acryl-datahub 1.2.0.8rc2__py3-none-any.whl → 1.2.0.8rc4__py3-none-any.whl

{acryl_datahub-1.2.0.8rc2.dist-info → acryl_datahub-1.2.0.8rc4.dist-info}/RECORD

@@ -1,7 +1,7 @@
-acryl_datahub-1.2.0.8rc2.dist-info/licenses/LICENSE,sha256=9xNHpsD0uYF5ONzXsKDCuHHB-xbiCrSbueWXqrTNsxk,11365
+acryl_datahub-1.2.0.8rc4.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=61ZxWUlQVKM0CF2BBOi-9OpFZENqh_B4oxFCZYQSJBc,323
+datahub/_version.py,sha256=uu8Uaqh56VQOGHwbtlo282g9v9lvEV1QwtbdKz41OI0,323
 datahub/entrypoints.py,sha256=9Qf-37rNnTzbGlx8S75OCDazIclFp6zWNcCEL1zCZto,9015
 datahub/errors.py,sha256=p5rFAdAGVCk4Lqolol1YvthceadUSwpaCxLXRcyCCFQ,676
 datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -148,10 +148,10 @@ datahub/ingestion/api/incremental_properties_helper.py,sha256=KzdxdrQtaMV2XMHfPs
 datahub/ingestion/api/ingestion_job_checkpointing_provider_base.py,sha256=3lLdkkxVqE9MVc26cdXImPeWy16az5BwgcorWxeBV50,1759
 datahub/ingestion/api/pipeline_run_listener.py,sha256=5uBP__LbMQxJ2utlf07cIzQINqPbUOKiZyOJta6a0og,713
 datahub/ingestion/api/registry.py,sha256=LbdZr89465Lj7ptQRVB4vI1JR1igWABvQFj9-WX63bI,7454
-datahub/ingestion/api/report.py,sha256=OuVZAgNkzSGkKhpOhpqebd9_bEsBCTeoWR1VcanPeD4,18509
+datahub/ingestion/api/report.py,sha256=1w63Y2yN49IaDLZaIvXEjRU3yVb_9t3wzymSI-fumZM,18959
 datahub/ingestion/api/report_helpers.py,sha256=WbUC1kQeaKqIagGV3XzfPmPs7slAT1mfNY4og2BH2A8,994
-datahub/ingestion/api/sink.py,sha256=GZt48PV56FAhNoma-V5EwwRZvezhb40YH_zprm8_Yo0,4961
-datahub/ingestion/api/source.py,sha256=uf0fNbiOy0bS_aKFOcNv6NvuZe0LSDIDdNza9hraP7s,21857
+datahub/ingestion/api/sink.py,sha256=nfal7nsYY1AT2WQRjqO48uAHitpjax7TsRVzYXnqbeM,4918
+datahub/ingestion/api/source.py,sha256=JASs7WygVB6g-tcwtchaftzv3lNtlVM31lEa242pn44,21853
 datahub/ingestion/api/source_helpers.py,sha256=XT9y5HgfVeF52jrX39vlLn1SdXpLVyT2Su8oGNsddYo,21148
 datahub/ingestion/api/transform.py,sha256=X0GpjMJzYkLuZx8MTWxH50cWGm9rGsnn3k188mmC8J8,582
 datahub/ingestion/api/workunit.py,sha256=e8n8RfSjHZZm2R4ShNH0UuMtUkMjyqqM2j2t7oL74lo,6327
@@ -193,7 +193,7 @@ datahub/ingestion/reporting/file_reporter.py,sha256=tiWukmMxHrTQI3rOAumsq6lRlw8T
 datahub/ingestion/reporting/reporting_provider_registry.py,sha256=jTYSh3T4sensjnHQfPLiIcbA2dG8w0px9ghChAJjGdU,310
 datahub/ingestion/run/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/run/connection.py,sha256=mngNzr5aRLUDa5Izqxa0xkdDEqEqcDuacWSKIlkdvPc,1483
-datahub/ingestion/run/pipeline.py,sha256=xKjH5jUTCwJ-rCAnOEXLmZqVLKz_qxcX2JJxcO5AGmE,32046
+datahub/ingestion/run/pipeline.py,sha256=WO9K4yUihmoL6crtilO_cpRAeE6yVa1jd2tzf6spb-A,32541
 datahub/ingestion/run/pipeline_config.py,sha256=joG1j9OlwJhb8zqv4TY6_FSzOaKOx6xsBu255A5lP8g,4101
 datahub/ingestion/run/sink_callback.py,sha256=xZAzaKkR0dcQP838pMJWsA52oaQXV5BiqXkpxEpJ_9U,2856
 datahub/ingestion/sink/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -201,7 +201,7 @@ datahub/ingestion/sink/blackhole.py,sha256=-jYcWo4i8q7312bCIoHrGr7nT9JdPvA7c4jvS
 datahub/ingestion/sink/console.py,sha256=TZfhA0Ec2eNCrMH7RRy2JOdUE-U-hkoIQrPm1CmKLQs,591
 datahub/ingestion/sink/datahub_kafka.py,sha256=_cjuXu5I6G0zJ2UK7hMbaKjMPZXeIwRMgm7CVeTiNtc,2578
 datahub/ingestion/sink/datahub_lite.py,sha256=7u2aWm7ENLshKHl-PkjJg6Mrw4bWs8sTfKIBz4mm8Ak,1879
-datahub/ingestion/sink/datahub_rest.py,sha256=QrtR-hJ6yljN1quXcjoUHdAmJueZclrFZFrhU7c4YJM,13563
+datahub/ingestion/sink/datahub_rest.py,sha256=xw4JIaXOyFTLN3_wYOd3-dKv2ZwrcSxY_oXbeWpsXhs,13189
 datahub/ingestion/sink/file.py,sha256=SxXJPJpkIGoaqRjCcSmj2ZE3xE4rLlBABBGwpTj5LWI,3271
 datahub/ingestion/sink/sink_registry.py,sha256=JRBWx8qEYg0ubSTyhqwgSWctgxwyp6fva9GoN2LwBao,490
 datahub/ingestion/source/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -278,8 +278,8 @@ datahub/ingestion/source/bigquery_v2/queries_extractor.py,sha256=_5cAXVU8b8T_nAP
 datahub/ingestion/source/bigquery_v2/usage.py,sha256=A9c-ofclaRk0NSnc4IRaqJYqMPv6ecCld_TPy3V2qFs,40748
 datahub/ingestion/source/cassandra/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/cassandra/cassandra.py,sha256=pNy61Z4kTqL_wGcWIYee5fnZiuJDseDcRcQwsxeAssk,14487
-datahub/ingestion/source/cassandra/cassandra_api.py,sha256=b7MApc3_tEfHoj-6ub6snkcv_DweL1wi_TGJjAA1-yU,13516
-datahub/ingestion/source/cassandra/cassandra_config.py,sha256=Ga9915cDZukR5-u2tMNx5Jkf8eza2oAE5YS_sQIVEVQ,4222
+datahub/ingestion/source/cassandra/cassandra_api.py,sha256=wCJx-1ZByGMgPkORBO420sGucKkxXXE4pOLWXxdpMIw,14222
+datahub/ingestion/source/cassandra/cassandra_config.py,sha256=w9LBiT8XrGvXlrvpcAU_xm82GiE4nUfEg-VKIX6MRMY,4446
 datahub/ingestion/source/cassandra/cassandra_profiling.py,sha256=FdgPK_8s8otTOJDqNM4rpF6Mn4lFWbnjTaKEChzn2iE,11011
 datahub/ingestion/source/cassandra/cassandra_utils.py,sha256=j-LidYkaCTmGnpUVNLsax_c3z32PsQbsbHeYojygd1s,5105
 datahub/ingestion/source/common/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -333,7 +333,7 @@ datahub/ingestion/source/excel/report.py,sha256=oEkeI8J6is7zB9iz4RqASu_-Q5xl36lA
 datahub/ingestion/source/excel/source.py,sha256=w_vOz4UD7BcXBBDKoo81_6-QFeOPITuXqkfjIMHCQj4,23827
 datahub/ingestion/source/excel/util.py,sha256=YYmadYuCiT-4_MfQM0YSE7wuDcE0k8o2KrlOKM9Z6eI,406
 datahub/ingestion/source/fivetran/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/source/fivetran/config.py,sha256=00jc9srBZUQ18RSSCnHiOoJ4_F6I_rp--z-V7d9wXNY,9031
+datahub/ingestion/source/fivetran/config.py,sha256=A_ZFZ1PAfBeSY2k7wU-ixCFg1FOO7rgPGTd2wxKqbEg,9032
 datahub/ingestion/source/fivetran/data_classes.py,sha256=ecdUJH5BEze0yv-uFpKWPNaNmV1gORDA2XMFk0zhcBw,595
 datahub/ingestion/source/fivetran/fivetran.py,sha256=s8wcECtmuugUoZ0Zdthq0SIPpTLvziZXuhhUX9bJ5N4,14492
 datahub/ingestion/source/fivetran/fivetran_log_api.py,sha256=PNzuykiiFTU8FhBIfUbW6udURZpz_35aq7rfffbpIfA,13010
@@ -366,7 +366,7 @@ datahub/ingestion/source/hex/mapper.py,sha256=IyDAE-TzZUji3ICI_9gkYC3dQN3gl6kERR
 datahub/ingestion/source/hex/model.py,sha256=eri4aRo1eXcE2SWjzCnPFMhzPTiJ8w8zC4GN7Lgpr74,1864
 datahub/ingestion/source/hex/query_fetcher.py,sha256=r9UvF_qwswkRlNY7AI8p46eqAYSxVtjVE2e7eO4XagA,13384
 datahub/ingestion/source/iceberg/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/source/iceberg/iceberg.py,sha256=2E3mhvsIDSHDUd1Prb0nlZnGIsQLIuwNeFRxJPYyS-0,37042
+datahub/ingestion/source/iceberg/iceberg.py,sha256=KM9IDdWxW2VwOv0Iv3sMV6a60FNkgDEMS8vZaFHTOyA,37040
 datahub/ingestion/source/iceberg/iceberg_common.py,sha256=CD_yHQ_wEgivyLQUTRO9BZJB29S7j5fUVllki-BPwUU,12292
 datahub/ingestion/source/iceberg/iceberg_profiler.py,sha256=9iwp2vpQTi4OMbIKoDZV5lAdvjMR0ls6Llpck9grJIE,9875
 datahub/ingestion/source/identity/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -640,7 +640,7 @@ datahub/metadata/schema.avsc,sha256=HluHCVmYg7RpOaw9xUMigEJBxlHF5WLdNcqVBKPeoOU,
 datahub/metadata/schema_classes.py,sha256=tPT8iHCak4IsZi_oL0nirbPpI8ETTPTZzapqLRpeKU4,1326
 datahub/metadata/urns.py,sha256=nfrCTExR-k2P9w272WVtWSN3xW1VUJngPwP3xnvULjU,1217
 datahub/metadata/_urns/__init__.py,sha256=cOF3GHMDgPhmbLKbN02NPpuLGHSu0qNgQyBRv08eqF0,243
-datahub/metadata/_urns/urn_defs.py,sha256=Rl2wjTuHxpOk4rDAfqUHeBxCxTFA2JGFg1oCc9yR10s,143154
+datahub/metadata/_urns/urn_defs.py,sha256=tBlEg7f0jaIWVQfpgzTe2gjkthP4janfAwJO7yx6-cw,143257
 datahub/metadata/com/__init__.py,sha256=gsAIuTxzfJdI7a9ybZlgMIHMAYksM1SxGxXjtySgKSc,202
 datahub/metadata/com/linkedin/__init__.py,sha256=gsAIuTxzfJdI7a9ybZlgMIHMAYksM1SxGxXjtySgKSc,202
 datahub/metadata/com/linkedin/events/__init__.py,sha256=s_dR0plZF-rOxxIbE8ojekJqwiHzl2WYR-Z3kW6kKS0,298
@@ -1093,7 +1093,7 @@ datahub/utilities/urns/field_paths.py,sha256=ra-o_fMGkBRLgzMewAJN5-HqAyo3PIpXQ0K
 datahub/utilities/urns/notebook_urn.py,sha256=CHqGrV45ReVODlFx7js2WUxjcXxt8B63-xsBZpujmtY,73
 datahub/utilities/urns/structured_properties_urn.py,sha256=fjA1Ysg7IQSly8IVYx1R8HnwnojQz6jZWbqfk_XVvno,271
 datahub/utilities/urns/tag_urn.py,sha256=MqEJdIaCnAyjYe_8VdNnUjOVV4TS8xMlv4pRsy8wwXY,63
-datahub/utilities/urns/urn.py,sha256=B4nYxiFT8s5DLA2NJsWg0KoiUDp9UWg1nvL0j7Sx-h8,218
+datahub/utilities/urns/urn.py,sha256=CDtOtVccQW2yj5MBNtR3aO1yEInTnAmSDMbGbjheGJY,1279
 datahub/utilities/urns/urn_iter.py,sha256=3LtmobKksKFbnNCUCjFxm8qqFLCPPHUW_Q3zc4PE5nY,4736
 datahub_provider/__init__.py,sha256=qyPbz00f8pgtLVyqHG1TSnTqBfXb1x-kUH10zOLoq2U,53
 datahub_provider/_airflow_compat.py,sha256=unmFDGP57xKHPIhkdw_qo1vW1GAYpZ1yCvCrkMdGJXM,98
@@ -1114,8 +1114,8 @@ 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-1.2.0.8rc2.dist-info/METADATA,sha256=Q8mmqp92zb_C5PbYaI7zQiAwkw9QrX0FUiCAGxtbzzg,186651
-acryl_datahub-1.2.0.8rc2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-acryl_datahub-1.2.0.8rc2.dist-info/entry_points.txt,sha256=qopCAD6qrsijaZ9mTw3UlPCKsE00C3t9MbkkWow7pi4,9943
-acryl_datahub-1.2.0.8rc2.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
-acryl_datahub-1.2.0.8rc2.dist-info/RECORD,,
+acryl_datahub-1.2.0.8rc4.dist-info/METADATA,sha256=w5nzJ9UysAV5hpiguzwq09DBTlLJwLyRbSj3sEEWa8Q,186651
+acryl_datahub-1.2.0.8rc4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+acryl_datahub-1.2.0.8rc4.dist-info/entry_points.txt,sha256=qopCAD6qrsijaZ9mTw3UlPCKsE00C3t9MbkkWow7pi4,9943
+acryl_datahub-1.2.0.8rc4.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
+acryl_datahub-1.2.0.8rc4.dist-info/RECORD,,

datahub/_version.py

@@ -1,6 +1,6 @@
 # Published at https://pypi.org/project/acryl-datahub/.
 __package_name__ = "acryl-datahub"
-__version__ = "1.2.0.8rc2"
+__version__ = "1.2.0.8rc4"
 
 
 def is_dev_mode() -> bool:

datahub/ingestion/api/report.py

@@ -29,6 +29,7 @@ from datahub.metadata.schema_classes import (
 )
 from datahub.utilities.file_backed_collections import FileBackedDict
 from datahub.utilities.lossy_collections import LossyList
+from datahub.utilities.urns.urn import guess_platform_name
 
 logger = logging.getLogger(__name__)
 LogLevel = Literal["ERROR", "WARNING", "INFO", "DEBUG"]
@@ -41,6 +42,15 @@ class SupportsAsObj(Protocol):
 
 @dataclass
 class Report(SupportsAsObj):
+    def __post_init__(self) -> None:
+        self.platform: Optional[str] = None
+
+    def set_platform(self, platform: str) -> None:
+        self.platform = platform
+
+    def get_platform(self) -> Optional[str]:
+        return self.platform
+
     @staticmethod
     def to_str(some_val: Any) -> str:
         if isinstance(some_val, Enum):
@@ -213,6 +223,7 @@ class ExamplesReport(Report, Closeable):
     _lineage_aspects_seen: Set[str] = field(default_factory=set)
 
     def __post_init__(self) -> None:
+        super().__post_init__()
         self._file_based_dict = FileBackedDict(
             tablename="urn_aspects",
             extra_columns={
@@ -347,6 +358,9 @@ class ExamplesReport(Report, Closeable):
         aspectName: str,
         mcp: Union[MetadataChangeProposalClass, MetadataChangeProposalWrapper],
     ) -> None:
+        platform_name = guess_platform_name(urn)
+        if platform_name != self.get_platform():
+            return
         if is_lineage_aspect(entityType, aspectName):
             self._lineage_aspects_seen.add(aspectName)
         has_fine_grained_lineage = self._has_fine_grained_lineage(mcp)

datahub/ingestion/api/sink.py

@@ -147,9 +147,6 @@ class Sink(Generic[SinkConfig, SinkReportType], Closeable, metaclass=ABCMeta):
     def close(self) -> None:
         pass
 
-    def flush(self) -> None:
-        pass
-
     def configured(self) -> str:
         """Override this method to output a human-readable and scrubbed version of the configured sink"""
         return ""

datahub/ingestion/api/source.py

@@ -531,9 +531,9 @@ class Source(Closeable, metaclass=ABCMeta):
             auto_status_aspect,
             auto_materialize_referenced_tags_terms,
             partial(
-                auto_fix_duplicate_schema_field_paths, platform=self._infer_platform()
+                auto_fix_duplicate_schema_field_paths, platform=self.infer_platform()
             ),
-            partial(auto_fix_empty_field_paths, platform=self._infer_platform()),
+            partial(auto_fix_empty_field_paths, platform=self.infer_platform()),
             browse_path_processor,
             partial(auto_workunit_reporter, self.get_report()),
             auto_patch_last_modified,
@@ -583,7 +583,7 @@ class Source(Closeable, metaclass=ABCMeta):
     def close(self) -> None:
         self.get_report().close()
 
-    def _infer_platform(self) -> Optional[str]:
+    def infer_platform(self) -> Optional[str]:
         config = self.get_config()
         platform = (
             getattr(config, "platform_name", None)
@@ -598,7 +598,7 @@ class Source(Closeable, metaclass=ABCMeta):
     def _get_browse_path_processor(self, dry_run: bool) -> MetadataWorkUnitProcessor:
         config = self.get_config()
 
-        platform = self._infer_platform()
+        platform = self.infer_platform()
         env = getattr(config, "env", None)
         browse_path_drop_dirs = [
             platform,

datahub/ingestion/run/pipeline.py

@@ -373,11 +373,13 @@ class Pipeline:
             )
             current_version = version_stats.client.current.version
 
-            logger.debug(f"""
+            logger.debug(
+                f"""
                 client_version: {current_version}
                 server_default_version: {server_default_version}
                 server_default_cli_ahead: True
-            """)
+            """
+            )
 
             self.source.get_report().warning(
                 title="Server default CLI version is ahead of CLI version",
@@ -440,7 +442,19 @@ class Pipeline:
             return True
         return False
 
+    def _set_platform(self) -> None:
+        platform = self.source.infer_platform()
+        if platform:
+            self.source.get_report().set_platform(platform)
+        else:
+            self.source.get_report().warning(
+                message="Platform not found",
+                title="Platform not found",
+                context="Platform not found",
+            )
+
     def run(self) -> None:
+        self._set_platform()
         self._warn_old_cli_version()
         with self.exit_stack, self.inner_exit_stack:
             if self.config.flags.generate_memory_profiles:
@@ -548,8 +562,9 @@ class Pipeline:
                 self._handle_uncaught_pipeline_exception(exc)
             finally:
                 clear_global_warnings()
-                self.sink.flush()
-                self._notify_reporters_on_ingestion_completion()
+
+        # This can't be in the finally part because this should happen after context manager exists
+        self._notify_reporters_on_ingestion_completion()
 
     def transform(self, records: Iterable[RecordEnvelope]) -> Iterable[RecordEnvelope]:
         """

datahub/ingestion/sink/datahub_rest.py

@@ -5,7 +5,6 @@ import functools
 import logging
 import os
 import threading
-import time
 import uuid
 from enum import auto
 from typing import List, Optional, Tuple, Union
@@ -349,17 +348,6 @@ class DatahubRestSink(Sink[DatahubRestSinkConfig, DataHubRestSinkReport]):
             RecordEnvelope(item, metadata={}), NoopWriteCallback()
         )
 
-    def flush(self) -> None:
-        """Wait for all pending records to be written."""
-        i = 0
-        while self.report.pending_requests > 0:
-            time.sleep(0.1)
-            i += 1
-            if i % 1000 == 0:
-                logger.info(
-                    f"Waiting for {self.report.pending_requests} records to be written"
-                )
-
     def close(self):
         with self.report.main_thread_blocking_timer:
             self.executor.shutdown()

datahub/ingestion/source/cassandra/cassandra_api.py

@@ -132,7 +132,23 @@ class CassandraAPI:
 
             ssl_context = None
             if self.config.ssl_ca_certs:
-                ssl_context = ssl.SSLContext(ssl.PROTOCOL_TLS_CLIENT)
+                # Map SSL version string to ssl module constant
+                ssl_version_map = {
+                    "TLS_CLIENT": ssl.PROTOCOL_TLS_CLIENT,
+                    "TLSv1": ssl.PROTOCOL_TLSv1,
+                    "TLSv1_1": ssl.PROTOCOL_TLSv1_1,
+                    "TLSv1_2": ssl.PROTOCOL_TLSv1_2,
+                    "TLSv1_3": ssl.PROTOCOL_TLSv1_2,  # Python's ssl module uses TLSv1_2 for TLS 1.3
+                }
+
+                ssl_protocol = (
+                    ssl_version_map.get(
+                        self.config.ssl_version, ssl.PROTOCOL_TLS_CLIENT
+                    )
+                    if self.config.ssl_version
+                    else ssl.PROTOCOL_TLS_CLIENT
+                )
+                ssl_context = ssl.SSLContext(ssl_protocol)
                 ssl_context.load_verify_locations(self.config.ssl_ca_certs)
                 if self.config.ssl_certfile and self.config.ssl_keyfile:
                     ssl_context.load_cert_chain(

datahub/ingestion/source/cassandra/cassandra_config.py

@@ -94,6 +94,11 @@ class CassandraSourceConfig(
         description="Path to the SSL key file for SSL connections.",
     )
 
+    ssl_version: Optional[str] = Field(
+        default="TLS_CLIENT",
+        description="SSL protocol version to use for connections. Options: TLS_CLIENT, TLSv1, TLSv1_1, TLSv1_2, TLSv1_3. Defaults to TLS_CLIENT.",
+    )
+
     keyspace_pattern: AllowDenyPattern = Field(
         default=AllowDenyPattern.allow_all(),
         description="Regex patterns to filter keyspaces for ingestion.",

datahub/ingestion/source/fivetran/config.py

@@ -194,7 +194,7 @@ class FivetranSourceConfig(StatefulIngestionConfigBase, DatasetSourceConfigMixin
 
     # Configuration for stateful ingestion
     stateful_ingestion: Optional[StatefulStaleMetadataRemovalConfig] = pydantic.Field(
-        default=None, description="Airbyte Stateful Ingestion Config."
+        default=None, description="Fivetran Stateful Ingestion Config."
     )
 
     # Fivetran connector all sources to platform instance mapping

datahub/ingestion/source/iceberg/iceberg.py

@@ -200,9 +200,9 @@ class IcebergSource(StatefulIngestionSourceBase):
             auto_lowercase_dataset_urns,
             auto_materialize_referenced_tags_terms,
             partial(
-                auto_fix_duplicate_schema_field_paths, platform=self._infer_platform()
+                auto_fix_duplicate_schema_field_paths, platform=self.infer_platform()
             ),
-            partial(auto_fix_empty_field_paths, platform=self._infer_platform()),
+            partial(auto_fix_empty_field_paths, platform=self.infer_platform()),
             partial(auto_workunit_reporter, self.get_report()),
             auto_patch_last_modified,
             EnsureAspectSizeProcessor(self.get_report()).ensure_aspect_size,

datahub/metadata/_urns/urn_defs.py

@@ -2904,6 +2904,10 @@ class DataJobUrn(_SpecificUrn):
     def get_data_flow_urn(self) -> "DataFlowUrn":
         return DataFlowUrn.from_string(self.flow)
 
+    @property
+    def orchestrator(self) -> str:
+        return self.get_data_flow_urn().orchestrator
+
     @deprecated(reason="Use .job_id instead")
     def get_job_id(self) -> str:
         return self.job_id

datahub/utilities/urns/urn.py

@@ -1,8 +1,47 @@
-from datahub.metadata.urns import Urn
+from typing import Optional
 
-__all__ = ["Urn", "guess_entity_type"]
+from datahub.metadata.urns import (
+    DataPlatformUrn,
+    Urn,
+)
+
+__all__ = ["Urn", "guess_entity_type", "guess_platform_name"]
 
 
 def guess_entity_type(urn: str) -> str:
     assert urn.startswith("urn:li:"), "urns must start with urn:li:"
     return urn.split(":")[2]
+
+
+def guess_platform_name(urn: str) -> Optional[str]:
+    """Extract platform from URN using a mapping dictionary."""
+    urn_obj = Urn.from_string(urn)
+
+    try:
+        platform = None
+        try:
+            platform = urn_obj.platform  # type: ignore[attr-defined]
+            platform_name = DataPlatformUrn.from_string(
+                platform
+            ).get_entity_id_as_string()
+            return platform_name
+        except AttributeError:
+            pass
+        try:
+            return urn_obj.orchestrator  # type: ignore[attr-defined]
+        except AttributeError:
+            pass
+        try:
+            return urn_obj.dashboard_tool  # type: ignore[attr-defined]
+        except AttributeError:
+            pass
+        try:
+            return urn_obj.ml_model_tool  # type: ignore[attr-defined]
+        except AttributeError:
+            pass
+
+        if platform is None:
+            return None
+    except AttributeError:
+        pass
+    return None