acryl-datahub 1.1.0rc4__py3-none-any.whl → 1.1.1rc1__py3-none-any.whl

{acryl_datahub-1.1.0rc4.dist-info → acryl_datahub-1.1.1rc1.dist-info}/RECORD

@@ -1,7 +1,7 @@
-acryl_datahub-1.1.0rc4.dist-info/licenses/LICENSE,sha256=9xNHpsD0uYF5ONzXsKDCuHHB-xbiCrSbueWXqrTNsxk,11365
+acryl_datahub-1.1.1rc1.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=Zj4PSLmf7HigsG8FgXH1eFwPxvms3fDGDTUKLIYIASE,321
+datahub/_version.py,sha256=IwGjyypssXwDAKXhZI-PNpHN8GGxWeoJGSvg98k1_G0,321
 datahub/entrypoints.py,sha256=H-YFTvxTJOgpWsFBVlxyb1opjkq-hjTzNmjy5Fq3RHg,8992
 datahub/errors.py,sha256=p5rFAdAGVCk4Lqolol1YvthceadUSwpaCxLXRcyCCFQ,676
 datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -226,7 +226,7 @@ datahub/ingestion/source/abs/profiling.py,sha256=yKNCKpr6w7qpCH-baeSkNE9VjkN6eBo
 datahub/ingestion/source/abs/report.py,sha256=CkRjsNn0Pab-ZPllxz3IUJI_r3x0T6urJePa_hJKi5U,586
 datahub/ingestion/source/abs/source.py,sha256=cuMezUzr-Smp5tok2ceYor5I5jp52NDMjfeN8kfIbvg,24816
 datahub/ingestion/source/apply/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/source/apply/datahub_apply.py,sha256=2BLOaJxgyYYR6YblHQfDAEz0syPfh85vELuKp5UXNi8,7703
+datahub/ingestion/source/apply/datahub_apply.py,sha256=xTD-Iq3UHhxcz61RwNuI2kJjRrnQEfZFSgvS1X6loV4,7703
 datahub/ingestion/source/aws/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/aws/aws_common.py,sha256=DfdQgkJ_s2isFx8WvqKTlAcBk4KE8SgfpmA5BgC3fgY,17716
 datahub/ingestion/source/aws/glue.py,sha256=JHvxgufm94pvlefcgARpLwIjAtHcA-EvbFAMR0YLAAA,58087
@@ -277,7 +277,8 @@ datahub/ingestion/source/common/gcp_credentials_config.py,sha256=_NapGkAqZMbXNCl
 datahub/ingestion/source/common/subtypes.py,sha256=iJ9IfuiLK_T6yJ9ovY1HA83ujaVWichAgR-rUlBXemk,3238
 datahub/ingestion/source/data_lake_common/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/data_lake_common/config.py,sha256=qUk83B01hjuBKHvVz8SmXnVCy5eFj-2-2QLEOrAdbgk,359
-datahub/ingestion/source/data_lake_common/data_lake_utils.py,sha256=nxu7osuzqxScPFc-1ODA2M1c_xPNPpRH_SMMU7zKOIE,6212
+datahub/ingestion/source/data_lake_common/data_lake_utils.py,sha256=55mK0nsehqGDTUOol9Oi3jZs8Pb04PIsHdC2WPP0dkg,6576
+datahub/ingestion/source/data_lake_common/object_store.py,sha256=T7onzor_15MR-7ecxqcd6YnGY0-bGXHJzseg6nfo2Og,20302
 datahub/ingestion/source/data_lake_common/path_spec.py,sha256=8teyhkfFTy6QJdwH11tKw3cd5XV2wJOBctAtAiksoPM,24423
 datahub/ingestion/source/datahub/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/datahub/config.py,sha256=JohcVz2pYnHbmJd0SGcIDH7Lp-K6MIJlswkid0vTQO4,4762
@@ -322,8 +323,8 @@ datahub/ingestion/source/gc/dataprocess_cleanup.py,sha256=mUWcMt-_FL1SYGIgI4lGZD
 datahub/ingestion/source/gc/execution_request_cleanup.py,sha256=y-9ZIs_DZPUzYH1CI6HmaAZg3olNNA7MjT8HrCqAI0k,11159
 datahub/ingestion/source/gc/soft_deleted_entity_cleanup.py,sha256=4-qQR_2HGIYU8kC2hRIsJyKKMb9lKq4B6paJm_abUk4,12628
 datahub/ingestion/source/gcs/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/source/gcs/gcs_source.py,sha256=5EZkrDqjRNQz_aUL1MLp0PTFm0Ztubmk0NYJGZTRLjU,6276
-datahub/ingestion/source/gcs/gcs_utils.py,sha256=_78KM863XXgkVLmZLtYGF5PJNnZas1go-XRtOq-79lo,1047
+datahub/ingestion/source/gcs/gcs_source.py,sha256=ZdEmfy9TPd8nR8zlTUBvWDb9_cNFxksp-34O53L_ATE,6798
+datahub/ingestion/source/gcs/gcs_utils.py,sha256=Kd2usZYIMFeSuE6_tJ4OoHGOdvG8mWaScFuAcIkC6P0,1789
 datahub/ingestion/source/git/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/git/git_import.py,sha256=5CT6vMDb0MDctCtShnxb3JVihULtvkYGr9judHJFsOk,4143
 datahub/ingestion/source/grafana/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -428,7 +429,7 @@ datahub/ingestion/source/s3/config.py,sha256=lElFXgEpKDT9SVoiXvtx98wV6Gp880qP4pL
 datahub/ingestion/source/s3/datalake_profiler_config.py,sha256=FfrcgK-JEF94vw-l3q6pN6FENXb-wZzW2w1VUZVkwW8,3620
 datahub/ingestion/source/s3/profiling.py,sha256=yKNCKpr6w7qpCH-baeSkNE9VjkN6eBot_weD-2_Jxzk,17579
 datahub/ingestion/source/s3/report.py,sha256=fzkTdTewYlWrTk4f2Cyl-e8RV4qw9wEVtm0cdKD-Xgw,542
-datahub/ingestion/source/s3/source.py,sha256=mqyLUfsPA6zqzC42AhDBWEMhpv1Jt4J4oO9UUOFx7oU,47492
+datahub/ingestion/source/s3/source.py,sha256=7rWHKDrKd3aq_PlzNILuWHcQUXdsx20ZJh1eyKrbSzI,50115
 datahub/ingestion/source/sac/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/sac/sac.py,sha256=zPSO9ukuyhvNaaVzeAYpA-_sFma_XMcCQMPaGvDWuTk,30226
 datahub/ingestion/source/sac/sac_common.py,sha256=-xQTDBtgH56AnpRXWGDnlmQqUuLRx-7wF1U1kQFWtX8,998
@@ -459,8 +460,8 @@ datahub/ingestion/source/snowflake/snowflake_connection.py,sha256=3-nP3HHCblUnUH
 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=PY4Wy6i89nqRl92ARwXNqWwm-ifagkKbKKtxYWeswkk,29209
-datahub/ingestion/source/snowflake/snowflake_query.py,sha256=OJDF4x5OCa0PTTdkVdO5_Wzox4039Vhf7zwDoHHW-W4,38410
+datahub/ingestion/source/snowflake/snowflake_queries.py,sha256=PyuooQO6_OoUno3kzi7m_n1tv5V3GKSmJPXBk29pgWw,30480
+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
 datahub/ingestion/source/snowflake/snowflake_schema_gen.py,sha256=_37-AQyI4uGt4fu-d3v2eAWzQ3uG835ZQxMjFwGYCng,57193
@@ -476,7 +477,7 @@ datahub/ingestion/source/sql/clickhouse.py,sha256=9Fvaic9FZufRKdhVz2EcPUnEt5cA9V
 datahub/ingestion/source/sql/cockroachdb.py,sha256=XaD7eae34plU9ISRC6PzYX9q6RdT2qkzjH6CpTOgkx4,1443
 datahub/ingestion/source/sql/druid.py,sha256=_tzgTa5jhPUXk6WCmS7p10feCwJm6yUFcOgMZA-OcE8,2922
 datahub/ingestion/source/sql/hana.py,sha256=0PIvcX0Rz59NyR7Ag5Bv1MBV_UbJwxl9UAopo_xe_CA,1342
-datahub/ingestion/source/sql/hive.py,sha256=E5ZuGHoJmLQDMpUQFXPUc69Zbjv9QxGqtocFu_S4hbw,31590
+datahub/ingestion/source/sql/hive.py,sha256=Sh3Us1vjUcbgQ4NxLKcXEdGtck5fJHIwOdjbDhBGfCw,31575
 datahub/ingestion/source/sql/hive_metastore.py,sha256=qpX9eCRm-zq3DKC49MaZP9vzGot9QIDfaaeFgXGbOuM,36283
 datahub/ingestion/source/sql/mariadb.py,sha256=Hm102kmfs_1rd4lsTYhzVMZq5S3B6cyfvpHSzJjqvMw,737
 datahub/ingestion/source/sql/mysql.py,sha256=nDWK4YbqomcJgnit9b8geUGrp_3eix4bt0_k94o7g-0,3350
@@ -945,7 +946,7 @@ 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=Hg8dyrZtsCjQEDLIFddnvYg8KBAumrj3QHP6xSKfAio,71427
+datahub/sql_parsing/sql_parsing_aggregator.py,sha256=tqFZsE-7owUiU0q49nmkTt50CU4vn8ffUbNcTv9nRbc,71431
 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=6tuVv64MPO4i2VsmO9pjvP5IBWLEGollT3Ayubj6MU4,58668
@@ -961,8 +962,10 @@ datahub/testing/check_str_enum.py,sha256=yqk0XXHOGteN-IGqCp5JHy0Kca13BnI09ZqKc4N
 datahub/testing/compare_metadata_json.py,sha256=mTU5evu7KLS3cx8OLOC1fFxj0eY1J1CGV2PEQZmapos,5361
 datahub/testing/docker_utils.py,sha256=tptUfpuKJzEeVz1GVC3Ud-P4cR7ctm9mALwSpCkIXGw,2565
 datahub/testing/doctest.py,sha256=1_8WEhHZ2eRQtw8vsXKzr9L5zzvs0Tcr6q4mnkyyvtw,295
+datahub/testing/mce_helpers.py,sha256=4_ehqyNria2FD8tFuR6bpIxYUs4CANxedaRWSGfh1Ws,14024
 datahub/testing/mcp_diff.py,sha256=1BpQ3hST46cOQi1SmKdsto3j6x6Sk6yHm0vG1w9IDL0,10749
 datahub/testing/pytest_hooks.py,sha256=eifmj0M68AIfjTn_-0vtaBkKl75vNKMjsbYX-pJqmGY,1417
+datahub/testing/sdk_v2_helpers.py,sha256=1E0OIRYn7poyrB3eAP3ObMBwHzzP_KKIR4kqODCjwXc,312
 datahub/upgrade/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/upgrade/upgrade.py,sha256=K90hxU5bdi3j45s6KR5iLSfywJ_wXyZl4ADND8BKz7k,16732
 datahub/utilities/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -1054,8 +1057,8 @@ 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-1.1.0rc4.dist-info/METADATA,sha256=3XahqTI4zJWz9-fZoCbLaQhf3PRiTpHZnMrIw4IKWC4,180529
-acryl_datahub-1.1.0rc4.dist-info/WHEEL,sha256=Nw36Djuh_5VDukK0H78QzOX-_FQEo6V37m3nkm96gtU,91
-acryl_datahub-1.1.0rc4.dist-info/entry_points.txt,sha256=o3mDeJXSKhsy7XLkuogihraiabBdLn9HaizYXPrxmk0,9710
-acryl_datahub-1.1.0rc4.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
-acryl_datahub-1.1.0rc4.dist-info/RECORD,,
+acryl_datahub-1.1.1rc1.dist-info/METADATA,sha256=v6G_1JMFsv5K3HmNOlog_1nopU51zU4UddxpzGFCnbg,180529
+acryl_datahub-1.1.1rc1.dist-info/WHEEL,sha256=Nw36Djuh_5VDukK0H78QzOX-_FQEo6V37m3nkm96gtU,91
+acryl_datahub-1.1.1rc1.dist-info/entry_points.txt,sha256=o3mDeJXSKhsy7XLkuogihraiabBdLn9HaizYXPrxmk0,9710
+acryl_datahub-1.1.1rc1.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
+acryl_datahub-1.1.1rc1.dist-info/RECORD,,

datahub/_version.py

@@ -1,6 +1,6 @@
 # Published at https://pypi.org/project/acryl-datahub/.
 __package_name__ = "acryl-datahub"
-__version__ = "1.1.0rc4"
+__version__ = "1.1.1rc1"
 
 
 def is_dev_mode() -> bool:

datahub/ingestion/source/apply/datahub_apply.py

@@ -96,7 +96,7 @@ def apply_association_to_container(
 class DomainApplyConfig(ConfigModel):
     assets: List[str] = Field(
         default_factory=list,
-        description="List of assets to apply domain hierarchichaly. Currently only containers and datasets are supported",
+        description="List of assets to apply domain hierarchically. Currently only containers and datasets are supported",
     )
     domain_urn: str = Field(default="")
 
@@ -104,7 +104,7 @@ class DomainApplyConfig(ConfigModel):
 class TagApplyConfig(ConfigModel):
     assets: List[str] = Field(
         default_factory=list,
-        description="List of assets to apply tag hierarchichaly. Currently only containers and datasets are supported",
+        description="List of assets to apply tag hierarchically. Currently only containers and datasets are supported",
     )
     tag_urn: str = Field(default="")
 
@@ -112,7 +112,7 @@ class TagApplyConfig(ConfigModel):
 class TermApplyConfig(ConfigModel):
     assets: List[str] = Field(
         default_factory=list,
-        description="List of assets to apply term hierarchichaly. Currently only containers and datasets are supported",
+        description="List of assets to apply term hierarchically. Currently only containers and datasets are supported",
     )
     term_urn: str = Field(default="")
 
@@ -120,7 +120,7 @@ class TermApplyConfig(ConfigModel):
 class OwnerApplyConfig(ConfigModel):
     assets: List[str] = Field(
         default_factory=list,
-        description="List of assets to apply owner hierarchichaly. Currently only containers and datasets are supported",
+        description="List of assets to apply owner hierarchically. Currently only containers and datasets are supported",
     )
     owner_urn: str = Field(default="")
 

datahub/ingestion/source/data_lake_common/data_lake_utils.py

@@ -11,20 +11,21 @@ from datahub.emitter.mcp_builder import (
 )
 from datahub.ingestion.api.workunit import MetadataWorkUnit
 from datahub.ingestion.source.aws.s3_util import (
-    get_bucket_name,
     get_bucket_relative_path,
     get_s3_prefix,
     is_s3_uri,
 )
 from datahub.ingestion.source.azure.abs_utils import (
     get_abs_prefix,
-    get_container_name,
     get_container_relative_path,
     is_abs_uri,
 )
 from datahub.ingestion.source.common.subtypes import DatasetContainerSubTypes
+from datahub.ingestion.source.data_lake_common.object_store import (
+    get_object_store_bucket_name,
+    get_object_store_for_uri,
+)
 from datahub.ingestion.source.gcs.gcs_utils import (
-    get_gcs_bucket_name,
     get_gcs_prefix,
     is_gcs_uri,
 )
@@ -87,6 +88,13 @@ class ContainerWUCreator:
 
     @staticmethod
     def get_protocol(path: str) -> str:
+        object_store = get_object_store_for_uri(path)
+        if object_store:
+            prefix = object_store.get_prefix(path)
+            if prefix:
+                return prefix
+
+        # Legacy fallback
         protocol: Optional[str] = None
         if is_s3_uri(path):
             protocol = get_s3_prefix(path)
@@ -104,13 +112,12 @@ class ContainerWUCreator:
 
     @staticmethod
     def get_bucket_name(path: str) -> str:
-        if is_s3_uri(path):
-            return get_bucket_name(path)
-        elif is_gcs_uri(path):
-            return get_gcs_bucket_name(path)
-        elif is_abs_uri(path):
-            return get_container_name(path)
-        raise ValueError(f"Unable to get bucket name from path: {path}")
+        """
+        Get the bucket/container name from any supported object store URI.
+
+        Delegates to the abstract get_object_store_bucket_name function.
+        """
+        return get_object_store_bucket_name(path)
 
     def get_sub_types(self) -> str:
         if self.platform == PLATFORM_S3:
@@ -122,6 +129,11 @@ class ContainerWUCreator:
         raise ValueError(f"Unable to sub type for platform: {self.platform}")
 
     def get_base_full_path(self, path: str) -> str:
+        object_store = get_object_store_for_uri(path)
+        if object_store:
+            return object_store.get_object_key(path)
+
+        # Legacy fallback
         if self.platform == "s3" or self.platform == "gcs":
             return get_bucket_relative_path(path)
         elif self.platform == "abs":