Merge branch 'main' of github.com:HTTPArchive/almanac.httparchive.org…

… into production
HTTPArchive · Jun 5, 2024 · 4538b65 · 4538b65
2 parents 5e8aba5 + f661a30
commit 4538b65
Show file tree

Hide file tree

Showing 85 changed files with 2,688 additions and 194 deletions.
diff --git a/sql/2023/performance/bfcache_cachecontrol_nostore.sql b/sql/2023/performance/bfcache_cachecontrol_nostore.sql
@@ -0,0 +1,29 @@
+CREATE TEMP FUNCTION HAS_NO_STORE_DIRECTIVE(cache_control STRING) RETURNS BOOL AS (
+  REGEXP_CONTAINS(cache_control, r'(?i)\bno-store\b')
+);
+
+WITH requests AS (
+  SELECT
+    client,
+    LOGICAL_OR(HAS_NO_STORE_DIRECTIVE(JSON_VALUE(payload, '$._cacheControl'))) AS includes_ccns
+  FROM
+    `httparchive.all.requests`
+  WHERE
+    date = '2023-10-01' AND
+    is_main_document
+  GROUP BY
+    client,
+    page
+)
+
+SELECT
+  client,
+  COUNTIF(includes_ccns) AS pages,
+  COUNT(0) AS total,
+  COUNTIF(includes_ccns) / COUNT(0) AS pct
+FROM
+  requests
+GROUP BY
+  client
+ORDER BY
+  client
diff --git a/sql/2023/performance/bfcache_unload.sql b/sql/2023/performance/bfcache_unload.sql
@@ -0,0 +1,31 @@
+WITH lh AS (
+  SELECT
+    client,
+    page,
+    rank,
+    JSON_VALUE(lighthouse, '$.audits.no-unload-listeners.score') = '0' AS has_unload
+  FROM
+    `httparchive.all.pages`
+  WHERE
+    date = '2023-10-01' AND
+    is_root_page
+)
+
+
+SELECT
+  client,
+  _rank AS rank,
+  COUNTIF(has_unload) AS pages,
+  COUNT(0) AS total,
+  COUNTIF(has_unload) / COUNT(0) AS pct
+FROM
+  lh,
+  UNNEST([1000, 10000, 100000, 1000000, 10000000, 100000000]) AS _rank
+WHERE
+  rank <= _rank
+GROUP BY
+  client,
+  rank
+ORDER BY
+  rank,
+  client
diff --git a/sql/2023/performance/cls_animations.sql b/sql/2023/performance/cls_animations.sql
@@ -0,0 +1,28 @@
+WITH lh AS (
+  SELECT
+    client,
+    ARRAY_LENGTH(JSON_QUERY_ARRAY(lighthouse, '$.audits.non-composited-animations.details.items')) AS num_animations
+  FROM
+    `httparchive.all.pages`
+  WHERE
+    date = '2023-10-01' AND
+    is_root_page
+)
+
+
+SELECT
+  percentile,
+  client,
+  APPROX_QUANTILES(num_animations, 1000)[OFFSET(percentile * 10)] AS num_animations,
+  COUNTIF(num_animations > 0) AS pages,
+  COUNT(0) AS total,
+  COUNTIF(num_animations > 0) / COUNT(0) AS pct
+FROM
+  lh,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2023/performance/cls_unsized_image_height.sql b/sql/2023/performance/cls_unsized_image_height.sql
@@ -0,0 +1,27 @@
+WITH lh AS (
+  SELECT
+    client,
+    CAST(JSON_VALUE(unsized_image, '$.node.boundingRect.height') AS INT64) AS height
+  FROM
+    `httparchive.all.pages`,
+    UNNEST(JSON_QUERY_ARRAY(lighthouse, '$.audits.unsized-images.details.items')) AS unsized_image
+  WHERE
+    date = '2023-10-01' AND
+    is_root_page
+)
+
+
+SELECT
+  percentile,
+  client,
+  APPROX_QUANTILES(height, 1000)[OFFSET(percentile * 10)] AS height,
+  COUNT(0) AS unsized_images
+FROM
+  lh,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2023/performance/cls_unsized_images.sql b/sql/2023/performance/cls_unsized_images.sql
@@ -0,0 +1,28 @@
+WITH lh AS (
+  SELECT
+    client,
+    ARRAY_LENGTH(JSON_QUERY_ARRAY(lighthouse, '$.audits.unsized-images.details.items')) AS num_unsized_images
+  FROM
+    `httparchive.all.pages`
+  WHERE
+    date = '2023-10-01' AND
+    is_root_page
+)
+
+
+SELECT
+  percentile,
+  client,
+  APPROX_QUANTILES(num_unsized_images, 1000)[OFFSET(percentile * 10)] AS num_unsized_images,
+  COUNTIF(num_unsized_images > 0) AS pages,
+  COUNT(0) AS total,
+  COUNTIF(num_unsized_images > 0) / COUNT(0) AS pct
+FROM
+  lh,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2023/performance/inp_long_tasks.sql b/sql/2023/performance/inp_long_tasks.sql
@@ -0,0 +1,36 @@
+WITH long_tasks AS (
+  SELECT
+    client,
+    page,
+    ANY_VALUE(httparchive.core_web_vitals.GET_CRUX_INP(payload)) AS inp,
+    SUM(CAST(JSON_QUERY(item, '$.duration') AS FLOAT64)) AS long_tasks
+  FROM
+    `httparchive.all.pages`,
+    UNNEST(JSON_QUERY_ARRAY(lighthouse, '$.audits.long-tasks.details.items')) AS item
+  WHERE
+    date = '2023-10-01' AND
+    is_root_page
+  GROUP BY
+    client,
+    page
+),
+
+meta AS (
+  SELECT
+    *,
+    COUNT(0) OVER (PARTITION BY client) AS n,
+    ROW_NUMBER() OVER (PARTITION BY client ORDER BY inp) AS row
+  FROM
+    long_tasks
+  WHERE
+    inp IS NOT NULL
+)
+
+SELECT
+  client,
+  long_tasks,
+  inp
+FROM
+  meta
+WHERE
+  MOD(row, CAST(FLOOR(n / 1000) AS INT64)) = 0
diff --git a/sql/2023/performance/inp_tbt.sql b/sql/2023/performance/inp_tbt.sql
@@ -0,0 +1,16 @@
+SELECT
+  percentile,
+  client,
+  APPROX_QUANTILES(CAST(JSON_QUERY(lighthouse, '$.audits.total-blocking-time.numericValue') AS FLOAT64), 1000)[OFFSET(percentile * 10)] AS tbt
+FROM
+  `httparchive.all.pages`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+WHERE
+  date = '2023-10-01' AND
+  is_root_page
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2023/performance/js_bytes_rank.sql b/sql/2023/performance/js_bytes_rank.sql
@@ -0,0 +1,17 @@
+SELECT
+  IF(_rank < 100000000, CAST(_rank AS STRING), 'all') AS rank,
+  client,
+  APPROX_QUANTILES(CAST(JSON_VALUE(summary, '$.bytesJS') AS INT64), 1000)[OFFSET(500)] / 1024 AS js_kbytes
+FROM
+  `httparchive.all.pages`,
+  UNNEST([1000, 10000, 100000, 1000000, 10000000, 100000000]) AS _rank
+WHERE
+  date = '2023-10-01' AND
+  is_root_page AND
+  rank <= _rank
+GROUP BY
+  rank,
+  client
+ORDER BY
+  rank,
+  client
diff --git a/sql/2023/performance/lcp_bytes_distribution.sql b/sql/2023/performance/lcp_bytes_distribution.sql
@@ -0,0 +1,42 @@
+WITH pages AS (
+  SELECT
+    client,
+    page,
+    JSON_VALUE(custom_metrics, '$.performance.lcp_elem_stats.url') AS url
+  FROM
+    `httparchive.all.pages`
+  WHERE
+    date = '2023-10-01' AND
+    is_root_page
+),
+
+requests AS (
+  SELECT
+    client,
+    page,
+    url,
+    CAST(JSON_VALUE(summary, '$.respSize') AS INT64) / 1024 AS kbytes
+  FROM
+    `httparchive.all.requests`
+  WHERE
+    date = '2023-10-01' AND
+    is_root_page
+)
+
+SELECT
+  percentile,
+  client,
+  APPROX_QUANTILES(kbytes, 1000)[OFFSET(percentile * 10)] AS kbytes
+FROM
+  pages
+JOIN
+  requests
+USING
+  (client, page, url),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2023/performance/lcp_bytes_histogram.sql b/sql/2023/performance/lcp_bytes_histogram.sql
@@ -0,0 +1,43 @@
+WITH pages AS (
+  SELECT
+    client,
+    page,
+    JSON_VALUE(custom_metrics, '$.performance.lcp_elem_stats.url') AS url
+  FROM
+    `httparchive.all.pages`
+  WHERE
+    date = '2023-10-01' AND
+    is_root_page
+),
+
+requests AS (
+  SELECT
+    client,
+    page,
+    url,
+    CAST(JSON_VALUE(summary, '$.respSize') AS INT64) / 1024 AS kbytes
+  FROM
+    `httparchive.all.requests`
+  WHERE
+    date = '2023-10-01' AND
+    is_root_page
+)
+
+SELECT
+  client,
+  IF(CEILING(kbytes / 100) * 100 < 1000, CAST(CEILING(kbytes / 100) * 100 AS STRING), '1000+') AS kbytes,
+  COUNT(0) AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct
+FROM
+  pages
+JOIN
+  requests
+USING
+  (client, page, url)
+GROUP BY
+  client,
+  kbytes
+ORDER BY
+  client,
+  kbytes