Bug 1860814 - fix amo_prod__desktop_addons_by_client (#4481)

* quick fix * fix spread out groupby * move out sourcetable query --------- Co-authored-by: Frank Bertsch <[email protected]>
mozilla · Dec 11, 2023 · ae0b290 · ae0b290
1 parent 461cfb0
commit ae0b290
Showing 1 changed file with 74 additions and 20 deletions.
diff --git a/sql/moz-fx-data-shared-prod/amo_prod/desktop_addons_by_client_v1/query.sql b/sql/moz-fx-data-shared-prod/amo_prod/desktop_addons_by_client_v1/query.sql
@@ -15,26 +15,27 @@ try {
 }
 """;
 
-WITH per_client AS (
+WITH filtered_main AS (
+  SELECT
+    submission_timestamp,
+    client_id,
+    sample_id,
+    payload.info.addons,
+    application.version,
+    normalized_country_code,
+    environment.settings.locale,
+    normalized_os
+  FROM
+    telemetry.main
+  WHERE
+    DATE(submission_timestamp) = @submission_date
+    AND client_id IS NOT NULL
+),
+per_clients_without_addons AS (
   SELECT
     DATE(submission_timestamp) AS submission_date,
     client_id,
     sample_id,
-    ARRAY_CONCAT_AGG(
-      -- As of 2020-07-17, we parse addon info from the payload.info.addons string
-      -- because it provides more complete information compared to environment.addons.active_addons
-      -- See https://bugzilla.mozilla.org/show_bug.cgi?id=1653499
-      -- Previous logic in https://github.com/mozilla/bigquery-etl/blob/55a429924beee3c31aca4fee0063d655f1d527f2/sql/amo_prod/desktop_addons_by_client_v1/query.sql
-      ARRAY(
-        SELECT AS STRUCT
-          decode_uri_component(REGEXP_EXTRACT(TRIM(addon), "(.+):")) AS id,
-          REGEXP_EXTRACT(TRIM(addon), ":(.+)") AS version
-        FROM
-          UNNEST(SPLIT(payload.info.addons, ",")) AS addon
-      )
-      ORDER BY
-        submission_timestamp
-    ) AS addons,
     -- We always want to take the most recent seen version per
     -- https://bugzilla.mozilla.org/show_bug.cgi?id=1693308
     ARRAY_AGG(
@@ -50,14 +51,67 @@ WITH per_client AS (
     ) AS locale,
     mozfun.stats.mode_last(ARRAY_AGG(normalized_os ORDER BY submission_timestamp)) AS app_os,
   FROM
-    telemetry.main
-  WHERE
-    DATE(submission_timestamp) = @submission_date
-    AND client_id IS NOT NULL
+    filtered_main
+  GROUP BY
+    submission_date,
+    sample_id,
+    client_id
+),
+per_clients_just_addons_base AS (
+  SELECT
+    DATE(submission_timestamp) AS submission_date,
+    client_id,
+    sample_id,
+    payload.info.addons
+  FROM
+    filtered_main
+  GROUP BY
+    submission_date,
+    client_id,
+    sample_id,
+    addons
+),
+per_clients_just_addons AS (
+  SELECT
+    submission_date,
+    client_id,
+    sample_id,
+    ARRAY_CONCAT_AGG(
+      -- As of 2020-07-17, we parse addon info from the payload.info.addons string
+      -- because it provides more complete information compared to environment.addons.active_addons
+      -- See https://bugzilla.mozilla.org/show_bug.cgi?id=1653499
+      -- Previous logic in https://github.com/mozilla/bigquery-etl/blob/55a429924beee3c31aca4fee0063d655f1d527f2/sql/amo_prod/desktop_addons_by_client_v1/query.sql
+      ARRAY(
+        SELECT AS STRUCT
+          decode_uri_component(REGEXP_EXTRACT(TRIM(addon), "(.+):")) AS id,
+          REGEXP_EXTRACT(TRIM(addon), ":(.+)") AS version
+        FROM
+          UNNEST(SPLIT(addons, ",")) AS addon
+      )
+    ) AS addons
+  FROM
+    per_clients_just_addons_base
   GROUP BY
     submission_date,
     sample_id,
     client_id
+),
+per_client AS (
+  SELECT
+    submission_date,
+    client_id,
+    sample_id,
+    addons,
+    app_version,
+    country,
+    locale,
+    app_os,
+  FROM
+    per_clients_without_addons
+  INNER JOIN
+    per_clients_just_addons
+  USING
+    (submission_date, client_id, sample_id)
 )
 SELECT
   * EXCEPT (addons),