split by CMS

sql/2019/14_CMS/14_13d.sql

@@ -0,0 +1,25 @@
+#standardSQL
+# 14_13d: Distribution of image stats per CMS
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  app,
+  COUNT(DISTINCT url) AS pages,
+  APPROX_QUANTILES(reqImg, 1000)[OFFSET(percentile * 10)] AS image_count,
+  ROUND(APPROX_QUANTILES(bytesImg, 1000)[OFFSET(percentile * 10)] / 1024, 2) AS image_kbytes
+FROM
+  `httparchive.summary_pages.2019_07_01_*`
+JOIN
+  `httparchive.technologies.2019_07_01_*`
+USING (_TABLE_SUFFIX, url),
+UNNEST([10, 25, 50, 75, 90]) AS percentile
+WHERE
+  category = 'CMS'
+GROUP BY
+  percentile,
+  client,
+  app
+ORDER BY
+  percentile,
+  client,
+  pages DESC

sql/2019/14_CMS/14_14b.sql

@@ -0,0 +1,23 @@
+#standardSQL
+# 14_14b: Distribution of HTML kilobytes per CMS per page
+SELECT
+  _TABLE_SUFFIX AS client,
+  app,
+  COUNT(DISTINCT url) AS pages,
+  ROUND(APPROX_QUANTILES(bytesHtml, 1000)[OFFSET(100)] / 1024, 2) AS p10,
+  ROUND(APPROX_QUANTILES(bytesHtml, 1000)[OFFSET(250)] / 1024, 2) AS p25,
+  ROUND(APPROX_QUANTILES(bytesHtml, 1000)[OFFSET(500)] / 1024, 2) AS p50,
+  ROUND(APPROX_QUANTILES(bytesHtml, 1000)[OFFSET(750)] / 1024, 2) AS p75,
+  ROUND(APPROX_QUANTILES(bytesHtml, 1000)[OFFSET(900)] / 1024, 2) AS p90
+FROM
+  `httparchive.summary_pages.2019_07_01_*`
+JOIN
+  `httparchive.technologies.2019_07_01_*`
+USING (_TABLE_SUFFIX, url)
+WHERE
+  category = 'CMS'
+GROUP BY
+  client,
+  app
+ORDER BY
+  pages DESC

sql/2019/14_CMS/14_18b.sql

@@ -0,0 +1,23 @@
+#standardSQL
+# 14_18b: Lighthouse indexability scores by CMS
+SELECT
+  app,
+  COUNTIF(crawlable = '1') AS passing,
+  COUNT(0) AS total,
+  ROUND(COUNTIF(crawlable = '1') * 100 / COUNT(0) , 2) AS pct
+FROM
+  `httparchive.technologies.2019_07_01_mobile`
+JOIN (
+  SELECT
+    url,
+    JSON_EXTRACT_SCALAR(report, '$.audits.is-crawlable.score') AS crawlable
+  FROM
+    `httparchive.lighthouse.2019_07_01_mobile`)
+USING (url)
+WHERE
+  category = 'CMS' AND
+  crawlable IS NOT NULL
+GROUP BY
+  app
+ORDER BY
+  total DESC

sql/2019/14_CMS/14_19b.sql

@@ -0,0 +1,17 @@
+#standardSQL
+# 14_19b: Lighthouse PWA scores by CMS
+SELECT
+  app,
+  APPROX_QUANTILES(CAST(JSON_EXTRACT_SCALAR(report, "$.categories.pwa.score") AS NUMERIC), 1000)[OFFSET(501)] AS median_pwa_score,
+  COUNT(0) AS pages
+FROM
+  `httparchive.lighthouse.2019_07_01_mobile`
+LEFT JOIN
+  `httparchive.technologies.2019_07_01_mobile`
+USING (url)
+WHERE
+  category = 'CMS'
+GROUP BY
+  app
+ORDER BY
+  pages DESC