Tech Report: Category total origins by client (#50)

* by client * clients in a record * fix aggregation * Update definitions/output/reports/cwv_tech_categories.js Co-authored-by: Barry Pollard <[email protected]> * Update definitions/output/reports/cwv_tech_categories.js Co-authored-by: Barry Pollard <[email protected]> * consistent naming * category sort * cross-platform origins * distinct technologies --------- Co-authored-by: Barry Pollard <[email protected]>
HTTPArchive · Jan 20, 2025 · 68fce03 · 68fce03
1 parent 295a45d
commit 68fce03
Showing 1 changed file with 38 additions and 22 deletions.
diff --git a/definitions/output/reports/cwv_tech_categories.js b/definitions/output/reports/cwv_tech_categories.js
@@ -7,53 +7,69 @@ publish('cwv_tech_categories', {
 }).query(ctx => `
 /* {"dataform_trigger": "report_cwv_tech_complete", "name": "categories", "type": "dict"} */
 WITH pages AS (
-  SELECT
+  SELECT DISTINCT
+    client,
     root_page,
     technologies
   FROM ${ctx.ref('crawl', 'pages')}
   WHERE
-    date = '${pastMonth}' AND
-    client = 'mobile'
+    date = '${pastMonth}'
     ${constants.devRankFilter}
-), categories AS (
+),
+
+category_descriptions AS (
   SELECT
     name AS category,
     description
   FROM ${ctx.ref('wappalyzer', 'categories')}
-), category_stats AS (
+),
+
+category_stats AS (
   SELECT
     category,
-    COUNT(DISTINCT root_page) AS origins
-  FROM pages,
-    UNNEST(technologies) AS t,
-    UNNEST(t.categories) AS category
+    STRUCT(
+      COALESCE(MAX(IF(client = 'desktop', origins, 0))) AS desktop,
+      COALESCE(MAX(IF(client = 'mobile', origins, 0))) AS mobile
+    ) AS origins
+  FROM (
+    SELECT
+      client,
+      category,
+      COUNT(DISTINCT root_page) AS origins
+    FROM pages
+    LEFT JOIN pages.technologies AS tech
+    LEFT JOIN tech.categories AS category
+    GROUP BY
+      client,
+      category
+  )
   GROUP BY category
-), technology_stats AS (
+),
+
+technology_stats AS (
   SELECT
-    category,
     technology,
-    COUNT(DISTINCT root_page) AS origins
-  FROM pages,
-    UNNEST(technologies) AS t,
-    UNNEST(t.categories) AS category
+    category_obj AS categories,
+    SUM(origins) AS total_origins
+  FROM ${ctx.ref('reports', 'cwv_tech_technologies')}
   GROUP BY
-    category,
-    technology
+    technology,
+    categories
 )
 
 SELECT
   category,
   description,
-  category_stats.origins,
-  ARRAY_AGG(technology IGNORE NULLS ORDER BY technology_stats.origins DESC) AS technologies
+  origins,
+  ARRAY_AGG(technology IGNORE NULLS ORDER BY technology_stats.total_origins DESC) AS technologies
 FROM category_stats
 INNER JOIN technology_stats
-USING (category)
-LEFT JOIN categories
+ON category_stats.category IN UNNEST(technology_stats.categories)
+INNER JOIN category_descriptions
 USING (category)
 GROUP BY
   category,
   description,
   origins
-ORDER BY origins DESC
+ORDER BY category ASC
 `)