Escape single quotes in the value to prevent injection
|
|
750
|
+
escaped_value = value.replace("'", "''") if value else ""
|
|
751
|
+
max_partition_filters.append(
|
|
752
|
+
f"{self._casted_partition_key(sanitized_key)} = '{escaped_value}'"
|
|
753
|
+
)
|
|
754
|
+
max_partition = str(partition.max_partition)
|
|
755
|
+
return (
|
|
756
|
+
max_partition,
|
|
757
|
+
f'SELECT * FROM "{sanitized_schema}"."{sanitized_table}" WHERE {" AND ".join(max_partition_filters)}',
|
|
685
758
|
)
|
|
686
|
-
|
|
687
|
-
|
|
688
|
-
|
|
689
|
-
|
|
690
|
-
|
|
759
|
+
except ValueError as e:
|
|
760
|
+
# If sanitization fails due to malicious identifiers,
|
|
761
|
+
# return None to disable partition profiling for this table
|
|
762
|
+
# rather than crashing the entire ingestion
|
|
763
|
+
logger.warning(
|
|
764
|
+
f"Failed to generate partition profiler query for {schema}.{table} due to unsafe identifiers: {e}. "
|
|
765
|
+
f"Partition profiling disabled for this table."
|
|
766
|
+
)
|
|
767
|
+
return None, None
|
|
691
768
|
return None, None
|
|
692
769
|
|
|
693
770
|
def close(self):
|