queries: add blacklist to exlucde bots users (pingcap#1646)

wsxiaoys · Sep 26, 2023 · d58a6d5 · d58a6d5
1 parent 660c00a
commit d58a6d5
Show file tree

Hide file tree

Showing 10 changed files with 24 additions and 12 deletions.
diff --git a/configs/queries/orgs/participants/active/ranking/template.sql b/configs/queries/orgs/participants/active/ranking/template.sql
@@ -21,7 +21,8 @@ WHERE
     {% endcase %}
     {% if excludeBots %}
     -- Exclude bot users.
-    AND user_login NOT LIKE '%bot%'
+    AND LOWER(user_login) NOT LIKE '%bot%'
+    AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
     {% endif %}
 GROUP BY user_login
 ORDER BY 2 DESC

diff --git a/configs/queries/orgs/participants/active/total/template.sql b/configs/queries/orgs/participants/active/total/template.sql
@@ -24,7 +24,8 @@ WITH repos AS (
         {% endcase %}
         {% if excludeBots %}
         -- Exclude bot users.
-        AND user_login NOT LIKE '%bot%'
+        AND LOWER(user_login) NOT LIKE '%bot%'
+        AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
         {% endif %}
 ), past_period_active_participants AS (
     SELECT
@@ -44,7 +45,8 @@ WITH repos AS (
         {% endcase %}
         {% if excludeBots %}
         -- Exclude bot users.
-        AND user_login NOT LIKE '%bot%'
+        AND LOWER(user_login) NOT LIKE '%bot%'
+        AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
         {% endif %}
 )
 SELECT

diff --git a/configs/queries/orgs/participants/active/trends/template.sql b/configs/queries/orgs/participants/active/trends/template.sql
@@ -76,7 +76,8 @@ WITH RECURSIVE seq(idx, current_period_day, past_period_day) AS (
             {% endcase %}
             {% if excludeBots %}
             -- Exclude bot users.
-            AND user_login NOT LIKE '%bot%'
+            AND LOWER(user_login) NOT LIKE '%bot%'
+            AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
             {% endif %}
         GROUP BY 1
     ) sub

diff --git a/configs/queries/orgs/participants/engagements/template.sql b/configs/queries/orgs/participants/engagements/template.sql
@@ -23,7 +23,8 @@ WITH repos AS (
         {% endcase %}
         {% if excludeBots %}
         -- Exclude bot users.
-        AND user_login NOT LIKE '%bot%'
+        AND LOWER(user_login) NOT LIKE '%bot%'
+        AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
         {% endif %}
     GROUP BY user_login
 )

diff --git a/configs/queries/orgs/participants/locations/template.sql b/configs/queries/orgs/participants/locations/template.sql
@@ -55,7 +55,8 @@ WITH repos AS (
         {% endcase %}
         {% if excludeBots %}
         -- Exclude bot users.
-        AND ge.actor_login NOT LIKE '%bot%'
+        AND LOWER(ge.actor_login) NOT LIKE '%bot%'
+        AND ge.actor_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
         {% endif %}
         {% if excludeUnknown %}
         -- Exclude users with no country code.

diff --git a/configs/queries/orgs/participants/new/ranking/template.sql b/configs/queries/orgs/participants/new/ranking/template.sql
@@ -25,7 +25,8 @@ WHERE
     {% endcase %}
     {% if excludeBots %}
     -- Exclude bot users.
-    AND user_login NOT LIKE '%bot%'
+    AND LOWER(user_login) NOT LIKE '%bot%'
+    AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
     {% endif %}
 GROUP BY user_login
 ORDER BY first_participated_at DESC

diff --git a/configs/queries/orgs/participants/new/total/template.sql b/configs/queries/orgs/participants/new/total/template.sql
@@ -24,7 +24,8 @@ WITH repos AS (
         {% endcase %}
         {% if excludeBots %}
         -- Exclude bot users.
-        AND mrp.user_login NOT LIKE '%bot%'
+        AND LOWER(user_login) NOT LIKE '%bot%'
+        AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
         {% endif %}
 ), past_period_new_participants AS (
     SELECT
@@ -44,7 +45,8 @@ WITH repos AS (
         {% endcase %}
         {% if excludeBots %}
         -- Exclude bot users.
-        AND mrp.user_login NOT LIKE '%bot%'
+        AND LOWER(user_login) NOT LIKE '%bot%'
+        AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
         {% endif %}
 )
 SELECT

diff --git a/configs/queries/orgs/participants/new/trends/template.sql b/configs/queries/orgs/participants/new/trends/template.sql
@@ -90,7 +90,8 @@ WITH RECURSIVE seq(idx, current_period_day, past_period_day) AS (
             {% endcase %}
             {% if excludeBots %}
             -- Exclude bot users.
-            AND user_login NOT LIKE '%bot%'
+            AND LOWER(user_login) NOT LIKE '%bot%'
+            AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
             {% endif %}
             AND np.first_engagement_at >= day
         GROUP BY 1

diff --git a/configs/queries/orgs/participants/organizations/template.sql b/configs/queries/orgs/participants/organizations/template.sql
@@ -59,7 +59,8 @@ WITH repos AS (
         {% endcase %}
         {% if excludeBots %}
         -- Exclude bot users.
-        AND ge.actor_login NOT LIKE '%bot%'
+        AND LOWER(ge.actor_login) NOT LIKE '%bot%'
+        AND ge.actor_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
         {% endif %}
         {% if excludeUnknown %}
         -- Exclude users with no organization.

diff --git a/configs/queries/orgs/participants/trends/template.sql b/configs/queries/orgs/participants/trends/template.sql
@@ -76,7 +76,8 @@ WITH RECURSIVE seq(idx, current_period_day, past_period_day) AS (
             {% endcase %}
             {% if excludeBots %}
             -- Exclude bot users.
-            AND user_login NOT LIKE '%bot%'
+            AND LOWER(user_login) NOT LIKE '%bot%'
+            AND user_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
             {% endif %}
         GROUP BY 1
     ) sub