[sc-28036] Oracle query logs need to have _id and sql_hash (#945)

* [sc-28036] Oracle query logs need to have _id and sql_hash

* bump version

metaphor/oracle/extractor.py

@@ -5,7 +5,7 @@
 from metaphor.common.entity_id import dataset_normalized_name
 from metaphor.common.event_util import ENTITY_TYPES
 from metaphor.common.logger import get_logger
-from metaphor.common.utils import start_of_day, to_utc_time
+from metaphor.common.utils import md5_digest, start_of_day, to_utc_time
 from metaphor.database.extractor import GenericDatabaseExtractor
 from metaphor.models.crawler_run_metadata import Platform
 from metaphor.models.metadata_change_event import (
@@ -202,12 +202,14 @@ def _extract_query_logs(self, inspector: Inspector, excluded_users: List[str]):
             """
 
             cursor = connection.execute(text(sql))
-            for user, query, start, duration, sql_id in cursor:
+            for user, query, start, duration, query_id in cursor:
                 yield QueryLog(
-                    query_id=sql_id,
+                    id=f"{DataPlatform.ORACLE.name}:{query_id}",
+                    query_id=query_id,
                     platform=DataPlatform.ORACLE,
                     user_id=user,
                     sql=query,
+                    sql_hash=md5_digest(query.encode("utf-8")),
                     duration=float(duration),
                     start_time=to_utc_time(start),
                 )

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "metaphor-connectors"
-version = "0.14.73"
+version = "0.14.74"
 license = "Apache-2.0"
 description = "A collection of Python-based 'connectors' that extract metadata from various sources to ingest into the Metaphor app."
 authors = ["Metaphor <[email protected]>"]

tests/oracle/expected_query_logs.json

@@ -1,10 +1,12 @@
 [
   {
+    "_id": "ORACLE:sql-id",
     "duration": 10.0,
     "platform": "ORACLE",
     "queryId": "sql-id",
     "sql": "SELECT...",
     "startTime": "2024-07-30T15:31:33+00:00",
+    "sqlHash": "191df6d782898cbb739c413fa5868422",
     "userId": "DEV"
   }
 ]