chore: include excluded repositories (CM-1006) (#3887)

joanagmaia · web-flow · commit 9d3e92ee1363 · 2026-03-09T16:50:09.000Z
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_0.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_0.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_1.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_1.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_2.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_2.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_3.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_3.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_4.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_4.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_5.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_5.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_6.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_6.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_7.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_7.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_8.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_8.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_9.pipe b/services/libs/tinybird/pipes/activityRelations_bucket_clean_enrich_copy_pipe_9.pipe
@@ -40,11 +40,7 @@ SQL >
                     SELECT r.url, r.segmentId
                     FROM repositories r FINAL
                     INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id
-                    WHERE
-                        isNull (r.deletedAt)
-                        AND isNull (i.deletedAt)
-                        AND r.excluded = false
-                        AND r.enabled = true
+                    WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true
                 )
             )
             OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')
diff --git a/services/libs/tinybird/pipes/activityRepositories_filtered.pipe b/services/libs/tinybird/pipes/activityRepositories_filtered.pipe
@@ -15,11 +15,7 @@ SQL >
     select i.id as "projectId", i.name as "projectName", i.slug as "projectSlug", r.url as repo
     from insightsProjects i final
     JOIN
-        repositories r FINAL
-        ON r.insightsProjectId = i.id
-        AND isNull (r.deletedAt)
-        AND r.enabled = true
-        AND r.excluded = false
+        repositories r FINAL ON r.insightsProjectId = i.id AND isNull (r.deletedAt) AND r.enabled = true
     where
         isNull (i.deletedAt)
         {% if defined(repos) %}
diff --git a/services/libs/tinybird/pipes/health_score_active_contributors.pipe b/services/libs/tinybird/pipes/health_score_active_contributors.pipe
@@ -11,6 +11,7 @@ SQL >
             memberId != ''
             AND (type, platform) IN (SELECT activityType, platform FROM activityTypes_filtered)
             AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND timestamp
@@ -29,6 +30,7 @@ SQL >
             AND (type, platform) IN (SELECT activityType, platform FROM activityTypes_filtered)
             AND timestamp >= toStartOfQuarter(now() - toIntervalQuarter(1))
             AND timestamp < toStartOfQuarter(now())
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         GROUP BY segmentId
     {% end %}
 
diff --git a/services/libs/tinybird/pipes/health_score_active_days.pipe b/services/libs/tinybird/pipes/health_score_active_days.pipe
@@ -6,6 +6,7 @@ SQL >
         FROM activityRelations_bucket_routing
         WHERE
             segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND timestamp
@@ -19,7 +20,10 @@ SQL >
     {% else %}
         SELECT segmentId, countDistinct(DATE(timestamp)) AS activeDaysCount
         FROM activityRelations_deduplicated_cleaned_bucket_union
-        WHERE timestamp >= toStartOfDay(now() - toIntervalDay(365)) AND timestamp < toStartOfDay(now())
+        WHERE
+            timestamp >= toStartOfDay(now() - toIntervalDay(365))
+            AND timestamp < toStartOfDay(now())
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         GROUP BY segmentId
     {% end %}
 
diff --git a/services/libs/tinybird/pipes/health_score_contributions_outside_work_hours.pipe b/services/libs/tinybird/pipes/health_score_contributions_outside_work_hours.pipe
@@ -12,6 +12,7 @@ SQL >
         1 = 1
         {% if defined(project) %}
             AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND timestamp
@@ -24,6 +25,7 @@ SQL >
         {% else %}
             AND timestamp >= toStartOfDay(now() - toIntervalDay(365))
             AND timestamp < toStartOfDay(now() + toIntervalDay(1))
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         {% end %}
     GROUP BY segmentId
 
diff --git a/services/libs/tinybird/pipes/health_score_contributor_dependency.pipe b/services/libs/tinybird/pipes/health_score_contributor_dependency.pipe
@@ -8,6 +8,7 @@ SQL >
             memberId != ''
             AND (type, platform) IN (SELECT activityType, platform FROM activityTypes_filtered)
             AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND timestamp
@@ -27,6 +28,7 @@ SQL >
             AND (type, platform) IN (SELECT activityType, platform FROM activityTypes_filtered)
             AND timestamp >= toStartOfDay(now() - INTERVAL 365 DAY)
             AND timestamp < toStartOfDay(now() + INTERVAL 1 DAY)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         GROUP BY segmentId, memberId
         ORDER by contributionCount DESC
     {% end %}
diff --git a/services/libs/tinybird/pipes/health_score_forks.pipe b/services/libs/tinybird/pipes/health_score_forks.pipe
@@ -8,7 +8,9 @@ SQL >
         SELECT segmentId, count() AS forks
         FROM activityRelations_bucket_routing
         WHERE
-            type = 'fork' AND segmentId = (SELECT segmentId FROM segments_filtered)
+            type = 'fork'
+            AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND timestamp
@@ -22,7 +24,7 @@ SQL >
     {% else %}
         SELECT segmentId, count() AS forks
         FROM activityRelations_deduplicated_cleaned_bucket_union
-        WHERE type = 'fork'
+        WHERE type = 'fork' AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         GROUP BY segmentId
     {% end %}
 
diff --git a/services/libs/tinybird/pipes/health_score_issues_resolution.pipe b/services/libs/tinybird/pipes/health_score_issues_resolution.pipe
@@ -15,7 +15,9 @@ SQL >
         SELECT segmentId, round(avg(closedInSeconds) / (60 * 60 * 24)) AS issueResolution
         FROM issues_analyzed
         WHERE
-            segmentId = (SELECT segmentId FROM segments_filtered) AND closedAt IS NOT NULL
+            segmentId = (SELECT segmentId FROM segments_filtered)
+            AND closedAt IS NOT NULL
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND openedAt
@@ -32,6 +34,7 @@ SQL >
             openedAt >= toStartOfDay(now()) - INTERVAL 365 DAY
             AND openedAt < toStartOfDay(now()) + INTERVAL 1 DAY
             AND closedAt IS NOT NULL
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         GROUP BY segmentId
     {% end %}
 
diff --git a/services/libs/tinybird/pipes/health_score_merge_lead_time.pipe b/services/libs/tinybird/pipes/health_score_merge_lead_time.pipe
@@ -10,6 +10,7 @@ SQL >
         1 = 1
         {% if defined(project) %}
             AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND openedAt
@@ -21,6 +22,7 @@ SQL >
         {% else %}
             AND openedAt >= toStartOfDay(now() - toIntervalDay(365))
             AND openedAt < toStartOfDay(now() + toIntervalDay(1))
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         {% end %}
     GROUP BY segmentId
 
diff --git a/services/libs/tinybird/pipes/health_score_organization_dependency.pipe b/services/libs/tinybird/pipes/health_score_organization_dependency.pipe
@@ -8,6 +8,7 @@ SQL >
             organizationId != ''
             AND (type, platform) IN (SELECT activityType, platform FROM activityTypes_filtered)
             AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND timestamp
@@ -26,6 +27,7 @@ SQL >
             AND (type, platform) IN (SELECT activityType, platform FROM activityTypes_filtered)
             AND timestamp >= toStartOfDay(now() - INTERVAL 365 DAY)
             AND timestamp < toStartOfDay(now() + INTERVAL 1 DAY)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         GROUP BY segmentId, organizationId
     {% end %}
 
diff --git a/services/libs/tinybird/pipes/health_score_pull_requests.pipe b/services/libs/tinybird/pipes/health_score_pull_requests.pipe
@@ -14,6 +14,7 @@ SQL >
                 OR type = 'changeset-created'
             )
             AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND timestamp
@@ -35,6 +36,7 @@ SQL >
             )
             AND timestamp >= toStartOfDay(now() - toIntervalDay(365))
             AND timestamp < toStartOfDay(now() + toIntervalDay(1))
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         GROUP BY segmentId
     {% end %}
 
diff --git a/services/libs/tinybird/pipes/health_score_retention.pipe b/services/libs/tinybird/pipes/health_score_retention.pipe
@@ -5,7 +5,9 @@ SQL >
         SELECT segmentId, groupUniqArray(memberId) AS currentQuarterMembers
         FROM activityRelations_bucket_routing
         WHERE
-            memberId != '' AND segmentId = (SELECT segmentId FROM segments_filtered)
+            memberId != ''
+            AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(endDate) %}
                 AND timestamp >= toStartOfQuarter(
@@ -28,7 +30,7 @@ SQL >
         SELECT segmentId, groupUniqArray(memberId) AS currentQuarterMembers
         FROM activityRelations_deduplicated_cleaned_bucket_union
         WHERE
-            memberId != ''
+            memberId != '' AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(endDate) %}
                 AND timestamp >= toStartOfQuarter(
                     parseDateTimeBestEffort(
@@ -55,7 +57,9 @@ SQL >
         SELECT segmentId, groupUniqArray(memberId) AS previousQuarterMembers
         FROM activityRelations_bucket_routing
         WHERE
-            memberId != '' AND segmentId = (SELECT segmentId FROM segments_filtered)
+            memberId != ''
+            AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(endDate) %}
                 AND timestamp >= toStartOfQuarter(
                     parseDateTimeBestEffort(
@@ -78,7 +82,7 @@ SQL >
         SELECT segmentId, groupUniqArray(memberId) AS previousQuarterMembers
         FROM activityRelations_deduplicated_cleaned_bucket_union
         WHERE
-            memberId != ''
+            memberId != '' AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(endDate) %}
                 AND timestamp >= toStartOfQuarter(
                     parseDateTimeBestEffort(
diff --git a/services/libs/tinybird/pipes/health_score_security.pipe b/services/libs/tinybird/pipes/health_score_security.pipe
@@ -5,7 +5,7 @@ SQL >
     FROM insightsProjects i FINAL
     JOIN repositories r FINAL ON r.insightsProjectId = i.id
     WHERE
-        isNull (i.deletedAt) AND isNull (r.deletedAt) AND r.enabled = true
+        isNull (i.deletedAt) AND isNull (r.deletedAt) AND r.enabled = true AND r.excluded = false
         {% if defined(project) %}
             AND i.id = (SELECT insightsProjectId FROM segments_filtered)
             AND r.url in (select arrayJoin(repositories) from segments_filtered)
diff --git a/services/libs/tinybird/pipes/health_score_stars.pipe b/services/libs/tinybird/pipes/health_score_stars.pipe
@@ -8,7 +8,9 @@ SQL >
         SELECT segmentId, count() AS stars
         FROM activityRelations_bucket_routing
         WHERE
-            type = 'star' AND segmentId = (SELECT segmentId FROM segments_filtered)
+            type = 'star'
+            AND segmentId = (SELECT segmentId FROM segments_filtered)
+            AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
             {% if defined(repos) %} AND channel IN (SELECT channel FROM repos_to_channels) {% end %}
             {% if defined(startDate) %}
                 AND timestamp
@@ -22,7 +24,7 @@ SQL >
     {% else %}
         SELECT segmentId, count() AS stars
         FROM activityRelations_deduplicated_cleaned_bucket_union
-        WHERE type = 'star'
+        WHERE type = 'star' AND channel NOT IN (SELECT channel FROM repos_to_channels(excluded = True))
         GROUP BY segmentId
     {% end %}
 
diff --git a/services/libs/tinybird/pipes/insights_projects_populated_copy.pipe b/services/libs/tinybird/pipes/insights_projects_populated_copy.pipe
@@ -69,11 +69,7 @@ SQL >
     SELECT i.id, r.url AS repository
     FROM insightsProjects i FINAL
     JOIN
-        repositories r FINAL
-        ON r.insightsProjectId = i.id
-        AND isNull (r.deletedAt)
-        AND r.enabled = true
-        AND r.excluded = false
+        repositories r FINAL ON r.insightsProjectId = i.id AND isNull (r.deletedAt) AND r.enabled = true
     WHERE isNull (i.deletedAt)
 
 NODE insights_projects_populated_copy_repository_criticality
diff --git a/services/libs/tinybird/pipes/repos_to_channels.pipe b/services/libs/tinybird/pipes/repos_to_channels.pipe
diff --git a/services/libs/tinybird/pipes/segments_filtered.pipe b/services/libs/tinybird/pipes/segments_filtered.pipe

Original file line number	Diff line number	Diff line change
`@@ -40,11 +40,7 @@ SQL >`
`40`	`40`	`SELECT r.url, r.segmentId`
`41`	`41`	`FROM repositories r FINAL`
`42`	`42`	`INNER JOIN insightsProjects i FINAL ON r.insightsProjectId = i.id`
`43`		`- WHERE`
`44`		`- isNull (r.deletedAt)`
`45`		`- AND isNull (i.deletedAt)`
`46`		`- AND r.excluded = false`
`47`		`- AND r.enabled = true`
	`43`	`+ WHERE isNull (r.deletedAt) AND isNull (i.deletedAt) AND r.enabled = true`
`48`	`44`	`)`
`49`	`45`	`)`
`50`	`46`	`OR platform NOT IN ('git', 'gerrit', 'github', 'gitlab')`