Merge branch 'telemetry-fix' into 'enterprise'

fix(analytics): simplify source datapoint - address review feedback

See merge request dkinternal/testgen/dataops-testgen!210

Author: ci bot

README.md

@@ -135,6 +135,9 @@ export TG_DECRYPT_SALT=<encryption_salt>
 export TESTGEN_USERNAME=<username>
 export TESTGEN_PASSWORD=<password>
 
+# Set an arbitrary base64-encoded string to be used for signing authentication tokens
+export TG_JWT_HASHING_KEY=<base64_key>
+
 # Set an accessible path for storing application logs
 export TESTGEN_LOG_FILE_PATH=<path_for_logs>
 ```

docs/local_development.md

@@ -64,6 +64,7 @@ Create a `local.env` file with the following environment variables, replacing th
 export TESTGEN_DEBUG=yes
 export TESTGEN_LOG_TO_FILE=no
 export TG_ANALYTICS=no
+export TG_JWT_HASHING_KEY=<base64_key>
 export TESTGEN_USERNAME=<username>
 export TESTGEN_PASSWORD=<password>
 export TG_DECRYPT_SALT=<decrypt_salt>

testgen/__main__.py

@@ -113,19 +113,12 @@ def cli(ctx: Context, verbose: bool):
     help="The identifier for the table group used during a profile run. Use a table_group_id shown in list-table-groups.",
     default=None,
 )
-@click.option(
-    "-s",
-    "--source",
-    required=False,
-    type=click.STRING,
-    default="cli",
-)
-def run_profile(configuration: Configuration, table_group_id: str, source: str):
+def run_profile(configuration: Configuration, table_group_id: str):
     click.echo(f"run-profile with table_group_id: {table_group_id}")
     spinner = None
     if not configuration.verbose:
         spinner = MoonSpinner("Processing ... ")
-    message = run_profiling_queries(table_group_id, spinner=spinner, source=source)
+    message = run_profiling_queries(table_group_id, spinner=spinner)
     click.echo("\n" + message)
 
 
@@ -152,17 +145,10 @@ def run_profile(configuration: Configuration, table_group_id: str, source: str):
     required=False,
     default=None,
 )
-@click.option(
-    "-s",
-    "--source",
-    required=False,
-    type=click.STRING,
-    default="cli",
-)
 @pass_configuration
-def run_test_generation(configuration: Configuration, table_group_id: str, test_suite_key: str, generation_set: str, source: str):
+def run_test_generation(configuration: Configuration, table_group_id: str, test_suite_key: str, generation_set: str):
     LOG.info("CurrentStep: Generate Tests - Main Procedure")
-    message = run_test_gen_queries(table_group_id, test_suite_key, generation_set, source)
+    message = run_test_gen_queries(table_group_id, test_suite_key, generation_set)
     LOG.info("Current Step: Generate Tests - Main Procedure Complete")
     click.echo("\n" + message)
 
@@ -184,20 +170,13 @@ def run_test_generation(configuration: Configuration, table_group_id: str, test_
     required=False,
     default=settings.DEFAULT_TEST_SUITE_KEY,
 )
-@click.option(
-    "-s",
-    "--source",
-    required=False,
-    type=click.STRING,
-    default="cli",
-)
 @pass_configuration
-def run_tests(configuration: Configuration, project_key: str, test_suite_key: str, source: str):
+def run_tests(configuration: Configuration, project_key: str, test_suite_key: str):
     click.echo(f"run-tests for suite: {test_suite_key}")
     spinner = None
     if not configuration.verbose:
         spinner = MoonSpinner("Processing ... ")
-    message = run_execution_steps(project_key, test_suite_key, spinner=spinner, source=source)
+    message = run_execution_steps(project_key, test_suite_key, spinner=spinner)
     click.echo("\n" + message)
 
 
@@ -660,6 +639,7 @@ def run_ui():
                 "--",
                 f"{'--debug' if settings.IS_DEBUG else ''}",
             ],
+            env={**os.environ, "TG_JOB_SOURCE": "UI"}
         )
     except Exception:
         LOG.exception(f"Testgen UI exited with status code {status_code}")

testgen/commands/run_execute_cat_tests.py

@@ -1,6 +1,7 @@
 import logging
 from datetime import UTC, datetime
 
+from testgen import settings
 from testgen.commands.queries.execute_cat_tests_query import CCATExecutionSQL
 from testgen.commands.run_refresh_score_cards_results import run_refresh_score_cards_results
 from testgen.common import (
@@ -62,20 +63,13 @@ def ParseCATResults(clsCATExecute):
     RunActionQueryList("DKTG", [strQuery])
 
 
-def FinalizeTestRun(clsCATExecute: CCATExecutionSQL, source: str):
+def FinalizeTestRun(clsCATExecute: CCATExecutionSQL):
     _, row_counts = RunActionQueryList(("DKTG"), [
         clsCATExecute.FinalizeTestResultsSQL(),
         clsCATExecute.PushTestRunStatusUpdateSQL(),
         clsCATExecute.FinalizeTestSuiteUpdateSQL(),
     ])
-
-    MixpanelService().send_event(
-        "run-tests",
-        source=source,
-        sql_flavor=clsCATExecute.flavor,
-        test_count=row_counts[0],
-        duration=(datetime.now(UTC) - date_service.parse_now(clsCATExecute.run_date)).total_seconds(),
-    )
+    end_time = datetime.now(UTC)
 
     RunActionQueryList(("DKTG"), [
         clsCATExecute.CalcPrevalenceTestResultsSQL(),
@@ -88,9 +82,18 @@ def FinalizeTestRun(clsCATExecute: CCATExecutionSQL, source: str):
         refresh_date=date_service.parse_now(clsCATExecute.run_date),
     )
 
+    MixpanelService().send_event(
+        "run-tests",
+        source=settings.ANALYTICS_JOB_SOURCE,
+        sql_flavor=clsCATExecute.flavor,
+        test_count=row_counts[0],
+        run_duration=(end_time - date_service.parse_now(clsCATExecute.run_date)).total_seconds(),
+        scoring_duration=(datetime.now(UTC) - end_time).total_seconds(),
+    )
+
 
 def run_cat_test_queries(
-    dctParms, strTestRunID, strTestTime, strProjectCode, strTestSuite, error_msg, minutes_offset=0, spinner=None, source=None
+    dctParms, strTestRunID, strTestTime, strProjectCode, strTestSuite, error_msg, minutes_offset=0, spinner=None
 ):
     booErrors = False
 
@@ -160,4 +163,4 @@ def run_cat_test_queries(
 
     finally:
         LOG.info("Finalizing test run")
-        FinalizeTestRun(clsCATExecute, source)
+        FinalizeTestRun(clsCATExecute)

testgen/commands/run_execute_tests.py

@@ -104,7 +104,6 @@ def run_execution_steps_in_background(project_code, test_suite):
         background_thread = threading.Thread(
             target=run_execution_steps,
             args=(project_code, test_suite),
-            kwargs={"source": "ui"},
         )
         background_thread.start()
     else:
@@ -118,7 +117,6 @@ def run_execution_steps(
     test_suite: str,
     minutes_offset: int=0,
     spinner: Spinner=None,
-    source: str | None=None,
 ) -> str:
     # Initialize required parms for all steps
     has_errors = False
@@ -169,7 +167,7 @@ def run_execution_steps(
 
     LOG.info("CurrentStep: Execute Step - CAT Test Execution")
     if run_cat_test_queries(
-        test_exec_params, test_run_id, test_time, project_code, test_suite, error_msg, minutes_offset, spinner, source
+        test_exec_params, test_run_id, test_time, project_code, test_suite, error_msg, minutes_offset, spinner
     ):
         has_errors = True
 

testgen/commands/run_generate_tests.py

@@ -1,13 +1,14 @@
 import logging
 
+from testgen import settings
 from testgen.commands.queries.generate_tests_query import CDeriveTestsSQL
 from testgen.common import AssignConnectParms, RetrieveDBResultsToDictList, RetrieveTestGenParms, RunActionQueryList
 from testgen.common.mixpanel_service import MixpanelService
 
 LOG = logging.getLogger("testgen")
 
 
-def run_test_gen_queries(strTableGroupsID, strTestSuite, strGenerationSet=None, source=None):
+def run_test_gen_queries(strTableGroupsID, strTestSuite, strGenerationSet=None):
     if strTableGroupsID is None:
         raise ValueError("Table Group ID was not specified")
 
@@ -112,7 +113,7 @@ def run_test_gen_queries(strTableGroupsID, strTestSuite, strGenerationSet=None,
 
     MixpanelService().send_event(
         "generate-tests",
-        source=source,
+        source=settings.ANALYTICS_JOB_SOURCE,
         sql_flavor=clsTests.sql_flavor,
         generation_set=clsTests.generation_set,
     )

testgen/commands/run_profiling_bridge.py

@@ -238,8 +238,7 @@ def run_profiling_in_background(table_group_id):
         empty_cache()
         background_thread = threading.Thread(
             target=run_profiling_queries,
-            args=(table_group_id),
-            kwargs={"source": "ui"},
+            args=(table_group_id,),
         )
         background_thread.start()
     else:
@@ -248,7 +247,7 @@ def run_profiling_in_background(table_group_id):
         subprocess.Popen(script)  # NOQA S603
 
 
-def run_profiling_queries(strTableGroupsID, spinner=None, source=None):
+def run_profiling_queries(strTableGroupsID, spinner=None):
     if strTableGroupsID is None:
         raise ValueError("Table Group ID was not specified")
 
@@ -504,16 +503,7 @@ def run_profiling_queries(strTableGroupsID, spinner=None, source=None):
         RunActionQueryList("DKTG", [
             clsProfiling.GetProfileRunInfoRecordUpdateQuery(),
         ])
-
-        MixpanelService().send_event(
-            "run-profiling",
-            source=source,
-            sql_flavor=clsProfiling.flavor,
-            sampling=clsProfiling.profile_use_sampling == "Y",
-            table_count=table_count,
-            column_count=column_count,
-            duration=(datetime.now(UTC) - date_service.parse_now(clsProfiling.run_date)).total_seconds(),
-        )
+        end_time = datetime.now(UTC)
 
         RunActionQueryList("DKTG", [
             clsProfiling.GetAnomalyScoringRollupRunQuery(),
@@ -525,6 +515,17 @@ def run_profiling_queries(strTableGroupsID, spinner=None, source=None):
             refresh_date=date_service.parse_now(clsProfiling.run_date),
         )
 
+        MixpanelService().send_event(
+            "run-profiling",
+            source=settings.ANALYTICS_JOB_SOURCE,
+            sql_flavor=clsProfiling.flavor,
+            sampling=clsProfiling.profile_use_sampling == "Y",
+            table_count=table_count,
+            column_count=column_count,
+            run_duration=(end_time - date_service.parse_now(clsProfiling.run_date)).total_seconds(),
+            scoring_duration=(datetime.now(UTC) - end_time).total_seconds(),
+        )
+
     return f"""
         Profiling completed {"with errors. Check log for details." if has_errors else "successfully."}
         Run ID: {profiling_run_id}

testgen/common/mixpanel_service.py

@@ -35,7 +35,7 @@ def instance_id(self):
 
     @cached_property
     def distinct_id(self):
-        return self._hash_value(session.username or self.instance_id)
+        return self._hash_value(session.username or "")
 
     def _hash_value(self, value: bytes | str, digest_size: int = 8) -> str:
         if isinstance(value, str):

testgen/scheduler/cli_scheduler.py

@@ -1,4 +1,5 @@
 import logging
+import os
 import signal
 import subprocess
 import sys
@@ -84,7 +85,7 @@ def start_job(self, job: CliJob, triggering_time: datetime) -> None:
 
         LOG.info("Executing  '%s'", " ".join(exec_cmd))
 
-        proc = subprocess.Popen(exec_cmd, start_new_session=True)  # noqa: S603
+        proc = subprocess.Popen(exec_cmd, start_new_session=True, env={**os.environ, "TG_JOB_SOURCE": "SCHEDULER"})  # noqa: S603
         threading.Thread(target=self._proc_wrapper, args=(proc,)).start()
 
     def _proc_wrapper(self, proc: subprocess.Popen):

testgen/settings.py

@@ -485,6 +485,11 @@
 Disables sending usage data when set to any value except "true" and "yes". Defaults to "yes"
 """
 
+ANALYTICS_JOB_SOURCE: str = os.getenv("TG_JOB_SOURCE", "CLI")
+"""
+Identifies the job trigger for analytics purposes.
+"""
+
 JWT_HASHING_KEY_B64: str = os.getenv("TG_JWT_HASHING_KEY")
 """
 Random key used to sign/verify the authentication token