acryl-datahub 0.15.0.2rc3__py3-none-any.whl → 0.15.0.2rc5__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-0.15.0.2rc3.dist-info → acryl_datahub-0.15.0.2rc5.dist-info}/METADATA +2460 -2460
- {acryl_datahub-0.15.0.2rc3.dist-info → acryl_datahub-0.15.0.2rc5.dist-info}/RECORD +58 -54
- datahub/__init__.py +1 -1
- datahub/cli/delete_cli.py +3 -3
- datahub/cli/migrate.py +2 -2
- datahub/emitter/mcp_builder.py +27 -0
- datahub/emitter/rest_emitter.py +1 -1
- datahub/ingestion/api/source.py +2 -2
- datahub/ingestion/graph/client.py +4 -2
- datahub/ingestion/source/aws/glue.py +14 -1
- datahub/ingestion/source/aws/s3_util.py +24 -1
- datahub/ingestion/source/delta_lake/source.py +0 -5
- datahub/ingestion/source/demo_data.py +1 -1
- datahub/ingestion/source/fivetran/fivetran.py +1 -6
- datahub/ingestion/source/gc/execution_request_cleanup.py +31 -6
- datahub/ingestion/source/gc/soft_deleted_entity_cleanup.py +14 -1
- datahub/ingestion/source/iceberg/iceberg.py +10 -3
- datahub/ingestion/source/iceberg/iceberg_common.py +49 -9
- datahub/ingestion/source/iceberg/iceberg_profiler.py +3 -1
- datahub/ingestion/source/kafka_connect/kafka_connect.py +1 -6
- datahub/ingestion/source/metabase.py +1 -6
- datahub/ingestion/source/mlflow.py +0 -5
- datahub/ingestion/source/nifi.py +0 -5
- datahub/ingestion/source/powerbi_report_server/report_server.py +1 -1
- datahub/ingestion/source/redash.py +0 -5
- datahub/ingestion/source/redshift/redshift.py +1 -0
- datahub/ingestion/source/s3/source.py +10 -14
- datahub/ingestion/source/snowflake/snowflake_config.py +13 -0
- datahub/ingestion/source/snowflake/snowflake_schema.py +5 -2
- datahub/ingestion/source/snowflake/snowflake_schema_gen.py +112 -20
- datahub/ingestion/source/snowflake/snowflake_tag.py +14 -4
- datahub/ingestion/source/snowflake/snowflake_v2.py +0 -6
- datahub/ingestion/source/sql/sql_types.py +1 -1
- datahub/ingestion/source/sql/sql_utils.py +5 -0
- datahub/ingestion/source/superset.py +1 -6
- datahub/ingestion/source/tableau/tableau.py +0 -6
- datahub/metadata/_schema_classes.py +316 -43
- datahub/metadata/_urns/urn_defs.py +69 -15
- datahub/metadata/com/linkedin/pegasus2avro/common/__init__.py +2 -0
- datahub/metadata/com/linkedin/pegasus2avro/metadata/key/__init__.py +2 -0
- datahub/metadata/com/linkedin/pegasus2avro/versionset/__init__.py +17 -0
- datahub/metadata/schema.avsc +296 -87
- datahub/metadata/schemas/DataFlowKey.avsc +1 -0
- datahub/metadata/schemas/DataJobKey.avsc +1 -0
- datahub/metadata/schemas/DatasetKey.avsc +2 -1
- datahub/metadata/schemas/MLFeatureProperties.avsc +51 -0
- datahub/metadata/schemas/MLModelDeploymentProperties.avsc +51 -0
- datahub/metadata/schemas/MLModelGroupProperties.avsc +96 -23
- datahub/metadata/schemas/MLModelKey.avsc +2 -1
- datahub/metadata/schemas/MLModelProperties.avsc +96 -48
- datahub/metadata/schemas/MLPrimaryKeyProperties.avsc +51 -0
- datahub/metadata/schemas/MetadataChangeEvent.avsc +98 -71
- datahub/metadata/schemas/VersionProperties.avsc +216 -0
- datahub/metadata/schemas/VersionSetKey.avsc +26 -0
- datahub/metadata/schemas/VersionSetProperties.avsc +49 -0
- {acryl_datahub-0.15.0.2rc3.dist-info → acryl_datahub-0.15.0.2rc5.dist-info}/WHEEL +0 -0
- {acryl_datahub-0.15.0.2rc3.dist-info → acryl_datahub-0.15.0.2rc5.dist-info}/entry_points.txt +0 -0
- {acryl_datahub-0.15.0.2rc3.dist-info → acryl_datahub-0.15.0.2rc5.dist-info}/top_level.txt +0 -0
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
datahub/__init__.py,sha256=
|
|
1
|
+
datahub/__init__.py,sha256=ytRckpG4lVjyN5pUIlOJTPXvNHOEyI-zqLb7oyer--A,576
|
|
2
2
|
datahub/__main__.py,sha256=pegIvQ9hzK7IhqVeUi1MeADSZ2QlP-D3K0OQdEg55RU,106
|
|
3
3
|
datahub/entrypoints.py,sha256=3-qSfXAx3Z0FEkBV5tlO8fQr4xk4ySeDRMVTpS5Xd6A,7793
|
|
4
4
|
datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
@@ -61,7 +61,7 @@ datahub/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
|
61
61
|
datahub/cli/check_cli.py,sha256=9dXNyzZayHeoFjwFjLkMVyx6DiCZfeESyI-sYtGA6bE,12850
|
|
62
62
|
datahub/cli/cli_utils.py,sha256=d_Q9vPZTPxO7XyyghD-i1Nkr4DX0M8cs2IWrMUQAu0c,13539
|
|
63
63
|
datahub/cli/config_utils.py,sha256=yuXw7RzpRY5x_-MAoqWbv46qUkIeRNAJL4_OeJpYdBE,4879
|
|
64
|
-
datahub/cli/delete_cli.py,sha256=
|
|
64
|
+
datahub/cli/delete_cli.py,sha256=rJpyQhuRb_BnA1Fyot3Yu6-_x_3CoSsgjrpyJJCwEJY,23050
|
|
65
65
|
datahub/cli/docker_check.py,sha256=rED4wHXqxcQ_qNFyIgFEZ85BHT9ZTE5YC-oUKqbRqi0,9432
|
|
66
66
|
datahub/cli/docker_cli.py,sha256=QGoWFp8ZZsXOSMbgu0Q4snMmMmtP3epWAN-fYglUNEc,36491
|
|
67
67
|
datahub/cli/env_utils.py,sha256=RQzjg4JE29hjPt4v7p-RuqoOr99w8E3DBHWiN2Sm7T4,252
|
|
@@ -70,7 +70,7 @@ datahub/cli/get_cli.py,sha256=VV80BCXfZ0-C8fr2k43SIuN9DB-fOYP9StWsTHnXwFw,2327
|
|
|
70
70
|
datahub/cli/ingest_cli.py,sha256=nRoZvVpsGPXmEZCvSOBfsZ61Ep1fCqYRVp79RBnHSnI,22393
|
|
71
71
|
datahub/cli/json_file.py,sha256=nWo-VVthaaW4Do1eUqgrzk0fShb29MjiKXvZVOTq76c,943
|
|
72
72
|
datahub/cli/lite_cli.py,sha256=UmlMMquce6lHiPaKUBBT0XQtqR9SHEmrGlJyKV9YY60,13030
|
|
73
|
-
datahub/cli/migrate.py,sha256=
|
|
73
|
+
datahub/cli/migrate.py,sha256=1ngS-jT59v-a9AnfSB44mN0mBbzhGqIyteG142Ui77c,17937
|
|
74
74
|
datahub/cli/migration_utils.py,sha256=0qHo_9eSR4buyV_K_tdcHSLBufKphBWwwwT1iK_I4S8,9382
|
|
75
75
|
datahub/cli/put_cli.py,sha256=4ol9aLdidX1VXjVxMG2tkfEMPyjLpgOk2pfl0Gvb8iU,3841
|
|
76
76
|
datahub/cli/quickstart_versioning.py,sha256=MyWvw92s4b84wIEizjSUZjoMClwLbhpgMdHeDav-x2o,5713
|
|
@@ -116,10 +116,10 @@ datahub/emitter/generic_emitter.py,sha256=i37ZFm9VR_tmiZm9kIypEkQEB_cLKbzj_tJvVi
|
|
|
116
116
|
datahub/emitter/kafka_emitter.py,sha256=Uix1W1WaXF8VqUTUfzdRZKca2XrR1w50Anx2LVkROlc,5822
|
|
117
117
|
datahub/emitter/mce_builder.py,sha256=IqHOm0cpzdVC_mQOqk0yEVJUEj9xn8am2OFAwwQeX_8,16342
|
|
118
118
|
datahub/emitter/mcp.py,sha256=hAAYziDdkwjazQU0DtWMbQWY8wS09ACrKJbqxoWXdgc,9637
|
|
119
|
-
datahub/emitter/mcp_builder.py,sha256=
|
|
119
|
+
datahub/emitter/mcp_builder.py,sha256=AHSeMfcFxvJl2PXyDQ5HsnWbk6HkJqUtppNKjQtIbUI,10791
|
|
120
120
|
datahub/emitter/mcp_patch_builder.py,sha256=oonC8iGOvDzqj890CxOjWlBdDEF1RnwvbSZy1sivlTY,4572
|
|
121
121
|
datahub/emitter/request_helper.py,sha256=33ORG3S3OVy97_jlWBRn7yUM5XCIkRN6WSdJvN7Ofcg,670
|
|
122
|
-
datahub/emitter/rest_emitter.py,sha256=
|
|
122
|
+
datahub/emitter/rest_emitter.py,sha256=v-A4eR_GSbXg-dsUgHAMcUd68qNEF5KO2MYlyhAYn8I,17880
|
|
123
123
|
datahub/emitter/serialization_helper.py,sha256=q12Avmf70Vy4ttQGMJoTKlE5EsybMKNg2w3MQeZiHvk,3652
|
|
124
124
|
datahub/emitter/sql_parsing_builder.py,sha256=Cr5imZrm3dYDSCACt5MFscgHCtVbHTD6IjUmsvsKoEs,11991
|
|
125
125
|
datahub/emitter/synchronized_file_emitter.py,sha256=s4ATuxalI4GDAkrZTaGSegxBdvvNPZ9jRSdtElU0kNs,1805
|
|
@@ -138,7 +138,7 @@ datahub/ingestion/api/registry.py,sha256=LGElUdzhNQoEr-k2SN23mJaIYnA1PYfF97LQxBm
|
|
|
138
138
|
datahub/ingestion/api/report.py,sha256=zb5Y_9ogmWm00KqX7_64sIMT24Wfpk7txRwEfKacw5I,4652
|
|
139
139
|
datahub/ingestion/api/report_helpers.py,sha256=WbUC1kQeaKqIagGV3XzfPmPs7slAT1mfNY4og2BH2A8,994
|
|
140
140
|
datahub/ingestion/api/sink.py,sha256=3jw7-x9gXGreOPwn49wG5fT3C8pYhaNMQITdMN6kbag,4478
|
|
141
|
-
datahub/ingestion/api/source.py,sha256=
|
|
141
|
+
datahub/ingestion/api/source.py,sha256=yLx_7TCyhflo0hloYzC4y2ovh3TWEVcDh1agvh8AQwI,19036
|
|
142
142
|
datahub/ingestion/api/source_helpers.py,sha256=AVO0ogiCKgYmX1ubJaSs6L30TCCgOIalp6awXPF5XM0,19643
|
|
143
143
|
datahub/ingestion/api/transform.py,sha256=X0GpjMJzYkLuZx8MTWxH50cWGm9rGsnn3k188mmC8J8,582
|
|
144
144
|
datahub/ingestion/api/workunit.py,sha256=e8n8RfSjHZZm2R4ShNH0UuMtUkMjyqqM2j2t7oL74lo,6327
|
|
@@ -164,7 +164,7 @@ datahub/ingestion/glossary/classifier.py,sha256=zp8Fe3he80H5Zz1EwymKjThUPkTpw6Pg
|
|
|
164
164
|
datahub/ingestion/glossary/classifier_registry.py,sha256=yFOYLQhDgCLqXYMG3L1BquXafeLcZDcmp8meyw6k9ts,307
|
|
165
165
|
datahub/ingestion/glossary/datahub_classifier.py,sha256=8VhwuLDhyOqqOr0jqAPIgorb4eAOnvTr4m13Y2Wy1-E,7515
|
|
166
166
|
datahub/ingestion/graph/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
167
|
-
datahub/ingestion/graph/client.py,sha256=
|
|
167
|
+
datahub/ingestion/graph/client.py,sha256=tXAzgeUnNfawgKDwKZsAuSDP0zAkhTKQb07Y3gNksgY,65056
|
|
168
168
|
datahub/ingestion/graph/config.py,sha256=_oha8Je7P80ZmrkZUAaRHyYbdMmTkMI5JkYjEP2Ri1Q,751
|
|
169
169
|
datahub/ingestion/graph/connections.py,sha256=9462L0ZWGKURyypAln25eMPhK3pcufBar9tNDoqspXs,741
|
|
170
170
|
datahub/ingestion/graph/filters.py,sha256=UeUZQHoimavIYx-jXLA0WGkOUe10TaO8uEZkfa-QgNE,6188
|
|
@@ -187,7 +187,7 @@ datahub/ingestion/sink/sink_registry.py,sha256=JRBWx8qEYg0ubSTyhqwgSWctgxwyp6fva
|
|
|
187
187
|
datahub/ingestion/source/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
188
188
|
datahub/ingestion/source/confluent_schema_registry.py,sha256=_h9D8bUXoaGcwgwB94dX6aTyLY5ve7XGdcVFSJHGSJc,18804
|
|
189
189
|
datahub/ingestion/source/csv_enricher.py,sha256=AIxQFkmSzFgCa_Fzt2EiFMyojQMFKmnPt878WypSPa4,29491
|
|
190
|
-
datahub/ingestion/source/demo_data.py,sha256=
|
|
190
|
+
datahub/ingestion/source/demo_data.py,sha256=PbtCHlZx3wrKlOPPgkWhDQuPm7ZfIx2neXJUzbUi9YY,1305
|
|
191
191
|
datahub/ingestion/source/elastic_search.py,sha256=uT4I0GyqSiD16BURqsXWyPN9wNBc3wLomz1nG-OxHec,22634
|
|
192
192
|
datahub/ingestion/source/feast.py,sha256=uZpeUkJsiNlvZcUkARiEuZT_3n6sbGc0yFzwqhtnefA,18103
|
|
193
193
|
datahub/ingestion/source/file.py,sha256=pH-Qkjh5FQ2XvyYPE7Z8XEY4vUk_SUHxm8p8IxG12tU,15879
|
|
@@ -195,20 +195,20 @@ datahub/ingestion/source/ge_data_profiler.py,sha256=7-ciHphLU8O259OU2WMDfCDpoqvD
|
|
|
195
195
|
datahub/ingestion/source/ge_profiling_config.py,sha256=P-9pd20koFvpxeEL_pqFvKWWz-qnpZ6XkELUyBKr7is,10807
|
|
196
196
|
datahub/ingestion/source/glue_profiling_config.py,sha256=vpMJH4Lf_qgR32BZy58suabri1yV5geaAPjzg2eORDc,2559
|
|
197
197
|
datahub/ingestion/source/ldap.py,sha256=Vnzg8tpwBYeyM-KBVVsUJvGZGBMJiCJ_i_FhxaFRQ9A,18627
|
|
198
|
-
datahub/ingestion/source/metabase.py,sha256=
|
|
199
|
-
datahub/ingestion/source/mlflow.py,sha256
|
|
198
|
+
datahub/ingestion/source/metabase.py,sha256=m9Gfhrs8F1z23ci8CIxdE5cW--25stgxg_IQTKwkFrk,31532
|
|
199
|
+
datahub/ingestion/source/mlflow.py,sha256=IPG_l2HH9Ec8wxu0MYb3QaOPmw1kB1gcS3t4wf9bZLs,12134
|
|
200
200
|
datahub/ingestion/source/mode.py,sha256=cq1KIpLxuplETF7sUW0hoMQIZG1cgga5BGHP54a28wE,63467
|
|
201
201
|
datahub/ingestion/source/mongodb.py,sha256=vZue4Nz0xaBoCUsQr3_0OIRkWRxeE_IH_Y_QKZ1s7S0,21077
|
|
202
|
-
datahub/ingestion/source/nifi.py,sha256=
|
|
202
|
+
datahub/ingestion/source/nifi.py,sha256=ODXmZRxGq5V0R6PCYyy-_-dDWOb-cCkPzAVf-38-ACM,55965
|
|
203
203
|
datahub/ingestion/source/openapi.py,sha256=3ea2ORz1cuq4e7L2hSjxG9Cw3__pVoJ5UNYTJS3EnKU,17386
|
|
204
204
|
datahub/ingestion/source/openapi_parser.py,sha256=1_68wHWe_SzWYEyC1YVDw9vxoadKjW1yv8DecvyIhwY,13606
|
|
205
205
|
datahub/ingestion/source/preset.py,sha256=fByqamRLnXxsfCGdLPzWN_5LJR_s2_G2f_zwSKUc8EA,3981
|
|
206
206
|
datahub/ingestion/source/pulsar.py,sha256=7rTOEqYmeOuRZl5DG8d5OFkb4l9H6-1bETZfa-4DfmI,20163
|
|
207
|
-
datahub/ingestion/source/redash.py,sha256=
|
|
207
|
+
datahub/ingestion/source/redash.py,sha256=GH0MGV_huvKio9hMQ-jKdYIxYcHN6WnivrhWCw3I03E,29880
|
|
208
208
|
datahub/ingestion/source/salesforce.py,sha256=S6LSM6mzl8-zKbrJPoINhM1SCpYfM244Xb74pbEI-J0,31792
|
|
209
209
|
datahub/ingestion/source/source_registry.py,sha256=a2mLjJPLkSI-gYCTb_7U7Jo4D8jGknNQ_yScPIihXFk,1208
|
|
210
210
|
datahub/ingestion/source/sql_queries.py,sha256=Ip7UZub7fgMh7P5jL_zJPY7lSkc9GGTy8GJ8lqZrcsE,9502
|
|
211
|
-
datahub/ingestion/source/superset.py,sha256
|
|
211
|
+
datahub/ingestion/source/superset.py,sha256=-_90rfZtKG5vf5OSFS8lhqI-nGGtKPRwYYNAS_m1xmY,24592
|
|
212
212
|
datahub/ingestion/source/abs/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
213
213
|
datahub/ingestion/source/abs/config.py,sha256=Doecl1mA6JshJTNar7oTVR7wnWl4gMu64MBHp3hIVJc,6737
|
|
214
214
|
datahub/ingestion/source/abs/datalake_profiler_config.py,sha256=qh38q-Zw8TUTZD5RF0_hSoEfR6BilNGXyKPRsq1KQKE,3600
|
|
@@ -217,9 +217,9 @@ datahub/ingestion/source/abs/report.py,sha256=fzkTdTewYlWrTk4f2Cyl-e8RV4qw9wEVtm
|
|
|
217
217
|
datahub/ingestion/source/abs/source.py,sha256=pzxW-R_cWGKPneEhX8JWdTZiX2k1kAZOPKgMxp9mAEI,24533
|
|
218
218
|
datahub/ingestion/source/aws/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
219
219
|
datahub/ingestion/source/aws/aws_common.py,sha256=DfdQgkJ_s2isFx8WvqKTlAcBk4KE8SgfpmA5BgC3fgY,17716
|
|
220
|
-
datahub/ingestion/source/aws/glue.py,sha256=
|
|
220
|
+
datahub/ingestion/source/aws/glue.py,sha256=DcGZzaeY6hVa1wN74o6spNH1B6OFu_6Mn4uoRdlZa44,57647
|
|
221
221
|
datahub/ingestion/source/aws/s3_boto_utils.py,sha256=Wyp9k9tapsCuw9dyH4FCXJr_wmeLaYFoCtKvrV6SEDk,3892
|
|
222
|
-
datahub/ingestion/source/aws/s3_util.py,sha256=
|
|
222
|
+
datahub/ingestion/source/aws/s3_util.py,sha256=pikTe9SuiKdN-TZ8eOhB0PYq0aUgUPDpxwtTLsVofRs,2834
|
|
223
223
|
datahub/ingestion/source/aws/sagemaker.py,sha256=Bl2tkBYnrindgx61VHYgNovUF_Kp_fXNcivQn28vC2w,5254
|
|
224
224
|
datahub/ingestion/source/aws/sagemaker_processors/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
225
225
|
datahub/ingestion/source/aws/sagemaker_processors/common.py,sha256=NvYfI8LHgDvhEZE7qp6qF1NSZ0_SQKhg3ivtdjsdpFg,2172
|
|
@@ -281,7 +281,7 @@ datahub/ingestion/source/delta_lake/__init__.py,sha256=u5oqUeus81ONAtdl6o9Puw33O
|
|
|
281
281
|
datahub/ingestion/source/delta_lake/config.py,sha256=bVBwGjCPiXyjbCLQsamt4hAsKJMtMuxupKjwZEwtU78,3374
|
|
282
282
|
datahub/ingestion/source/delta_lake/delta_lake_utils.py,sha256=VqIDPEXepOnlk4oWMeRaneSpQBlWmlCKAa1wGUl1sfk,1525
|
|
283
283
|
datahub/ingestion/source/delta_lake/report.py,sha256=uqWWivPltlZ7dwpOOluTvHOKKsSusqihn67clCAwxoM,467
|
|
284
|
-
datahub/ingestion/source/delta_lake/source.py,sha256=
|
|
284
|
+
datahub/ingestion/source/delta_lake/source.py,sha256=5VyE_ZYrop4JCTVhoXLjRXb1MRfWbIj0lMMmvNxsb80,13362
|
|
285
285
|
datahub/ingestion/source/dremio/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
286
286
|
datahub/ingestion/source/dremio/dremio_api.py,sha256=am8o_mQq7zteI4zasnkRb9B9-_BFrchTIA_oJkqRagA,33470
|
|
287
287
|
datahub/ingestion/source/dremio/dremio_aspects.py,sha256=3VeHzCw9q1ytngmsq_K4Ll9tWD2V8EDFySBImHdhPAw,18287
|
|
@@ -298,14 +298,14 @@ datahub/ingestion/source/dynamodb/dynamodb.py,sha256=wcEQSfQak45yPNZN7pCUEQFmjyW
|
|
|
298
298
|
datahub/ingestion/source/fivetran/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
299
299
|
datahub/ingestion/source/fivetran/config.py,sha256=kTfsu7oC4BSzkcFtetZr0UmiQ-B2af-_jbNYSFbDim4,8749
|
|
300
300
|
datahub/ingestion/source/fivetran/data_classes.py,sha256=ecdUJH5BEze0yv-uFpKWPNaNmV1gORDA2XMFk0zhcBw,595
|
|
301
|
-
datahub/ingestion/source/fivetran/fivetran.py,sha256=
|
|
301
|
+
datahub/ingestion/source/fivetran/fivetran.py,sha256=mJ3gi4LWYqul0NyHdZ0U4fDv3WuKEl_yxc2oOd3q6bw,13318
|
|
302
302
|
datahub/ingestion/source/fivetran/fivetran_log_api.py,sha256=EAak3hJpe75WZSgz6wP_CyAT5Cian2N4a-lb8x1NKHk,12776
|
|
303
303
|
datahub/ingestion/source/fivetran/fivetran_query.py,sha256=vLrTj7e-0NxZ2U4bWTB57pih42WirqPlUvwtIRfStlQ,5275
|
|
304
304
|
datahub/ingestion/source/gc/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
305
305
|
datahub/ingestion/source/gc/datahub_gc.py,sha256=6O-TxU2uCJ1Y8NNzJDufUd3ymapo--E3hTeIuy_QDtY,12763
|
|
306
306
|
datahub/ingestion/source/gc/dataprocess_cleanup.py,sha256=86Tm3NNWMf0xM4TklNIEeNOjEingKpYy-XvCPeaAb4k,17125
|
|
307
|
-
datahub/ingestion/source/gc/execution_request_cleanup.py,sha256=
|
|
308
|
-
datahub/ingestion/source/gc/soft_deleted_entity_cleanup.py,sha256=
|
|
307
|
+
datahub/ingestion/source/gc/execution_request_cleanup.py,sha256=VbZ-Xzryl5TMRapu7nlxlsXS8T8lFZcHK9AJnEadJ8Q,11111
|
|
308
|
+
datahub/ingestion/source/gc/soft_deleted_entity_cleanup.py,sha256=PTx1XmD4Jv9SzXzHqUbFpT3vKGCtkD01MeBUaq1p4no,12083
|
|
309
309
|
datahub/ingestion/source/gcs/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
310
310
|
datahub/ingestion/source/gcs/gcs_source.py,sha256=iwvj4JwjyVWRP1Vq106sUtQhh0GuOYVSu9zCa1wCZN0,6189
|
|
311
311
|
datahub/ingestion/source/gcs/gcs_utils.py,sha256=_78KM863XXgkVLmZLtYGF5PJNnZas1go-XRtOq-79lo,1047
|
|
@@ -314,9 +314,9 @@ datahub/ingestion/source/git/git_import.py,sha256=5CT6vMDb0MDctCtShnxb3JVihULtvk
|
|
|
314
314
|
datahub/ingestion/source/grafana/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
315
315
|
datahub/ingestion/source/grafana/grafana_source.py,sha256=3pU3xodPgS5lmnjuQ_u7F0XPzD_Y8MnPlMxRJ86qz4g,4960
|
|
316
316
|
datahub/ingestion/source/iceberg/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
317
|
-
datahub/ingestion/source/iceberg/iceberg.py,sha256=
|
|
318
|
-
datahub/ingestion/source/iceberg/iceberg_common.py,sha256=
|
|
319
|
-
datahub/ingestion/source/iceberg/iceberg_profiler.py,sha256=
|
|
317
|
+
datahub/ingestion/source/iceberg/iceberg.py,sha256=tRPoThF6f8PlV_EbpIHjZPo8lcAzaiwE7G6E2m9rY-w,27485
|
|
318
|
+
datahub/ingestion/source/iceberg/iceberg_common.py,sha256=LEZaJleL5KJt1u_pLRUkeCqPEsthzH7tG8FgBwd9MC8,10218
|
|
319
|
+
datahub/ingestion/source/iceberg/iceberg_profiler.py,sha256=CkBB5fryMVoqqCM6eLSIeb4yP85ABHONNRm0QqZKrnw,9977
|
|
320
320
|
datahub/ingestion/source/identity/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
321
321
|
datahub/ingestion/source/identity/azure_ad.py,sha256=GdmJFD4UMsb5353Z7phXRf-YsXR2woGLRJwBXUkgXq0,28809
|
|
322
322
|
datahub/ingestion/source/identity/okta.py,sha256=PnRokWLG8wSoNZlXJiRZiW6APTEHO09q4n2j_l6m3V0,30756
|
|
@@ -325,7 +325,7 @@ datahub/ingestion/source/kafka/kafka.py,sha256=9SR7bqp9J0rPYde5IClhnAuVNy9ItsB8-
|
|
|
325
325
|
datahub/ingestion/source/kafka/kafka_schema_registry_base.py,sha256=13XjSwqyVhH1CJUFHAbWdmmv_Rw0Ju_9HQdBmIzPNNA,566
|
|
326
326
|
datahub/ingestion/source/kafka_connect/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
327
327
|
datahub/ingestion/source/kafka_connect/common.py,sha256=Ekb1K_J1eTgiH7LSP1AbEIf7NQh_2Vyu1lYX_Ggcqk4,7049
|
|
328
|
-
datahub/ingestion/source/kafka_connect/kafka_connect.py,sha256=
|
|
328
|
+
datahub/ingestion/source/kafka_connect/kafka_connect.py,sha256=NwBZQypX7TfZHm_W3uhdMI9YclxTxazJoxQqL8Nfv4Y,14054
|
|
329
329
|
datahub/ingestion/source/kafka_connect/sink_connectors.py,sha256=ESuJE5SFLLvss9OwDEIB8SAko4rhzaWZ-4dKY0Dh0N8,12900
|
|
330
330
|
datahub/ingestion/source/kafka_connect/source_connectors.py,sha256=_765fSMDAWAe0Cf_F4VNHfOWKNhtqBA1Ep2jL3rf-qc,21263
|
|
331
331
|
datahub/ingestion/source/looker/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
@@ -376,7 +376,7 @@ datahub/ingestion/source/powerbi/rest_api_wrapper/profiling_utils.py,sha256=bgcP
|
|
|
376
376
|
datahub/ingestion/source/powerbi/rest_api_wrapper/query.py,sha256=VNw1Uvli6g0pnu9FpigYmnCdEPbVEipz7vdZU_WmHf4,616
|
|
377
377
|
datahub/ingestion/source/powerbi_report_server/__init__.py,sha256=N9fGcrHXBbuPmx9rpGjd_jkMC3smXmfiwISDP1QZapk,324
|
|
378
378
|
datahub/ingestion/source/powerbi_report_server/constants.py,sha256=i_hXcvPHjwk3VpTT5ef7s8dN9F6pJzPyRUiG5UzCCYI,3544
|
|
379
|
-
datahub/ingestion/source/powerbi_report_server/report_server.py,sha256=
|
|
379
|
+
datahub/ingestion/source/powerbi_report_server/report_server.py,sha256=Xsvu_FeteYNyLW_U0pER94-zQLLGUzU5tUEkhsLTQ2Y,20176
|
|
380
380
|
datahub/ingestion/source/powerbi_report_server/report_server_domain.py,sha256=bBYr9fz4zPEFeZZOkldzKm4SBMQdisdp-MMtaYI0n3s,11783
|
|
381
381
|
datahub/ingestion/source/profiling/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
382
382
|
datahub/ingestion/source/profiling/common.py,sha256=4sZ58AeBV64KRfKAgjkg-UyNjAc3YERahQMmW4algAw,1426
|
|
@@ -393,7 +393,7 @@ datahub/ingestion/source/redshift/lineage.py,sha256=bUy0uJowrqSc33Z50fIxFlJkyhe-
|
|
|
393
393
|
datahub/ingestion/source/redshift/lineage_v2.py,sha256=OcVW_27sSaZOYZPTd2j-LS9SzFQ1kXz6cMzM2ZDWhJQ,16751
|
|
394
394
|
datahub/ingestion/source/redshift/profile.py,sha256=T4H79ycq2tPobLM1tTLRtu581Qa8LlKxEok49m0AirU,4294
|
|
395
395
|
datahub/ingestion/source/redshift/query.py,sha256=bY1D9RoOHaw89LgcXal7GYlJN0RG7PxXRRC-YKIdC8E,43105
|
|
396
|
-
datahub/ingestion/source/redshift/redshift.py,sha256=
|
|
396
|
+
datahub/ingestion/source/redshift/redshift.py,sha256=tBM4r9PiYDtRdZDccy4dwTTte842GADUS5Sfl9t5VHg,44436
|
|
397
397
|
datahub/ingestion/source/redshift/redshift_data_reader.py,sha256=zc69jwXHdF-w8J4Hq-ZQ6BjHQ75Ij2iNDMpoRJlcmlU,1724
|
|
398
398
|
datahub/ingestion/source/redshift/redshift_schema.py,sha256=9IYeUsnISenq3eVB3k-s7zK8nInWDAYViFnDrNjtkb0,19149
|
|
399
399
|
datahub/ingestion/source/redshift/report.py,sha256=M19aUHBkd9n-BVBX4fRhyRNdVkN2b9Es6ZqInRx5ZGI,2958
|
|
@@ -403,7 +403,7 @@ datahub/ingestion/source/s3/config.py,sha256=Zs1nrBZKLImteZreIcSMMRLj8vBGgxakNDs
|
|
|
403
403
|
datahub/ingestion/source/s3/datalake_profiler_config.py,sha256=k7S9Xcmgr3-CvWrd5NEX-V8JSrcAwkm7vbHPTVZicow,3620
|
|
404
404
|
datahub/ingestion/source/s3/profiling.py,sha256=yKNCKpr6w7qpCH-baeSkNE9VjkN6eBot_weD-2_Jxzk,17579
|
|
405
405
|
datahub/ingestion/source/s3/report.py,sha256=fzkTdTewYlWrTk4f2Cyl-e8RV4qw9wEVtm0cdKD-Xgw,542
|
|
406
|
-
datahub/ingestion/source/s3/source.py,sha256=
|
|
406
|
+
datahub/ingestion/source/s3/source.py,sha256=GXnyvMdr6dKaXdGros8GHEFwh1jwWiX3pKebvWxO2n8,47264
|
|
407
407
|
datahub/ingestion/source/sac/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
408
408
|
datahub/ingestion/source/sac/sac.py,sha256=zPSO9ukuyhvNaaVzeAYpA-_sFma_XMcCQMPaGvDWuTk,30226
|
|
409
409
|
datahub/ingestion/source/sac/sac_common.py,sha256=-xQTDBtgH56AnpRXWGDnlmQqUuLRx-7wF1U1kQFWtX8,998
|
|
@@ -429,7 +429,7 @@ datahub/ingestion/source/snowflake/constants.py,sha256=22n-0r04nuy-ImxWFFpmbrt_G
|
|
|
429
429
|
datahub/ingestion/source/snowflake/oauth_config.py,sha256=ol9D3RmruGStJAeL8PYSQguSqcD2HfkjPkMF2AB_eZs,1277
|
|
430
430
|
datahub/ingestion/source/snowflake/oauth_generator.py,sha256=fu2VnREGuJXeTqIV2jx4TwieVnznf83HQkrE0h2DGGM,3423
|
|
431
431
|
datahub/ingestion/source/snowflake/snowflake_assertion.py,sha256=_l3k4aI9wvioE81xxdeizJn9nJCZ_nMIXgk9N6pEk5o,4803
|
|
432
|
-
datahub/ingestion/source/snowflake/snowflake_config.py,sha256=
|
|
432
|
+
datahub/ingestion/source/snowflake/snowflake_config.py,sha256=kNzNQhpnEO4dGvqQudqO4NAX0MypujvM23XN4ZrA5Es,18949
|
|
433
433
|
datahub/ingestion/source/snowflake/snowflake_connection.py,sha256=yzv-01FdmfDSCJY5rqKNNodXxzg3SS5DF7oA4WXArOA,17793
|
|
434
434
|
datahub/ingestion/source/snowflake/snowflake_data_reader.py,sha256=ffR5E2uhD71FUMXd3XOg2rHwrp1rbbGEFTAbqKcmI2s,2195
|
|
435
435
|
datahub/ingestion/source/snowflake/snowflake_lineage_v2.py,sha256=FBmiONx4EGHWV8RNJT6zHZyntKinPFFyd2oKbTUIbhE,21319
|
|
@@ -437,14 +437,14 @@ datahub/ingestion/source/snowflake/snowflake_profiler.py,sha256=0DJiSwII6FY34url
|
|
|
437
437
|
datahub/ingestion/source/snowflake/snowflake_queries.py,sha256=jTpnFWRqqFId6DKJvvAbNuFPxyNi1oQxxDUyMvh1iu4,26968
|
|
438
438
|
datahub/ingestion/source/snowflake/snowflake_query.py,sha256=5po2FWz41UVowykJYbTFGxsltbmlHBCPcHG20VOhdOE,38469
|
|
439
439
|
datahub/ingestion/source/snowflake/snowflake_report.py,sha256=9Jjrie9XhD1JsIL2Wgx6pVPCNi9HuuAg6nuS0OgbLoE,6331
|
|
440
|
-
datahub/ingestion/source/snowflake/snowflake_schema.py,sha256=
|
|
441
|
-
datahub/ingestion/source/snowflake/snowflake_schema_gen.py,sha256=
|
|
440
|
+
datahub/ingestion/source/snowflake/snowflake_schema.py,sha256=qzQxugJJiDuF8LfSo9mTdshCexYyhjl_LlUwrwAcs8k,21806
|
|
441
|
+
datahub/ingestion/source/snowflake/snowflake_schema_gen.py,sha256=U0ELcn0YGlI16wv8VmTM1_UotcIDCSUwkAb0oUhAGCs,46209
|
|
442
442
|
datahub/ingestion/source/snowflake/snowflake_shares.py,sha256=maZyFkfrbVogEFM0tTKRiNp9c_1muv6YfleSd3q0umI,6341
|
|
443
443
|
datahub/ingestion/source/snowflake/snowflake_summary.py,sha256=kTmuCtRnvHqM8WBYhWeK4XafJq3ssFL9kcS03jEeWT4,5506
|
|
444
|
-
datahub/ingestion/source/snowflake/snowflake_tag.py,sha256=
|
|
444
|
+
datahub/ingestion/source/snowflake/snowflake_tag.py,sha256=uWhNsaV9gK9i65w3Ii3UeMJQ5aWy7rjh2ItxyIk-V4U,6539
|
|
445
445
|
datahub/ingestion/source/snowflake/snowflake_usage_v2.py,sha256=OI7MdARZcN1WjzRMF-hNIRSZyeVshlDVMD9Ga13W-SE,24846
|
|
446
446
|
datahub/ingestion/source/snowflake/snowflake_utils.py,sha256=MoI8-DR9tuMuHMBQcpDo4GFjvcoQZWLNkdFZsTkgK-M,12786
|
|
447
|
-
datahub/ingestion/source/snowflake/snowflake_v2.py,sha256=
|
|
447
|
+
datahub/ingestion/source/snowflake/snowflake_v2.py,sha256=dgvEx0TazQN17hruoe0kqqfJzAZF7mzUKk6LR0pGfjM,31949
|
|
448
448
|
datahub/ingestion/source/sql/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
449
449
|
datahub/ingestion/source/sql/athena.py,sha256=G3cIY8H_76lIUAzQWW2kLnZOEsfbakmojxbiHb3dYZ8,24059
|
|
450
450
|
datahub/ingestion/source/sql/clickhouse.py,sha256=jzvaXP5Wr0SMhj2rtuvVE821xnfpKiXhO3cm0xblgHs,27299
|
|
@@ -463,8 +463,8 @@ datahub/ingestion/source/sql/sql_config.py,sha256=M-l_uXau0ODolLZHBzAXhy-Rq5yYxv
|
|
|
463
463
|
datahub/ingestion/source/sql/sql_generic.py,sha256=9AERvkK8kdJUeDOzCYJDb93xdv6Z4DGho0NfeHj5Uyg,2740
|
|
464
464
|
datahub/ingestion/source/sql/sql_generic_profiler.py,sha256=oLjqgsxVKGerj5dZnCCRMremrxjp-kr5_P45gFOM4Pg,11602
|
|
465
465
|
datahub/ingestion/source/sql/sql_report.py,sha256=gw-OPHSExp_b6DRjvwqE1U6BpkwekxGrsvNMGYSGDio,2671
|
|
466
|
-
datahub/ingestion/source/sql/sql_types.py,sha256=
|
|
467
|
-
datahub/ingestion/source/sql/sql_utils.py,sha256=
|
|
466
|
+
datahub/ingestion/source/sql/sql_types.py,sha256=vuivhVDO27Hu_05Q1aYzsCuyCYXmdprW3gLt-fP_Yyk,15045
|
|
467
|
+
datahub/ingestion/source/sql/sql_utils.py,sha256=q-Bsk6WxlsRtrw9RXBxvqI3zuaMTC_F25T2VrCziR9I,8418
|
|
468
468
|
datahub/ingestion/source/sql/sqlalchemy_data_reader.py,sha256=FvHZ4JEK3aR2DYOBZiT_ZsAy12RjTu4t_KIR_92B11k,2644
|
|
469
469
|
datahub/ingestion/source/sql/sqlalchemy_uri_mapper.py,sha256=KOpbmDIE2h1hyYEsbVHJi2B7FlsyUMTXZx4diyzltQg,1826
|
|
470
470
|
datahub/ingestion/source/sql/teradata.py,sha256=M2txTIttQPXDrQfBwFfhPGqwWCZb2ei2yjxxtSodP1w,32499
|
|
@@ -491,7 +491,7 @@ datahub/ingestion/source/state_provider/datahub_ingestion_checkpointing_provider
|
|
|
491
491
|
datahub/ingestion/source/state_provider/file_ingestion_checkpointing_provider.py,sha256=xsH7Ao_05VTjqpkzLkhdf5B1ULMzFoD8vkJJIJU9w-U,4077
|
|
492
492
|
datahub/ingestion/source/state_provider/state_provider_registry.py,sha256=SVq4mIyGNmLXE9OZx1taOiNPqDoQp03-Ot9rYnB5F3k,401
|
|
493
493
|
datahub/ingestion/source/tableau/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
494
|
-
datahub/ingestion/source/tableau/tableau.py,sha256=
|
|
494
|
+
datahub/ingestion/source/tableau/tableau.py,sha256=MVqu7X7TOhVOyfRlYFSx9_ZqlcXuhKGzy7q2VibMx7w,145707
|
|
495
495
|
datahub/ingestion/source/tableau/tableau_common.py,sha256=a3Nu0Upy6_pnrd7XpSMcYHdnYca1JBW7H0jMqkYr0ME,26871
|
|
496
496
|
datahub/ingestion/source/tableau/tableau_constant.py,sha256=ZcAeHsQUXVVL26ORly0ByZk_GJAFbxaKuJAlX_sYMac,2686
|
|
497
497
|
datahub/ingestion/source/tableau/tableau_server_wrapper.py,sha256=nSyx9RzC6TCQDm-cTVJ657qT8iDwzk_8JMKpohhmOc4,1046
|
|
@@ -566,12 +566,12 @@ datahub/lite/lite_registry.py,sha256=bpH0kasP-LtwwUFNA2QsOIehfekAYfJtN-AkQLmSWnw
|
|
|
566
566
|
datahub/lite/lite_server.py,sha256=p9Oa2nNs65mqcssSIVOr7VOzWqfVstz6ZQEdT4f82S0,1949
|
|
567
567
|
datahub/lite/lite_util.py,sha256=pgBpT3vTO1YCQ2njZRNyicSkHYeEmQCt41BaXU8WvMo,4503
|
|
568
568
|
datahub/metadata/__init__.py,sha256=AjhXPjI6cnpdcrBRrE5gOWo15vv2TTl2ctU4UAnUN7A,238
|
|
569
|
-
datahub/metadata/_schema_classes.py,sha256=
|
|
570
|
-
datahub/metadata/schema.avsc,sha256=
|
|
569
|
+
datahub/metadata/_schema_classes.py,sha256=yqNgYrOGdIQ5XsTPo1XEFCjWDsThT3Qjl4IRUTSCYh0,975069
|
|
570
|
+
datahub/metadata/schema.avsc,sha256=US1MViEE4IunE2Q9zFmDsj1YZm8FwqUI5tNqp68UBUk,735960
|
|
571
571
|
datahub/metadata/schema_classes.py,sha256=X5Jl5EaSxyHdXOQv14pJ5WkQALun4MRpJ4q12wVFE18,1299
|
|
572
572
|
datahub/metadata/urns.py,sha256=nfrCTExR-k2P9w272WVtWSN3xW1VUJngPwP3xnvULjU,1217
|
|
573
573
|
datahub/metadata/_urns/__init__.py,sha256=cOF3GHMDgPhmbLKbN02NPpuLGHSu0qNgQyBRv08eqF0,243
|
|
574
|
-
datahub/metadata/_urns/urn_defs.py,sha256=
|
|
574
|
+
datahub/metadata/_urns/urn_defs.py,sha256=7wLzbGE-UnPZiJlCm8RcU3hROhHJ3QtwxJLGFLLjJlw,109984
|
|
575
575
|
datahub/metadata/com/__init__.py,sha256=gsAIuTxzfJdI7a9ybZlgMIHMAYksM1SxGxXjtySgKSc,202
|
|
576
576
|
datahub/metadata/com/linkedin/__init__.py,sha256=gsAIuTxzfJdI7a9ybZlgMIHMAYksM1SxGxXjtySgKSc,202
|
|
577
577
|
datahub/metadata/com/linkedin/events/__init__.py,sha256=s_dR0plZF-rOxxIbE8ojekJqwiHzl2WYR-Z3kW6kKS0,298
|
|
@@ -581,7 +581,7 @@ datahub/metadata/com/linkedin/pegasus2avro/access/token/__init__.py,sha256=P9M7N
|
|
|
581
581
|
datahub/metadata/com/linkedin/pegasus2avro/assertion/__init__.py,sha256=PgK5O-6pVRaEcvmwXAsSkwRLe8NjGiLH8AVBXeArqK8,5751
|
|
582
582
|
datahub/metadata/com/linkedin/pegasus2avro/businessattribute/__init__.py,sha256=N8kO-eUi0_Rt7weizIExxlnJ2_kZRtPrZLWCC1xtDMA,653
|
|
583
583
|
datahub/metadata/com/linkedin/pegasus2avro/chart/__init__.py,sha256=RNyyHLBNp_fxgFcBOLWO2UsXR1ofD_JczcBdPEQSusg,848
|
|
584
|
-
datahub/metadata/com/linkedin/pegasus2avro/common/__init__.py,sha256=
|
|
584
|
+
datahub/metadata/com/linkedin/pegasus2avro/common/__init__.py,sha256=x3AG1BxTAQijzsm_eKaP0P9VFraUT32d0glfvbQBrVI,5618
|
|
585
585
|
datahub/metadata/com/linkedin/pegasus2avro/common/fieldtransformer/__init__.py,sha256=FN63vLiB3FCmIRqBjTA-0Xt7M6i7h5NhaVzbA1ysv18,396
|
|
586
586
|
datahub/metadata/com/linkedin/pegasus2avro/connection/__init__.py,sha256=qRtw-dB14pzVzgQ0pDK8kyBplNdpRxVKNj4D70e_FqI,564
|
|
587
587
|
datahub/metadata/com/linkedin/pegasus2avro/container/__init__.py,sha256=3yWt36KqDKFhRc9pzvt0AMnbMTlhKurGvT3BUvc25QU,510
|
|
@@ -609,7 +609,7 @@ datahub/metadata/com/linkedin/pegasus2avro/identity/__init__.py,sha256=1U583fdMT
|
|
|
609
609
|
datahub/metadata/com/linkedin/pegasus2avro/incident/__init__.py,sha256=HEnbvzkz1KpPcomySyJFkuHfSUjH2d2mHQXhXU5uY7Q,735
|
|
610
610
|
datahub/metadata/com/linkedin/pegasus2avro/ingestion/__init__.py,sha256=1bfG2naq4iS_pwU4J-BVer_gfL0hDbJbnH0gh1MPNgA,871
|
|
611
611
|
datahub/metadata/com/linkedin/pegasus2avro/metadata/__init__.py,sha256=gsAIuTxzfJdI7a9ybZlgMIHMAYksM1SxGxXjtySgKSc,202
|
|
612
|
-
datahub/metadata/com/linkedin/pegasus2avro/metadata/key/__init__.py,sha256=
|
|
612
|
+
datahub/metadata/com/linkedin/pegasus2avro/metadata/key/__init__.py,sha256=kooj_lICFtq7GjWmYFtQOGuXYy9u4QomKrhTTVKFnDg,4812
|
|
613
613
|
datahub/metadata/com/linkedin/pegasus2avro/metadata/query/__init__.py,sha256=gsAIuTxzfJdI7a9ybZlgMIHMAYksM1SxGxXjtySgKSc,202
|
|
614
614
|
datahub/metadata/com/linkedin/pegasus2avro/metadata/query/filter/__init__.py,sha256=DBP_QtxkFmC5q_kuk4dGjb4uOKbB4xKgqTWXGxmNbBQ,532
|
|
615
615
|
datahub/metadata/com/linkedin/pegasus2avro/metadata/snapshot/__init__.py,sha256=OPboF8SV11wGnjvWQB-rxtB0otMdCsE7Tcy7xkOUgz8,2358
|
|
@@ -641,6 +641,7 @@ datahub/metadata/com/linkedin/pegasus2avro/test/__init__.py,sha256=Z4DlDtf-NELFp
|
|
|
641
641
|
datahub/metadata/com/linkedin/pegasus2avro/timeseries/__init__.py,sha256=6Pbit2drar8n99RFNQiXfYj7PhIzrO1SIpsGELZR4oA,637
|
|
642
642
|
datahub/metadata/com/linkedin/pegasus2avro/upgrade/__init__.py,sha256=o3U2TuzRSU1uPL-4AOMCPDqEwngqRb6g4-CBFY7eSvQ,525
|
|
643
643
|
datahub/metadata/com/linkedin/pegasus2avro/usage/__init__.py,sha256=vhCBrCM6hTXcl_U_OIexO9vdK8OKrGImAmi9BxTDNeY,602
|
|
644
|
+
datahub/metadata/com/linkedin/pegasus2avro/versionset/__init__.py,sha256=TJ9PXc6rctPWNTY1yYW5lwIh91jD5EDoKex2WYxOLXM,406
|
|
644
645
|
datahub/metadata/com/linkedin/pegasus2avro/view/__init__.py,sha256=-Le-jOqUJKv3ppwMhJHFFV3WwKTKjj1ETTkl9r7fY0o,498
|
|
645
646
|
datahub/metadata/schemas/Access.avsc,sha256=gdEfWJLkvjIz-jzlceK4Dl5pBDdCHG423Ba_EYGQgUk,1562
|
|
646
647
|
datahub/metadata/schemas/Actors.avsc,sha256=M76L2_Dlp7VyhVtu9__jhnh8rBNvNobtNJUfvl7bcPE,1188
|
|
@@ -678,7 +679,7 @@ datahub/metadata/schemas/DataContractKey.avsc,sha256=Oceu7P26--E0812IFrX3RiEY0Kt
|
|
|
678
679
|
datahub/metadata/schemas/DataContractProperties.avsc,sha256=RCxuJMlZwqEE0iHTpuXvcH6zRFoOt7ysQFPrJRp3RqE,4763
|
|
679
680
|
datahub/metadata/schemas/DataContractStatus.avsc,sha256=5yvT43AIB13Dn_h0-4s7fsL7BTuXhkK5pi2KJug4_qg,1029
|
|
680
681
|
datahub/metadata/schemas/DataFlowInfo.avsc,sha256=tDRTd1rA3v_7kwUVbQbb-cuo6D-t3pcuE4fiRz4D8f0,4682
|
|
681
|
-
datahub/metadata/schemas/DataFlowKey.avsc,sha256=
|
|
682
|
+
datahub/metadata/schemas/DataFlowKey.avsc,sha256=4QVVgv4SICNkRMQbQjxGtSXuMHqlKEno9BgTstDxNvU,1323
|
|
682
683
|
datahub/metadata/schemas/DataHubAccessTokenInfo.avsc,sha256=WS77M5w7GJFxUAiyXaxUvBqO0XFV2FnKPxXSXYbXHTE,1646
|
|
683
684
|
datahub/metadata/schemas/DataHubAccessTokenKey.avsc,sha256=3EspNIxgb_I4WwV0a2o4NJOB5yODVr9J-wZzkZanEgo,483
|
|
684
685
|
datahub/metadata/schemas/DataHubActionKey.avsc,sha256=bjiKcoyvUPQKaGUi2ICBMJ_ukwnt7dh0szJS4WBZE0A,448
|
|
@@ -705,7 +706,7 @@ datahub/metadata/schemas/DataHubViewInfo.avsc,sha256=U3fBIoG9ietLUpOknfQGNekqBdP
|
|
|
705
706
|
datahub/metadata/schemas/DataHubViewKey.avsc,sha256=p53axIdSVbubo3r23Vpsed7NqRcQBMGveVikEHAVAok,424
|
|
706
707
|
datahub/metadata/schemas/DataJobInfo.avsc,sha256=--obUbt_4X2paB39EeRKP13sBSiK-r0nq070EamoV1w,7212
|
|
707
708
|
datahub/metadata/schemas/DataJobInputOutput.avsc,sha256=H1O8eAzZV34tvULdu67iBSWkdn08rt7wS208b8Nisbk,15268
|
|
708
|
-
datahub/metadata/schemas/DataJobKey.avsc,sha256=
|
|
709
|
+
datahub/metadata/schemas/DataJobKey.avsc,sha256=fVlqnz1Ljan3YKzDTrMRFD3z09Ge7Gqt4aor5K_AgkY,1602
|
|
709
710
|
datahub/metadata/schemas/DataPlatformInfo.avsc,sha256=WGPFumBNHbR75vsLrivnRCbBc8vSCuxDw2UlylMieh4,2686
|
|
710
711
|
datahub/metadata/schemas/DataPlatformInstance.avsc,sha256=SNd3v_YyyLaDflv8Rd5cQR9GrVuky_cDTkYM6FqJiM8,1058
|
|
711
712
|
datahub/metadata/schemas/DataPlatformInstanceKey.avsc,sha256=sXUV5EMT6N-x8d6s8ebcJ5JdFIOsJCtiiU5Jtm-ncIk,800
|
|
@@ -727,7 +728,7 @@ datahub/metadata/schemas/DataTypeKey.avsc,sha256=Gs5uc_azwg10e36ZbwDTFQMevr0IfiF
|
|
|
727
728
|
datahub/metadata/schemas/DatahubIngestionCheckpoint.avsc,sha256=m2Zyrx3ZWDc5gHuwbmBSRJ3JN4NFkpUhDEKM2Yeuqrw,5681
|
|
728
729
|
datahub/metadata/schemas/DatahubIngestionRunSummary.avsc,sha256=_Ek7NqfJVTLqlM0NR9BRA57N9_ejwDdQvz7B1tVxSEE,9367
|
|
729
730
|
datahub/metadata/schemas/DatasetDeprecation.avsc,sha256=ucXxaDcAUib9_y0k5qOINMn5VK2X3trHK2dcpNcsR2Q,1256
|
|
730
|
-
datahub/metadata/schemas/DatasetKey.avsc,sha256=
|
|
731
|
+
datahub/metadata/schemas/DatasetKey.avsc,sha256=ACxwY5rfK695wdd7Z2rlCs15TSPW7GeG79dB30z9H9g,3163
|
|
731
732
|
datahub/metadata/schemas/DatasetProfile.avsc,sha256=3ZCU9JD6l2razACp0AY6LLMgnkMTj6D_5Xk9np6WWRM,9965
|
|
732
733
|
datahub/metadata/schemas/DatasetProperties.avsc,sha256=DFJn75feqaoQk84zin_o_lqsFFhqkwya5LGC5LLJXbU,4209
|
|
733
734
|
datahub/metadata/schemas/DatasetUpstreamLineage.avsc,sha256=PjAWPbsqwH7FjX2kFDy0dE6ENYOwRynH9vJerWisr2A,5365
|
|
@@ -788,22 +789,22 @@ datahub/metadata/schemas/IntendedUse.avsc,sha256=IKZSWdvc0uAyyT-FtdQOGbMC-P7RS9c
|
|
|
788
789
|
datahub/metadata/schemas/InviteToken.avsc,sha256=8k_9MxHu9GVf7gvS0SlnQu7tJfpbXsRFdz6lQrFKPNc,737
|
|
789
790
|
datahub/metadata/schemas/InviteTokenKey.avsc,sha256=MuQUlQaeVjaBkjSshB9gsx5Fm0civYgWD8UhCiRLdOQ,434
|
|
790
791
|
datahub/metadata/schemas/MLFeatureKey.avsc,sha256=sm4S8_G_YGwllESFk4LFwGS11MWrr84gkV9R68gQhzw,1110
|
|
791
|
-
datahub/metadata/schemas/MLFeatureProperties.avsc,sha256=
|
|
792
|
+
datahub/metadata/schemas/MLFeatureProperties.avsc,sha256=HpF7VcnH2FvDsqy2g5AMJPqU7upkTVADW1ps-9bjINo,6893
|
|
792
793
|
datahub/metadata/schemas/MLFeatureTableKey.avsc,sha256=VUG2pxBfyMJNf8J_yCyQVBGh9gpxsUN_8jXVPYadAus,1384
|
|
793
794
|
datahub/metadata/schemas/MLFeatureTableProperties.avsc,sha256=BtrqcsxoQXObPZXSGRNYtIBJCoeHkMK_Zr_imBWF2Zk,2008
|
|
794
795
|
datahub/metadata/schemas/MLHyperParam.avsc,sha256=dE6i5r6LTYMNrQe9yy-jKoP09GOJUf__1bO69ldpydc,833
|
|
795
796
|
datahub/metadata/schemas/MLMetric.avsc,sha256=y8WPVVwjhu3YGtqpFFJYNYK8w778RRL_d2sHG1Dc7uM,804
|
|
796
797
|
datahub/metadata/schemas/MLModelDeploymentKey.avsc,sha256=gmXaUYxII8BVLnXOFdlPmyhD1rUhrw455R_hL77foSU,2406
|
|
797
|
-
datahub/metadata/schemas/MLModelDeploymentProperties.avsc,sha256=
|
|
798
|
+
datahub/metadata/schemas/MLModelDeploymentProperties.avsc,sha256=I3v-uNOeYxO4hooPHOjafWWHuVyeGvG90oma0tzpNFg,5409
|
|
798
799
|
datahub/metadata/schemas/MLModelFactorPrompts.avsc,sha256=8kX-P4F4mVLFT980z3MwIautt1_6uA-c_Z87nYNDK-k,2712
|
|
799
800
|
datahub/metadata/schemas/MLModelGroupKey.avsc,sha256=eb4qdIdQLvvQ7u1e_FHIHTkWrvIwYQji0yKoGrxCiWI,2460
|
|
800
|
-
datahub/metadata/schemas/MLModelGroupProperties.avsc,sha256=
|
|
801
|
-
datahub/metadata/schemas/MLModelKey.avsc,sha256=
|
|
802
|
-
datahub/metadata/schemas/MLModelProperties.avsc,sha256=
|
|
801
|
+
datahub/metadata/schemas/MLModelGroupProperties.avsc,sha256=zMl6ab6zfcYJmt31f-AUrrfeqfLoaSZQpfB3_S9JFFQ,6534
|
|
802
|
+
datahub/metadata/schemas/MLModelKey.avsc,sha256=NeUHF1dNWpBc4422fp3YDqHCaDIlggQWuTD1PSfCMCM,2829
|
|
803
|
+
datahub/metadata/schemas/MLModelProperties.avsc,sha256=hDCBHxGe-cmCBeU1k0ANuQlKjtZsDcTfl2X_jWmtFqo,12355
|
|
803
804
|
datahub/metadata/schemas/MLPrimaryKeyKey.avsc,sha256=mX4CQcoN3FC_VQDBCkhlmJk4pfQKDrSeuqqCTTXTmq8,1092
|
|
804
|
-
datahub/metadata/schemas/MLPrimaryKeyProperties.avsc,sha256=
|
|
805
|
+
datahub/metadata/schemas/MLPrimaryKeyProperties.avsc,sha256=URIuOpS93RVk8MZVcbZ-dmTwu_cN3KSOKxSR8fm-eTo,6744
|
|
805
806
|
datahub/metadata/schemas/MLTrainingRunProperties.avsc,sha256=WGgj0MuQrGD4UgvyHCJHzTnHja2LlJTOr1gLu8SySj0,4269
|
|
806
|
-
datahub/metadata/schemas/MetadataChangeEvent.avsc,sha256=
|
|
807
|
+
datahub/metadata/schemas/MetadataChangeEvent.avsc,sha256=mk-ggA71SFW1H_BdzaLXMgH01wVzImCNPviGiH_v94Y,372781
|
|
807
808
|
datahub/metadata/schemas/MetadataChangeLog.avsc,sha256=mpdodpx25E6M1Gq_7slEcPAm-1Es5xPsoqV60HgO7zg,12167
|
|
808
809
|
datahub/metadata/schemas/MetadataChangeProposal.avsc,sha256=EMfQrYsuHf1p6UvBjoLtfdTHGe-vGNJaCFEHz8hdKU0,9698
|
|
809
810
|
datahub/metadata/schemas/Metrics.avsc,sha256=O7DJGjOwmHbb1x_Zj7AuM_HaHKjBvkfJKfUsX8icXD4,690
|
|
@@ -853,6 +854,9 @@ datahub/metadata/schemas/TrainingData.avsc,sha256=7p7sFBA_UyV5IbNU5qLgS3vVu70yev
|
|
|
853
854
|
datahub/metadata/schemas/UpstreamLineage.avsc,sha256=iaeFRbL2aVSYFwj-HQHyfIVaHRrK3kLbkkLXgIfJTsk,10639
|
|
854
855
|
datahub/metadata/schemas/UsageAggregation.avsc,sha256=QaF6lyWGUq8IlRel2h4qIXOXCMxBhrwjoaUELsd-I6g,4538
|
|
855
856
|
datahub/metadata/schemas/VersionInfo.avsc,sha256=9gMcZ8tjuhgcZiq2gOAp_EOV9q9jvuOgfph6m6v_X7c,1189
|
|
857
|
+
datahub/metadata/schemas/VersionProperties.avsc,sha256=knGNZxD_tOdnzFXqtrUy_JyXue3uJS9JqsxYy7CLKG4,7328
|
|
858
|
+
datahub/metadata/schemas/VersionSetKey.avsc,sha256=psjGNNcFua3Zs9Xlh4HnUHNmBEU74uYdJR5g20NhRJU,659
|
|
859
|
+
datahub/metadata/schemas/VersionSetProperties.avsc,sha256=yrhhVNioD11nFlDO7IfUbxAQjhA9Tr-4wnAYH5I9W74,1172
|
|
856
860
|
datahub/metadata/schemas/ViewProperties.avsc,sha256=3HhcbH5493dJUnEUtFMYMVfbYQ52aDedm5L4j77Nym4,1032
|
|
857
861
|
datahub/metadata/schemas/__init__.py,sha256=uvLNC3VyCkWA_v8e9FdA1leFf46NFKDD0AajCfihepI,581
|
|
858
862
|
datahub/secret/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
@@ -986,8 +990,8 @@ datahub_provider/operators/datahub_assertion_operator.py,sha256=uvTQ-jk2F0sbqqxp
|
|
|
986
990
|
datahub_provider/operators/datahub_assertion_sensor.py,sha256=lCBj_3x1cf5GMNpHdfkpHuyHfVxsm6ff5x2Z5iizcAo,140
|
|
987
991
|
datahub_provider/operators/datahub_operation_operator.py,sha256=aevDp2FzX7FxGlXrR0khoHNbxbhKR2qPEX5e8O2Jyzw,174
|
|
988
992
|
datahub_provider/operators/datahub_operation_sensor.py,sha256=8fcdVBCEPgqy1etTXgLoiHoJrRt_nzFZQMdSzHqSG7M,168
|
|
989
|
-
acryl_datahub-0.15.0.
|
|
990
|
-
acryl_datahub-0.15.0.
|
|
991
|
-
acryl_datahub-0.15.0.
|
|
992
|
-
acryl_datahub-0.15.0.
|
|
993
|
-
acryl_datahub-0.15.0.
|
|
993
|
+
acryl_datahub-0.15.0.2rc5.dist-info/METADATA,sha256=MBQ2EWJtuUyLlzfHsMfnf6niYbVl3nBvZJhV3qJ5e10,173662
|
|
994
|
+
acryl_datahub-0.15.0.2rc5.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
|
|
995
|
+
acryl_datahub-0.15.0.2rc5.dist-info/entry_points.txt,sha256=xnPSPLK3bJGADxe4TDS4wL4u0FT_PGlahDa-ENYdYCQ,9512
|
|
996
|
+
acryl_datahub-0.15.0.2rc5.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
|
|
997
|
+
acryl_datahub-0.15.0.2rc5.dist-info/RECORD,,
|
datahub/__init__.py
CHANGED
datahub/cli/delete_cli.py
CHANGED
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
import logging
|
|
2
|
+
import random
|
|
2
3
|
from concurrent.futures import ThreadPoolExecutor, as_completed
|
|
3
4
|
from dataclasses import dataclass
|
|
4
5
|
from datetime import datetime
|
|
5
|
-
from random import choices
|
|
6
6
|
from typing import Dict, List, Optional
|
|
7
7
|
|
|
8
8
|
import click
|
|
@@ -457,11 +457,11 @@ def by_filter(
|
|
|
457
457
|
click.echo("Found urns of multiple entity types")
|
|
458
458
|
for entity_type, entity_urns in urns_by_type.items():
|
|
459
459
|
click.echo(
|
|
460
|
-
f"- {len(entity_urns)} {entity_type} urn(s). Sample: {
|
|
460
|
+
f"- {len(entity_urns)} {entity_type} urn(s). Sample: {random.sample(entity_urns, k=min(5, len(entity_urns)))}"
|
|
461
461
|
)
|
|
462
462
|
else:
|
|
463
463
|
click.echo(
|
|
464
|
-
f"Found {len(urns)} {entity_type} urn(s). Sample: {
|
|
464
|
+
f"Found {len(urns)} {entity_type} urn(s). Sample: {random.sample(urns, k=min(5, len(urns)))}"
|
|
465
465
|
)
|
|
466
466
|
|
|
467
467
|
if not force and not dry_run:
|
datahub/cli/migrate.py
CHANGED
|
@@ -179,7 +179,7 @@ def dataplatform2instance_func(
|
|
|
179
179
|
|
|
180
180
|
if not force and not dry_run:
|
|
181
181
|
# get a confirmation from the operator before proceeding if this is not a dry run
|
|
182
|
-
sampled_urns_to_migrate = random.
|
|
182
|
+
sampled_urns_to_migrate = random.sample(
|
|
183
183
|
urns_to_migrate, k=min(10, len(urns_to_migrate))
|
|
184
184
|
)
|
|
185
185
|
sampled_new_urns: List[str] = [
|
|
@@ -193,7 +193,7 @@ def dataplatform2instance_func(
|
|
|
193
193
|
if key
|
|
194
194
|
]
|
|
195
195
|
click.echo(
|
|
196
|
-
f"Will migrate {len(urns_to_migrate)} urns such as {random.
|
|
196
|
+
f"Will migrate {len(urns_to_migrate)} urns such as {random.sample(urns_to_migrate, k=min(10, len(urns_to_migrate)))}"
|
|
197
197
|
)
|
|
198
198
|
click.echo(f"New urns will look like {sampled_new_urns}")
|
|
199
199
|
click.confirm("Ok to proceed?", abort=True)
|
datahub/emitter/mcp_builder.py
CHANGED
|
@@ -31,9 +31,12 @@ from datahub.metadata.schema_classes import (
|
|
|
31
31
|
OwnershipClass,
|
|
32
32
|
OwnershipTypeClass,
|
|
33
33
|
StatusClass,
|
|
34
|
+
StructuredPropertiesClass,
|
|
35
|
+
StructuredPropertyValueAssignmentClass,
|
|
34
36
|
SubTypesClass,
|
|
35
37
|
TagAssociationClass,
|
|
36
38
|
)
|
|
39
|
+
from datahub.metadata.urns import StructuredPropertyUrn
|
|
37
40
|
|
|
38
41
|
# In https://github.com/datahub-project/datahub/pull/11214, we added a
|
|
39
42
|
# new env field to container properties. However, populating this field
|
|
@@ -187,12 +190,31 @@ def add_tags_to_entity_wu(
|
|
|
187
190
|
).as_workunit()
|
|
188
191
|
|
|
189
192
|
|
|
193
|
+
def add_structured_properties_to_entity_wu(
|
|
194
|
+
entity_urn: str, structured_properties: Dict[StructuredPropertyUrn, str]
|
|
195
|
+
) -> Iterable[MetadataWorkUnit]:
|
|
196
|
+
aspect = StructuredPropertiesClass(
|
|
197
|
+
properties=[
|
|
198
|
+
StructuredPropertyValueAssignmentClass(
|
|
199
|
+
propertyUrn=urn.urn(),
|
|
200
|
+
values=[value],
|
|
201
|
+
)
|
|
202
|
+
for urn, value in structured_properties.items()
|
|
203
|
+
]
|
|
204
|
+
)
|
|
205
|
+
yield MetadataChangeProposalWrapper(
|
|
206
|
+
entityUrn=entity_urn,
|
|
207
|
+
aspect=aspect,
|
|
208
|
+
).as_workunit()
|
|
209
|
+
|
|
210
|
+
|
|
190
211
|
def gen_containers(
|
|
191
212
|
container_key: KeyType,
|
|
192
213
|
name: str,
|
|
193
214
|
sub_types: List[str],
|
|
194
215
|
parent_container_key: Optional[ContainerKey] = None,
|
|
195
216
|
extra_properties: Optional[Dict[str, str]] = None,
|
|
217
|
+
structured_properties: Optional[Dict[StructuredPropertyUrn, str]] = None,
|
|
196
218
|
domain_urn: Optional[str] = None,
|
|
197
219
|
description: Optional[str] = None,
|
|
198
220
|
owner_urn: Optional[str] = None,
|
|
@@ -282,6 +304,11 @@ def gen_containers(
|
|
|
282
304
|
tags=sorted(tags),
|
|
283
305
|
)
|
|
284
306
|
|
|
307
|
+
if structured_properties:
|
|
308
|
+
yield from add_structured_properties_to_entity_wu(
|
|
309
|
+
entity_urn=container_urn, structured_properties=structured_properties
|
|
310
|
+
)
|
|
311
|
+
|
|
285
312
|
|
|
286
313
|
def add_dataset_to_container(
|
|
287
314
|
container_key: KeyType, dataset_urn: str
|
datahub/emitter/rest_emitter.py
CHANGED
|
@@ -374,7 +374,7 @@ class DataHubRestEmitter(Closeable, Emitter):
|
|
|
374
374
|
# the size when chunking, and again for the actual request.
|
|
375
375
|
payload_dict: dict = {"proposals": mcp_obj_chunk}
|
|
376
376
|
if async_flag is not None:
|
|
377
|
-
payload_dict["async"] =
|
|
377
|
+
payload_dict["async"] = "true" if async_flag else "false"
|
|
378
378
|
|
|
379
379
|
payload = json.dumps(payload_dict)
|
|
380
380
|
self._emit_generic(url, payload)
|
datahub/ingestion/api/source.py
CHANGED
|
@@ -23,7 +23,7 @@ from typing import (
|
|
|
23
23
|
)
|
|
24
24
|
|
|
25
25
|
from pydantic import BaseModel
|
|
26
|
-
from typing_extensions import LiteralString
|
|
26
|
+
from typing_extensions import LiteralString, Self
|
|
27
27
|
|
|
28
28
|
from datahub.configuration.common import ConfigModel
|
|
29
29
|
from datahub.configuration.source_common import PlatformInstanceConfigMixin
|
|
@@ -400,7 +400,7 @@ class Source(Closeable, metaclass=ABCMeta):
|
|
|
400
400
|
ctx: PipelineContext
|
|
401
401
|
|
|
402
402
|
@classmethod
|
|
403
|
-
def create(cls, config_dict: dict, ctx: PipelineContext) ->
|
|
403
|
+
def create(cls, config_dict: dict, ctx: PipelineContext) -> Self:
|
|
404
404
|
# Technically, this method should be abstract. However, the @config_class
|
|
405
405
|
# decorator automatically generates a create method at runtime if one is
|
|
406
406
|
# not defined. Python still treats the class as abstract because it thinks
|
|
@@ -248,9 +248,11 @@ class DataHubGraph(DatahubRestEmitter):
|
|
|
248
248
|
with DatahubRestSink(PipelineContext(run_id=run_id), sink_config) as sink:
|
|
249
249
|
yield sink
|
|
250
250
|
if sink.report.failures:
|
|
251
|
+
logger.error(
|
|
252
|
+
f"Failed to emit {len(sink.report.failures)} records\n{sink.report.as_string()}"
|
|
253
|
+
)
|
|
251
254
|
raise OperationalError(
|
|
252
|
-
f"Failed to emit {len(sink.report.failures)} records"
|
|
253
|
-
info=sink.report.as_obj(),
|
|
255
|
+
f"Failed to emit {len(sink.report.failures)} records"
|
|
254
256
|
)
|
|
255
257
|
|
|
256
258
|
def emit_all(
|
|
@@ -248,6 +248,9 @@ class GlueSourceReport(StaleEntityRemovalSourceReport):
|
|
|
248
248
|
"Enabled by default when stateful ingestion is turned on.",
|
|
249
249
|
)
|
|
250
250
|
@capability(SourceCapability.LINEAGE_COARSE, "Enabled by default")
|
|
251
|
+
@capability(
|
|
252
|
+
SourceCapability.LINEAGE_FINE, "Support via the `emit_s3_lineage` config field"
|
|
253
|
+
)
|
|
251
254
|
class GlueSource(StatefulIngestionSourceBase):
|
|
252
255
|
"""
|
|
253
256
|
Note: if you also have files in S3 that you'd like to ingest, we recommend you use Glue's built-in data catalog. See [here](../../../../docs/generated/ingestion/sources/s3.md) for a quick guide on how to set up a crawler on Glue and ingest the outputs with DataHub.
|
|
@@ -284,12 +287,22 @@ class GlueSource(StatefulIngestionSourceBase):
|
|
|
284
287
|
"Action": [
|
|
285
288
|
"glue:GetDataflowGraph",
|
|
286
289
|
"glue:GetJobs",
|
|
290
|
+
"s3:GetObject",
|
|
287
291
|
],
|
|
288
292
|
"Resource": "*"
|
|
289
293
|
}
|
|
290
294
|
```
|
|
291
295
|
|
|
292
|
-
|
|
296
|
+
For profiling datasets, the following additional permissions are required:
|
|
297
|
+
```json
|
|
298
|
+
{
|
|
299
|
+
"Effect": "Allow",
|
|
300
|
+
"Action": [
|
|
301
|
+
"glue:GetPartitions",
|
|
302
|
+
],
|
|
303
|
+
"Resource": "*"
|
|
304
|
+
}
|
|
305
|
+
```
|
|
293
306
|
|
|
294
307
|
"""
|
|
295
308
|
|
|
@@ -1,6 +1,11 @@
|
|
|
1
1
|
import logging
|
|
2
2
|
import os
|
|
3
|
-
from
|
|
3
|
+
from collections import defaultdict
|
|
4
|
+
from typing import TYPE_CHECKING, Dict, Iterable, List, Optional
|
|
5
|
+
|
|
6
|
+
if TYPE_CHECKING:
|
|
7
|
+
from mypy_boto3_s3.service_resource import ObjectSummary
|
|
8
|
+
|
|
4
9
|
|
|
5
10
|
S3_PREFIXES = ["s3://", "s3n://", "s3a://"]
|
|
6
11
|
|
|
@@ -68,3 +73,21 @@ def get_key_prefix(s3_uri: str) -> str:
|
|
|
68
73
|
f"Not an S3 URI. Must start with one of the following prefixes: {str(S3_PREFIXES)}"
|
|
69
74
|
)
|
|
70
75
|
return strip_s3_prefix(s3_uri).split("/", maxsplit=1)[1]
|
|
76
|
+
|
|
77
|
+
|
|
78
|
+
def group_s3_objects_by_dirname(
|
|
79
|
+
s3_objects: Iterable["ObjectSummary"],
|
|
80
|
+
) -> Dict[str, List["ObjectSummary"]]:
|
|
81
|
+
"""
|
|
82
|
+
Groups S3 objects by their directory name.
|
|
83
|
+
|
|
84
|
+
If a s3_object in the root directory (i.e., s3://bucket/file.txt), it is grouped under '/'.
|
|
85
|
+
"""
|
|
86
|
+
grouped_s3_objs = defaultdict(list)
|
|
87
|
+
for obj in s3_objects:
|
|
88
|
+
if "/" in obj.key:
|
|
89
|
+
dirname = obj.key.rsplit("/", 1)[0]
|
|
90
|
+
else:
|
|
91
|
+
dirname = "/"
|
|
92
|
+
grouped_s3_objs[dirname].append(obj)
|
|
93
|
+
return grouped_s3_objs
|
|
@@ -122,11 +122,6 @@ class DeltaLakeSource(Source):
|
|
|
122
122
|
config_report,
|
|
123
123
|
)
|
|
124
124
|
|
|
125
|
-
@classmethod
|
|
126
|
-
def create(cls, config_dict: dict, ctx: PipelineContext) -> "Source":
|
|
127
|
-
config = DeltaLakeSourceConfig.parse_obj(config_dict)
|
|
128
|
-
return cls(config, ctx)
|
|
129
|
-
|
|
130
125
|
def _parse_datatype(self, raw_field_json_str: str) -> List[SchemaFieldClass]:
|
|
131
126
|
raw_field_json = json.loads(raw_field_json_str)
|
|
132
127
|
|
|
@@ -29,7 +29,7 @@ class DemoDataSource(Source):
|
|
|
29
29
|
|
|
30
30
|
def __init__(self, ctx: PipelineContext, config: DemoDataConfig):
|
|
31
31
|
file_config = FileSourceConfig(path=str(download_sample_data()))
|
|
32
|
-
self.file_source = GenericFileSource(ctx, file_config)
|
|
32
|
+
self.file_source: GenericFileSource = GenericFileSource(ctx, file_config)
|
|
33
33
|
|
|
34
34
|
def get_workunits(self) -> Iterable[MetadataWorkUnit]:
|
|
35
35
|
yield from self.file_source.get_workunits()
|