acryl-datahub 1.3.0.1rc7__py3-none-any.whl → 1.3.0.1rc8__py3-none-any.whl

{acryl_datahub-1.3.0.1rc7.dist-info → acryl_datahub-1.3.0.1rc8.dist-info}/RECORD

@@ -1,7 +1,7 @@
-acryl_datahub-1.3.0.1rc7.dist-info/licenses/LICENSE,sha256=9xNHpsD0uYF5ONzXsKDCuHHB-xbiCrSbueWXqrTNsxk,11365
+acryl_datahub-1.3.0.1rc8.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=661Az3oTCAOu_uY9mbHEJZPDcekVKKHGjBByQxBb0CM,323
+datahub/_version.py,sha256=3eJGiXMGVkMi1lpj9ZplnwELXbksY0tQWUHU5_VpRwU,323
 datahub/entrypoints.py,sha256=VcbU6Z47b_JKW1zI-WJMYIngm05FSogKLiuvFNtyNcI,9088
 datahub/errors.py,sha256=p5rFAdAGVCk4Lqolol1YvthceadUSwpaCxLXRcyCCFQ,676
 datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -545,7 +545,7 @@ datahub/ingestion/source/sql/two_tier_sql_source.py,sha256=xlpQ9thbEn6uta6eVMzsH
 datahub/ingestion/source/sql/vertica.py,sha256=LUpIzDpzFRtMcXCSG2EC5CBZbveBOvnB27P_pDrXC9o,33501
 datahub/ingestion/source/sql/mssql/__init__.py,sha256=1agpl8S_uDW40olkhCX_W19dbr5GO9qgjS3R7pLRZSk,87
 datahub/ingestion/source/sql/mssql/job_models.py,sha256=ozEMc86e2oWRRF8Vjasjj32onVaVOQ98f00mbCYKmY8,9401
-datahub/ingestion/source/sql/mssql/source.py,sha256=N3ccX621byKLhthQw1-_aCkrn5teV4H1hc8u1gUMcy0,44762
+datahub/ingestion/source/sql/mssql/source.py,sha256=6dVWAFaaXDR4z2LJZozusg4lEFjlGLn2ELK0JoJbfYA,46342
 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
@@ -1133,8 +1133,8 @@ 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-1.3.0.1rc7.dist-info/METADATA,sha256=9QVlP4lgQfARG_mUrqs38beedc4VXd6o4r7YiXWiF2w,184619
-acryl_datahub-1.3.0.1rc7.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-acryl_datahub-1.3.0.1rc7.dist-info/entry_points.txt,sha256=pzsBoTx-D-iTcmpX8oCGCyzlHP2112EygUMzZWz56M8,10105
-acryl_datahub-1.3.0.1rc7.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
-acryl_datahub-1.3.0.1rc7.dist-info/RECORD,,
+acryl_datahub-1.3.0.1rc8.dist-info/METADATA,sha256=lOSilMTQwZ3rvmq-XCw-gmoitXmWUW_muDrbc-GqlKY,184619
+acryl_datahub-1.3.0.1rc8.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+acryl_datahub-1.3.0.1rc8.dist-info/entry_points.txt,sha256=pzsBoTx-D-iTcmpX8oCGCyzlHP2112EygUMzZWz56M8,10105
+acryl_datahub-1.3.0.1rc8.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
+acryl_datahub-1.3.0.1rc8.dist-info/RECORD,,

datahub/_version.py

@@ -1,6 +1,6 @@
 # Published at https://pypi.org/project/acryl-datahub/.
 __package_name__ = "acryl-datahub"
-__version__ = "1.3.0.1rc7"
+__version__ = "1.3.0.1rc8"
 
 
 def is_dev_mode() -> bool:

datahub/ingestion/source/sql/mssql/source.py

@@ -135,6 +135,10 @@ class SQLServerConfig(BasicSQLAlchemyConfig):
         default=False,
         description="Represent a schema identifiers combined with quoting preferences. See [sqlalchemy quoted_name docs](https://docs.sqlalchemy.org/en/20/core/sqlelement.html#sqlalchemy.sql.expression.quoted_name).",
     )
+    is_aws_rds: Optional[bool] = Field(
+        default=None,
+        description="Indicates if the SQL Server instance is running on AWS RDS. When None (default), automatic detection will be attempted using server name analysis.",
+    )
 
     @pydantic.validator("uri_args")
     def passwords_match(cls, v, values, **kwargs):
@@ -367,18 +371,42 @@ class SQLServerSource(SQLAlchemySource):
     def _detect_rds_environment(self, conn: Connection) -> bool:
         """
         Detect if we're running in an RDS/managed environment vs on-premises.
+        Uses explicit configuration if provided, otherwise attempts automatic detection.
         Returns True if RDS/managed, False if on-premises.
         """
+        if self.config.is_aws_rds is not None:
+            logger.info(
+                f"Using explicit is_aws_rds configuration: {self.config.is_aws_rds}"
+            )
+            return self.config.is_aws_rds
+
         try:
-            # Try to access system tables directly - this typically fails in RDS
-            conn.execute("SELECT TOP 1 * FROM msdb.dbo.sysjobs")
-            logger.debug(
-                "Direct table access successful - likely on-premises environment"
+            result = conn.execute("SELECT @@servername AS server_name")
+            server_name_row = result.fetchone()
+            if server_name_row:
+                server_name = server_name_row["server_name"].lower()
+
+                aws_indicators = ["amazon", "amzn", "amaz", "ec2", "rds.amazonaws.com"]
+                is_rds = any(indicator in server_name for indicator in aws_indicators)
+                if is_rds:
+                    logger.info(f"AWS RDS detected based on server name: {server_name}")
+                else:
+                    logger.info(
+                        f"Non-RDS environment detected based on server name: {server_name}"
+                    )
+
+                return is_rds
+            else:
+                logger.warning(
+                    "Could not retrieve server name, assuming non-RDS environment"
+                )
+                return False
+
+        except Exception as e:
+            logger.warning(
+                f"Failed to detect RDS/managed vs on-prem env, assuming non-RDS environment ({e})"
             )
             return False
-        except Exception:
-            logger.debug("Direct table access failed - likely RDS/managed environment")
-            return True
 
     def _get_jobs(self, conn: Connection, db_name: str) -> Dict[str, Dict[str, Any]]:
         """
@@ -453,7 +481,10 @@ class SQLServerSource(SQLAlchemySource):
         jobs_result = conn.execute("EXEC msdb.dbo.sp_help_job")
         jobs_data = {}
 
-        for row in jobs_result:
+        # SQLAlchemy 1.3 support was dropped in Sept 2023 (PR #8810)
+        # SQLAlchemy 1.4+ returns LegacyRow objects that don't support dictionary-style .get() method
+        # Use .mappings() to get MappingResult with dictionary-like rows that support .get()
+        for row in jobs_result.mappings():
             job_id = str(row["job_id"])
             jobs_data[job_id] = {
                 "job_id": job_id,
@@ -473,7 +504,8 @@ class SQLServerSource(SQLAlchemySource):
                 )
 
                 job_steps = {}
-                for step_row in steps_result:
+                # Use .mappings() for dictionary-like access (SQLAlchemy 1.4+ compatibility)
+                for step_row in steps_result.mappings():
                     # Only include steps that run against our target database
                     step_database = step_row.get("database_name", "")
                     if step_database.lower() == db_name.lower() or not step_database: