optimize query

getsentry · Dec 19, 2024 · 9b4396c · 9b4396c
1 parent fe64d03
commit 9b4396c
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 16 deletions.
diff --git a/snuba/manual_jobs/scrub_users_from_eap_spans.py b/snuba/manual_jobs/scrub_users_from_eap_spans.py
@@ -29,14 +29,7 @@ def _get_query(self, cluster_name: str | None) -> str:
         on_cluster = f"ON CLUSTER '{cluster_name}'" if cluster_name else ""
         return f"""ALTER TABLE eap_spans_2_local
 {on_cluster}
-UPDATE `attr_str_2` = mapApply((k, v) -> (k, if(k = 'user' AND startsWith(v, '{_IP_PREFIX}'), concat(
-                    '{_IP_PREFIX}',
-                    if(
-                        isIPv4String(substring(v, 4)) OR isIPv6String(substring(v, 4)),
-                        '{_SCRUBBED}',
-                        substring(v, 4)
-                    )
-                ), v)), `attr_str_2`)
+UPDATE `attr_str_2` = mapApply((k, v) -> (k, if(k = 'user' AND startsWith(v, '{_IP_PREFIX}') AND (isIPv4String(substring(v, 4)) OR isIPv6String(substring(v, 4))), 'ip:scrubbed', v)), `attr_str_2`)
 WHERE organization_id IN [{organization_ids}]
 AND _sort_timestamp >= toDateTime('{start_datetime}')
 AND _sort_timestamp < toDateTime('{end_datetime}')"""

diff --git a/tests/manual_jobs/test_scrub_users_from_eap_spans.py b/tests/manual_jobs/test_scrub_users_from_eap_spans.py
@@ -111,14 +111,7 @@ def test_generate_query() -> None:
         job._get_query(None)
         == """ALTER TABLE eap_spans_2_local
 
-UPDATE `attr_str_2` = mapApply((k, v) -> (k, if(k = 'user' AND startsWith(v, 'ip:'), concat(
-                    'ip:',
-                    if(
-                        isIPv4String(substring(v, 4)) OR isIPv6String(substring(v, 4)),
-                        'scrubbed',
-                        substring(v, 4)
-                    )
-                ), v)), `attr_str_2`)
+UPDATE `attr_str_2` = mapApply((k, v) -> (k, if(k = 'user' AND startsWith(v, 'ip:') AND (isIPv4String(substring(v, 4)) OR isIPv6String(substring(v, 4))), 'ip:scrubbed', v)), `attr_str_2`)
 WHERE organization_id IN [1,3,5,6]
 AND _sort_timestamp >= toDateTime('2024-12-01T00:00:00')
 AND _sort_timestamp < toDateTime('2024-12-10T00:00:00')"""