acryl-datahub 1.0.0.4rc6__py3-none-any.whl → 1.0.0.4rc8__py3-none-any.whl

{acryl_datahub-1.0.0.4rc6.dist-info → acryl_datahub-1.0.0.4rc8.dist-info}/RECORD

@@ -1,7 +1,7 @@
-acryl_datahub-1.0.0.4rc6.dist-info/licenses/LICENSE,sha256=9xNHpsD0uYF5ONzXsKDCuHHB-xbiCrSbueWXqrTNsxk,11365
+acryl_datahub-1.0.0.4rc8.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=lFUyplnN9ea9C0jyvH1T6H0kVRxk5DIQVEdHH_34ut4,323
+datahub/_version.py,sha256=6c6YqNuY3b6Obtyadrf104RzQBKN4sFE4aGrxFv1FGw,323
 datahub/entrypoints.py,sha256=AQN5MzCe6q3LKI4SS6WmwN56kgjF6AC1ld7yELWVP2w,8953
 datahub/errors.py,sha256=p5rFAdAGVCk4Lqolol1YvthceadUSwpaCxLXRcyCCFQ,676
 datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -207,7 +207,7 @@ datahub/ingestion/source/glue_profiling_config.py,sha256=vpMJH4Lf_qgR32BZy58suab
 datahub/ingestion/source/ldap.py,sha256=PKoA5pVjuIxFfW1TcbYNIWSm7-C7shK2FDn7Zo5mrVM,18705
 datahub/ingestion/source/metabase.py,sha256=j8DRV2GvisezidL1JZ5HJLF_hdFdtvaoyDoEdEyh0Ks,32603
 datahub/ingestion/source/mlflow.py,sha256=fh7izN9jlSwbpGIrEyJktlmwFZR5vNG9z9L5VQ31k_4,33141
-datahub/ingestion/source/mode.py,sha256=Hs_qi-ntsuW_DIJ2wvMsFjir6P_QkRo4mq5Ldk7c1iU,66709
+datahub/ingestion/source/mode.py,sha256=YOsNnnSSenScbsUpWUaiflZ9zZHqLlw-LbEWVB4SUWA,70911
 datahub/ingestion/source/mongodb.py,sha256=2C2Cxn8DXL53IbNiywIuKt8UT_EMcPg9f8su-OPSNGU,21237
 datahub/ingestion/source/nifi.py,sha256=D1gBXxdpLuUQ0eurwofIR_SGg1rHGhwk3qxsWI1PT9c,56882
 datahub/ingestion/source/openapi.py,sha256=zx976zstg6M2KoTz_iKKgU9VETDeX2rnw6BofiHXbDc,18669
@@ -266,8 +266,8 @@ 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=lKvPP0Uahi9xw_yh9cArPPtwvAauXolaEk-6f-jhpz4,14558
-datahub/ingestion/source/cassandra/cassandra_api.py,sha256=UVGQTsk6O57Q6wrWo54bQPLtStTWhw_Fq6fgW3Bjgk8,12515
-datahub/ingestion/source/cassandra/cassandra_config.py,sha256=vIMUOzazWTGi03B51vI0-YMxaMJHUGmCxJJgd8pKhC8,3791
+datahub/ingestion/source/cassandra/cassandra_api.py,sha256=b7MApc3_tEfHoj-6ub6snkcv_DweL1wi_TGJjAA1-yU,13516
+datahub/ingestion/source/cassandra/cassandra_config.py,sha256=Ga9915cDZukR5-u2tMNx5Jkf8eza2oAE5YS_sQIVEVQ,4222
 datahub/ingestion/source/cassandra/cassandra_profiling.py,sha256=ZqsAY8NFsrrLqOduV7Aem2eJLtc2_OU9tW4tc_dh0V8,10984
 datahub/ingestion/source/cassandra/cassandra_utils.py,sha256=j-LidYkaCTmGnpUVNLsax_c3z32PsQbsbHeYojygd1s,5105
 datahub/ingestion/source/common/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -331,10 +331,10 @@ datahub/ingestion/source/grafana/grafana_source.py,sha256=3pU3xodPgS5lmnjuQ_u7F0
 datahub/ingestion/source/hex/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/hex/api.py,sha256=OVQNI_11NJJcNCT6OzSDEtVjNcom0vmes_KkjgzWCcI,11806
 datahub/ingestion/source/hex/constants.py,sha256=8hUTMWyG5keTNfXoLu_Dh413Hw_mGGJX1atiiDZyKtg,271
-datahub/ingestion/source/hex/hex.py,sha256=PIRl8fPkKtlHV7cqR4H8RKVYdTLgEFXHFzc3QAqJLhE,12733
+datahub/ingestion/source/hex/hex.py,sha256=hpMCkwH_RWdb0MG1U2-N71wc6vPUVQ3fAkTiudplzrQ,12968
 datahub/ingestion/source/hex/mapper.py,sha256=N3mTlEcrOmhv9ia1dnHGFgFJD2ddyTtU3H5IUbb-UxU,13344
 datahub/ingestion/source/hex/model.py,sha256=S9bUhfFcjzuio2dBS6HzSyRVPiSJvRvMQ0qyVrjV5-E,1766
-datahub/ingestion/source/hex/query_fetcher.py,sha256=ZaRrja05mK0VlIvpsFi-8-EBoJr0GSLbUxBUjycibIU,12505
+datahub/ingestion/source/hex/query_fetcher.py,sha256=m9A9xzxJipGD-nJU_lItyLF7Q5VYOm11wuM3fMyvm4I,12594
 datahub/ingestion/source/iceberg/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/iceberg/iceberg.py,sha256=-8uaBerljvonaT7Gn9Evokq6-SSDiMRf8kKo7Hg1qY4,35414
 datahub/ingestion/source/iceberg/iceberg_common.py,sha256=CD_yHQ_wEgivyLQUTRO9BZJB29S7j5fUVllki-BPwUU,12292
@@ -482,7 +482,7 @@ datahub/ingestion/source/sql/mariadb.py,sha256=Hm102kmfs_1rd4lsTYhzVMZq5S3B6cyfv
 datahub/ingestion/source/sql/mysql.py,sha256=nDWK4YbqomcJgnit9b8geUGrp_3eix4bt0_k94o7g-0,3350
 datahub/ingestion/source/sql/oracle.py,sha256=ftnrk3iiEelyv9PBHPYbairuP1WgxZbi1gu6YdqY69E,29965
 datahub/ingestion/source/sql/postgres.py,sha256=uC1kYEI8VdxiZ1Y9IxMWzwmg11wtMqYN0e2fkok1rxo,11972
-datahub/ingestion/source/sql/presto.py,sha256=PB-CS5MX2dSRFRHjlxfkLHGXLZXFNCsVAAyRBtY6HMg,3611
+datahub/ingestion/source/sql/presto.py,sha256=tATa0M2q0PjUC_E9W_jSUsmKTP7cVJayLgrFMzG_eao,4223
 datahub/ingestion/source/sql/sql_common.py,sha256=jsweel_-vesNtcPonnfS11OUrlcZnS3wGt5r0dYTPnM,48637
 datahub/ingestion/source/sql/sql_config.py,sha256=u3nGZYYl1WtaxfNsDU5bglgZ5Jq3Fxk9xei_CUIAXB0,8222
 datahub/ingestion/source/sql/sql_generic.py,sha256=9AERvkK8kdJUeDOzCYJDb93xdv6Z4DGho0NfeHj5Uyg,2740
@@ -494,12 +494,12 @@ datahub/ingestion/source/sql/sqlalchemy_data_reader.py,sha256=FvHZ4JEK3aR2DYOBZi
 datahub/ingestion/source/sql/sqlalchemy_uri.py,sha256=u0ZvgdJjXZdo_vl7YIQfYuuWbGwpnH6OSozI2e8ZV4I,858
 datahub/ingestion/source/sql/sqlalchemy_uri_mapper.py,sha256=KOpbmDIE2h1hyYEsbVHJi2B7FlsyUMTXZx4diyzltQg,1826
 datahub/ingestion/source/sql/teradata.py,sha256=9WdrxDy02lRJi9IZgsAATFsmxcQnIw5Gr6yCqHJQy5k,33507
-datahub/ingestion/source/sql/trino.py,sha256=gSLDyETKavSVR8l9wdebrfoc41cqAWz6ApqIicW0BF8,17892
+datahub/ingestion/source/sql/trino.py,sha256=zIfQ6GvW8Sbw4sxqsTcnibT51STka_nzNYvmld6HfHw,18947
 datahub/ingestion/source/sql/two_tier_sql_source.py,sha256=AB3Gtx4omAy_08zadHQpmUGmIGufkZ6o_ihWNnfvzYc,5783
 datahub/ingestion/source/sql/vertica.py,sha256=_9OgSgIgqBml0av063rb8nACiT3SAmzpw0ouyF91wv8,33382
 datahub/ingestion/source/sql/mssql/__init__.py,sha256=1agpl8S_uDW40olkhCX_W19dbr5GO9qgjS3R7pLRZSk,87
 datahub/ingestion/source/sql/mssql/job_models.py,sha256=nAo3rciu-w2-dXCz6_ekDEbGMEjCMEfh8WvSfXoF2l0,9359
-datahub/ingestion/source/sql/mssql/source.py,sha256=qaHOtZzCu_pWM8b5B9o8U5ELM01YJy1QRSkaYA1EDv8,32806
+datahub/ingestion/source/sql/mssql/source.py,sha256=BDkeJV64K_IA8a5t4emviYc9tgphkx1GMAp2gGPS60M,34230
 datahub/ingestion/source/sql/stored_procedures/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/sql/stored_procedures/base.py,sha256=n0l5OaTuW-m3TRvkxs3TqvgMeWF6BagzW3tjyWUcC1A,8631
 datahub/ingestion/source/sql/stored_procedures/lineage.py,sha256=fryLhuAlsjr9SHIjHJ-PmtCMx89bjzWVnJZ3f1bwQVU,1905
@@ -597,12 +597,12 @@ datahub/lite/lite_registry.py,sha256=bpH0kasP-LtwwUFNA2QsOIehfekAYfJtN-AkQLmSWnw
 datahub/lite/lite_server.py,sha256=p9Oa2nNs65mqcssSIVOr7VOzWqfVstz6ZQEdT4f82S0,1949
 datahub/lite/lite_util.py,sha256=G0LQHKkyEb1pc_q183g6hflShclGx7kikgMaOxtVVcs,4545
 datahub/metadata/__init__.py,sha256=AjhXPjI6cnpdcrBRrE5gOWo15vv2TTl2ctU4UAnUN7A,238
-datahub/metadata/_internal_schema_classes.py,sha256=zzpnbnA4K-3yrR-84VhwrdWuZ45lsYsmIxmsqN9UrYQ,1013133
-datahub/metadata/schema.avsc,sha256=Jxgo9sneW02r13VtiSduSNQCzY3_Vgsg5BSpfJdktTU,672600
+datahub/metadata/_internal_schema_classes.py,sha256=tl0FUKLebvTk766QFVUerWKFf0OpaSSY0o-w-vqPNsA,1013133
+datahub/metadata/schema.avsc,sha256=F4cl2V81M75_GVEu-zHL8a8KoM_dI62irB-1lAI147w,701582
 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=ZQGfW0P_u4GTN0uwXCWhCRHYW5JDUgjT75GFfPvloiA,136344
+datahub/metadata/_urns/urn_defs.py,sha256=NVHoYdf8wGZZpQfALAO01yrKiXKLpz0ST1iqaWMEYuo,136344
 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
@@ -899,7 +899,7 @@ datahub/metadata/schemas/VersionProperties.avsc,sha256=ME8V01JzG8lEsLXgYWnSYCehm
 datahub/metadata/schemas/VersionSetKey.avsc,sha256=psjGNNcFua3Zs9Xlh4HnUHNmBEU74uYdJR5g20NhRJU,659
 datahub/metadata/schemas/VersionSetProperties.avsc,sha256=fxNxEMxGdUDi_-T0sd6KJks5BWEo5AzboQxpZYKLbiQ,1434
 datahub/metadata/schemas/ViewProperties.avsc,sha256=3HhcbH5493dJUnEUtFMYMVfbYQ52aDedm5L4j77Nym4,1032
-datahub/metadata/schemas/__init__.py,sha256=kCcak_fBn_KyuysZTJIoipAzZ8EO44Amk4DWSEvplEY,581
+datahub/metadata/schemas/__init__.py,sha256=uvLNC3VyCkWA_v8e9FdA1leFf46NFKDD0AajCfihepI,581
 datahub/pydantic/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/pydantic/compat.py,sha256=TUEo4kSEeOWVAhV6LQtst1phrpVgGtK4uif4OI5vQ2M,1937
 datahub/sdk/__init__.py,sha256=QeutS6Th8K4E4ZxXuoGrmvahN6zA9Oh9asKk5mw9AIk,1670
@@ -945,10 +945,10 @@ 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=avoE5HGEZOUW-mUU1wm-9xclNFOe14sc0ScZf8MsdNE,70719
+datahub/sql_parsing/sql_parsing_aggregator.py,sha256=Hg8dyrZtsCjQEDLIFddnvYg8KBAumrj3QHP6xSKfAio,71427
 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=l0kT8MuRIg96X7BNJaboMznF54b-yvM2nMTLyF2d0Nw,47446
+datahub/sql_parsing/sqlglot_lineage.py,sha256=3pvKpARKPZG_IIqM7JjY8u7rQz6IjeFGQFVCPbm1O9c,54475
 datahub/sql_parsing/sqlglot_utils.py,sha256=TI11oBu1wrGeUuUGBg7hGTr6lTvztahdqiqXNJYRfbQ,14823
 datahub/sql_parsing/tool_meta_extractor.py,sha256=EV_g7sOchTSUm2p6wluNJqND7-rDYokVTqqFCM7hQ6c,7599
 datahub/telemetry/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -1054,8 +1054,8 @@ 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-1.0.0.4rc6.dist-info/METADATA,sha256=Ry6qydIyyjAptNjSQr46P_yzOTtYyW2WaIzk8SQI6Rw,179949
-acryl_datahub-1.0.0.4rc6.dist-info/WHEEL,sha256=DnLRTWE75wApRYVsjgc6wsVswC54sMSJhAEd4xhDpBk,91
-acryl_datahub-1.0.0.4rc6.dist-info/entry_points.txt,sha256=o3mDeJXSKhsy7XLkuogihraiabBdLn9HaizYXPrxmk0,9710
-acryl_datahub-1.0.0.4rc6.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
-acryl_datahub-1.0.0.4rc6.dist-info/RECORD,,
+acryl_datahub-1.0.0.4rc8.dist-info/METADATA,sha256=Q7oTvNGUsY5WFRrh4NKQlFK-3E2zBgiRrxgtWLkMCCk,179949
+acryl_datahub-1.0.0.4rc8.dist-info/WHEEL,sha256=DnLRTWE75wApRYVsjgc6wsVswC54sMSJhAEd4xhDpBk,91
+acryl_datahub-1.0.0.4rc8.dist-info/entry_points.txt,sha256=o3mDeJXSKhsy7XLkuogihraiabBdLn9HaizYXPrxmk0,9710
+acryl_datahub-1.0.0.4rc8.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
+acryl_datahub-1.0.0.4rc8.dist-info/RECORD,,

datahub/_version.py

@@ -1,6 +1,6 @@
 # Published at https://pypi.org/project/acryl-datahub/.
 __package_name__ = "acryl-datahub"
-__version__ = "1.0.0.4rc6"
+__version__ = "1.0.0.4rc8"
 
 
 def is_dev_mode() -> bool:

datahub/ingestion/source/cassandra/cassandra_api.py

@@ -1,3 +1,4 @@
+import ssl
 from dataclasses import dataclass, field
 from typing import Any, Dict, List, Optional
 
@@ -128,6 +129,23 @@ class CassandraAPI:
 
                 self._cassandra_session = cluster.connect()
                 return True
+
+            ssl_context = None
+            if self.config.ssl_ca_certs:
+                ssl_context = ssl.SSLContext(ssl.PROTOCOL_TLS_CLIENT)
+                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(
+                        certfile=self.config.ssl_certfile,
+                        keyfile=self.config.ssl_keyfile,
+                    )
+                elif self.config.ssl_certfile or self.config.ssl_keyfile:
+                    # If one is provided, the other must be too.
+                    # This is a simplification; real-world scenarios might allow one without the other depending on setup.
+                    raise ValueError(
+                        "Both ssl_certfile and ssl_keyfile must be provided if one is specified."
+                    )
+
             if self.config.username and self.config.password:
                 auth_provider = PlainTextAuthProvider(
                     username=self.config.username, password=self.config.password
@@ -136,12 +154,14 @@ class CassandraAPI:
                     [self.config.contact_point],
                     port=self.config.port,
                     auth_provider=auth_provider,
+                    ssl_context=ssl_context,
                     load_balancing_policy=None,
                 )
             else:
                 cluster = Cluster(
                     [self.config.contact_point],
                     port=self.config.port,
+                    ssl_context=ssl_context,
                     load_balancing_policy=None,
                 )
 

datahub/ingestion/source/cassandra/cassandra_config.py

@@ -79,6 +79,21 @@ class CassandraSourceConfig(
         description="Configuration for cloud-based Cassandra, such as DataStax Astra DB.",
     )
 
+    ssl_ca_certs: Optional[str] = Field(
+        default=None,
+        description="Path to the CA certificate file for SSL connections.",
+    )
+
+    ssl_certfile: Optional[str] = Field(
+        default=None,
+        description="Path to the SSL certificate file for SSL connections.",
+    )
+
+    ssl_keyfile: Optional[str] = Field(
+        default=None,
+        description="Path to the SSL key file for SSL connections.",
+    )
+
     keyspace_pattern: AllowDenyPattern = Field(
         default=AllowDenyPattern.allow_all(),
         description="Regex patterns to filter keyspaces for ingestion.",

datahub/ingestion/source/hex/hex.py

@@ -283,7 +283,12 @@ class HexSource(StatefulIngestionSourceBase):
                         self.report.report_warning(
                             title="Missing project for lineage",
                             message="Lineage missed because missed project, likely due to filter patterns or deleted project.",
-                            context=str(query_metadata),
+                            context=str(
+                                dict(
+                                    query_urn=query_metadata.urn,
+                                    hex_project_id=query_metadata.hex_project_id,
+                                )
+                            ),
                         )
 
         with self.report.new_stage("Emit"):

datahub/ingestion/source/hex/query_fetcher.py

@@ -213,6 +213,7 @@ class HexQueryFetcher:
 
         if not match:
             self.report.filtered_out_queries_no_match += 1
+            logger.debug(f"No match for Hex metadata in SQL statement: {sql_statement}")
             return None
 
         try: