From 8ca1031bd81b73fbcb859f5be6be3a12500ebde9 Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Tue, 31 Aug 2021 01:56:52 +0530
Subject: [PATCH 01/23] add requests, async, module queries

---
 sql/2021/javascript/async_and_defer.sql       | 25 +++++++++++++
 ...ipts_using_async_defer_module_nomodule.sql | 36 +++++++++++++++++++
 ...ipts_using_async_defer_module_nomodule.sql | 34 ++++++++++++++++++
 sql/2021/javascript/requests_2019.sql         | 16 +++++++++
 sql/2021/javascript/requests_2020.sql         | 15 ++++++++
 sql/2021/javascript/requests_2021.sql         | 16 +++++++++
 6 files changed, 142 insertions(+)
 create mode 100644 sql/2021/javascript/async_and_defer.sql
 create mode 100644 sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql
 create mode 100644 sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql
 create mode 100644 sql/2021/javascript/requests_2019.sql
 create mode 100644 sql/2021/javascript/requests_2020.sql
 create mode 100644 sql/2021/javascript/requests_2021.sql

diff --git a/sql/2021/javascript/async_and_defer.sql b/sql/2021/javascript/async_and_defer.sql
new file mode 100644
index 00000000000..718e7201bd8
--- /dev/null
+++ b/sql/2021/javascript/async_and_defer.sql
@@ -0,0 +1,25 @@
+SELECT
+  client,
+  COUNT(DISTINCT IF(async, page, NULL)) AS async,
+  COUNT(DISTINCT IF(defer, page, NULL)) AS defer,
+  COUNT(DISTINCT IF(async AND defer, page, NULL)) AS both,
+  COUNT(DISTINCT page) AS total,
+  COUNT(DISTINCT IF(async, page, NULL)) / COUNT(DISTINCT page) AS pct_async,
+  COUNT(DISTINCT IF(defer, page, NULL)) / COUNT(DISTINCT page) AS pct_defer,
+  COUNT(DISTINCT IF(async AND defer, page, NULL)) / COUNT(DISTINCT page) AS pct_both
+FROM (
+  SELECT
+    client,
+    page,
+    script,
+    REGEXP_CONTAINS(script, r'(?i)\basync\b') AS async,
+    REGEXP_CONTAINS(script, r'(?i)\bdefer\b') AS defer
+  FROM
+    `httparchive.sample_data.summary_response_bodies`  -- TODO: update to the original table name
+  LEFT JOIN
+    UNNEST(REGEXP_EXTRACT_ALL(body, r'(?i)(<script[^>]*>)')) AS script
+  WHERE
+    date = '2021-08-01' AND
+    firstHtml)
+GROUP BY
+  client
diff --git a/sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql b/sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql
new file mode 100644
index 00000000000..1e06046cf91
--- /dev/null
+++ b/sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql
@@ -0,0 +1,36 @@
+#standardSQL
+# Average Percent per page of external scripts using Async, Defer, Module or NoModule attributes.
+SELECT
+  client,
+  AVG(pct_external_async) AS avg_pct_external_async,
+  AVG(pct_external_defer) AS avg_pct_external_defer,
+  AVG(pct_external_module) AS avg_pct_external_module,
+  AVG(pct_external_nomodule) AS avg_pct_external_nomodule
+FROM (
+  SELECT
+    client,
+    page,
+    COUNT(0) AS external_scripts,
+    COUNTIF(REGEXP_CONTAINS(script, r'\basync\b')) / COUNT(0) AS pct_external_async,
+    COUNTIF(REGEXP_CONTAINS(script, r'\bdefer\b')) / COUNT(0) AS pct_external_defer,
+    COUNTIF(REGEXP_CONTAINS(script, r'\bmodule\b')) / COUNT(0) AS pct_external_module,
+    COUNTIF(REGEXP_CONTAINS(script, r'\bnomodule\b')) / COUNT(0) AS pct_external_nomodule
+  FROM (
+    SELECT
+      client,
+      page,
+      url,
+      REGEXP_EXTRACT_ALL(body, "(?i)(<script [^>]*)") AS scripts
+    FROM
+      `httparchive.sample_data.summary_response_bodies`  -- TODO: update to the original table name
+    WHERE
+      date = '2021-08-01' AND
+      firstHtml),
+    UNNEST(scripts) AS script
+  WHERE
+    REGEXP_CONTAINS(script, r'\bsrc\b')
+  GROUP BY
+    client,
+    page)
+GROUP BY
+  client
diff --git a/sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql b/sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql
new file mode 100644
index 00000000000..2f43c8dd535
--- /dev/null
+++ b/sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql
@@ -0,0 +1,34 @@
+#standardSQL
+# Breakdown of scripts using Async, Defer, Module or NoModule attributes.  Also breakdown of inline vs external scripts
+SELECT
+  client,
+  COUNT(0) AS total_scripts,
+  SUM(IF(script NOT LIKE "%src%", 1, 0)) AS inline_script,
+  SUM(IF(script LIKE "%src%", 1, 0)) AS external_script,
+  SUM(IF(script LIKE "%src%", 1, 0)) / COUNT(0) AS pct_external_script,
+  SUM(IF(script NOT LIKE "%src%", 1, 0)) / COUNT(0) AS pct_inline_script,
+  SUM(IF(script LIKE "%async%", 1, 0)) AS async,
+  SUM(IF(script LIKE "%defer%", 1, 0)) AS defer,
+  SUM(IF(script LIKE "%module%", 1, 0)) AS module,
+  SUM(IF(script LIKE "%nomodule%", 1, 0)) AS nomodule,
+  SUM(IF(script LIKE "%async%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_async,
+  SUM(IF(script LIKE "%defer%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_defer,
+  SUM(IF(script LIKE "%module%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_module,
+  SUM(IF(script LIKE "%nomodule%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_nomodule
+FROM
+  (
+    SELECT
+      client,
+      page,
+      url,
+      REGEXP_EXTRACT_ALL(LOWER(body), "(<script [^>]*)") AS scripts
+    FROM
+      `httparchive.sample_data.summary_response_bodies`  -- TODO: update to the original table name
+    WHERE
+      date = '2021-08-01' AND
+      firstHtml
+  )
+CROSS JOIN
+  UNNEST(scripts) AS script
+GROUP BY
+  client
diff --git a/sql/2021/javascript/requests_2019.sql b/sql/2021/javascript/requests_2019.sql
new file mode 100644
index 00000000000..9d6feb06e9a
--- /dev/null
+++ b/sql/2021/javascript/requests_2019.sql
@@ -0,0 +1,16 @@
+#standardSQL
+# Sum of JS requests per page (2019)
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(reqJS, 1000)[OFFSET(percentile * 10)] AS js_requests
+FROM
+  `httparchive.summary_pages.2019_08_01_*`,
+  -- `httparchive.latest.summary_pages_*`
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/requests_2020.sql b/sql/2021/javascript/requests_2020.sql
new file mode 100644
index 00000000000..087b21ae176
--- /dev/null
+++ b/sql/2021/javascript/requests_2020.sql
@@ -0,0 +1,15 @@
+#standardSQL
+# Sum of JS requests per page (2020)
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(reqJS, 1000)[OFFSET(percentile * 10)] AS js_requests
+FROM
+  `httparchive.summary_pages.2020_08_01_*`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/requests_2021.sql b/sql/2021/javascript/requests_2021.sql
new file mode 100644
index 00000000000..ea9da50d48e
--- /dev/null
+++ b/sql/2021/javascript/requests_2021.sql
@@ -0,0 +1,16 @@
+#standardSQL
+# Sum of JS requests per page (2021)
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(reqJS, 1000)[OFFSET(percentile * 10)] AS js_requests
+FROM
+  `httparchive.latest.summary_pages_*`,
+  -- `httparchive.sample_data.summary_pages_*`
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client

From 36b0b4581499c48e2796fa498c5352ec0b5627cf Mon Sep 17 00:00:00 2001
From: Pankaj Parkar <parkarpankaj3@gmail.com>
Date: Sat, 11 Sep 2021 04:50:38 +0800
Subject: [PATCH 02/23] Added ajax request per page percentile based query.

---
 sql/2021/javascript/ajax_request_per_page.sql | 22 +++++++++++++++++++
 1 file changed, 22 insertions(+)
 create mode 100644 sql/2021/javascript/ajax_request_per_page.sql

diff --git a/sql/2021/javascript/ajax_request_per_page.sql b/sql/2021/javascript/ajax_request_per_page.sql
new file mode 100644
index 00000000000..142368f09a1
--- /dev/null
+++ b/sql/2021/javascript/ajax_request_per_page.sql
@@ -0,0 +1,22 @@
+
+SELECT
+  client,
+  percentile,
+  APPROX_QUANTILES(ajax_requests_total, 1000)[OFFSET(percentile * 10)] AS ajax_requests_total
+FROM
+  (
+    SELECT
+      _TABLE_SUFFIX AS client,
+      CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload, '$._javascript'),'$.ajax_requests.total') AS INT64) AS ajax_requests_total
+    FROM
+      `httparchive.sample_data.pages_*`
+  ),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+WHERE
+  ajax_requests_total > 0
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
\ No newline at end of file

From 7774a0a8d9a735489e9dcc68249eb58d9a790862 Mon Sep 17 00:00:00 2001
From: Pankaj Parkar <parkarpankaj3@gmail.com>
Date: Sat, 11 Sep 2021 04:58:09 +0800
Subject: [PATCH 03/23] Added most requested content-types.

---
 sql/2021/javascript/most_requested_content_type.sql | 6 ++++++
 1 file changed, 6 insertions(+)
 create mode 100644 sql/2021/javascript/most_requested_content_type.sql

diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
new file mode 100644
index 00000000000..60ff2fcacdb
--- /dev/null
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -0,0 +1,6 @@
+SELECT
+    resp_content_type,
+    COUNT(1)
+FROM `httparchive.sample_data.summary_requests_*`
+GROUP BY 1
+ORDER BY 2 DESC

From de19ca1fa5e10cfed50125f0dda606eac24ac171 Mon Sep 17 00:00:00 2001
From: Pankaj Parkar <parkarpankaj3@gmail.com>
Date: Sun, 12 Sep 2021 01:17:00 +0800
Subject: [PATCH 04/23] Formatted query.

---
 sql/2021/javascript/ajax_request_per_page.sql | 21 ++++++++++---------
 .../most_requested_content_type.sql           | 13 +++++++-----
 2 files changed, 19 insertions(+), 15 deletions(-)

diff --git a/sql/2021/javascript/ajax_request_per_page.sql b/sql/2021/javascript/ajax_request_per_page.sql
index 142368f09a1..96736e72c53 100644
--- a/sql/2021/javascript/ajax_request_per_page.sql
+++ b/sql/2021/javascript/ajax_request_per_page.sql
@@ -1,16 +1,17 @@
-
 SELECT
   client,
   percentile,
-  APPROX_QUANTILES(ajax_requests_total, 1000)[OFFSET(percentile * 10)] AS ajax_requests_total
-FROM
-  (
-    SELECT
-      _TABLE_SUFFIX AS client,
-      CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload, '$._javascript'),'$.ajax_requests.total') AS INT64) AS ajax_requests_total
-    FROM
-      `httparchive.sample_data.pages_*`
-  ),
+  APPROX_QUANTILES(ajax_requests_total, 1000)[
+OFFSET
+  (percentile * 10)] AS ajax_requests_total
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+          '$._javascript'),
+        '$.ajax_requests.total') AS INT64) AS ajax_requests_total
+  FROM
+    `httparchive.sample_data.pages_*` ),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 WHERE
   ajax_requests_total > 0
diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index 60ff2fcacdb..b0ce283d6c8 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -1,6 +1,9 @@
 SELECT
-    resp_content_type,
-    COUNT(1)
-FROM `httparchive.sample_data.summary_requests_*`
-GROUP BY 1
-ORDER BY 2 DESC
+  resp_content_type,
+  COUNT(1)
+FROM
+  `httparchive.sample_data.summary_requests_*`
+GROUP BY
+  1
+ORDER BY
+  2 DESC
\ No newline at end of file

From bc6b137c232d94a174fef0b122495c9b3c61ba5e Mon Sep 17 00:00:00 2001
From: Pankaj Parkar <parkarpankaj3@gmail.com>
Date: Mon, 13 Sep 2021 02:05:39 +0800
Subject: [PATCH 05/23] Added query for iframe per page.

---
 sql/2021/javascript/iframe_per_page.sql | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)
 create mode 100644 sql/2021/javascript/iframe_per_page.sql

diff --git a/sql/2021/javascript/iframe_per_page.sql b/sql/2021/javascript/iframe_per_page.sql
new file mode 100644
index 00000000000..b127b97b2e1
--- /dev/null
+++ b/sql/2021/javascript/iframe_per_page.sql
@@ -0,0 +1,23 @@
+SELECT
+  client,
+  percentile,
+  APPROX_QUANTILES(iframe_total, 1000)[
+OFFSET
+  (percentile * 10)] AS iframe_total
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+          '$._javascript'),
+        '$.iframe') AS INT64) AS iframe_total
+  FROM
+    `httparchive.sample_data.pages_*` ),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+WHERE
+  iframe_total > 0
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
\ No newline at end of file

From 895f031ecf3be0f96885d3742dc37f49ec32e833 Mon Sep 17 00:00:00 2001
From: Pankaj Parkar <parkarpankaj3@gmail.com>
Date: Mon, 13 Sep 2021 02:38:41 +0800
Subject: [PATCH 06/23] Percentage usage of different ajax APIs.

---
 ...ercentage_usage_of_different_ajax_apis.sql | 33 +++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 sql/2020/javascript/percentage_usage_of_different_ajax_apis.sql

diff --git a/sql/2020/javascript/percentage_usage_of_different_ajax_apis.sql b/sql/2020/javascript/percentage_usage_of_different_ajax_apis.sql
new file mode 100644
index 00000000000..b6a30ebd7bf
--- /dev/null
+++ b/sql/2020/javascript/percentage_usage_of_different_ajax_apis.sql
@@ -0,0 +1,33 @@
+SELECT
+  client,
+  percentile,
+  APPROX_QUANTILES(xmlhttprequest_total, 1000)[
+OFFSET
+  (percentile * 10)] AS xmlhttprequest_total,
+  APPROX_QUANTILES(fetch_total, 1000)[
+OFFSET
+  (percentile * 10)] AS fetch_total,
+  APPROX_QUANTILES(beacon_total, 1000)[
+OFFSET
+  (percentile * 10)] AS beacon_total
+FROM (
+    SELECT
+    _TABLE_SUFFIX AS client,
+    JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+        '$.ajax_requests.xmlhttprequest') AS xmlhttprequest_total,
+    JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+        '$.ajax_requests.fetch') AS fetch_total,
+    JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+        '$.ajax_requests.beacon') AS beacon_total
+    FROM
+    `httparchive.sample_data.pages_*`),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
\ No newline at end of file

From bf828ca127edb7813c91a95f44e4e709b39c8c70 Mon Sep 17 00:00:00 2001
From: Pankaj Parkar <parkarpankaj3@gmail.com>
Date: Mon, 13 Sep 2021 02:47:46 +0800
Subject: [PATCH 07/23] Added h1 vs h2 vs h3 script.

---
 sql/2021/javascript/h1_vs_h2_vs_h3.sql        | 42 +++++++++++++++++++
 ...ercentage_usage_of_different_ajax_apis.sql |  0
 2 files changed, 42 insertions(+)
 create mode 100644 sql/2021/javascript/h1_vs_h2_vs_h3.sql
 rename sql/{2020 => 2021}/javascript/percentage_usage_of_different_ajax_apis.sql (100%)

diff --git a/sql/2021/javascript/h1_vs_h2_vs_h3.sql b/sql/2021/javascript/h1_vs_h2_vs_h3.sql
new file mode 100644
index 00000000000..61ce229d356
--- /dev/null
+++ b/sql/2021/javascript/h1_vs_h2_vs_h3.sql
@@ -0,0 +1,42 @@
+SELECT
+  client,
+  percentile,
+  APPROX_QUANTILES((ajax_h1 + resources_h1), 1000)[
+OFFSET
+  (percentile * 10)] AS h1_request,
+  APPROX_QUANTILES((ajax_h2 + resources_h2), 1000)[
+OFFSET
+  (percentile * 10)] AS h2_request,
+  APPROX_QUANTILES((ajax_h3 + resources_h3), 1000)[
+OFFSET
+  (percentile * 10)] AS h3_request
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+     CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+      '$.requests_protocol.ajax_h1') AS INT64) AS ajax_h1,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+      '$.requests_protocol.resources_h1') AS INT64) AS resources_h1,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+      '$.requests_protocol.ajax_h2') AS INT64) AS ajax_h2,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+      '$.requests_protocol.resources_h2') AS INT64) AS resources_h2,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+      '$.requests_protocol.ajax_h3') AS INT64) AS ajax_h3,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+      '$.requests_protocol.resources_h3') AS INT64) AS resources_h3,
+  FROM
+    `httparchive.sample_data.pages_*`),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
\ No newline at end of file
diff --git a/sql/2020/javascript/percentage_usage_of_different_ajax_apis.sql b/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
similarity index 100%
rename from sql/2020/javascript/percentage_usage_of_different_ajax_apis.sql
rename to sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql

From c87eea5f89898c9a056a4bd85bace93dd0bfc6b5 Mon Sep 17 00:00:00 2001
From: Pankaj Parkar <parkarpankaj3@gmail.com>
Date: Mon, 13 Sep 2021 02:57:43 +0800
Subject: [PATCH 08/23] Added web_components_sepc sql.

---
 sql/2021/javascript/web_components_specs.sql | 33 ++++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 sql/2021/javascript/web_components_specs.sql

diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
new file mode 100644
index 00000000000..02da1a075a5
--- /dev/null
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -0,0 +1,33 @@
+SELECT
+  client,
+  percentile,
+  APPROX_QUANTILES(custom_elements, 1000)[
+OFFSET
+  (percentile * 10)] AS custom_elements,
+    APPROX_QUANTILES(shadow_roots, 1000)[
+OFFSET
+  (percentile * 10)] AS shadow_roots,
+    APPROX_QUANTILES(template, 1000)[
+OFFSET
+  (percentile * 10)] AS template
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+          '$._javascript'),
+        '$.web_component_specs.custom_elements') AS INT64) AS custom_elements,
+        CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+          '$._javascript'),
+        '$.web_component_specs.shadow_roots') AS INT64) AS shadow_roots,
+        CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+          '$._javascript'),
+        '$.web_component_specs.template') AS INT64) AS template
+  FROM
+    `httparchive.sample_data.pages_*` ),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
\ No newline at end of file

From e4c7b3083e74a4d89e96ad5890d4fd037a406504 Mon Sep 17 00:00:00 2001
From: Max Ostapenko <bvz2001@gmail.com>
Date: Thu, 16 Sep 2021 01:49:41 +0200
Subject: [PATCH 09/23] linting

---
 sql/2021/javascript/ajax_request_per_page.sql |  5 ++--
 sql/2021/javascript/h1_vs_h2_vs_h3.sql        | 15 +++++------
 sql/2021/javascript/iframe_per_page.sql       |  5 ++--
 .../most_requested_content_type.sql           |  2 +-
 ...ercentage_usage_of_different_ajax_apis.sql | 25 ++++++++-----------
 sql/2021/javascript/web_components_specs.sql  | 19 ++++++--------
 6 files changed, 30 insertions(+), 41 deletions(-)

diff --git a/sql/2021/javascript/ajax_request_per_page.sql b/sql/2021/javascript/ajax_request_per_page.sql
index 96736e72c53..93ba5343c0b 100644
--- a/sql/2021/javascript/ajax_request_per_page.sql
+++ b/sql/2021/javascript/ajax_request_per_page.sql
@@ -2,8 +2,7 @@ SELECT
   client,
   percentile,
   APPROX_QUANTILES(ajax_requests_total, 1000)[
-OFFSET
-  (percentile * 10)] AS ajax_requests_total
+    OFFSET(percentile * 10)] AS ajax_requests_total
 FROM (
   SELECT
     _TABLE_SUFFIX AS client,
@@ -20,4 +19,4 @@ GROUP BY
   client
 ORDER BY
   percentile,
-  client
\ No newline at end of file
+  client
diff --git a/sql/2021/javascript/h1_vs_h2_vs_h3.sql b/sql/2021/javascript/h1_vs_h2_vs_h3.sql
index 61ce229d356..5c4dad4c21a 100644
--- a/sql/2021/javascript/h1_vs_h2_vs_h3.sql
+++ b/sql/2021/javascript/h1_vs_h2_vs_h3.sql
@@ -2,18 +2,15 @@ SELECT
   client,
   percentile,
   APPROX_QUANTILES((ajax_h1 + resources_h1), 1000)[
-OFFSET
-  (percentile * 10)] AS h1_request,
+    OFFSET(percentile * 10)] AS h1_request,
   APPROX_QUANTILES((ajax_h2 + resources_h2), 1000)[
-OFFSET
-  (percentile * 10)] AS h2_request,
+    OFFSET(percentile * 10)] AS h2_request,
   APPROX_QUANTILES((ajax_h3 + resources_h3), 1000)[
-OFFSET
-  (percentile * 10)] AS h3_request
+    OFFSET(percentile * 10)] AS h3_request
 FROM (
   SELECT
     _TABLE_SUFFIX AS client,
-     CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
         '$._javascript'),
       '$.requests_protocol.ajax_h1') AS INT64) AS ajax_h1,
     CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
@@ -30,7 +27,7 @@ FROM (
       '$.requests_protocol.ajax_h3') AS INT64) AS ajax_h3,
     CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
         '$._javascript'),
-      '$.requests_protocol.resources_h3') AS INT64) AS resources_h3,
+      '$.requests_protocol.resources_h3') AS INT64) AS resources_h3
   FROM
     `httparchive.sample_data.pages_*`),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
@@ -39,4 +36,4 @@ GROUP BY
   client
 ORDER BY
   percentile,
-  client
\ No newline at end of file
+  client
diff --git a/sql/2021/javascript/iframe_per_page.sql b/sql/2021/javascript/iframe_per_page.sql
index b127b97b2e1..d80c8b07545 100644
--- a/sql/2021/javascript/iframe_per_page.sql
+++ b/sql/2021/javascript/iframe_per_page.sql
@@ -2,8 +2,7 @@ SELECT
   client,
   percentile,
   APPROX_QUANTILES(iframe_total, 1000)[
-OFFSET
-  (percentile * 10)] AS iframe_total
+    OFFSET(percentile * 10)] AS iframe_total
 FROM (
   SELECT
     _TABLE_SUFFIX AS client,
@@ -20,4 +19,4 @@ GROUP BY
   client
 ORDER BY
   percentile,
-  client
\ No newline at end of file
+  client
diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index b0ce283d6c8..9ce2f2005e5 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -6,4 +6,4 @@ FROM
 GROUP BY
   1
 ORDER BY
-  2 DESC
\ No newline at end of file
+  2 DESC
diff --git a/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql b/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
index b6a30ebd7bf..beb5ca5a2b2 100644
--- a/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
+++ b/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
@@ -2,32 +2,29 @@ SELECT
   client,
   percentile,
   APPROX_QUANTILES(xmlhttprequest_total, 1000)[
-OFFSET
-  (percentile * 10)] AS xmlhttprequest_total,
+    OFFSET(percentile * 10)] AS xmlhttprequest_total,
   APPROX_QUANTILES(fetch_total, 1000)[
-OFFSET
-  (percentile * 10)] AS fetch_total,
+    OFFSET(percentile * 10)] AS fetch_total,
   APPROX_QUANTILES(beacon_total, 1000)[
-OFFSET
-  (percentile * 10)] AS beacon_total
+    OFFSET(percentile * 10)] AS beacon_total
 FROM (
     SELECT
-    _TABLE_SUFFIX AS client,
-    JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+      _TABLE_SUFFIX AS client,
+      JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
         '$._javascript'),
         '$.ajax_requests.xmlhttprequest') AS xmlhttprequest_total,
-    JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-        '$._javascript'),
+      JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+          '$._javascript'),
         '$.ajax_requests.fetch') AS fetch_total,
-    JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-        '$._javascript'),
+      JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+          '$._javascript'),
         '$.ajax_requests.beacon') AS beacon_total
     FROM
-    `httparchive.sample_data.pages_*`),
+      `httparchive.sample_data.pages_*`),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
   client
 ORDER BY
   percentile,
-  client
\ No newline at end of file
+  client
diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
index 02da1a075a5..db9cf094386 100644
--- a/sql/2021/javascript/web_components_specs.sql
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -2,24 +2,21 @@ SELECT
   client,
   percentile,
   APPROX_QUANTILES(custom_elements, 1000)[
-OFFSET
-  (percentile * 10)] AS custom_elements,
-    APPROX_QUANTILES(shadow_roots, 1000)[
-OFFSET
-  (percentile * 10)] AS shadow_roots,
-    APPROX_QUANTILES(template, 1000)[
-OFFSET
-  (percentile * 10)] AS template
+    OFFSET(percentile * 10)] AS custom_elements,
+  APPROX_QUANTILES(shadow_roots, 1000)[
+    OFFSET(percentile * 10)] AS shadow_roots,
+  APPROX_QUANTILES(template, 1000)[
+    OFFSET(percentile * 10)] AS template
 FROM (
   SELECT
     _TABLE_SUFFIX AS client,
     CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
         '$.web_component_specs.custom_elements') AS INT64) AS custom_elements,
-        CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
         '$.web_component_specs.shadow_roots') AS INT64) AS shadow_roots,
-        CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
         '$.web_component_specs.template') AS INT64) AS template
   FROM
@@ -30,4 +27,4 @@ GROUP BY
   client
 ORDER BY
   percentile,
-  client
\ No newline at end of file
+  client

From 5017cb66000a33f4dad39bae4c3f98c36c19d453 Mon Sep 17 00:00:00 2001
From: Max Ostapenko <bvz2001@gmail.com>
Date: Thu, 16 Sep 2021 02:00:24 +0200
Subject: [PATCH 10/23] consistent syntax to count number of rows

---
 sql/2021/javascript/most_requested_content_type.sql | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index 9ce2f2005e5..b2649e99ce5 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -1,6 +1,6 @@
 SELECT
   resp_content_type,
-  COUNT(1)
+  COUNT(0) AS count
 FROM
   `httparchive.sample_data.summary_requests_*`
 GROUP BY

From 812e8dc825386704b669b64b96515a6b9dd57648 Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Sat, 25 Sep 2021 02:49:36 +0530
Subject: [PATCH 11/23] frameworks, sourcemaps queries

---
 sql/2021/javascript/compression_method.sql    | 47 ++++++++++++++++
 .../javascript/compression_method_by_3p.sql   | 52 ++++++++++++++++++
 .../javascript/compression_none_by_bytes.sql  | 55 +++++++++++++++++++
 .../frameworks-bytes-by-framework.sql         | 35 ++++++++++++
 sql/2021/javascript/frameworks_libraries.sql  | 30 ++++++++++
 .../frameworks_libraries_by_version.sql       | 32 +++++++++++
 .../frameworks_libraries_combos.sql           | 43 +++++++++++++++
 sql/2021/javascript/requests_2019.sql         |  1 -
 sql/2021/javascript/sourcemap_header.sql      | 48 ++++++++++++++++
 .../javascript/sourcemap_header_by_3p.sql     | 54 ++++++++++++++++++
 .../unused_js_bytes_distribution.sql          | 12 ++++
 11 files changed, 408 insertions(+), 1 deletion(-)
 create mode 100644 sql/2021/javascript/compression_method.sql
 create mode 100644 sql/2021/javascript/compression_method_by_3p.sql
 create mode 100644 sql/2021/javascript/compression_none_by_bytes.sql
 create mode 100644 sql/2021/javascript/frameworks-bytes-by-framework.sql
 create mode 100644 sql/2021/javascript/frameworks_libraries.sql
 create mode 100644 sql/2021/javascript/frameworks_libraries_by_version.sql
 create mode 100644 sql/2021/javascript/frameworks_libraries_combos.sql
 create mode 100644 sql/2021/javascript/sourcemap_header.sql
 create mode 100644 sql/2021/javascript/sourcemap_header_by_3p.sql
 create mode 100644 sql/2021/javascript/unused_js_bytes_distribution.sql

diff --git a/sql/2021/javascript/compression_method.sql b/sql/2021/javascript/compression_method.sql
new file mode 100644
index 00000000000..273e0e49551
--- /dev/null
+++ b/sql/2021/javascript/compression_method.sql
@@ -0,0 +1,47 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  compression,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total_js_requests,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct_js_requests
+FROM (
+  SELECT
+    client,
+    page,
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'Content-Encoding') AS compression
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script')
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  compression
+ORDER BY
+  pct_js_requests DESC
diff --git a/sql/2021/javascript/compression_method_by_3p.sql b/sql/2021/javascript/compression_method_by_3p.sql
new file mode 100644
index 00000000000..e9d00d1d453
--- /dev/null
+++ b/sql/2021/javascript/compression_method_by_3p.sql
@@ -0,0 +1,52 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  host,
+  compression,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client, host) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client, host) AS pct
+FROM (
+  SELECT
+    client,
+    page,
+    IF(NET.HOST(url) IN (
+      SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+    ), 'third party', 'first party') AS host,
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'Content-Encoding') AS compression
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script')
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  host,
+  compression
+ORDER BY
+  pct DESC
diff --git a/sql/2021/javascript/compression_none_by_bytes.sql b/sql/2021/javascript/compression_none_by_bytes.sql
new file mode 100644
index 00000000000..592248c542f
--- /dev/null
+++ b/sql/2021/javascript/compression_none_by_bytes.sql
@@ -0,0 +1,55 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  host,
+  IF(kbytes < 100, FLOOR(kbytes / 5) * 5, 100) AS kbytes,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client, host) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client, host) AS pct
+FROM (
+  SELECT
+    client,
+    page,
+    IF(NET.HOST(url) IN (
+      SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+    ), 'third party', 'first party') AS host,
+    respSize / 1024 AS kbytes
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script' AND
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'Content-Encoding') IS NULL)
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  host,
+  kbytes
+ORDER BY
+  client,
+  host,
+  kbytes
diff --git a/sql/2021/javascript/frameworks-bytes-by-framework.sql b/sql/2021/javascript/frameworks-bytes-by-framework.sql
new file mode 100644
index 00000000000..17d3c9a8535
--- /dev/null
+++ b/sql/2021/javascript/frameworks-bytes-by-framework.sql
@@ -0,0 +1,35 @@
+#standardSQL
+# Sum of JS request bytes per page by framework (2020)
+SELECT
+  percentile,
+  client,
+  app AS js_framework,
+  COUNT(DISTINCT page) AS pages,
+  APPROX_QUANTILES(bytesJs / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    bytesJs
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`)
+JOIN (
+  SELECT DISTINCT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    app
+  FROM
+    `httparchive.technologies.2021_07_01_*`
+  WHERE
+    category = 'JavaScript frameworks')
+USING
+  (client, page),
+  UNNEST([10, 25, 50, 75, 90]) AS percentile
+GROUP BY
+  percentile,
+  client,
+  js_framework
+ORDER BY
+  percentile,
+  client,
+  pages DESC
diff --git a/sql/2021/javascript/frameworks_libraries.sql b/sql/2021/javascript/frameworks_libraries.sql
new file mode 100644
index 00000000000..27b0627455f
--- /dev/null
+++ b/sql/2021/javascript/frameworks_libraries.sql
@@ -0,0 +1,30 @@
+#standardSQL
+# Top JS frameworks and libraries
+SELECT
+  _TABLE_SUFFIX AS client,
+  category,
+  app,
+  COUNT(DISTINCT url) AS pages,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.technologies.2021_07_01_*`
+JOIN (
+  SELECT
+    _TABLE_SUFFIX,
+    COUNT(0) AS total
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    _TABLE_SUFFIX)
+USING
+  (_TABLE_SUFFIX)
+WHERE
+  category IN ('JavaScript frameworks', 'JavaScript libraries')
+GROUP BY
+  client,
+  category,
+  app,
+  total
+ORDER BY
+  pct DESC
diff --git a/sql/2021/javascript/frameworks_libraries_by_version.sql b/sql/2021/javascript/frameworks_libraries_by_version.sql
new file mode 100644
index 00000000000..9d56b59fe2e
--- /dev/null
+++ b/sql/2021/javascript/frameworks_libraries_by_version.sql
@@ -0,0 +1,32 @@
+#standardSQL
+# Top JS frameworks and libraries by version
+SELECT
+  _TABLE_SUFFIX AS client,
+  category,
+  app,
+  info AS version,
+  COUNT(DISTINCT url) AS pages,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.technologies.2021_07_01_*`
+JOIN (
+  SELECT
+    _TABLE_SUFFIX,
+    COUNT(0) AS total
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    _TABLE_SUFFIX)
+USING
+  (_TABLE_SUFFIX)
+WHERE
+  app IN ('jQuery', 'jQuery Migrate', 'jQuery UI', 'Modernizr', 'FancyBox', 'Slick', 'Lightbox', 'Moment.js', 'Underscore.js', 'Lodash', 'React')
+GROUP BY
+  client,
+  category,
+  app,
+  info,
+  total
+ORDER BY
+  pct DESC
diff --git a/sql/2021/javascript/frameworks_libraries_combos.sql b/sql/2021/javascript/frameworks_libraries_combos.sql
new file mode 100644
index 00000000000..24bea43f43a
--- /dev/null
+++ b/sql/2021/javascript/frameworks_libraries_combos.sql
@@ -0,0 +1,43 @@
+#standardSQL
+# Top JS frameworks and libraries combinations
+SELECT
+  *
+FROM (
+  SELECT
+    client,
+    apps,
+    COUNT(DISTINCT page) AS pages,
+    total,
+    COUNT(DISTINCT page) / total AS pct
+  FROM (
+    SELECT
+      _TABLE_SUFFIX AS client,
+      url AS page,
+      total,
+      ARRAY_TO_STRING(ARRAY_AGG(app ORDER BY app), ', ') AS apps
+    FROM
+      `httparchive.technologies.2021_07_01_*`
+    JOIN (
+      SELECT
+        _TABLE_SUFFIX,
+        COUNT(0) AS total
+      FROM
+        `httparchive.summary_pages.2021_07_01_*`
+      GROUP BY
+        _TABLE_SUFFIX)
+    USING
+      (_TABLE_SUFFIX)
+    WHERE
+      category IN ('JavaScript frameworks', 'JavaScript libraries')
+    GROUP BY
+      client,
+      url,
+      total)
+  GROUP BY
+    client,
+    apps,
+    total)
+WHERE
+  pages >= 10000
+ORDER BY
+  pct DESC
diff --git a/sql/2021/javascript/requests_2019.sql b/sql/2021/javascript/requests_2019.sql
index 9d6feb06e9a..c4d2ccd8934 100644
--- a/sql/2021/javascript/requests_2019.sql
+++ b/sql/2021/javascript/requests_2019.sql
@@ -6,7 +6,6 @@ SELECT
   APPROX_QUANTILES(reqJS, 1000)[OFFSET(percentile * 10)] AS js_requests
 FROM
   `httparchive.summary_pages.2019_08_01_*`,
-  -- `httparchive.latest.summary_pages_*`
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
diff --git a/sql/2021/javascript/sourcemap_header.sql b/sql/2021/javascript/sourcemap_header.sql
new file mode 100644
index 00000000000..2788df59861
--- /dev/null
+++ b/sql/2021/javascript/sourcemap_header.sql
@@ -0,0 +1,48 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  has_sourcemap_header,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total_js_requests,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct_js_requests
+FROM (
+  SELECT
+    client,
+    page,
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'SourceMap') IS NOT NULL AS has_sourcemap_header
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script')
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  has_sourcemap_header
+ORDER BY
+  client,
+  has_sourcemap_header
diff --git a/sql/2021/javascript/sourcemap_header_by_3p.sql b/sql/2021/javascript/sourcemap_header_by_3p.sql
new file mode 100644
index 00000000000..c863a5371a1
--- /dev/null
+++ b/sql/2021/javascript/sourcemap_header_by_3p.sql
@@ -0,0 +1,54 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  host,
+  has_sourcemap_header,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total_js_requests,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct_js_requests
+FROM (
+  SELECT
+    client,
+    page,
+    IF(NET.HOST(url) IN (
+      SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+    ), 'third party', 'first party') AS host,
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'SourceMap') IS NOT NULL AS has_sourcemap_header
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script')
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  host,
+  has_sourcemap_header
+ORDER BY
+  client,
+  host,
+  has_sourcemap_header
diff --git a/sql/2021/javascript/unused_js_bytes_distribution.sql b/sql/2021/javascript/unused_js_bytes_distribution.sql
new file mode 100644
index 00000000000..62872729613
--- /dev/null
+++ b/sql/2021/javascript/unused_js_bytes_distribution.sql
@@ -0,0 +1,12 @@
+#standardSQL
+# Distribution of unused JS request bytes per page
+SELECT
+  percentile,
+  APPROX_QUANTILES(CAST(JSON_EXTRACT_SCALAR(report, "$.audits.unused-javascript.details.overallSavingsBytes") AS INT64) / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM
+  `httparchive.lighthouse.2021_07_01_mobile`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile
+ORDER BY
+  percentile

From b95883bc45391153074a6e0b7a2df7ea6c52e1ef Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Sat, 25 Sep 2021 02:56:44 +0530
Subject: [PATCH 12/23] 1p vs 3p

---
 sql/2021/javascript/bytes_2019.sql  | 15 +++++++++++++
 sql/2021/javascript/bytes_2020.sql  | 15 +++++++++++++
 sql/2021/javascript/bytes_2021.sql  | 15 +++++++++++++
 sql/2021/javascript/bytes_by_3p.sql | 33 +++++++++++++++++++++++++++++
 4 files changed, 78 insertions(+)
 create mode 100644 sql/2021/javascript/bytes_2019.sql
 create mode 100644 sql/2021/javascript/bytes_2020.sql
 create mode 100644 sql/2021/javascript/bytes_2021.sql
 create mode 100644 sql/2021/javascript/bytes_by_3p.sql

diff --git a/sql/2021/javascript/bytes_2019.sql b/sql/2021/javascript/bytes_2019.sql
new file mode 100644
index 00000000000..3348585ad43
--- /dev/null
+++ b/sql/2021/javascript/bytes_2019.sql
@@ -0,0 +1,15 @@
+#standardSQL
+# Sum of JS request bytes per page (2019)
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(bytesJs / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM
+  `httparchive.summary_pages.2019_07_01_*`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/bytes_2020.sql b/sql/2021/javascript/bytes_2020.sql
new file mode 100644
index 00000000000..839c6139ca8
--- /dev/null
+++ b/sql/2021/javascript/bytes_2020.sql
@@ -0,0 +1,15 @@
+#standardSQL
+# Sum of JS request bytes per page (2020)
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(bytesJs / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM
+  `httparchive.summary_pages.2020_08_01_*`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/bytes_2021.sql b/sql/2021/javascript/bytes_2021.sql
new file mode 100644
index 00000000000..25f82a8ca4c
--- /dev/null
+++ b/sql/2021/javascript/bytes_2021.sql
@@ -0,0 +1,15 @@
+#standardSQL
+# Sum of JS request bytes per page (2020)
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(bytesJs / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM
+  `httparchive.summary_pages.2021_07_01_*`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/bytes_by_3p.sql b/sql/2021/javascript/bytes_by_3p.sql
new file mode 100644
index 00000000000..f5bde4b9d9f
--- /dev/null
+++ b/sql/2021/javascript/bytes_by_3p.sql
@@ -0,0 +1,33 @@
+#standardSQL
+# Distribution of 1P/3P JS bytes
+SELECT
+  percentile,
+  client,
+  host,
+  APPROX_QUANTILES(kbytes, 1000)[OFFSET(percentile * 10)] AS kbytes
+FROM (
+  SELECT
+    client,
+    page,
+    IF(NET.HOST(url) IN (
+      SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+    ), 'third party', 'first party') AS host,
+    SUM(respSize) / 1024 AS kbytes
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script'
+  GROUP BY
+    client,
+    page,
+    host),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client,
+  host
+ORDER BY
+  percentile,
+  client,
+  host

From 7a70e3e3f448791db444dfbaa7daaa498bb181b6 Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Sat, 25 Sep 2021 03:01:54 +0530
Subject: [PATCH 13/23] adds unminified, preload/prefetch queries

---
 .../javascript/lighthouse_unminified_js.sql   | 18 ++++++
 .../lighthouse_unminified_js_by_3p.sql        | 34 +++++++++++
 .../lighthouse_unminified_js_bytes.sql        | 31 ++++++++++
 .../lighthouse_unminified_js_requests.sql     | 31 ++++++++++
 ...urce-hints-prefetch-preload-percentage.sql | 49 ++++++++++++++++
 ...ce-hints-preload-prefetch-distribution.sql | 57 +++++++++++++++++++
 sql/2021/javascript/resource_hints.sql        | 46 +++++++++++++++
 .../javascript/resource_hints_per_page.sql    | 54 ++++++++++++++++++
 8 files changed, 320 insertions(+)
 create mode 100644 sql/2021/javascript/lighthouse_unminified_js.sql
 create mode 100644 sql/2021/javascript/lighthouse_unminified_js_by_3p.sql
 create mode 100644 sql/2021/javascript/lighthouse_unminified_js_bytes.sql
 create mode 100644 sql/2021/javascript/lighthouse_unminified_js_requests.sql
 create mode 100644 sql/2021/javascript/resource-hints-prefetch-preload-percentage.sql
 create mode 100644 sql/2021/javascript/resource-hints-preload-prefetch-distribution.sql
 create mode 100644 sql/2021/javascript/resource_hints.sql
 create mode 100644 sql/2021/javascript/resource_hints_per_page.sql

diff --git a/sql/2021/javascript/lighthouse_unminified_js.sql b/sql/2021/javascript/lighthouse_unminified_js.sql
new file mode 100644
index 00000000000..a5e48687936
--- /dev/null
+++ b/sql/2021/javascript/lighthouse_unminified_js.sql
@@ -0,0 +1,18 @@
+#standardSQL
+# Pages with unminified JS
+SELECT
+  score,
+  COUNT(0) AS pages,
+  SUM(COUNT(0)) OVER () AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER () AS pct
+FROM (
+  SELECT
+    JSON_EXTRACT_SCALAR(report, "$.audits['unminified-javascript'].score") AS score
+  FROM
+    `httparchive.lighthouse.2021_07_01_mobile`)
+WHERE
+  score IS NOT NULL
+GROUP BY
+  score
+ORDER BY
+  score
diff --git a/sql/2021/javascript/lighthouse_unminified_js_by_3p.sql b/sql/2021/javascript/lighthouse_unminified_js_by_3p.sql
new file mode 100644
index 00000000000..e93e86bd657
--- /dev/null
+++ b/sql/2021/javascript/lighthouse_unminified_js_by_3p.sql
@@ -0,0 +1,34 @@
+#standardSQL
+# Pages with unminified JS by 1P/3P
+CREATE TEMPORARY FUNCTION getUnminifiedJsUrls(audit STRING)
+RETURNS ARRAY<STRUCT<url STRING, wastedBytes INT64>> LANGUAGE js AS '''
+try {
+  var $ = JSON.parse(audit);
+  return $.details.items.map(({url, wastedBytes}) => {
+    return {url, wastedBytes};
+  });
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  AVG(pct_1p_wasted_bytes) AS avg_pct_1p_wasted_bytes,
+  AVG(pct_3p_wasted_bytes) AS avg_pct_3p_wasted_bytes
+FROM (
+  SELECT
+    page,
+    SUM(IF(is_3p, 0, wasted_bytes)) / SUM(wasted_bytes) AS pct_1p_wasted_bytes,
+    SUM(IF(is_3p, wasted_bytes, 0)) / SUM(wasted_bytes) AS pct_3p_wasted_bytes
+  FROM (
+    SELECT
+      test.url AS page,
+      NET.HOST(unminified.url) IS NOT NULL AND NET.HOST(unminified.url) IN (
+        SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+      ) AS is_3p,
+      unminified.wastedBytes AS wasted_bytes
+    FROM
+      `httparchive.lighthouse.2021_07_01_mobile` AS test,
+      UNNEST(getUnminifiedJsUrls(JSON_EXTRACT(report, "$.audits['unminified-javascript']"))) AS unminified)
+  GROUP BY
+    page)
diff --git a/sql/2021/javascript/lighthouse_unminified_js_bytes.sql b/sql/2021/javascript/lighthouse_unminified_js_bytes.sql
new file mode 100644
index 00000000000..923cd16e87c
--- /dev/null
+++ b/sql/2021/javascript/lighthouse_unminified_js_bytes.sql
@@ -0,0 +1,31 @@
+#standardSQL
+# Histogram of wasted bytes per page
+CREATE TEMPORARY FUNCTION getUnminifiedJsBytes(audit STRING)
+RETURNS ARRAY<INT64> LANGUAGE js AS '''
+try {
+  var $ = JSON.parse(audit);
+  return $.details.items.map(({wastedBytes}) => wastedBytes);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  IF(unminified_js_kbytes <= 200, CEIL(unminified_js_kbytes / 10) * 10, 200) AS unminified_js_kbytes,
+  COUNT(0) AS pages,
+  SUM(COUNT(0)) OVER () AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER () AS pct
+FROM (
+  SELECT
+    test.url AS page,
+    SUM(IFNULL(unminified_js_bytes, 0)) / 1024 AS unminified_js_kbytes
+  FROM
+    `httparchive.lighthouse.2021_07_01_mobile` AS test
+  LEFT JOIN
+    UNNEST(getUnminifiedJsBytes(JSON_EXTRACT(report, "$.audits['unminified-javascript']"))) AS unminified_js_bytes
+  GROUP BY
+    page)
+GROUP BY
+  unminified_js_kbytes
+ORDER BY
+  unminified_js_kbytes
diff --git a/sql/2021/javascript/lighthouse_unminified_js_requests.sql b/sql/2021/javascript/lighthouse_unminified_js_requests.sql
new file mode 100644
index 00000000000..950bfb505be
--- /dev/null
+++ b/sql/2021/javascript/lighthouse_unminified_js_requests.sql
@@ -0,0 +1,31 @@
+#standardSQL
+# Percent of requests with and without minification
+CREATE TEMPORARY FUNCTION getUnminifiedJsBytes(audit STRING)
+RETURNS ARRAY<INT64> LANGUAGE js AS '''
+try {
+  var $ = JSON.parse(audit);
+  return $.details.items.map(({wastedBytes}) => wastedBytes);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  IF(unminified_js_kbytes <= 200, CEIL(unminified_js_kbytes / 10) * 10, 200) AS unminified_js_kbytes,
+  COUNT(0) AS pages,
+  SUM(COUNT(0)) OVER () AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER () AS pct
+FROM (
+  SELECT
+    test.url AS page,
+    SUM(IFNULL(unminified_js_bytes, 0)) / 1024 AS unminified_js_kbytes
+  FROM
+    `httparchive.lighthouse.2021_07_01_mobile` AS test
+  LEFT JOIN
+    UNNEST(getUnminifiedJsBytes(JSON_EXTRACT(report, "$.audits['unminified-javascript']"))) AS unminified_js_bytes
+  GROUP BY
+    page)
+GROUP BY
+  unminified_js_kbytes
+ORDER BY
+  unminified_js_kbytes
diff --git a/sql/2021/javascript/resource-hints-prefetch-preload-percentage.sql b/sql/2021/javascript/resource-hints-prefetch-preload-percentage.sql
new file mode 100644
index 00000000000..ca9c29468e5
--- /dev/null
+++ b/sql/2021/javascript/resource-hints-prefetch-preload-percentage.sql
@@ -0,0 +1,49 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getResourceHintAttrs(payload STRING)
+RETURNS ARRAY<STRUCT<name STRING, attribute STRING, value STRING>>
+LANGUAGE js AS '''
+var hints = new Set(['preload', 'prefetch']);
+var attributes = ['as'];
+try {
+  var $ = JSON.parse(payload);
+  var almanac = JSON.parse($._almanac);
+  return almanac['link-nodes'].nodes.reduce((results, link) => {
+    var hint = link.rel.toLowerCase();
+    if (!hints.has(hint)) {
+      return results;
+    }
+    attributes.forEach(attribute => {
+      var value = link[attribute];
+      results.push({
+        name: hint,
+        attribute: attribute,
+        // Support empty strings.
+        value: typeof value == 'string' ? value : null
+      });
+    });
+    return results;
+  }, []);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  client,
+  COUNT(DISTINCT IF(prefetch_hint, page, NULL)) AS prefetch_pages,
+  COUNT(DISTINCT page) AS total,
+  COUNT(DISTINCT IF(prefetch_hint, page, NULL)) / COUNT(DISTINCT page) AS prefetch_pct,
+  COUNT(DISTINCT IF(preload_hint, page, NULL)) AS preload_pages,
+  COUNT(DISTINCT IF(preload_hint, page, NULL)) / COUNT(DISTINCT page) AS preload_pct
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    hint.name = 'prefetch' AND hint.value = 'script' AS prefetch_hint,
+    hint.name = 'preload' AND hint.value = 'script' AS preload_hint
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(getResourceHintAttrs(payload)) AS hint)
+GROUP BY
+  client
diff --git a/sql/2021/javascript/resource-hints-preload-prefetch-distribution.sql b/sql/2021/javascript/resource-hints-preload-prefetch-distribution.sql
new file mode 100644
index 00000000000..c3ba28337da
--- /dev/null
+++ b/sql/2021/javascript/resource-hints-preload-prefetch-distribution.sql
@@ -0,0 +1,57 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getResourceHintAttrs(payload STRING)
+RETURNS ARRAY<STRUCT<name STRING, attribute STRING, value STRING>>
+LANGUAGE js AS '''
+var hints = new Set(['preload', 'prefetch']);
+var attributes = ['as'];
+try {
+  var $ = JSON.parse(payload);
+  var almanac = JSON.parse($._almanac);
+  return almanac['link-nodes'].nodes.reduce((results, link) => {
+    var hint = link.rel.toLowerCase();
+    if (!hints.has(hint)) {
+      return results;
+    }
+    attributes.forEach(attribute => {
+      var value = link[attribute];
+      results.push({
+        name: hint,
+        attribute: attribute,
+        // Support empty strings.
+        value: typeof value == 'string' ? value : null
+      });
+    });
+    return results;
+  }, []);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  percentile,
+  client,
+  APPROX_QUANTILES(prefetch_hint, 1000)[OFFSET(percentile * 10)] AS prefetch_hints_per_page,
+  APPROX_QUANTILES(IF(prefetch_hint = 0, NULL, prefetch_hint), 1000 IGNORE NULLS)[OFFSET(percentile * 10)] AS prefetch_hints_per_page_with_hints,
+  APPROX_QUANTILES(preload_hint, 1000)[OFFSET(percentile * 10)] AS preload_hints_per_page,
+  APPROX_QUANTILES(IF(preload_hint = 0, NULL, preload_hint), 1000 IGNORE NULLS)[OFFSET(percentile * 10)] AS preload_hints_per_page_with_hints
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    COUNTIF(hint.name = 'prefetch' AND hint.value = 'script') AS prefetch_hint,
+    COUNTIF(hint.name = 'preload' AND hint.value = 'script') AS preload_hint
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(getResourceHintAttrs(payload)) AS hint
+  GROUP BY
+    client,
+    page),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/resource_hints.sql b/sql/2021/javascript/resource_hints.sql
new file mode 100644
index 00000000000..10488f3d410
--- /dev/null
+++ b/sql/2021/javascript/resource_hints.sql
@@ -0,0 +1,46 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getResourceHintAttrs(payload STRING)
+RETURNS ARRAY<STRUCT<name STRING, attribute STRING, value STRING>>
+LANGUAGE js AS '''
+var hints = new Set(['preload', 'prefetch']);
+var attributes = ['as'];
+try {
+  var $ = JSON.parse(payload);
+  var almanac = JSON.parse($._almanac);
+  return almanac['link-nodes'].nodes.reduce((results, link) => {
+    var hint = link.rel.toLowerCase();
+    if (!hints.has(hint)) {
+      return results;
+    }
+    attributes.forEach(attribute => {
+      var value = link[attribute];
+      results.push({
+        name: hint,
+        attribute: attribute,
+        // Support empty strings.
+        value: typeof value == 'string' ? value : null
+      });
+    });
+    return results;
+  }, []);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  client,
+  COUNT(DISTINCT IF(script_hint, page, NULL)) AS pages,
+  COUNT(DISTINCT page) AS total,
+  COUNT(DISTINCT IF(script_hint, page, NULL)) / COUNT(DISTINCT page) AS pct
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    hint.name IN ('prefetch', 'preload') AND hint.value = 'script' AS script_hint
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(getResourceHintAttrs(payload)) AS hint)
+GROUP BY
+  client
diff --git a/sql/2021/javascript/resource_hints_per_page.sql b/sql/2021/javascript/resource_hints_per_page.sql
new file mode 100644
index 00000000000..6fa1cec60cf
--- /dev/null
+++ b/sql/2021/javascript/resource_hints_per_page.sql
@@ -0,0 +1,54 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getResourceHintAttrs(payload STRING)
+RETURNS ARRAY<STRUCT<name STRING, attribute STRING, value STRING>>
+LANGUAGE js AS '''
+var hints = new Set(['preload', 'prefetch']);
+var attributes = ['as'];
+try {
+  var $ = JSON.parse(payload);
+  var almanac = JSON.parse($._almanac);
+  return almanac['link-nodes'].nodes.reduce((results, link) => {
+    var hint = link.rel.toLowerCase();
+    if (!hints.has(hint)) {
+      return results;
+    }
+    attributes.forEach(attribute => {
+      var value = link[attribute];
+      results.push({
+        name: hint,
+        attribute: attribute,
+        // Support empty strings.
+        value: typeof value == 'string' ? value : null
+      });
+    });
+    return results;
+  }, []);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  percentile,
+  client,
+  APPROX_QUANTILES(script_hint, 1000)[OFFSET(percentile * 10)] AS hints_per_page,
+  APPROX_QUANTILES(IF(script_hint = 0, NULL, script_hint), 1000 IGNORE NULLS)[OFFSET(percentile * 10)] AS hints_per_page_with_hints
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    COUNTIF(hint.name IN ('prefetch', 'preload') AND hint.value = 'script') AS script_hint
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(getResourceHintAttrs(payload)) AS hint
+  GROUP BY
+    client,
+    page),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client

From 396e90e1f770f9e16b8eb0ec71f7a03d5c0d7c75 Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Wed, 6 Oct 2021 02:18:01 +0530
Subject: [PATCH 14/23] PR feedback accomodate & fixes

---
 sql/2021/javascript/ajax_request_per_page.sql |  2 +-
 sql/2021/javascript/async_and_defer.sql       | 25 -------------
 ...ipts_using_async_defer_module_nomodule.sql | 36 -------------------
 ...ipts_using_async_defer_module_nomodule.sql | 34 ------------------
 sql/2021/javascript/h1_vs_h2_vs_h3.sql        |  2 +-
 sql/2021/javascript/iframe_per_page.sql       |  2 +-
 .../most_requested_content_type.sql           | 12 ++++---
 ...ercentage_usage_of_different_ajax_apis.sql |  2 +-
 sql/2021/javascript/requests_2019.sql         |  2 +-
 sql/2021/javascript/requests_2021.sql         |  3 +-
 sql/2021/javascript/web_components_specs.sql  |  2 +-
 11 files changed, 15 insertions(+), 107 deletions(-)
 delete mode 100644 sql/2021/javascript/async_and_defer.sql
 delete mode 100644 sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql
 delete mode 100644 sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql

diff --git a/sql/2021/javascript/ajax_request_per_page.sql b/sql/2021/javascript/ajax_request_per_page.sql
index 93ba5343c0b..46fae7d771f 100644
--- a/sql/2021/javascript/ajax_request_per_page.sql
+++ b/sql/2021/javascript/ajax_request_per_page.sql
@@ -10,7 +10,7 @@ FROM (
           '$._javascript'),
         '$.ajax_requests.total') AS INT64) AS ajax_requests_total
   FROM
-    `httparchive.sample_data.pages_*` ),
+    `httparchive.pages.2021_07_01_*` ),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 WHERE
   ajax_requests_total > 0
diff --git a/sql/2021/javascript/async_and_defer.sql b/sql/2021/javascript/async_and_defer.sql
deleted file mode 100644
index 718e7201bd8..00000000000
--- a/sql/2021/javascript/async_and_defer.sql
+++ /dev/null
@@ -1,25 +0,0 @@
-SELECT
-  client,
-  COUNT(DISTINCT IF(async, page, NULL)) AS async,
-  COUNT(DISTINCT IF(defer, page, NULL)) AS defer,
-  COUNT(DISTINCT IF(async AND defer, page, NULL)) AS both,
-  COUNT(DISTINCT page) AS total,
-  COUNT(DISTINCT IF(async, page, NULL)) / COUNT(DISTINCT page) AS pct_async,
-  COUNT(DISTINCT IF(defer, page, NULL)) / COUNT(DISTINCT page) AS pct_defer,
-  COUNT(DISTINCT IF(async AND defer, page, NULL)) / COUNT(DISTINCT page) AS pct_both
-FROM (
-  SELECT
-    client,
-    page,
-    script,
-    REGEXP_CONTAINS(script, r'(?i)\basync\b') AS async,
-    REGEXP_CONTAINS(script, r'(?i)\bdefer\b') AS defer
-  FROM
-    `httparchive.sample_data.summary_response_bodies`  -- TODO: update to the original table name
-  LEFT JOIN
-    UNNEST(REGEXP_EXTRACT_ALL(body, r'(?i)(<script[^>]*>)')) AS script
-  WHERE
-    date = '2021-08-01' AND
-    firstHtml)
-GROUP BY
-  client
diff --git a/sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql b/sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql
deleted file mode 100644
index 1e06046cf91..00000000000
--- a/sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql
+++ /dev/null
@@ -1,36 +0,0 @@
-#standardSQL
-# Average Percent per page of external scripts using Async, Defer, Module or NoModule attributes.
-SELECT
-  client,
-  AVG(pct_external_async) AS avg_pct_external_async,
-  AVG(pct_external_defer) AS avg_pct_external_defer,
-  AVG(pct_external_module) AS avg_pct_external_module,
-  AVG(pct_external_nomodule) AS avg_pct_external_nomodule
-FROM (
-  SELECT
-    client,
-    page,
-    COUNT(0) AS external_scripts,
-    COUNTIF(REGEXP_CONTAINS(script, r'\basync\b')) / COUNT(0) AS pct_external_async,
-    COUNTIF(REGEXP_CONTAINS(script, r'\bdefer\b')) / COUNT(0) AS pct_external_defer,
-    COUNTIF(REGEXP_CONTAINS(script, r'\bmodule\b')) / COUNT(0) AS pct_external_module,
-    COUNTIF(REGEXP_CONTAINS(script, r'\bnomodule\b')) / COUNT(0) AS pct_external_nomodule
-  FROM (
-    SELECT
-      client,
-      page,
-      url,
-      REGEXP_EXTRACT_ALL(body, "(?i)(<script [^>]*)") AS scripts
-    FROM
-      `httparchive.sample_data.summary_response_bodies`  -- TODO: update to the original table name
-    WHERE
-      date = '2021-08-01' AND
-      firstHtml),
-    UNNEST(scripts) AS script
-  WHERE
-    REGEXP_CONTAINS(script, r'\bsrc\b')
-  GROUP BY
-    client,
-    page)
-GROUP BY
-  client
diff --git a/sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql b/sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql
deleted file mode 100644
index 2f43c8dd535..00000000000
--- a/sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql
+++ /dev/null
@@ -1,34 +0,0 @@
-#standardSQL
-# Breakdown of scripts using Async, Defer, Module or NoModule attributes.  Also breakdown of inline vs external scripts
-SELECT
-  client,
-  COUNT(0) AS total_scripts,
-  SUM(IF(script NOT LIKE "%src%", 1, 0)) AS inline_script,
-  SUM(IF(script LIKE "%src%", 1, 0)) AS external_script,
-  SUM(IF(script LIKE "%src%", 1, 0)) / COUNT(0) AS pct_external_script,
-  SUM(IF(script NOT LIKE "%src%", 1, 0)) / COUNT(0) AS pct_inline_script,
-  SUM(IF(script LIKE "%async%", 1, 0)) AS async,
-  SUM(IF(script LIKE "%defer%", 1, 0)) AS defer,
-  SUM(IF(script LIKE "%module%", 1, 0)) AS module,
-  SUM(IF(script LIKE "%nomodule%", 1, 0)) AS nomodule,
-  SUM(IF(script LIKE "%async%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_async,
-  SUM(IF(script LIKE "%defer%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_defer,
-  SUM(IF(script LIKE "%module%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_module,
-  SUM(IF(script LIKE "%nomodule%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_nomodule
-FROM
-  (
-    SELECT
-      client,
-      page,
-      url,
-      REGEXP_EXTRACT_ALL(LOWER(body), "(<script [^>]*)") AS scripts
-    FROM
-      `httparchive.sample_data.summary_response_bodies`  -- TODO: update to the original table name
-    WHERE
-      date = '2021-08-01' AND
-      firstHtml
-  )
-CROSS JOIN
-  UNNEST(scripts) AS script
-GROUP BY
-  client
diff --git a/sql/2021/javascript/h1_vs_h2_vs_h3.sql b/sql/2021/javascript/h1_vs_h2_vs_h3.sql
index 5c4dad4c21a..59645440827 100644
--- a/sql/2021/javascript/h1_vs_h2_vs_h3.sql
+++ b/sql/2021/javascript/h1_vs_h2_vs_h3.sql
@@ -29,7 +29,7 @@ FROM (
         '$._javascript'),
       '$.requests_protocol.resources_h3') AS INT64) AS resources_h3
   FROM
-    `httparchive.sample_data.pages_*`),
+    `httparchive.pages.2021_07_01_*`),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
diff --git a/sql/2021/javascript/iframe_per_page.sql b/sql/2021/javascript/iframe_per_page.sql
index d80c8b07545..684630f0053 100644
--- a/sql/2021/javascript/iframe_per_page.sql
+++ b/sql/2021/javascript/iframe_per_page.sql
@@ -10,7 +10,7 @@ FROM (
           '$._javascript'),
         '$.iframe') AS INT64) AS iframe_total
   FROM
-    `httparchive.sample_data.pages_*` ),
+    `httparchive.pages.2021_07_01_*` ),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 WHERE
   iframe_total > 0
diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index b2649e99ce5..ec8dd32ea9b 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -1,9 +1,13 @@
 SELECT
   resp_content_type,
-  COUNT(0) AS count
+  COUNT(0) AS count,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total,
+  ROUND((COUNT(0) * 100 / total), 2) AS pct,
 FROM
-  `httparchive.sample_data.summary_requests_*`
+  `httparchive.almanac.requests`
+WHERE
+  date = '2021-07-01'
 GROUP BY
-  1
+  client
 ORDER BY
-  2 DESC
+  client DESC
diff --git a/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql b/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
index beb5ca5a2b2..db4e8c65adc 100644
--- a/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
+++ b/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
@@ -20,7 +20,7 @@ FROM (
           '$._javascript'),
         '$.ajax_requests.beacon') AS beacon_total
     FROM
-      `httparchive.sample_data.pages_*`),
+      `httparchive.pages.2021_07_01_*`),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
diff --git a/sql/2021/javascript/requests_2019.sql b/sql/2021/javascript/requests_2019.sql
index c4d2ccd8934..f9947aece20 100644
--- a/sql/2021/javascript/requests_2019.sql
+++ b/sql/2021/javascript/requests_2019.sql
@@ -5,7 +5,7 @@ SELECT
   _TABLE_SUFFIX AS client,
   APPROX_QUANTILES(reqJS, 1000)[OFFSET(percentile * 10)] AS js_requests
 FROM
-  `httparchive.summary_pages.2019_08_01_*`,
+  `httparchive.summary_pages.2019_07_01_*`,
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
diff --git a/sql/2021/javascript/requests_2021.sql b/sql/2021/javascript/requests_2021.sql
index ea9da50d48e..fef2951e58d 100644
--- a/sql/2021/javascript/requests_2021.sql
+++ b/sql/2021/javascript/requests_2021.sql
@@ -5,8 +5,7 @@ SELECT
   _TABLE_SUFFIX AS client,
   APPROX_QUANTILES(reqJS, 1000)[OFFSET(percentile * 10)] AS js_requests
 FROM
-  `httparchive.latest.summary_pages_*`,
-  -- `httparchive.sample_data.summary_pages_*`
+  `httparchive.summary_pages.2021_07_01_*`,
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
index db9cf094386..1e7b54566e6 100644
--- a/sql/2021/javascript/web_components_specs.sql
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -20,7 +20,7 @@ FROM (
           '$._javascript'),
         '$.web_component_specs.template') AS INT64) AS template
   FROM
-    `httparchive.sample_data.pages_*` ),
+    `httparchive.pages.2021_07_01_*` ),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,

From 25343710bf8f1baf703ee1dd4b61c38c8e34c713 Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Wed, 6 Oct 2021 02:23:48 +0530
Subject: [PATCH 15/23] link fix

---
 sql/2021/javascript/most_requested_content_type.sql | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index ec8dd32ea9b..6f170ec778f 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -2,7 +2,7 @@ SELECT
   resp_content_type,
   COUNT(0) AS count,
   SUM(COUNT(0)) OVER (PARTITION BY client) AS total,
-  ROUND((COUNT(0) * 100 / total), 2) AS pct,
+  ROUND((COUNT(0) * 100 / total), 2) AS pct
 FROM
   `httparchive.almanac.requests`
 WHERE

From 04a401159547675fa5047807911bd8b2fc9cde27 Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Thu, 7 Oct 2021 00:21:38 +0530
Subject: [PATCH 16/23] Feedback II update

---
 .../lighthouse_unminified_js_requests.sql     | 31 -------------------
 .../most_requested_content_type.sql           |  2 +-
 2 files changed, 1 insertion(+), 32 deletions(-)
 delete mode 100644 sql/2021/javascript/lighthouse_unminified_js_requests.sql

diff --git a/sql/2021/javascript/lighthouse_unminified_js_requests.sql b/sql/2021/javascript/lighthouse_unminified_js_requests.sql
deleted file mode 100644
index 950bfb505be..00000000000
--- a/sql/2021/javascript/lighthouse_unminified_js_requests.sql
+++ /dev/null
@@ -1,31 +0,0 @@
-#standardSQL
-# Percent of requests with and without minification
-CREATE TEMPORARY FUNCTION getUnminifiedJsBytes(audit STRING)
-RETURNS ARRAY<INT64> LANGUAGE js AS '''
-try {
-  var $ = JSON.parse(audit);
-  return $.details.items.map(({wastedBytes}) => wastedBytes);
-} catch (e) {
-  return [];
-}
-''';
-
-SELECT
-  IF(unminified_js_kbytes <= 200, CEIL(unminified_js_kbytes / 10) * 10, 200) AS unminified_js_kbytes,
-  COUNT(0) AS pages,
-  SUM(COUNT(0)) OVER () AS total,
-  COUNT(0) / SUM(COUNT(0)) OVER () AS pct
-FROM (
-  SELECT
-    test.url AS page,
-    SUM(IFNULL(unminified_js_bytes, 0)) / 1024 AS unminified_js_kbytes
-  FROM
-    `httparchive.lighthouse.2021_07_01_mobile` AS test
-  LEFT JOIN
-    UNNEST(getUnminifiedJsBytes(JSON_EXTRACT(report, "$.audits['unminified-javascript']"))) AS unminified_js_bytes
-  GROUP BY
-    page)
-GROUP BY
-  unminified_js_kbytes
-ORDER BY
-  unminified_js_kbytes
diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index 6f170ec778f..5561b338b24 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -2,7 +2,7 @@ SELECT
   resp_content_type,
   COUNT(0) AS count,
   SUM(COUNT(0)) OVER (PARTITION BY client) AS total,
-  ROUND((COUNT(0) * 100 / total), 2) AS pct
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct
 FROM
   `httparchive.almanac.requests`
 WHERE

From 387303ccd53b99117cedbe5b1dd73247eeaee985 Mon Sep 17 00:00:00 2001
From: Pankaj Parkar <parkarpankaj3@gmail.com>
Date: Tue, 12 Oct 2021 11:23:56 +0530
Subject: [PATCH 17/23] Fixed review comment related to web_component_specs.

---
 sql/2021/javascript/web_components_specs.sql | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
index 1e7b54566e6..383d0db5503 100644
--- a/sql/2021/javascript/web_components_specs.sql
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -12,13 +12,13 @@ FROM (
     _TABLE_SUFFIX AS client,
     CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
-        '$.web_component_specs.custom_elements') AS INT64) AS custom_elements,
+        '$.web_component_specs.custom_elements.length') AS INT64) AS custom_elements,
     CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
-        '$.web_component_specs.shadow_roots') AS INT64) AS shadow_roots,
+        '$.web_component_specs.shadow_roots.length') AS INT64) AS shadow_roots,
     CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
-        '$.web_component_specs.template') AS INT64) AS template
+        '$.web_component_specs.template.length') AS INT64) AS template
   FROM
     `httparchive.pages.2021_07_01_*` ),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile

From 4642675507e1c864b15ce429094ac41e91c4ab55 Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Tue, 19 Oct 2021 02:48:11 +0530
Subject: [PATCH 18/23] update content_type and webcomponent query

---
 .../most_requested_content_type.sql           | 19 +++++++-------
 sql/2021/javascript/web_components_specs.sql  | 25 +++++++++++++------
 2 files changed, 27 insertions(+), 17 deletions(-)

diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index 5561b338b24..a695d45967c 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -1,13 +1,14 @@
 SELECT
-  resp_content_type,
-  COUNT(0) AS count,
-  SUM(COUNT(0)) OVER (PARTITION BY client) AS total,
-  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct
-FROM
-  `httparchive.almanac.requests`
-WHERE
-  date = '2021-07-01'
+    client,
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%image%') as images,
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%font%') as font, 
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%css%') as css, 
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%javascript%') as javascript, 
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%json%') as json,
+    count(*) as total
+  FROM
+    `httparchive.almanac.requests`
 GROUP BY
   client
 ORDER BY
-  client DESC
+  client
\ No newline at end of file
diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
index 383d0db5503..a4463e2c5c7 100644
--- a/sql/2021/javascript/web_components_specs.sql
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -10,21 +10,30 @@ SELECT
 FROM (
   SELECT
     _TABLE_SUFFIX AS client,
-    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+    (CASE 
+     WHEN ARRAY_LENGTH(JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
-        '$.web_component_specs.custom_elements.length') AS INT64) AS custom_elements,
-    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$.web_component_specs.custom_elements'))) > 1 THEN 1
+    ELSE 0 
+    END) AS custom_elements,
+     (CASE 
+     WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
-        '$.web_component_specs.shadow_roots.length') AS INT64) AS shadow_roots,
-    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$.web_component_specs.shadow_roots'))) > 1 THEN 1
+    ELSE 0 
+    END) AS shadow_roots,
+    (CASE 
+     WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
-        '$.web_component_specs.template.length') AS INT64) AS template
+        '$.web_component_specs.template'))) > 1 THEN 1
+    ELSE 0 
+    END) AS template
   FROM
-    `httparchive.pages.2021_07_01_*` ),
+    `httparchive.pages.2021_09_01_*` ),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
   client
 ORDER BY
   percentile,
-  client
+  client
\ No newline at end of file

From fec59a1ff9e99559ee4d02b6155a624fafa2b5df Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Tue, 19 Oct 2021 03:17:06 +0530
Subject: [PATCH 19/23] Javascript sql 2021 (#2386)

* Bump prettier from 2.4.0 to 2.4.1 in /src (#2334)

Bumps [prettier](https://github.com/prettier/prettier) from 2.4.0 to 2.4.1.
- [Release notes](https://github.com/prettier/prettier/releases)
- [Changelog](https://github.com/prettier/prettier/blob/main/CHANGELOG.md)
- [Commits](https://github.com/prettier/prettier/compare/2.4.0...2.4.1)

---
updated-dependencies:
- dependency-name: prettier
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Turn off uses-passive-event-listeners Lighthouse audit (#2335)

* Bump node-watch from 0.7.1 to 0.7.2 in /src (#2336)

Bumps [node-watch](https://github.com/yuanchuan/node-watch) from 0.7.1 to 0.7.2.
- [Release notes](https://github.com/yuanchuan/node-watch/releases)
- [Changelog](https://github.com/yuanchuan/node-watch/blob/master/Changelog.md)
- [Commits](https://github.com/yuanchuan/node-watch/compare/0.7.1...v0.7.2)

---
updated-dependencies:
- dependency-name: node-watch
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* PWA 2021 queries (#2272)

* Initial conversion of queries to custom metrics

* Better Lighthouse queries

* Remove comma

* Add Service Worker over time for top 1k sites query

* Change adoption ranking to include all ranking categories

* Formatting

* Linting fixes

* Spacing improvements

* Missing queries

* Query updates

* Missed some

* Remove s

* Space

* Misc updates

* More Tweaks

* Spacing fixes

* Fix assestlinks query

* Missing DESC

* Group rankings properly

* Linting fixes

* Remove unnecessary extra join

* Spacing

* Additional queries

* Linting fixes

* Update Events to look at onEvents

* Better array merge

* Misc fixes

* Better JS

* Filtering false positives for install query

* Fixes to events query

* Adding comment for install query

* Expand on comment

* Review feedback

* Linting fixes

* Correct comment

* Apply suggestions from code review

Co-authored-by: David Fox <david@davidjfox.com>

* Review feedback

* Review feedback

* Change to COUNT(DISTINCT url)

Co-authored-by: demianrenzulli <demian77@gmail.com>
Co-authored-by: David Fox <david@davidjfox.com>

* Bump sqlfluff from 0.6.5 to 0.6.6 in /src (#2339)

Bumps [sqlfluff](https://github.com/sqlfluff/sqlfluff) from 0.6.5 to 0.6.6.
- [Release notes](https://github.com/sqlfluff/sqlfluff/releases)
- [Changelog](https://github.com/sqlfluff/sqlfluff/blob/main/CHANGELOG.md)
- [Commits](https://github.com/sqlfluff/sqlfluff/compare/0.6.5...0.6.6)

---
updated-dependencies:
- dependency-name: sqlfluff
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Add more 2 more PWA queries (#2340)

* Small SQL linting fixes to old media queries (#2341)

* Caching 2021 queries (#2254)

* Update README.md

* Top CDNs

* caching queries part 1

* LH queries

* caching sql part 2

* changes from review

* comment

* Delete sql/2021/third-parties directory (#2342)

* Delete sql/2021/third-parties directory

* Delete third-parties.md

* frameworks, sourcemaps queries

* 1p vs 3p

* adds unminified, preload/prefetch queries

* Bump actions/github-script from 4.1 to 5 (#2346)

Bumps [actions/github-script](https://github.com/actions/github-script) from 4.1 to 5.
- [Release notes](https://github.com/actions/github-script/releases)
- [Commits](https://github.com/actions/github-script/compare/v4.1...v5)

---
updated-dependencies:
- dependency-name: actions/github-script
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump actions/setup-node from 2.4.0 to 2.4.1 (#2347)

Bumps [actions/setup-node](https://github.com/actions/setup-node) from 2.4.0 to 2.4.1.
- [Release notes](https://github.com/actions/setup-node/releases)
- [Commits](https://github.com/actions/setup-node/compare/v2.4.0...v2.4.1)

---
updated-dependencies:
- dependency-name: actions/setup-node
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump github/super-linter from 4.7.3 to 4.8.0 (#2350)

* Bump github/super-linter from 4.7.3 to 4.8.0

Bumps [github/super-linter](https://github.com/github/super-linter) from 4.7.3 to 4.8.0.
- [Release notes](https://github.com/github/super-linter/releases)
- [Changelog](https://github.com/github/super-linter/blob/master/docs/release-process.md)
- [Commits](https://github.com/github/super-linter/compare/v4.7.3...v4.8.0)

---
updated-dependencies:
- dependency-name: github/super-linter
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>

* Fix queries failing linting

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Barry <barry@tunetheweb.com>

* PR feedback accomodate & fixes

* link fix

* Feedback II update

* Fixed review comment related to web_component_specs.

* update content_type and webcomponent query

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Barry Pollard <barry_pollard@hotmail.com>
Co-authored-by: demianrenzulli <demian77@gmail.com>
Co-authored-by: David Fox <david@davidjfox.com>
Co-authored-by: Rick Viscomi <rviscomi@users.noreply.github.com>
Co-authored-by: Barry <barry@tunetheweb.com>
Co-authored-by: Pankaj Parkar <parkarpankaj3@gmail.com>
---
 .github/lighthouse/lighthouse-config-dev.json |   1 +
 .../lighthouse/lighthouse-config-prod.json    |   1 +
 .github/workflows/linter.yml                  |   2 +-
 .github/workflows/predeploy.yml               |   2 +-
 .github/workflows/progress-tracker.yml        |   2 +-
 .github/workflows/test-template-changes.yml   |   4 +-
 .github/workflows/test_website.yml            |   4 +-
 sql/2019/media/04_03b.sql                     |   2 +-
 sql/2019/media/04_11a.sql                     |   2 +-
 sql/2020/media/bytes_per_pixel.sql            |   2 +-
 sql/2021/caching/README.md                    |  12 +--
 sql/2021/caching/cache_control_directives.sql |  71 +++++++++++++
 .../cache_control_directives_invalid.sql      |  33 ++++++
 .../cache_ttl_and_content_age_diff.sql        |  32 ++++++
 .../caching/cache_ttl_lighthouse_score.sql    |  16 +++
 .../caching/cache_wastedbytes_lighthouse.sql  |  16 +++
 .../caching/content_age_older_than_ttl.sql    |  30 ++++++
 .../content_age_older_than_ttl_by_party.sql   |  36 +++++++
 sql/2021/caching/cookie_attributes.sql        |  60 +++++++++++
 sql/2021/caching/cookie_names.sql             |  37 +++++++
 sql/2021/caching/header_trends.sql            |  86 +++++++++++++++
 ...lid_last_modified_and_expires_and_date.sql |  31 ++++++
 sql/2021/caching/max_age_distribution.sql     |  22 ++++
 .../non_cacheable_by_resource_type.sql        |  33 ++++++
 .../resource_age_party_and_type_wise.sql      |  35 ++++++
 ...esource_age_party_and_type_wise_groups.sql |  52 +++++++++
 sql/2021/caching/service_worker_rank.sql      |  20 ++++
 sql/2021/caching/set_cookie.sql               |  25 +++++
 sql/2021/caching/top_cdns.sql                 |  24 +++++
 .../caching/top_hosts_using_immutable.sql     |  21 ++++
 sql/2021/caching/ttl.sql                      |  27 +++++
 sql/2021/caching/ttl_by_resource.sql          |  21 ++++
 .../valid_if_modified_since_returns_304.sql   |  26 +++++
 .../valid_if_none_match_returns_304.sql       |  24 +++++
 sql/2021/caching/vary_directives.sql          |  43 ++++++++
 sql/2021/javascript/ajax_request_per_page.sql |   2 +-
 sql/2021/javascript/async_and_defer.sql       |  25 -----
 ...ipts_using_async_defer_module_nomodule.sql |  36 -------
 ...ipts_using_async_defer_module_nomodule.sql |  34 ------
 sql/2021/javascript/bytes_2019.sql            |  15 +++
 sql/2021/javascript/bytes_2020.sql            |  15 +++
 sql/2021/javascript/bytes_2021.sql            |  15 +++
 sql/2021/javascript/bytes_by_3p.sql           |  33 ++++++
 sql/2021/javascript/compression_method.sql    |  47 ++++++++
 .../javascript/compression_method_by_3p.sql   |  52 +++++++++
 .../javascript/compression_none_by_bytes.sql  |  55 ++++++++++
 .../frameworks-bytes-by-framework.sql         |  35 ++++++
 sql/2021/javascript/frameworks_libraries.sql  |  30 ++++++
 .../frameworks_libraries_by_version.sql       |  32 ++++++
 .../frameworks_libraries_combos.sql           |  43 ++++++++
 sql/2021/javascript/h1_vs_h2_vs_h3.sql        |   2 +-
 sql/2021/javascript/iframe_per_page.sql       |   2 +-
 .../javascript/lighthouse_unminified_js.sql   |  18 ++++
 .../lighthouse_unminified_js_by_3p.sql        |  34 ++++++
 .../lighthouse_unminified_js_bytes.sql        |  31 ++++++
 .../most_requested_content_type.sql           |  17 +--
 ...ercentage_usage_of_different_ajax_apis.sql |   2 +-
 sql/2021/javascript/requests_2019.sql         |   3 +-
 sql/2021/javascript/requests_2021.sql         |   3 +-
 ...urce-hints-prefetch-preload-percentage.sql |  49 +++++++++
 ...ce-hints-preload-prefetch-distribution.sql |  57 ++++++++++
 sql/2021/javascript/resource_hints.sql        |  46 ++++++++
 .../javascript/resource_hints_per_page.sql    |  54 ++++++++++
 sql/2021/javascript/sourcemap_header.sql      |  48 +++++++++
 .../javascript/sourcemap_header_by_3p.sql     |  54 ++++++++++
 .../unused_js_bytes_distribution.sql          |  12 +++
 sql/2021/javascript/web_components_specs.sql  |  25 +++--
 sql/2021/pwa/README.md                        |   8 ++
 sql/2021/pwa/assetlink_usage.sql              |  61 +++++++++++
 sql/2021/pwa/install_events.sql               |  69 ++++++++++++
 sql/2021/pwa/lighthouse_pwa_audits.sql        |  67 ++++++++++++
 sql/2021/pwa/lighthouse_pwa_score.sql         |  48 +++++++++
 .../pwa/manifests_and_service_workers.sql     |  21 ++++
 sql/2021/pwa/manifests_not_json_parsable.sql  |  52 +++++++++
 .../pwa/manifests_preferring_native_apps.sql  |  46 ++++++++
 sql/2021/pwa/popular_pwa_libraries.sql        | 100 ++++++++++++++++++
 sql/2021/pwa/popular_pwa_libraries_helper.sql |  48 +++++++++
 sql/2021/pwa/pwa_blink_usage.sql              |  21 ++++
 .../pwa/pwa_notification_acceptance_rates.sql |  47 ++++++++
 sql/2021/pwa/sw_adoption_over_time.sql        |  15 +++
 .../pwa/sw_adoption_over_time_ranking.sql     |  57 ++++++++++
 sql/2021/pwa/sw_events.sql                    |  46 ++++++++
 sql/2021/pwa/sw_methods.sql                   |  44 ++++++++
 sql/2021/pwa/sw_objects.sql                   |  47 ++++++++
 sql/2021/pwa/sw_objects_name_only.sql         |  47 ++++++++
 sql/2021/pwa/sw_registration_properties.sql   |  48 +++++++++
 .../sw_registration_properties_name_only.sql  |  48 +++++++++
 sql/2021/pwa/top_manifest_categories.sql      |  86 +++++++++++++++
 sql/2021/pwa/top_manifest_display_values.sql  |  56 ++++++++++
 sql/2021/pwa/top_manifest_icon_sizes.sql      |  88 +++++++++++++++
 sql/2021/pwa/top_manifest_orientations.sql    |  52 +++++++++
 sql/2021/pwa/top_manifest_properties.sql      |  92 ++++++++++++++++
 sql/2021/pwa/workbox_methods.sql              |  57 ++++++++++
 sql/2021/pwa/workbox_packages.sql             |  60 +++++++++++
 sql/2021/pwa/workbox_usage.sql                |  18 ++++
 sql/2021/pwa/workbox_versions.sql             |  60 +++++++++++
 sql/2021/third-parties/README.md              |  10 --
 src/content/en/2021/third-parties.md          |  20 ----
 src/package-lock.json                         |  28 ++---
 src/package.json                              |   4 +-
 src/requirements.txt                          |   2 +-
 101 files changed, 3173 insertions(+), 181 deletions(-)
 create mode 100644 sql/2021/caching/cache_control_directives.sql
 create mode 100644 sql/2021/caching/cache_control_directives_invalid.sql
 create mode 100644 sql/2021/caching/cache_ttl_and_content_age_diff.sql
 create mode 100644 sql/2021/caching/cache_ttl_lighthouse_score.sql
 create mode 100644 sql/2021/caching/cache_wastedbytes_lighthouse.sql
 create mode 100644 sql/2021/caching/content_age_older_than_ttl.sql
 create mode 100644 sql/2021/caching/content_age_older_than_ttl_by_party.sql
 create mode 100644 sql/2021/caching/cookie_attributes.sql
 create mode 100644 sql/2021/caching/cookie_names.sql
 create mode 100644 sql/2021/caching/header_trends.sql
 create mode 100644 sql/2021/caching/invalid_last_modified_and_expires_and_date.sql
 create mode 100644 sql/2021/caching/max_age_distribution.sql
 create mode 100644 sql/2021/caching/non_cacheable_by_resource_type.sql
 create mode 100644 sql/2021/caching/resource_age_party_and_type_wise.sql
 create mode 100644 sql/2021/caching/resource_age_party_and_type_wise_groups.sql
 create mode 100644 sql/2021/caching/service_worker_rank.sql
 create mode 100644 sql/2021/caching/set_cookie.sql
 create mode 100644 sql/2021/caching/top_cdns.sql
 create mode 100644 sql/2021/caching/top_hosts_using_immutable.sql
 create mode 100644 sql/2021/caching/ttl.sql
 create mode 100644 sql/2021/caching/ttl_by_resource.sql
 create mode 100644 sql/2021/caching/valid_if_modified_since_returns_304.sql
 create mode 100644 sql/2021/caching/valid_if_none_match_returns_304.sql
 create mode 100644 sql/2021/caching/vary_directives.sql
 delete mode 100644 sql/2021/javascript/async_and_defer.sql
 delete mode 100644 sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql
 delete mode 100644 sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql
 create mode 100644 sql/2021/javascript/bytes_2019.sql
 create mode 100644 sql/2021/javascript/bytes_2020.sql
 create mode 100644 sql/2021/javascript/bytes_2021.sql
 create mode 100644 sql/2021/javascript/bytes_by_3p.sql
 create mode 100644 sql/2021/javascript/compression_method.sql
 create mode 100644 sql/2021/javascript/compression_method_by_3p.sql
 create mode 100644 sql/2021/javascript/compression_none_by_bytes.sql
 create mode 100644 sql/2021/javascript/frameworks-bytes-by-framework.sql
 create mode 100644 sql/2021/javascript/frameworks_libraries.sql
 create mode 100644 sql/2021/javascript/frameworks_libraries_by_version.sql
 create mode 100644 sql/2021/javascript/frameworks_libraries_combos.sql
 create mode 100644 sql/2021/javascript/lighthouse_unminified_js.sql
 create mode 100644 sql/2021/javascript/lighthouse_unminified_js_by_3p.sql
 create mode 100644 sql/2021/javascript/lighthouse_unminified_js_bytes.sql
 create mode 100644 sql/2021/javascript/resource-hints-prefetch-preload-percentage.sql
 create mode 100644 sql/2021/javascript/resource-hints-preload-prefetch-distribution.sql
 create mode 100644 sql/2021/javascript/resource_hints.sql
 create mode 100644 sql/2021/javascript/resource_hints_per_page.sql
 create mode 100644 sql/2021/javascript/sourcemap_header.sql
 create mode 100644 sql/2021/javascript/sourcemap_header_by_3p.sql
 create mode 100644 sql/2021/javascript/unused_js_bytes_distribution.sql
 create mode 100644 sql/2021/pwa/assetlink_usage.sql
 create mode 100644 sql/2021/pwa/install_events.sql
 create mode 100644 sql/2021/pwa/lighthouse_pwa_audits.sql
 create mode 100644 sql/2021/pwa/lighthouse_pwa_score.sql
 create mode 100644 sql/2021/pwa/manifests_and_service_workers.sql
 create mode 100644 sql/2021/pwa/manifests_not_json_parsable.sql
 create mode 100644 sql/2021/pwa/manifests_preferring_native_apps.sql
 create mode 100644 sql/2021/pwa/popular_pwa_libraries.sql
 create mode 100644 sql/2021/pwa/popular_pwa_libraries_helper.sql
 create mode 100644 sql/2021/pwa/pwa_blink_usage.sql
 create mode 100644 sql/2021/pwa/pwa_notification_acceptance_rates.sql
 create mode 100644 sql/2021/pwa/sw_adoption_over_time.sql
 create mode 100644 sql/2021/pwa/sw_adoption_over_time_ranking.sql
 create mode 100644 sql/2021/pwa/sw_events.sql
 create mode 100644 sql/2021/pwa/sw_methods.sql
 create mode 100644 sql/2021/pwa/sw_objects.sql
 create mode 100644 sql/2021/pwa/sw_objects_name_only.sql
 create mode 100644 sql/2021/pwa/sw_registration_properties.sql
 create mode 100644 sql/2021/pwa/sw_registration_properties_name_only.sql
 create mode 100644 sql/2021/pwa/top_manifest_categories.sql
 create mode 100644 sql/2021/pwa/top_manifest_display_values.sql
 create mode 100644 sql/2021/pwa/top_manifest_icon_sizes.sql
 create mode 100644 sql/2021/pwa/top_manifest_orientations.sql
 create mode 100644 sql/2021/pwa/top_manifest_properties.sql
 create mode 100644 sql/2021/pwa/workbox_methods.sql
 create mode 100644 sql/2021/pwa/workbox_packages.sql
 create mode 100644 sql/2021/pwa/workbox_usage.sql
 create mode 100644 sql/2021/pwa/workbox_versions.sql
 delete mode 100644 sql/2021/third-parties/README.md
 delete mode 100644 src/content/en/2021/third-parties.md

diff --git a/.github/lighthouse/lighthouse-config-dev.json b/.github/lighthouse/lighthouse-config-dev.json
index 5b2dfe5537f..91573413c77 100644
--- a/.github/lighthouse/lighthouse-config-dev.json
+++ b/.github/lighthouse/lighthouse-config-dev.json
@@ -34,6 +34,7 @@
         "unused-css-rules": "off",
         "unused-javascript": "off",
         "uses-long-cache-ttl": "off",
+        "uses-passive-event-listeners": "off",
         "uses-rel-preconnect": "off",
         "uses-rel-preload": "off",
         "uses-text-compression": "off",
diff --git a/.github/lighthouse/lighthouse-config-prod.json b/.github/lighthouse/lighthouse-config-prod.json
index e1614b2540e..c5e5cfd3464 100644
--- a/.github/lighthouse/lighthouse-config-prod.json
+++ b/.github/lighthouse/lighthouse-config-prod.json
@@ -33,6 +33,7 @@
         "unused-css-rules": "off",
         "unused-javascript": "off",
         "uses-long-cache-ttl": "off",
+        "uses-passive-event-listeners": "off",
         "uses-rel-preconnect": "off",
         "uses-rel-preload": "off",
         "uses-responsive-images": "off",
diff --git a/.github/workflows/linter.yml b/.github/workflows/linter.yml
index 837de84efcc..211550e4d37 100644
--- a/.github/workflows/linter.yml
+++ b/.github/workflows/linter.yml
@@ -28,7 +28,7 @@ jobs:
         run: |
           echo "VALIDATE_ALL_CODEBASE=false" >> $GITHUB_ENV
       - name: Lint Code Base
-        uses: github/super-linter@v4.7.3
+        uses: github/super-linter@v4.8.0
         #uses: docker://github/super-linter:v4.7.1
         env:
           DEFAULT_BRANCH: main
diff --git a/.github/workflows/predeploy.yml b/.github/workflows/predeploy.yml
index 44a676568ef..21dce51637f 100644
--- a/.github/workflows/predeploy.yml
+++ b/.github/workflows/predeploy.yml
@@ -31,7 +31,7 @@ jobs:
     - name: Checkout branch
       uses: actions/checkout@v2.3.4
     - name: Setup Node.js for use with actions
-      uses: actions/setup-node@v2.4.0
+      uses: actions/setup-node@v2.4.1
       with:
         node-version: '15'
     - name: Set up Python 3.8
diff --git a/.github/workflows/progress-tracker.yml b/.github/workflows/progress-tracker.yml
index 853f179ccd2..862747712b4 100644
--- a/.github/workflows/progress-tracker.yml
+++ b/.github/workflows/progress-tracker.yml
@@ -18,7 +18,7 @@ jobs:
     if: github.repository == 'HTTPArchive/almanac.httparchive.org'
     runs-on: ubuntu-20.04
     steps:
-      - uses: actions/github-script@v4.1
+      - uses: actions/github-script@v5
         if: github.event_name == 'workflow_dispatch' || contains(github.event.issue.labels.*.name, env.FILTER_LABEL)
         with:
           script: |
diff --git a/.github/workflows/test-template-changes.yml b/.github/workflows/test-template-changes.yml
index 40e556e7419..82e67a4e9e3 100644
--- a/.github/workflows/test-template-changes.yml
+++ b/.github/workflows/test-template-changes.yml
@@ -33,13 +33,13 @@ jobs:
       with:
         fetch-depth: 0
     - name: Setup Node.js for use with actions
-      uses: actions/setup-node@v2.4.0
+      uses: actions/setup-node@v2.4.1
       with:
         node-version:  '15'
     - name: Test Template Changes
       run: ./src/tools/scripts/test_template_changes.sh
     - name: 'Comment PR'
-      uses: actions/github-script@v4.1
+      uses: actions/github-script@v5
       if: github.event_name == 'pull_request' && env.PR_COMMENT != ''
       with:
         github-token: ${{ secrets.GITHUB_TOKEN }}
diff --git a/.github/workflows/test_website.yml b/.github/workflows/test_website.yml
index f4c7ad7f27c..f0b03c74f4c 100644
--- a/.github/workflows/test_website.yml
+++ b/.github/workflows/test_website.yml
@@ -23,7 +23,7 @@ jobs:
         # Full git history is needed to get a proper list of changed files within `super-linter`
         fetch-depth: 0
     - name: Setup Node.js for use with actions
-      uses: actions/setup-node@v2.4.0
+      uses: actions/setup-node@v2.4.1
       with:
         node-version:  '15'
     - name: Set up Python 3.8
@@ -37,7 +37,7 @@ jobs:
     - name: Use more complete checks for generated HTML linting
       run: cp -f .github/linters/.htmlhintrc_morechecks .github/linters/.htmlhintrc
     - name: Lint Generated HTML
-      uses: github/super-linter@v4.7.3
+      uses: github/super-linter@v4.8.0
       env:
         DEFAULT_BRANCH: main
         FILTER_REGEX_INCLUDE: src/static/html/.*
diff --git a/sql/2019/media/04_03b.sql b/sql/2019/media/04_03b.sql
index 0938bb79291..0f7270c7ad0 100644
--- a/sql/2019/media/04_03b.sql
+++ b/sql/2019/media/04_03b.sql
@@ -43,7 +43,7 @@ FROM
           # many 404s ANDredirects show up as image/gif
           status = 200 AND
 
-        # we are trying to catch images. WPO populates the format for media but it uses a file extension guess.
+          # we are trying to catch images. WPO populates the format for media but it uses a file extension guess.
           #So we exclude mimetypes that aren't image or where the format couldn't be guessed by WPO
           (format <> '' OR mimetype LIKE 'image%') AND
 
diff --git a/sql/2019/media/04_11a.sql b/sql/2019/media/04_11a.sql
index dd6e6ddc252..48dd3b6fa8b 100644
--- a/sql/2019/media/04_11a.sql
+++ b/sql/2019/media/04_11a.sql
@@ -63,7 +63,7 @@ LEFT JOIN
       # many 404s and redirects show up as image/gif
       status = 200 AND
 
-        # we are trying to catch images. WPO populates the format for media but it uses a file extension guess.
+      # we are trying to catch images. WPO populates the format for media but it uses a file extension guess.
       #So we exclude mimetypes that aren't image or where the format couldn't be guessed by WPO
       (format <> '' OR mimetype LIKE 'image%') AND
 
diff --git a/sql/2020/media/bytes_per_pixel.sql b/sql/2020/media/bytes_per_pixel.sql
index 74a827d3be0..e4b26cbd350 100644
--- a/sql/2020/media/bytes_per_pixel.sql
+++ b/sql/2020/media/bytes_per_pixel.sql
@@ -62,7 +62,7 @@ LEFT JOIN
       # many 404s and redirects show up as image/gif
       status = 200 AND
 
-        # we are trying to catch images. WPO populates the format for media but it uses a file extension guess.
+      # we are trying to catch images. WPO populates the format for media but it uses a file extension guess.
       #So we exclude mimetypes that aren't image or where the format couldn't be guessed by WPO
       (format <> '' OR mimetype LIKE 'image%') AND
 
diff --git a/sql/2021/caching/README.md b/sql/2021/caching/README.md
index 64ac88c55bd..5791e24f2b2 100644
--- a/sql/2021/caching/README.md
+++ b/sql/2021/caching/README.md
@@ -1,10 +1,8 @@
 # 2021 Caching queries
 
-<!--
-  This directory contains all of the 2021 Caching chapter queries.
+Resources:
 
-  Each query should have a corresponding `metric_name.sql` file.
-  Note that readers are linked to this directory, so try to make the SQL file names descriptive for easy browsing.
-
-  Analysts: if helpful, you can use this README to give additional info about the queries.
--->
+- [Planning doc](https://docs.google.com/document/d/1BSdljOv86CPAP5T0A9jcV6bU7NJp1f2mTFa4ZZ7xxe8/edit)
+- [Results sheet](https://docs.google.com/spreadsheets/d/1-v3yR0LZIC3t4zWtqTgR3jJsKjjRMP-HATU2caP8e2c/edit)
+- [2019 chapter](https://almanac.httparchive.org/en/2019/caching)
+- [2020 chapter](https://almanac.httparchive.org/en/2020/caching)
diff --git a/sql/2021/caching/cache_control_directives.sql b/sql/2021/caching/cache_control_directives.sql
new file mode 100644
index 00000000000..67310139f00
--- /dev/null
+++ b/sql/2021/caching/cache_control_directives.sql
@@ -0,0 +1,71 @@
+#standardSQL
+# Use of Cache-Control directives
+SELECT
+  client,
+  COUNT(0) AS total_requests,
+  COUNTIF(uses_cache_control) AS total_using_cache_control,
+  COUNTIF(uses_max_age) AS total_using_max_age,
+  COUNTIF(uses_no_cache) AS total_using_no_cache,
+  COUNTIF(uses_public) AS total_using_public,
+  COUNTIF(uses_must_revalidate) AS total_using_must_revalidate,
+  COUNTIF(uses_no_store) AS total_using_no_store,
+  COUNTIF(uses_private) AS total_using_private,
+  COUNTIF(uses_proxy_revalidate) AS total_using_proxy_revalidate,
+  COUNTIF(uses_s_maxage) AS total_using_s_maxage,
+  COUNTIF(uses_no_transform) AS total_using_no_transform,
+  COUNTIF(uses_immutable) AS total_using_immutable,
+  COUNTIF(uses_stale_while_revalidate) AS total_using_stale_while_revalidate,
+  COUNTIF(uses_stale_if_error) AS total_using_stale_if_error,
+  COUNTIF(uses_no_store AND uses_no_cache AND uses_max_age_zero) AS total_using_no_store_and_no_cache_and_max_age_zero,
+  COUNTIF(uses_no_store AND uses_no_cache AND NOT uses_max_age_zero) AS total_using_no_store_and_no_cache_only,
+  COUNTIF(uses_no_store AND NOT uses_no_cache AND NOT uses_max_age_zero) AS total_using_no_store_only,
+  COUNTIF(uses_max_age_zero AND NOT uses_no_store) AS total_using_max_age_zero_without_no_store,
+  COUNTIF(uses_pre_check_zero AND uses_post_check_zero) AS total_using_pre_check_zero_and_post_check_zero,
+  COUNTIF(uses_pre_check_zero) AS total_using_pre_check_zero,
+  COUNTIF(uses_post_check_zero) AS total_using_post_check_zero,
+  COUNTIF(uses_cache_control AND NOT uses_max_age AND NOT uses_no_cache AND NOT uses_public AND NOT uses_must_revalidate AND NOT uses_no_store AND NOT uses_private AND NOT uses_proxy_revalidate AND NOT uses_s_maxage AND NOT uses_no_transform AND NOT uses_immutable AND NOT uses_stale_while_revalidate AND NOT uses_stale_if_error AND NOT uses_pre_check_zero AND NOT uses_post_check_zero) AS total_erroneous_directives,
+  COUNTIF(uses_cache_control) / COUNT(0) AS pct_using_cache_control,
+  COUNTIF(uses_max_age) / COUNT(0) AS pct_using_max_age,
+  COUNTIF(uses_no_cache) / COUNT(0) AS pct_using_no_cache,
+  COUNTIF(uses_public) / COUNT(0) AS pct_using_public,
+  COUNTIF(uses_must_revalidate) / COUNT(0) AS pct_using_must_revalidate,
+  COUNTIF(uses_no_store) / COUNT(0) AS pct_using_no_store,
+  COUNTIF(uses_private) / COUNT(0) AS pct_using_private,
+  COUNTIF(uses_proxy_revalidate) / COUNT(0) AS pct_using_proxy_revalidate,
+  COUNTIF(uses_s_maxage) / COUNT(0) AS pct_using_s_maxage,
+  COUNTIF(uses_no_transform) / COUNT(0) AS pct_using_no_transform,
+  COUNTIF(uses_immutable) / COUNT(0) AS pct_using_immutable,
+  COUNTIF(uses_stale_while_revalidate) / COUNT(0) AS pct_using_stale_while_revalidate,
+  COUNTIF(uses_stale_if_error) / COUNT(0) AS pct_using_stale_if_error,
+  COUNTIF(uses_no_store AND uses_no_cache AND uses_max_age_zero) / COUNT(0) AS pct_using_no_store_and_no_cache_and_max_age_zero,
+  COUNTIF(uses_no_store AND uses_no_cache AND NOT uses_max_age_zero) / COUNT(0) AS pct_using_no_store_and_no_cache_only,
+  COUNTIF(uses_no_store AND NOT uses_no_cache AND NOT uses_max_age_zero) / COUNT(0) AS pct_using_no_store_only,
+  COUNTIF(uses_max_age_zero AND NOT uses_no_store) / COUNT(0) AS pct_using_max_age_zero_without_no_store,
+  COUNTIF(uses_pre_check_zero AND uses_post_check_zero) / COUNT(0) AS pct_using_pre_check_zero_and_post_check_zero,
+  COUNTIF(uses_pre_check_zero) / COUNT(0) AS pct_using_pre_check_zero,
+  COUNTIF(uses_post_check_zero) / COUNT(0) AS pct_using_post_check_zero,
+  COUNTIF(uses_cache_control AND NOT uses_max_age AND NOT uses_no_cache AND NOT uses_public AND NOT uses_must_revalidate AND NOT uses_no_store AND NOT uses_private AND NOT uses_proxy_revalidate AND NOT uses_s_maxage AND NOT uses_no_transform AND NOT uses_immutable AND NOT uses_stale_while_revalidate AND NOT uses_stale_if_error AND NOT uses_pre_check_zero AND NOT uses_post_check_zero) / COUNT(0) AS pct_erroneous_directives
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    TRIM(resp_cache_control) != "" AS uses_cache_control,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)max-age\s*=\s*[0-9]+') AS uses_max_age,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)max-age\s*=\s*0') AS uses_max_age_zero,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)public') AS uses_public,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)no-cache') AS uses_no_cache,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)must-revalidate') AS uses_must_revalidate,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)no-store') AS uses_no_store,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)private') AS uses_private,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)proxy-revalidate') AS uses_proxy_revalidate,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)s-maxage\s*=\s*[0-9]+') AS uses_s_maxage,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)no-transform') AS uses_no_transform,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)immutable') AS uses_immutable,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)stale-while-revalidate\s*=\s*[0-9]+') AS uses_stale_while_revalidate,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)stale-if-error\s*=\s*[0-9]+') AS uses_stale_if_error,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)pre-check\s*=\s*0') AS uses_pre_check_zero,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)post-check\s*=\s*0') AS uses_post_check_zero
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+)
+GROUP BY
+  client
diff --git a/sql/2021/caching/cache_control_directives_invalid.sql b/sql/2021/caching/cache_control_directives_invalid.sql
new file mode 100644
index 00000000000..84dc0e4d8b1
--- /dev/null
+++ b/sql/2021/caching/cache_control_directives_invalid.sql
@@ -0,0 +1,33 @@
+#standardSQL
+# List of invalid Cache-Control directive names.
+SELECT
+  client,
+  total_directives,
+  total_using_cache_control,
+  directive_name,
+  directive_occurrences,
+  directive_occurrences / total_using_cache_control AS pct_of_cache_control,
+  directive_occurrences / total_directives AS pct_of_total_directives
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    directive_name,
+    COUNT(0) AS directive_occurrences,
+    SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total_directives,
+    SUM(COUNTIF(TRIM(resp_cache_control) != '')) OVER (PARTITION BY _TABLE_SUFFIX) AS total_using_cache_control
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(REGEXP_EXTRACT_ALL(LOWER(resp_cache_control), r'([a-z][^,\s="\']*)')) AS directive_name
+  GROUP BY
+    client,
+    directive_name)
+WHERE
+  directive_name NOT IN (
+    'max-age', 'public', 'no-cache', 'must-revalidate', 'no-store',
+    'private', 'proxy-revalidate', 's-maxage', 'no-transform',
+    'immutable', 'stale-while-revalidate', 'stale-if-error',
+    'pre-check', 'post-check')
+ORDER BY
+  client,
+  directive_occurrences DESC
diff --git a/sql/2021/caching/cache_ttl_and_content_age_diff.sql b/sql/2021/caching/cache_ttl_and_content_age_diff.sql
new file mode 100644
index 00000000000..d36db078c8c
--- /dev/null
+++ b/sql/2021/caching/cache_ttl_and_content_age_diff.sql
@@ -0,0 +1,32 @@
+#standardSQL
+# Difference between Cache TTL and the contents age
+CREATE TEMPORARY FUNCTION toTimestamp(date_string STRING)
+RETURNS INT64 LANGUAGE js AS '''
+  try {
+    var timestamp = Math.round(new Date(date_string).getTime() / 1000);
+    return isNaN(timestamp) || timestamp < 0 ? -1 : timestamp;
+  } catch (e) {
+    return null;
+  }
+''';
+
+SELECT
+  client,
+  percentile,
+  APPROX_QUANTILES(diff_in_days, 1000 IGNORE NULLS)[OFFSET(percentile * 10)] AS diff_in_days
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    ROUND((expAge - (startedDateTime - toTimestamp(resp_last_modified))) / (60 * 60 * 24), 2) AS diff_in_days
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+  WHERE
+    resp_last_modified <> "" AND
+    expAge > 0),
+  UNNEST([10, 25, 50, 75, 90]) AS percentile
+GROUP BY
+  client,
+  percentile
+ORDER BY
+  client,
+  percentile
diff --git a/sql/2021/caching/cache_ttl_lighthouse_score.sql b/sql/2021/caching/cache_ttl_lighthouse_score.sql
new file mode 100644
index 00000000000..fd7fa786907
--- /dev/null
+++ b/sql/2021/caching/cache_ttl_lighthouse_score.sql
@@ -0,0 +1,16 @@
+#standardSQL
+# Distribution of cache TTL Lighthouse scores
+SELECT
+  _TABLE_SUFFIX AS client,
+  JSON_EXTRACT_SCALAR(report, "$.audits.uses-long-cache-ttl.score") AS caching_score,
+  COUNT(0) AS num_pages,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct_pages
+FROM
+  `httparchive.lighthouse.2021_07_01_*`
+GROUP BY
+  client,
+  caching_score
+ORDER BY
+  client,
+  caching_score ASC
diff --git a/sql/2021/caching/cache_wastedbytes_lighthouse.sql b/sql/2021/caching/cache_wastedbytes_lighthouse.sql
new file mode 100644
index 00000000000..ee4cd21c5ad
--- /dev/null
+++ b/sql/2021/caching/cache_wastedbytes_lighthouse.sql
@@ -0,0 +1,16 @@
+#standardSQL
+# Distribution of bytes wasted (absence of adequate caching) from Lighthouse
+SELECT
+  _TABLE_SUFFIX AS client,
+  ROUND(CAST(JSON_EXTRACT_SCALAR(report, "$.audits.uses-long-cache-ttl.details.summary.wastedBytes") AS NUMERIC) / 1024 / 1024) AS mbyte_savings,
+  COUNT(0) AS num_pages,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct_pages
+FROM
+  `httparchive.lighthouse.2021_07_01_*`
+GROUP BY
+  client,
+  mbyte_savings
+ORDER BY
+  client,
+  mbyte_savings ASC
diff --git a/sql/2021/caching/content_age_older_than_ttl.sql b/sql/2021/caching/content_age_older_than_ttl.sql
new file mode 100644
index 00000000000..b7ef9f8873e
--- /dev/null
+++ b/sql/2021/caching/content_age_older_than_ttl.sql
@@ -0,0 +1,30 @@
+#standardSQL
+# Requests with a content age older than its TTL
+CREATE TEMPORARY FUNCTION toTimestamp(date_string STRING)
+RETURNS INT64 LANGUAGE js AS '''
+  try {
+    var timestamp = Math.round(new Date(date_string).getTime() / 1000);
+    return isNaN(timestamp) || timestamp < 0 ? -1 : timestamp;
+  } catch (e) {
+    return null;
+  }
+''';
+
+SELECT
+  client,
+  COUNT(0) AS total_req,
+  COUNTIF(diff < 0) AS req_too_short_cache,
+  COUNTIF(diff < 0) / COUNT(0) AS perc_req_too_short_cache
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    expAge - (startedDateTime - toTimestamp(resp_last_modified)) AS diff
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+  WHERE
+    resp_last_modified <> "" AND
+    expAge > 0)
+GROUP BY
+  client
+ORDER BY
+  client
diff --git a/sql/2021/caching/content_age_older_than_ttl_by_party.sql b/sql/2021/caching/content_age_older_than_ttl_by_party.sql
new file mode 100644
index 00000000000..028b971aee9
--- /dev/null
+++ b/sql/2021/caching/content_age_older_than_ttl_by_party.sql
@@ -0,0 +1,36 @@
+#standardSQL
+# Difference between Cache TTL and the content age for third party request
+CREATE TEMPORARY FUNCTION toTimestamp(date_string STRING)
+RETURNS INT64 LANGUAGE js AS '''
+  try {
+    var timestamp = Math.round(new Date(date_string).getTime() / 1000);
+    return isNaN(timestamp) || timestamp < 0 ? -1 : timestamp;
+  } catch (e) {
+    return null;
+  }
+''';
+
+SELECT
+  client,
+  party,
+  COUNT(0) AS total_req,
+  COUNTIF(diff < 0) AS req_too_short_cache,
+  COUNTIF(diff < 0) / COUNT(0) AS perc_req_too_short_cache
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    IF(NET.HOST(url) IN (
+      SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+    ), 'third party', 'first party') AS party,
+    requests.expAge - (requests.startedDateTime - toTimestamp(requests.resp_last_modified)) AS diff
+  FROM
+    `httparchive.summary_requests.2021_07_01_*` requests
+  WHERE
+    TRIM(requests.resp_last_modified) <> "" AND
+    expAge > 0)
+GROUP BY
+  client,
+  party
+ORDER BY
+  client,
+  party
diff --git a/sql/2021/caching/cookie_attributes.sql b/sql/2021/caching/cookie_attributes.sql
new file mode 100644
index 00000000000..1780069143d
--- /dev/null
+++ b/sql/2021/caching/cookie_attributes.sql
@@ -0,0 +1,60 @@
+#standardSQL
+# Popularity of top Set-Cookie attributes/directives
+CREATE TEMPORARY FUNCTION getCookieAttributes(headers STRING)
+RETURNS ARRAY<STRING> DETERMINISTIC LANGUAGE js AS '''
+try {
+  var $ = JSON.parse(headers);
+  return $.filter(header => {
+    return header.name.toLowerCase() == 'set-cookie';
+  }).flatMap(header => {
+    return Array.from(new Set(header.value.split(';').slice(1).map(attr => {
+      return attr.trim().split('=')[0].trim();
+    })));
+  });
+} catch (e) {
+  return [];
+}
+''';
+
+CREATE TEMPORARY FUNCTION countCookies(headers STRING)
+RETURNS INT64 DETERMINISTIC LANGUAGE js AS '''
+try {
+  var $ = JSON.parse(headers);
+  return $.filter(header => {
+    return header.name.toLowerCase() == 'set-cookie';
+  }).length;
+} catch (e) {
+  return 0;
+}
+''';
+
+SELECT
+  client,
+  attr.value AS attr,
+  attr.count AS freq,
+  total,
+  attr.count / total AS pct
+FROM (
+  SELECT
+    client,
+    APPROX_TOP_COUNT(attr, 100) AS attrs
+  FROM
+    `httparchive.almanac.requests`,
+    UNNEST(getCookieAttributes(response_headers)) AS attr
+  WHERE
+    date = '2021-07-01'
+  GROUP BY
+    client)
+JOIN (
+  SELECT
+    client,
+    SUM(countCookies(response_headers)) AS total
+  FROM
+    `httparchive.almanac.requests`
+  GROUP BY
+    client)
+USING
+  (client),
+  UNNEST(attrs) AS attr
+ORDER BY
+  pct DESC
diff --git a/sql/2021/caching/cookie_names.sql b/sql/2021/caching/cookie_names.sql
new file mode 100644
index 00000000000..b3fe1f9084a
--- /dev/null
+++ b/sql/2021/caching/cookie_names.sql
@@ -0,0 +1,37 @@
+#standardSQL
+# Popularity of top Set-Cookie names
+CREATE TEMPORARY FUNCTION getCookies(headers STRING)
+RETURNS ARRAY<STRING> DETERMINISTIC LANGUAGE js AS '''
+try {
+  var $ = JSON.parse(headers);
+  return $.filter(header => {
+    return header.name.toLowerCase() == 'set-cookie';
+  }).map(header => {
+    return header.value.split('=')[0].trim();
+  });
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  client,
+  cookie.value AS cookie,
+  cookie.count AS freq,
+  total,
+  cookie.count / total AS pct
+FROM (
+  SELECT
+    client,
+    APPROX_TOP_COUNT(cookie, 100) AS cookies,
+    COUNT(0) AS total
+  FROM
+    `httparchive.almanac.requests`,
+    UNNEST(getCookies(response_headers)) AS cookie
+  WHERE
+    date = '2021-07-01'
+  GROUP BY
+    client),
+  UNNEST(cookies) AS cookie
+ORDER BY
+  pct DESC
diff --git a/sql/2021/caching/header_trends.sql b/sql/2021/caching/header_trends.sql
new file mode 100644
index 00000000000..9e85bcc1d64
--- /dev/null
+++ b/sql/2021/caching/header_trends.sql
@@ -0,0 +1,86 @@
+#standardSQL
+# Use of Cache-Control, max-age in Cache-Control, and Expires
+WITH summary_requests AS (
+  SELECT
+    '2019' AS year,
+    _TABLE_SUFFIX AS client,
+    *
+  FROM
+    `httparchive.summary_requests.2019_07_01_*`
+  UNION ALL
+  SELECT
+    '2020' AS year,
+    _TABLE_SUFFIX AS client,
+    *
+  FROM
+    `httparchive.summary_requests.2020_08_01_*`
+  UNION ALL
+  SELECT
+    '2021' AS year,
+    _TABLE_SUFFIX AS client,
+    *
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+)
+
+SELECT
+  year,
+  client,
+  COUNT(0) AS total_requests,
+
+  COUNTIF(uses_cache_control) AS total_using_cache_control,
+  COUNTIF(uses_max_age) AS total_using_max_age,
+  COUNTIF(uses_expires) AS total_using_expires,
+  COUNTIF(uses_max_age AND uses_expires) AS total_using_max_age_and_expires,
+  COUNTIF(uses_cache_control AND uses_expires) AS total_using_both,
+  COUNTIF(NOT uses_cache_control AND NOT uses_expires) AS total_using_neither,
+  COUNTIF(uses_cache_control AND NOT uses_expires) AS total_using_only_cache_control,
+  COUNTIF(NOT uses_cache_control AND uses_expires) AS total_using_only_expires,
+
+  COUNTIF(uses_no_etag) AS total_using_no_etag,
+  COUNTIF(uses_etag) AS total_using_etag,
+  COUNTIF(uses_weak_etag) AS total_using_weak_etag,
+  COUNTIF(uses_strong_etag) AS total_using_strong_etag,
+  COUNTIF(NOT uses_weak_etag AND NOT uses_strong_etag AND uses_etag) AS total_using_invalid_etag,
+  COUNTIF(uses_last_modified) AS total_using_last_modified,
+  COUNTIF(uses_etag AND uses_last_modified) AS total_using_both,
+  COUNTIF(NOT uses_etag AND NOT uses_last_modified) AS total_using_neither,
+
+  COUNTIF(uses_cache_control) / COUNT(0) AS pct_cache_control,
+  COUNTIF(uses_max_age) / COUNT(0) AS pct_using_max_age,
+  COUNTIF(uses_expires) / COUNT(0) AS pct_using_expires,
+  COUNTIF(uses_max_age AND uses_expires) / COUNT(0) AS pct_using_max_age_and_expires,
+  COUNTIF(uses_cache_control AND uses_expires) / COUNT(0) AS pct_using_both,
+  COUNTIF(NOT uses_cache_control AND NOT uses_expires) / COUNT(0) AS pct_using_neither,
+  COUNTIF(uses_cache_control AND NOT uses_expires) / COUNT(0) AS pct_using_only_cache_control,
+  COUNTIF(NOT uses_cache_control AND uses_expires) / COUNT(0) AS pct_using_only_expires,
+
+  COUNTIF(uses_no_etag) / COUNT(0) AS pct_using_no_etag,
+  COUNTIF(uses_etag) / COUNT(0) AS pct_using_etag,
+  COUNTIF(uses_weak_etag) / COUNT(0) AS pct_using_weak_etag,
+  COUNTIF(uses_strong_etag) / COUNT(0) AS pct_using_strong_etag,
+  COUNTIF(NOT uses_weak_etag AND NOT uses_strong_etag AND uses_etag) / COUNT(0) AS pct_using_invalid_etag,
+  COUNTIF(uses_last_modified) / COUNT(0) AS pct_using_last_modified,
+  COUNTIF(uses_etag AND uses_last_modified) / COUNT(0) AS pct_using_both,
+  COUNTIF(NOT uses_etag AND NOT uses_last_modified) / COUNT(0) AS pct_using_neither
+FROM (
+  SELECT
+    year,
+    client,
+    TRIM(resp_expires) != "" AS uses_expires,
+    TRIM(resp_cache_control) != "" AS uses_cache_control,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)max-age\s*=\s*[0-9]+') AS uses_max_age,
+    TRIM(resp_etag) = "" AS uses_no_etag,
+    TRIM(resp_etag) != "" AS uses_etag,
+    TRIM(resp_last_modified) != "" AS uses_last_modified,
+    REGEXP_CONTAINS(TRIM(resp_etag), '^W/\".*\"') AS uses_weak_etag,
+    REGEXP_CONTAINS(TRIM(resp_etag), '^\".*\"') AS uses_strong_etag
+  FROM
+    summary_requests
+)
+GROUP BY
+  year,
+  client
+ORDER BY
+  year,
+  client
diff --git a/sql/2021/caching/invalid_last_modified_and_expires_and_date.sql b/sql/2021/caching/invalid_last_modified_and_expires_and_date.sql
new file mode 100644
index 00000000000..f8dedf039ca
--- /dev/null
+++ b/sql/2021/caching/invalid_last_modified_and_expires_and_date.sql
@@ -0,0 +1,31 @@
+#standardSQL
+# Valid date in Last-Modified, Expires, and Date headers
+SELECT
+  client,
+  COUNT(0) AS total_requests,
+  COUNTIF(uses_date) AS total_using_date,
+  COUNTIF(uses_last_modified) AS total_using_last_modified,
+  COUNTIF(uses_expires) AS total_using_expires,
+  COUNTIF(uses_date AND NOT has_valid_date) AS total_using_invalid_date,
+  COUNTIF(uses_last_modified AND NOT has_valid_last_modified) AS total_using_invalid_last_modified,
+  COUNTIF(uses_expires AND NOT has_valid_expires) AS total_using_invalid_expires,
+  COUNTIF(uses_date) / COUNT(0) AS pct_using_date,
+  COUNTIF(uses_last_modified) / COUNT(0) AS pct_using_last_modified,
+  COUNTIF(uses_expires) / COUNT(0) AS pct_using_expires,
+  COUNTIF(uses_date AND NOT has_valid_date) / COUNT(uses_date) AS pct_using_invalid_date,
+  COUNTIF(uses_last_modified AND NOT has_valid_last_modified) / COUNT(uses_last_modified) AS pct_using_invalid_last_modified,
+  COUNTIF(uses_expires AND NOT has_valid_expires) / COUNT(uses_expires) AS pct_using_invalid_expires
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    TRIM(resp_date) != "" AS uses_date,
+    TRIM(resp_last_modified) != "" AS uses_last_modified,
+    TRIM(resp_expires) != "" AS uses_expires,
+    REGEXP_CONTAINS(TRIM(resp_date), r'^(Mon|Tue|Wed|Thu|Fri|Sat|Sun), \d{1,2} (Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec) \d{4} \d{2}:\d{2}:\d{2} GMT$') AS has_valid_date,
+    REGEXP_CONTAINS(TRIM(resp_last_modified), r'^(Mon|Tue|Wed|Thu|Fri|Sat|Sun), \d{1,2} (Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec) \d{4} \d{2}:\d{2}:\d{2} GMT$') AS has_valid_last_modified,
+    REGEXP_CONTAINS(TRIM(resp_expires), r'^(Mon|Tue|Wed|Thu|Fri|Sat|Sun), \d{1,2} (Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec) \d{4} \d{2}:\d{2}:\d{2} GMT$') AS has_valid_expires
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+)
+GROUP BY
+  client
diff --git a/sql/2021/caching/max_age_distribution.sql b/sql/2021/caching/max_age_distribution.sql
new file mode 100644
index 00000000000..d524fec6e08
--- /dev/null
+++ b/sql/2021/caching/max_age_distribution.sql
@@ -0,0 +1,22 @@
+#standardSQL
+# Distribution of max-age values
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(CAST(max_age AS NUMERIC), 1000)[OFFSET(percentile * 10)] AS max_age
+FROM (
+  SELECT
+    _TABLE_SUFFIX,
+    REGEXP_EXTRACT(resp_cache_control, r'(?i)max-age\s*=\s*(\d+)') AS max_age
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+  ),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+WHERE
+  max_age IS NOT NULL
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/caching/non_cacheable_by_resource_type.sql b/sql/2021/caching/non_cacheable_by_resource_type.sql
new file mode 100644
index 00000000000..a0c7a5c44e2
--- /dev/null
+++ b/sql/2021/caching/non_cacheable_by_resource_type.sql
@@ -0,0 +1,33 @@
+#standardSQL
+# Non-cacheable content (no-store present) by resource type
+SELECT
+  client,
+  resource_type,
+  COUNT(0) AS total_requests,
+  COUNTIF(NOT uses_no_store) AS total_cacheable,
+  COUNTIF(uses_no_store) AS total_non_cacheable,
+  COUNTIF(NOT uses_cache_control AND NOT uses_expires) AS total_using_neither,
+  COUNTIF(NOT uses_no_store AND uses_max_age AND exp_age = 0) AS total_exp_age_zero,
+  COUNTIF(NOT uses_no_store AND uses_max_age AND exp_age > 0) AS total_exp_age_gt_zero,
+  COUNTIF(NOT uses_no_store) / COUNT(0) AS pct_cacheable,
+  COUNTIF(uses_no_store) / COUNT(0) AS pct_non_cacheable,
+  COUNTIF(NOT uses_cache_control AND NOT uses_expires) / COUNTIF(NOT uses_no_store) AS pct_using_neither,
+  COUNTIF(NOT uses_no_store AND uses_max_age AND exp_age = 0) / COUNTIF(NOT uses_no_store) AS pct_using_exp_age_zero,
+  COUNTIF(NOT uses_no_store AND uses_max_age AND exp_age > 0) / COUNTIF(NOT uses_no_store) AS pct_using_exp_age_gt_zero
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    type AS resource_type,
+    TRIM(resp_cache_control) != "" AS uses_cache_control,
+    TRIM(resp_expires) != "" AS uses_expires,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)no-store') AS uses_no_store,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)max-age\s*=\s*[0-9]+') AS uses_max_age,
+    expAge AS exp_age
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`)
+GROUP BY
+  client,
+  resource_type
+ORDER BY
+  client,
+  resource_type
diff --git a/sql/2021/caching/resource_age_party_and_type_wise.sql b/sql/2021/caching/resource_age_party_and_type_wise.sql
new file mode 100644
index 00000000000..0c062bf7154
--- /dev/null
+++ b/sql/2021/caching/resource_age_party_and_type_wise.sql
@@ -0,0 +1,35 @@
+#standardSQL
+# Age of resources party, type wise.
+CREATE TEMPORARY FUNCTION toTimestamp(date_string STRING)
+RETURNS INT64 LANGUAGE js AS '''
+  try {
+    var timestamp = Math.round(new Date(date_string).getTime() / 1000);
+    return isNaN(timestamp) || timestamp < 0 ? null : timestamp;
+  } catch (e) {
+    return null;
+  }
+''';
+
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  IF(NET.HOST(url) IN (
+    SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+  ), 'third party', 'first party') AS party,
+  type AS resource_type,
+  APPROX_QUANTILES(ROUND((startedDateTime - toTimestamp(resp_last_modified)) / (60 * 60 * 24 * 7)), 1000 IGNORE NULLS)[OFFSET(percentile * 10)] AS age_weeks
+FROM
+  `httparchive.summary_requests.2021_07_01_*`,
+  UNNEST([10, 25, 50, 75, 90]) AS percentile
+WHERE
+  TRIM(resp_last_modified) <> ""
+GROUP BY
+  percentile,
+  client,
+  party,
+  resource_type
+ORDER BY
+  percentile,
+  client,
+  party,
+  resource_type
diff --git a/sql/2021/caching/resource_age_party_and_type_wise_groups.sql b/sql/2021/caching/resource_age_party_and_type_wise_groups.sql
new file mode 100644
index 00000000000..a93542fdc5f
--- /dev/null
+++ b/sql/2021/caching/resource_age_party_and_type_wise_groups.sql
@@ -0,0 +1,52 @@
+#standardSQL
+# Age of resources party, type wise in groups.
+CREATE TEMPORARY FUNCTION toTimestamp(date_string STRING)
+RETURNS INT64 LANGUAGE js AS '''
+  try {
+    var timestamp = Math.round(new Date(date_string).getTime() / 1000);
+    return isNaN(timestamp) || timestamp < 0 ? null : timestamp;
+  } catch (e) {
+    return null;
+  }
+''';
+
+SELECT
+  client,
+  party,
+  resource_type,
+  COUNT(0) AS requests_total,
+  COUNTIF(age_weeks IS NOT NULL) AS requests_with_age,
+  COUNTIF(age_weeks < 0) AS age_neg,
+  COUNTIF(age_weeks = 0) AS age_0wk,
+  COUNTIF(age_weeks >= 1 AND age_weeks <= 7) AS age_1_to_7wk,
+  COUNTIF(age_weeks >= 8 AND age_weeks <= 52) AS age_8_to_52wk,
+  COUNTIF(age_weeks >= 53 AND age_weeks <= 104) AS age_gt_1y,
+  COUNTIF(age_weeks >= 105) AS age_gt_2y,
+  SAFE_DIVIDE(COUNTIF(age_weeks < 0), COUNTIF(age_weeks IS NOT NULL)) AS age_neg_pct,
+  SAFE_DIVIDE(COUNTIF(age_weeks = 0), COUNTIF(age_weeks IS NOT NULL)) AS age_0wk_pct,
+  SAFE_DIVIDE(COUNTIF(age_weeks >= 1 AND age_weeks <= 7), COUNTIF(age_weeks IS NOT NULL)) AS age_1_to_7wk_pct,
+  SAFE_DIVIDE(COUNTIF(age_weeks >= 8 AND age_weeks <= 52), COUNTIF(age_weeks IS NOT NULL)) AS age_8_to_52wk_pct,
+  SAFE_DIVIDE(COUNTIF(age_weeks >= 53 AND age_weeks <= 104), COUNTIF(age_weeks IS NOT NULL)) AS age_gt_1y_pct,
+  SAFE_DIVIDE(COUNTIF(age_weeks >= 105), COUNTIF(age_weeks IS NOT NULL)) AS age_gt_2y_pct
+FROM
+  (
+    SELECT
+      _TABLE_SUFFIX AS client,
+      IF(NET.HOST(url) IN (
+        SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+      ), 'third party', 'first party') AS party,
+      type AS resource_type,
+      ROUND((startedDateTime - toTimestamp(resp_last_modified)) / (60 * 60 * 24 * 7)) AS age_weeks
+    FROM
+      `httparchive.summary_requests.2021_07_01_*`
+    WHERE
+      TRIM(resp_last_modified) <> ""
+  )
+GROUP BY
+  client,
+  party,
+  resource_type
+ORDER BY
+  client,
+  party,
+  resource_type
diff --git a/sql/2021/caching/service_worker_rank.sql b/sql/2021/caching/service_worker_rank.sql
new file mode 100644
index 00000000000..bcfa57719a5
--- /dev/null
+++ b/sql/2021/caching/service_worker_rank.sql
@@ -0,0 +1,20 @@
+#standardSQL
+# Adoption of SW by CrUX rank
+SELECT
+  client,
+  rank_magnitude AS rank,
+  COUNT(DISTINCT IF(feature = 'ServiceWorkerControlledPage', url, NULL)) AS sw_pages,
+  COUNT(DISTINCT url) AS total,
+  COUNT(DISTINCT IF(feature = 'ServiceWorkerControlledPage', url, NULL)) / COUNT(DISTINCT url) AS pct
+FROM
+  `httparchive.blink_features.features`,
+  UNNEST([1e3, 1e4, 1e5, 1e6, 1e7]) AS rank_magnitude
+WHERE
+  yyyymmdd = '2021-07-01' AND
+  rank <= rank_magnitude
+GROUP BY
+  client,
+  rank
+ORDER BY
+  rank,
+  client
diff --git a/sql/2021/caching/set_cookie.sql b/sql/2021/caching/set_cookie.sql
new file mode 100644
index 00000000000..fc5eb03c726
--- /dev/null
+++ b/sql/2021/caching/set_cookie.sql
@@ -0,0 +1,25 @@
+#standardSQL
+# Responses with Set-cookie header, absence of no-store means cacheable (max-age, Expires, or heuristic)
+SELECT
+  client,
+  COUNT(0) AS total_requests,
+  COUNTIF(NOT uses_no_store) AS total_cacheable,
+  COUNTIF(NOT uses_no_store AND uses_cookies) AS total_cacheable_set_cookie,
+  COUNTIF(NOT uses_no_store AND NOT uses_cookies) AS total_cacheable_without_set_cookie,
+  COUNTIF(NOT uses_no_store AND uses_cookies AND uses_private) AS total_pvt_cacheable_set_cookie,
+  COUNTIF(NOT uses_no_store AND uses_cookies AND NOT uses_private) AS total_pvt_public_cacheable_set_cookie,
+  COUNTIF(NOT uses_no_store AND uses_cookies) / COUNTIF(NOT uses_no_store) AS pct_cacheable_set_cookie,
+  COUNTIF(NOT uses_no_store AND NOT uses_cookies) / COUNTIF(NOT uses_no_store) AS pct_cacheable_without_set_cookie,
+  COUNTIF(NOT uses_no_store AND uses_cookies AND uses_private) / COUNTIF(NOT uses_no_store AND uses_cookies) AS pct_pvt_cacheable_set_cookie,
+  COUNTIF(NOT uses_no_store AND uses_cookies AND NOT uses_private) / COUNTIF(NOT uses_no_store AND uses_cookies) AS pct_pvt_public_cacheable_set_cookie
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)no-store') AS uses_no_store,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)private') AS uses_private,
+    (reqCookieLen > 0) AS uses_cookies
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+)
+GROUP BY
+  client
diff --git a/sql/2021/caching/top_cdns.sql b/sql/2021/caching/top_cdns.sql
new file mode 100644
index 00000000000..c00c31bfed6
--- /dev/null
+++ b/sql/2021/caching/top_cdns.sql
@@ -0,0 +1,24 @@
+#standardSQL
+# Adoption of top CDNs
+SELECT
+  client,
+  cdn,
+  COUNT(0) AS freq,
+  total,
+  COUNT(0) / total AS pct
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total,
+    ARRAY_CONCAT_AGG(SPLIT(cdn, ', ')) AS cdn_list
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client),
+  UNNEST(cdn_list) AS cdn
+GROUP BY
+  client,
+  cdn,
+  total
+ORDER BY
+  pct DESC
diff --git a/sql/2021/caching/top_hosts_using_immutable.sql b/sql/2021/caching/top_hosts_using_immutable.sql
new file mode 100644
index 00000000000..4ff23637a1d
--- /dev/null
+++ b/sql/2021/caching/top_hosts_using_immutable.sql
@@ -0,0 +1,21 @@
+#standardSQL
+# The top domains to use immutable Cache-Control directive.
+SELECT
+  _TABLE_SUFFIX AS client,
+  NET.HOST(url) AS host,
+  COUNT(DISTINCT pageid) AS pages,
+  COUNT(0) AS requests,
+  SUM(COUNT(DISTINCT pageid)) OVER () AS total_pages,
+  SUM(COUNT(0)) OVER () AS total_requests,
+  COUNT(0) / SUM(COUNT(0)) OVER () AS pct_requests
+FROM
+  `httparchive.summary_requests.2021_07_01_*`
+WHERE
+  REGEXP_CONTAINS(resp_cache_control, r'(?i)immutable')
+GROUP BY
+  client,
+  domain
+ORDER BY
+  client,
+  pct_requests DESC
+LIMIT 200
diff --git a/sql/2021/caching/ttl.sql b/sql/2021/caching/ttl.sql
new file mode 100644
index 00000000000..a15a09e90f4
--- /dev/null
+++ b/sql/2021/caching/ttl.sql
@@ -0,0 +1,27 @@
+#standardSQL
+# TTL statistics for cacheable content (no-store absent)
+SELECT
+  client,
+  COUNT(0) AS total_requests,
+  COUNTIF(NOT uses_cache_control AND NOT uses_expires) AS total_using_neither,
+  COUNTIF(NOT uses_no_store AND uses_max_age AND exp_age = 0) AS total_exp_age_zero,
+  COUNTIF(NOT uses_no_store AND uses_max_age AND exp_age > 0) AS total_exp_age_gt_zero,
+  COUNTIF(uses_no_store) AS total_not_cacheable,
+  COUNTIF(NOT uses_no_store) AS total_cacheable,
+  COUNTIF(NOT uses_cache_control AND NOT uses_expires) / COUNTIF(NOT uses_no_store) AS pct_using_neither,
+  COUNTIF(NOT uses_no_store AND uses_max_age AND exp_age = 0) / COUNTIF(NOT uses_no_store) AS pct_using_exp_age_zero,
+  COUNTIF(NOT uses_no_store AND uses_max_age AND exp_age > 0) / COUNTIF(NOT uses_no_store) AS pct_using_exp_age_gt_zero,
+  COUNTIF(uses_no_store) / COUNT(0) AS pct_not_cacheable,
+  COUNTIF(NOT uses_no_store) / COUNT(0) AS pct_cacheable
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    TRIM(resp_cache_control) != "" AS uses_cache_control,
+    TRIM(resp_expires) != "" AS uses_expires,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)no-store') AS uses_no_store,
+    REGEXP_CONTAINS(resp_cache_control, r'(?i)max-age\s*=\s*[0-9]+') AS uses_max_age,
+    expAge AS exp_age
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`)
+GROUP BY
+  client
diff --git a/sql/2021/caching/ttl_by_resource.sql b/sql/2021/caching/ttl_by_resource.sql
new file mode 100644
index 00000000000..426cb301d8c
--- /dev/null
+++ b/sql/2021/caching/ttl_by_resource.sql
@@ -0,0 +1,21 @@
+#standardSQL
+# TTL by resource type for cacheable (no-store absent) content
+SELECT
+  _TABLE_SUFFIX AS client,
+  type AS response_type,
+  percentile,
+  APPROX_QUANTILES(expAge, 1000)[OFFSET(percentile * 10)] AS ttl
+FROM
+  `httparchive.summary_requests.2021_07_01_*`,
+  UNNEST([10, 25, 50, 75, 90]) AS percentile
+WHERE
+  NOT REGEXP_CONTAINS(resp_cache_control, r'(?i)no-store') AND
+  expAge > 0
+GROUP BY
+  client,
+  response_type,
+  percentile
+ORDER BY
+  client,
+  response_type,
+  percentile
diff --git a/sql/2021/caching/valid_if_modified_since_returns_304.sql b/sql/2021/caching/valid_if_modified_since_returns_304.sql
new file mode 100644
index 00000000000..bb40a8df41f
--- /dev/null
+++ b/sql/2021/caching/valid_if_modified_since_returns_304.sql
@@ -0,0 +1,26 @@
+#standardSQL
+# Whether making a If-Modified-Since request returns a 304 if the content have not changed (as seen from Last-Modified)
+SELECT
+  client,
+  COUNT(0) AS total_requests,
+  COUNTIF(status = 304) AS total_304,
+  COUNTIF(NOT uses_etag AND uses_last_modified AND uses_if_modified AND no_change) AS total_expected_304,
+  COUNTIF(NOT uses_etag AND uses_last_modified AND uses_if_modified AND no_change AND status = 304) AS total_actual_304,
+  COUNTIF(status = 304) / COUNT(0) AS pct_304,
+  COUNTIF(NOT uses_etag AND uses_last_modified AND uses_if_modified AND no_change) / COUNTIF(status = 304) AS pct_expected_304,
+  COUNTIF(NOT uses_etag AND uses_last_modified AND uses_if_modified AND no_change AND status = 304) / COUNTIF(NOT uses_etag AND uses_last_modified AND uses_if_modified AND no_change) AS pct_actual_304
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    status,
+    TRIM(resp_last_modified) = TRIM(req_if_modified_since) AS no_change,
+    TRIM(resp_last_modified) != "" AS uses_last_modified,
+    TRIM(req_if_modified_since) != "" AS uses_if_modified,
+    TRIM(resp_etag) != "" AS uses_etag
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+)
+GROUP BY
+  client
+ORDER BY
+  client
diff --git a/sql/2021/caching/valid_if_none_match_returns_304.sql b/sql/2021/caching/valid_if_none_match_returns_304.sql
new file mode 100644
index 00000000000..6c47a0de870
--- /dev/null
+++ b/sql/2021/caching/valid_if_none_match_returns_304.sql
@@ -0,0 +1,24 @@
+# Whether making a If-None-Match request returns a 304 if the content have not changed (as seen from ETag)
+SELECT
+  client,
+  COUNT(0) AS total_requests,
+  COUNTIF(status = 304) AS total_304,
+  COUNTIF(uses_etag AND uses_if_non_match AND no_change) AS total_expected_304,
+  COUNTIF(uses_etag AND uses_if_non_match AND no_change AND status = 304) AS total_actual_304,
+  COUNTIF(status = 304) / COUNT(0) AS pct_304,
+  COUNTIF(uses_etag AND uses_if_non_match AND no_change) / COUNTIF(status = 304) AS pct_expected_304,
+  COUNTIF(uses_etag AND uses_if_non_match AND no_change AND status = 304) / COUNTIF(uses_etag AND uses_if_non_match AND no_change) AS pct_actual_304
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    status,
+    TRIM(resp_etag) = TRIM(req_if_none_match) AS no_change,
+    TRIM(resp_etag) != "" AS uses_etag,
+    TRIM(req_if_none_match) != "" AS uses_if_non_match
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+)
+GROUP BY
+  client
+ORDER BY
+  client
diff --git a/sql/2021/caching/vary_directives.sql b/sql/2021/caching/vary_directives.sql
new file mode 100644
index 00000000000..f1c197937f1
--- /dev/null
+++ b/sql/2021/caching/vary_directives.sql
@@ -0,0 +1,43 @@
+#standardSQL
+# List of Vary directive names.
+SELECT
+  client,
+  ANY_VALUE(total_requests) AS total_requests,
+  ANY_VALUE(total_using_vary) AS total_using_vary,
+  vary_header,
+  ANY_VALUE(occurrences) AS occurrences,
+  ANY_VALUE(occurrences) / ANY_VALUE(total_using_vary) AS pct_of_vary,
+  ANY_VALUE(occurrences) / ANY_VALUE(total_requests) AS pct_of_total_requests,
+  ANY_VALUE(total_using_both) / ANY_VALUE(total_using_vary) AS pct_of_vary_with_cache_control,
+  ANY_VALUE(total_using_vary) / ANY_VALUE(total_requests) AS pct_using_vary
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_requests,
+    COUNTIF(TRIM(resp_vary) != '') AS total_using_vary,
+    COUNTIF(TRIM(resp_vary) != '' AND TRIM(resp_cache_control) != '') AS total_using_both
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+  GROUP BY
+    client)
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    vary_header,
+    COUNT(0) AS occurrences
+  FROM
+    `httparchive.summary_requests.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(REGEXP_EXTRACT_ALL(LOWER(resp_vary), r'([a-z][^,\s="\']*)')) AS vary_header
+  GROUP BY
+    client,
+    vary_header)
+USING
+  (client)
+WHERE
+  vary_header IS NOT NULL
+GROUP BY
+  client,
+  vary_header
+ORDER BY
+  occurrences DESC
diff --git a/sql/2021/javascript/ajax_request_per_page.sql b/sql/2021/javascript/ajax_request_per_page.sql
index 93ba5343c0b..46fae7d771f 100644
--- a/sql/2021/javascript/ajax_request_per_page.sql
+++ b/sql/2021/javascript/ajax_request_per_page.sql
@@ -10,7 +10,7 @@ FROM (
           '$._javascript'),
         '$.ajax_requests.total') AS INT64) AS ajax_requests_total
   FROM
-    `httparchive.sample_data.pages_*` ),
+    `httparchive.pages.2021_07_01_*` ),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 WHERE
   ajax_requests_total > 0
diff --git a/sql/2021/javascript/async_and_defer.sql b/sql/2021/javascript/async_and_defer.sql
deleted file mode 100644
index 718e7201bd8..00000000000
--- a/sql/2021/javascript/async_and_defer.sql
+++ /dev/null
@@ -1,25 +0,0 @@
-SELECT
-  client,
-  COUNT(DISTINCT IF(async, page, NULL)) AS async,
-  COUNT(DISTINCT IF(defer, page, NULL)) AS defer,
-  COUNT(DISTINCT IF(async AND defer, page, NULL)) AS both,
-  COUNT(DISTINCT page) AS total,
-  COUNT(DISTINCT IF(async, page, NULL)) / COUNT(DISTINCT page) AS pct_async,
-  COUNT(DISTINCT IF(defer, page, NULL)) / COUNT(DISTINCT page) AS pct_defer,
-  COUNT(DISTINCT IF(async AND defer, page, NULL)) / COUNT(DISTINCT page) AS pct_both
-FROM (
-  SELECT
-    client,
-    page,
-    script,
-    REGEXP_CONTAINS(script, r'(?i)\basync\b') AS async,
-    REGEXP_CONTAINS(script, r'(?i)\bdefer\b') AS defer
-  FROM
-    `httparchive.sample_data.summary_response_bodies`  -- TODO: update to the original table name
-  LEFT JOIN
-    UNNEST(REGEXP_EXTRACT_ALL(body, r'(?i)(<script[^>]*>)')) AS script
-  WHERE
-    date = '2021-08-01' AND
-    firstHtml)
-GROUP BY
-  client
diff --git a/sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql b/sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql
deleted file mode 100644
index 1e06046cf91..00000000000
--- a/sql/2021/javascript/avg_pct_per_page_scripts_using_async_defer_module_nomodule.sql
+++ /dev/null
@@ -1,36 +0,0 @@
-#standardSQL
-# Average Percent per page of external scripts using Async, Defer, Module or NoModule attributes.
-SELECT
-  client,
-  AVG(pct_external_async) AS avg_pct_external_async,
-  AVG(pct_external_defer) AS avg_pct_external_defer,
-  AVG(pct_external_module) AS avg_pct_external_module,
-  AVG(pct_external_nomodule) AS avg_pct_external_nomodule
-FROM (
-  SELECT
-    client,
-    page,
-    COUNT(0) AS external_scripts,
-    COUNTIF(REGEXP_CONTAINS(script, r'\basync\b')) / COUNT(0) AS pct_external_async,
-    COUNTIF(REGEXP_CONTAINS(script, r'\bdefer\b')) / COUNT(0) AS pct_external_defer,
-    COUNTIF(REGEXP_CONTAINS(script, r'\bmodule\b')) / COUNT(0) AS pct_external_module,
-    COUNTIF(REGEXP_CONTAINS(script, r'\bnomodule\b')) / COUNT(0) AS pct_external_nomodule
-  FROM (
-    SELECT
-      client,
-      page,
-      url,
-      REGEXP_EXTRACT_ALL(body, "(?i)(<script [^>]*)") AS scripts
-    FROM
-      `httparchive.sample_data.summary_response_bodies`  -- TODO: update to the original table name
-    WHERE
-      date = '2021-08-01' AND
-      firstHtml),
-    UNNEST(scripts) AS script
-  WHERE
-    REGEXP_CONTAINS(script, r'\bsrc\b')
-  GROUP BY
-    client,
-    page)
-GROUP BY
-  client
diff --git a/sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql b/sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql
deleted file mode 100644
index 2f43c8dd535..00000000000
--- a/sql/2021/javascript/breakdown_of_scripts_using_async_defer_module_nomodule.sql
+++ /dev/null
@@ -1,34 +0,0 @@
-#standardSQL
-# Breakdown of scripts using Async, Defer, Module or NoModule attributes.  Also breakdown of inline vs external scripts
-SELECT
-  client,
-  COUNT(0) AS total_scripts,
-  SUM(IF(script NOT LIKE "%src%", 1, 0)) AS inline_script,
-  SUM(IF(script LIKE "%src%", 1, 0)) AS external_script,
-  SUM(IF(script LIKE "%src%", 1, 0)) / COUNT(0) AS pct_external_script,
-  SUM(IF(script NOT LIKE "%src%", 1, 0)) / COUNT(0) AS pct_inline_script,
-  SUM(IF(script LIKE "%async%", 1, 0)) AS async,
-  SUM(IF(script LIKE "%defer%", 1, 0)) AS defer,
-  SUM(IF(script LIKE "%module%", 1, 0)) AS module,
-  SUM(IF(script LIKE "%nomodule%", 1, 0)) AS nomodule,
-  SUM(IF(script LIKE "%async%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_async,
-  SUM(IF(script LIKE "%defer%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_defer,
-  SUM(IF(script LIKE "%module%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_module,
-  SUM(IF(script LIKE "%nomodule%", 1, 0)) / SUM(IF(script LIKE "%src%", 1, 0)) AS pct_external_nomodule
-FROM
-  (
-    SELECT
-      client,
-      page,
-      url,
-      REGEXP_EXTRACT_ALL(LOWER(body), "(<script [^>]*)") AS scripts
-    FROM
-      `httparchive.sample_data.summary_response_bodies`  -- TODO: update to the original table name
-    WHERE
-      date = '2021-08-01' AND
-      firstHtml
-  )
-CROSS JOIN
-  UNNEST(scripts) AS script
-GROUP BY
-  client
diff --git a/sql/2021/javascript/bytes_2019.sql b/sql/2021/javascript/bytes_2019.sql
new file mode 100644
index 00000000000..3348585ad43
--- /dev/null
+++ b/sql/2021/javascript/bytes_2019.sql
@@ -0,0 +1,15 @@
+#standardSQL
+# Sum of JS request bytes per page (2019)
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(bytesJs / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM
+  `httparchive.summary_pages.2019_07_01_*`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/bytes_2020.sql b/sql/2021/javascript/bytes_2020.sql
new file mode 100644
index 00000000000..839c6139ca8
--- /dev/null
+++ b/sql/2021/javascript/bytes_2020.sql
@@ -0,0 +1,15 @@
+#standardSQL
+# Sum of JS request bytes per page (2020)
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(bytesJs / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM
+  `httparchive.summary_pages.2020_08_01_*`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/bytes_2021.sql b/sql/2021/javascript/bytes_2021.sql
new file mode 100644
index 00000000000..25f82a8ca4c
--- /dev/null
+++ b/sql/2021/javascript/bytes_2021.sql
@@ -0,0 +1,15 @@
+#standardSQL
+# Sum of JS request bytes per page (2020)
+SELECT
+  percentile,
+  _TABLE_SUFFIX AS client,
+  APPROX_QUANTILES(bytesJs / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM
+  `httparchive.summary_pages.2021_07_01_*`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/bytes_by_3p.sql b/sql/2021/javascript/bytes_by_3p.sql
new file mode 100644
index 00000000000..f5bde4b9d9f
--- /dev/null
+++ b/sql/2021/javascript/bytes_by_3p.sql
@@ -0,0 +1,33 @@
+#standardSQL
+# Distribution of 1P/3P JS bytes
+SELECT
+  percentile,
+  client,
+  host,
+  APPROX_QUANTILES(kbytes, 1000)[OFFSET(percentile * 10)] AS kbytes
+FROM (
+  SELECT
+    client,
+    page,
+    IF(NET.HOST(url) IN (
+      SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+    ), 'third party', 'first party') AS host,
+    SUM(respSize) / 1024 AS kbytes
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script'
+  GROUP BY
+    client,
+    page,
+    host),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client,
+  host
+ORDER BY
+  percentile,
+  client,
+  host
diff --git a/sql/2021/javascript/compression_method.sql b/sql/2021/javascript/compression_method.sql
new file mode 100644
index 00000000000..273e0e49551
--- /dev/null
+++ b/sql/2021/javascript/compression_method.sql
@@ -0,0 +1,47 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  compression,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total_js_requests,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct_js_requests
+FROM (
+  SELECT
+    client,
+    page,
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'Content-Encoding') AS compression
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script')
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  compression
+ORDER BY
+  pct_js_requests DESC
diff --git a/sql/2021/javascript/compression_method_by_3p.sql b/sql/2021/javascript/compression_method_by_3p.sql
new file mode 100644
index 00000000000..e9d00d1d453
--- /dev/null
+++ b/sql/2021/javascript/compression_method_by_3p.sql
@@ -0,0 +1,52 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  host,
+  compression,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client, host) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client, host) AS pct
+FROM (
+  SELECT
+    client,
+    page,
+    IF(NET.HOST(url) IN (
+      SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+    ), 'third party', 'first party') AS host,
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'Content-Encoding') AS compression
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script')
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  host,
+  compression
+ORDER BY
+  pct DESC
diff --git a/sql/2021/javascript/compression_none_by_bytes.sql b/sql/2021/javascript/compression_none_by_bytes.sql
new file mode 100644
index 00000000000..592248c542f
--- /dev/null
+++ b/sql/2021/javascript/compression_none_by_bytes.sql
@@ -0,0 +1,55 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  host,
+  IF(kbytes < 100, FLOOR(kbytes / 5) * 5, 100) AS kbytes,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client, host) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client, host) AS pct
+FROM (
+  SELECT
+    client,
+    page,
+    IF(NET.HOST(url) IN (
+      SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+    ), 'third party', 'first party') AS host,
+    respSize / 1024 AS kbytes
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script' AND
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'Content-Encoding') IS NULL)
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  host,
+  kbytes
+ORDER BY
+  client,
+  host,
+  kbytes
diff --git a/sql/2021/javascript/frameworks-bytes-by-framework.sql b/sql/2021/javascript/frameworks-bytes-by-framework.sql
new file mode 100644
index 00000000000..17d3c9a8535
--- /dev/null
+++ b/sql/2021/javascript/frameworks-bytes-by-framework.sql
@@ -0,0 +1,35 @@
+#standardSQL
+# Sum of JS request bytes per page by framework (2020)
+SELECT
+  percentile,
+  client,
+  app AS js_framework,
+  COUNT(DISTINCT page) AS pages,
+  APPROX_QUANTILES(bytesJs / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    bytesJs
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`)
+JOIN (
+  SELECT DISTINCT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    app
+  FROM
+    `httparchive.technologies.2021_07_01_*`
+  WHERE
+    category = 'JavaScript frameworks')
+USING
+  (client, page),
+  UNNEST([10, 25, 50, 75, 90]) AS percentile
+GROUP BY
+  percentile,
+  client,
+  js_framework
+ORDER BY
+  percentile,
+  client,
+  pages DESC
diff --git a/sql/2021/javascript/frameworks_libraries.sql b/sql/2021/javascript/frameworks_libraries.sql
new file mode 100644
index 00000000000..27b0627455f
--- /dev/null
+++ b/sql/2021/javascript/frameworks_libraries.sql
@@ -0,0 +1,30 @@
+#standardSQL
+# Top JS frameworks and libraries
+SELECT
+  _TABLE_SUFFIX AS client,
+  category,
+  app,
+  COUNT(DISTINCT url) AS pages,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.technologies.2021_07_01_*`
+JOIN (
+  SELECT
+    _TABLE_SUFFIX,
+    COUNT(0) AS total
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    _TABLE_SUFFIX)
+USING
+  (_TABLE_SUFFIX)
+WHERE
+  category IN ('JavaScript frameworks', 'JavaScript libraries')
+GROUP BY
+  client,
+  category,
+  app,
+  total
+ORDER BY
+  pct DESC
diff --git a/sql/2021/javascript/frameworks_libraries_by_version.sql b/sql/2021/javascript/frameworks_libraries_by_version.sql
new file mode 100644
index 00000000000..9d56b59fe2e
--- /dev/null
+++ b/sql/2021/javascript/frameworks_libraries_by_version.sql
@@ -0,0 +1,32 @@
+#standardSQL
+# Top JS frameworks and libraries by version
+SELECT
+  _TABLE_SUFFIX AS client,
+  category,
+  app,
+  info AS version,
+  COUNT(DISTINCT url) AS pages,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.technologies.2021_07_01_*`
+JOIN (
+  SELECT
+    _TABLE_SUFFIX,
+    COUNT(0) AS total
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    _TABLE_SUFFIX)
+USING
+  (_TABLE_SUFFIX)
+WHERE
+  app IN ('jQuery', 'jQuery Migrate', 'jQuery UI', 'Modernizr', 'FancyBox', 'Slick', 'Lightbox', 'Moment.js', 'Underscore.js', 'Lodash', 'React')
+GROUP BY
+  client,
+  category,
+  app,
+  info,
+  total
+ORDER BY
+  pct DESC
diff --git a/sql/2021/javascript/frameworks_libraries_combos.sql b/sql/2021/javascript/frameworks_libraries_combos.sql
new file mode 100644
index 00000000000..24bea43f43a
--- /dev/null
+++ b/sql/2021/javascript/frameworks_libraries_combos.sql
@@ -0,0 +1,43 @@
+#standardSQL
+# Top JS frameworks and libraries combinations
+SELECT
+  *
+FROM (
+  SELECT
+    client,
+    apps,
+    COUNT(DISTINCT page) AS pages,
+    total,
+    COUNT(DISTINCT page) / total AS pct
+  FROM (
+    SELECT
+      _TABLE_SUFFIX AS client,
+      url AS page,
+      total,
+      ARRAY_TO_STRING(ARRAY_AGG(app ORDER BY app), ', ') AS apps
+    FROM
+      `httparchive.technologies.2021_07_01_*`
+    JOIN (
+      SELECT
+        _TABLE_SUFFIX,
+        COUNT(0) AS total
+      FROM
+        `httparchive.summary_pages.2021_07_01_*`
+      GROUP BY
+        _TABLE_SUFFIX)
+    USING
+      (_TABLE_SUFFIX)
+    WHERE
+      category IN ('JavaScript frameworks', 'JavaScript libraries')
+    GROUP BY
+      client,
+      url,
+      total)
+  GROUP BY
+    client,
+    apps,
+    total)
+WHERE
+  pages >= 10000
+ORDER BY
+  pct DESC
diff --git a/sql/2021/javascript/h1_vs_h2_vs_h3.sql b/sql/2021/javascript/h1_vs_h2_vs_h3.sql
index 5c4dad4c21a..59645440827 100644
--- a/sql/2021/javascript/h1_vs_h2_vs_h3.sql
+++ b/sql/2021/javascript/h1_vs_h2_vs_h3.sql
@@ -29,7 +29,7 @@ FROM (
         '$._javascript'),
       '$.requests_protocol.resources_h3') AS INT64) AS resources_h3
   FROM
-    `httparchive.sample_data.pages_*`),
+    `httparchive.pages.2021_07_01_*`),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
diff --git a/sql/2021/javascript/iframe_per_page.sql b/sql/2021/javascript/iframe_per_page.sql
index d80c8b07545..684630f0053 100644
--- a/sql/2021/javascript/iframe_per_page.sql
+++ b/sql/2021/javascript/iframe_per_page.sql
@@ -10,7 +10,7 @@ FROM (
           '$._javascript'),
         '$.iframe') AS INT64) AS iframe_total
   FROM
-    `httparchive.sample_data.pages_*` ),
+    `httparchive.pages.2021_07_01_*` ),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 WHERE
   iframe_total > 0
diff --git a/sql/2021/javascript/lighthouse_unminified_js.sql b/sql/2021/javascript/lighthouse_unminified_js.sql
new file mode 100644
index 00000000000..a5e48687936
--- /dev/null
+++ b/sql/2021/javascript/lighthouse_unminified_js.sql
@@ -0,0 +1,18 @@
+#standardSQL
+# Pages with unminified JS
+SELECT
+  score,
+  COUNT(0) AS pages,
+  SUM(COUNT(0)) OVER () AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER () AS pct
+FROM (
+  SELECT
+    JSON_EXTRACT_SCALAR(report, "$.audits['unminified-javascript'].score") AS score
+  FROM
+    `httparchive.lighthouse.2021_07_01_mobile`)
+WHERE
+  score IS NOT NULL
+GROUP BY
+  score
+ORDER BY
+  score
diff --git a/sql/2021/javascript/lighthouse_unminified_js_by_3p.sql b/sql/2021/javascript/lighthouse_unminified_js_by_3p.sql
new file mode 100644
index 00000000000..e93e86bd657
--- /dev/null
+++ b/sql/2021/javascript/lighthouse_unminified_js_by_3p.sql
@@ -0,0 +1,34 @@
+#standardSQL
+# Pages with unminified JS by 1P/3P
+CREATE TEMPORARY FUNCTION getUnminifiedJsUrls(audit STRING)
+RETURNS ARRAY<STRUCT<url STRING, wastedBytes INT64>> LANGUAGE js AS '''
+try {
+  var $ = JSON.parse(audit);
+  return $.details.items.map(({url, wastedBytes}) => {
+    return {url, wastedBytes};
+  });
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  AVG(pct_1p_wasted_bytes) AS avg_pct_1p_wasted_bytes,
+  AVG(pct_3p_wasted_bytes) AS avg_pct_3p_wasted_bytes
+FROM (
+  SELECT
+    page,
+    SUM(IF(is_3p, 0, wasted_bytes)) / SUM(wasted_bytes) AS pct_1p_wasted_bytes,
+    SUM(IF(is_3p, wasted_bytes, 0)) / SUM(wasted_bytes) AS pct_3p_wasted_bytes
+  FROM (
+    SELECT
+      test.url AS page,
+      NET.HOST(unminified.url) IS NOT NULL AND NET.HOST(unminified.url) IN (
+        SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+      ) AS is_3p,
+      unminified.wastedBytes AS wasted_bytes
+    FROM
+      `httparchive.lighthouse.2021_07_01_mobile` AS test,
+      UNNEST(getUnminifiedJsUrls(JSON_EXTRACT(report, "$.audits['unminified-javascript']"))) AS unminified)
+  GROUP BY
+    page)
diff --git a/sql/2021/javascript/lighthouse_unminified_js_bytes.sql b/sql/2021/javascript/lighthouse_unminified_js_bytes.sql
new file mode 100644
index 00000000000..923cd16e87c
--- /dev/null
+++ b/sql/2021/javascript/lighthouse_unminified_js_bytes.sql
@@ -0,0 +1,31 @@
+#standardSQL
+# Histogram of wasted bytes per page
+CREATE TEMPORARY FUNCTION getUnminifiedJsBytes(audit STRING)
+RETURNS ARRAY<INT64> LANGUAGE js AS '''
+try {
+  var $ = JSON.parse(audit);
+  return $.details.items.map(({wastedBytes}) => wastedBytes);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  IF(unminified_js_kbytes <= 200, CEIL(unminified_js_kbytes / 10) * 10, 200) AS unminified_js_kbytes,
+  COUNT(0) AS pages,
+  SUM(COUNT(0)) OVER () AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER () AS pct
+FROM (
+  SELECT
+    test.url AS page,
+    SUM(IFNULL(unminified_js_bytes, 0)) / 1024 AS unminified_js_kbytes
+  FROM
+    `httparchive.lighthouse.2021_07_01_mobile` AS test
+  LEFT JOIN
+    UNNEST(getUnminifiedJsBytes(JSON_EXTRACT(report, "$.audits['unminified-javascript']"))) AS unminified_js_bytes
+  GROUP BY
+    page)
+GROUP BY
+  unminified_js_kbytes
+ORDER BY
+  unminified_js_kbytes
diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index b2649e99ce5..a695d45967c 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -1,9 +1,14 @@
 SELECT
-  resp_content_type,
-  COUNT(0) AS count
-FROM
-  `httparchive.sample_data.summary_requests_*`
+    client,
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%image%') as images,
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%font%') as font, 
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%css%') as css, 
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%javascript%') as javascript, 
+    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%json%') as json,
+    count(*) as total
+  FROM
+    `httparchive.almanac.requests`
 GROUP BY
-  1
+  client
 ORDER BY
-  2 DESC
+  client
\ No newline at end of file
diff --git a/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql b/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
index beb5ca5a2b2..db4e8c65adc 100644
--- a/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
+++ b/sql/2021/javascript/percentage_usage_of_different_ajax_apis.sql
@@ -20,7 +20,7 @@ FROM (
           '$._javascript'),
         '$.ajax_requests.beacon') AS beacon_total
     FROM
-      `httparchive.sample_data.pages_*`),
+      `httparchive.pages.2021_07_01_*`),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
diff --git a/sql/2021/javascript/requests_2019.sql b/sql/2021/javascript/requests_2019.sql
index 9d6feb06e9a..f9947aece20 100644
--- a/sql/2021/javascript/requests_2019.sql
+++ b/sql/2021/javascript/requests_2019.sql
@@ -5,8 +5,7 @@ SELECT
   _TABLE_SUFFIX AS client,
   APPROX_QUANTILES(reqJS, 1000)[OFFSET(percentile * 10)] AS js_requests
 FROM
-  `httparchive.summary_pages.2019_08_01_*`,
-  -- `httparchive.latest.summary_pages_*`
+  `httparchive.summary_pages.2019_07_01_*`,
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
diff --git a/sql/2021/javascript/requests_2021.sql b/sql/2021/javascript/requests_2021.sql
index ea9da50d48e..fef2951e58d 100644
--- a/sql/2021/javascript/requests_2021.sql
+++ b/sql/2021/javascript/requests_2021.sql
@@ -5,8 +5,7 @@ SELECT
   _TABLE_SUFFIX AS client,
   APPROX_QUANTILES(reqJS, 1000)[OFFSET(percentile * 10)] AS js_requests
 FROM
-  `httparchive.latest.summary_pages_*`,
-  -- `httparchive.sample_data.summary_pages_*`
+  `httparchive.summary_pages.2021_07_01_*`,
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
diff --git a/sql/2021/javascript/resource-hints-prefetch-preload-percentage.sql b/sql/2021/javascript/resource-hints-prefetch-preload-percentage.sql
new file mode 100644
index 00000000000..ca9c29468e5
--- /dev/null
+++ b/sql/2021/javascript/resource-hints-prefetch-preload-percentage.sql
@@ -0,0 +1,49 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getResourceHintAttrs(payload STRING)
+RETURNS ARRAY<STRUCT<name STRING, attribute STRING, value STRING>>
+LANGUAGE js AS '''
+var hints = new Set(['preload', 'prefetch']);
+var attributes = ['as'];
+try {
+  var $ = JSON.parse(payload);
+  var almanac = JSON.parse($._almanac);
+  return almanac['link-nodes'].nodes.reduce((results, link) => {
+    var hint = link.rel.toLowerCase();
+    if (!hints.has(hint)) {
+      return results;
+    }
+    attributes.forEach(attribute => {
+      var value = link[attribute];
+      results.push({
+        name: hint,
+        attribute: attribute,
+        // Support empty strings.
+        value: typeof value == 'string' ? value : null
+      });
+    });
+    return results;
+  }, []);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  client,
+  COUNT(DISTINCT IF(prefetch_hint, page, NULL)) AS prefetch_pages,
+  COUNT(DISTINCT page) AS total,
+  COUNT(DISTINCT IF(prefetch_hint, page, NULL)) / COUNT(DISTINCT page) AS prefetch_pct,
+  COUNT(DISTINCT IF(preload_hint, page, NULL)) AS preload_pages,
+  COUNT(DISTINCT IF(preload_hint, page, NULL)) / COUNT(DISTINCT page) AS preload_pct
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    hint.name = 'prefetch' AND hint.value = 'script' AS prefetch_hint,
+    hint.name = 'preload' AND hint.value = 'script' AS preload_hint
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(getResourceHintAttrs(payload)) AS hint)
+GROUP BY
+  client
diff --git a/sql/2021/javascript/resource-hints-preload-prefetch-distribution.sql b/sql/2021/javascript/resource-hints-preload-prefetch-distribution.sql
new file mode 100644
index 00000000000..c3ba28337da
--- /dev/null
+++ b/sql/2021/javascript/resource-hints-preload-prefetch-distribution.sql
@@ -0,0 +1,57 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getResourceHintAttrs(payload STRING)
+RETURNS ARRAY<STRUCT<name STRING, attribute STRING, value STRING>>
+LANGUAGE js AS '''
+var hints = new Set(['preload', 'prefetch']);
+var attributes = ['as'];
+try {
+  var $ = JSON.parse(payload);
+  var almanac = JSON.parse($._almanac);
+  return almanac['link-nodes'].nodes.reduce((results, link) => {
+    var hint = link.rel.toLowerCase();
+    if (!hints.has(hint)) {
+      return results;
+    }
+    attributes.forEach(attribute => {
+      var value = link[attribute];
+      results.push({
+        name: hint,
+        attribute: attribute,
+        // Support empty strings.
+        value: typeof value == 'string' ? value : null
+      });
+    });
+    return results;
+  }, []);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  percentile,
+  client,
+  APPROX_QUANTILES(prefetch_hint, 1000)[OFFSET(percentile * 10)] AS prefetch_hints_per_page,
+  APPROX_QUANTILES(IF(prefetch_hint = 0, NULL, prefetch_hint), 1000 IGNORE NULLS)[OFFSET(percentile * 10)] AS prefetch_hints_per_page_with_hints,
+  APPROX_QUANTILES(preload_hint, 1000)[OFFSET(percentile * 10)] AS preload_hints_per_page,
+  APPROX_QUANTILES(IF(preload_hint = 0, NULL, preload_hint), 1000 IGNORE NULLS)[OFFSET(percentile * 10)] AS preload_hints_per_page_with_hints
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    COUNTIF(hint.name = 'prefetch' AND hint.value = 'script') AS prefetch_hint,
+    COUNTIF(hint.name = 'preload' AND hint.value = 'script') AS preload_hint
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(getResourceHintAttrs(payload)) AS hint
+  GROUP BY
+    client,
+    page),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/resource_hints.sql b/sql/2021/javascript/resource_hints.sql
new file mode 100644
index 00000000000..10488f3d410
--- /dev/null
+++ b/sql/2021/javascript/resource_hints.sql
@@ -0,0 +1,46 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getResourceHintAttrs(payload STRING)
+RETURNS ARRAY<STRUCT<name STRING, attribute STRING, value STRING>>
+LANGUAGE js AS '''
+var hints = new Set(['preload', 'prefetch']);
+var attributes = ['as'];
+try {
+  var $ = JSON.parse(payload);
+  var almanac = JSON.parse($._almanac);
+  return almanac['link-nodes'].nodes.reduce((results, link) => {
+    var hint = link.rel.toLowerCase();
+    if (!hints.has(hint)) {
+      return results;
+    }
+    attributes.forEach(attribute => {
+      var value = link[attribute];
+      results.push({
+        name: hint,
+        attribute: attribute,
+        // Support empty strings.
+        value: typeof value == 'string' ? value : null
+      });
+    });
+    return results;
+  }, []);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  client,
+  COUNT(DISTINCT IF(script_hint, page, NULL)) AS pages,
+  COUNT(DISTINCT page) AS total,
+  COUNT(DISTINCT IF(script_hint, page, NULL)) / COUNT(DISTINCT page) AS pct
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    hint.name IN ('prefetch', 'preload') AND hint.value = 'script' AS script_hint
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(getResourceHintAttrs(payload)) AS hint)
+GROUP BY
+  client
diff --git a/sql/2021/javascript/resource_hints_per_page.sql b/sql/2021/javascript/resource_hints_per_page.sql
new file mode 100644
index 00000000000..6fa1cec60cf
--- /dev/null
+++ b/sql/2021/javascript/resource_hints_per_page.sql
@@ -0,0 +1,54 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getResourceHintAttrs(payload STRING)
+RETURNS ARRAY<STRUCT<name STRING, attribute STRING, value STRING>>
+LANGUAGE js AS '''
+var hints = new Set(['preload', 'prefetch']);
+var attributes = ['as'];
+try {
+  var $ = JSON.parse(payload);
+  var almanac = JSON.parse($._almanac);
+  return almanac['link-nodes'].nodes.reduce((results, link) => {
+    var hint = link.rel.toLowerCase();
+    if (!hints.has(hint)) {
+      return results;
+    }
+    attributes.forEach(attribute => {
+      var value = link[attribute];
+      results.push({
+        name: hint,
+        attribute: attribute,
+        // Support empty strings.
+        value: typeof value == 'string' ? value : null
+      });
+    });
+    return results;
+  }, []);
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  percentile,
+  client,
+  APPROX_QUANTILES(script_hint, 1000)[OFFSET(percentile * 10)] AS hints_per_page,
+  APPROX_QUANTILES(IF(script_hint = 0, NULL, script_hint), 1000 IGNORE NULLS)[OFFSET(percentile * 10)] AS hints_per_page_with_hints
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    COUNTIF(hint.name IN ('prefetch', 'preload') AND hint.value = 'script') AS script_hint
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  LEFT JOIN
+    UNNEST(getResourceHintAttrs(payload)) AS hint
+  GROUP BY
+    client,
+    page),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile,
+  client
+ORDER BY
+  percentile,
+  client
diff --git a/sql/2021/javascript/sourcemap_header.sql b/sql/2021/javascript/sourcemap_header.sql
new file mode 100644
index 00000000000..2788df59861
--- /dev/null
+++ b/sql/2021/javascript/sourcemap_header.sql
@@ -0,0 +1,48 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  has_sourcemap_header,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total_js_requests,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct_js_requests
+FROM (
+  SELECT
+    client,
+    page,
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'SourceMap') IS NOT NULL AS has_sourcemap_header
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script')
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  has_sourcemap_header
+ORDER BY
+  client,
+  has_sourcemap_header
diff --git a/sql/2021/javascript/sourcemap_header_by_3p.sql b/sql/2021/javascript/sourcemap_header_by_3p.sql
new file mode 100644
index 00000000000..c863a5371a1
--- /dev/null
+++ b/sql/2021/javascript/sourcemap_header_by_3p.sql
@@ -0,0 +1,54 @@
+#standardSQL
+CREATE TEMPORARY FUNCTION getHeader(headers STRING, headername STRING)
+RETURNS STRING
+DETERMINISTIC
+LANGUAGE js AS '''
+  const parsed_headers = JSON.parse(headers);
+  const matching_headers = parsed_headers.filter(h => h.name.toLowerCase() == headername.toLowerCase());
+  if (matching_headers.length > 0) {
+    return matching_headers[0].value;
+  }
+  return null;
+''';
+
+SELECT
+  client,
+  host,
+  has_sourcemap_header,
+  COUNT(DISTINCT page) AS pages,
+  ANY_VALUE(total_pages) AS total_pages,
+  COUNT(DISTINCT page) / ANY_VALUE(total_pages) AS pct_pages,
+  COUNT(0) AS js_requests,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total_js_requests,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct_js_requests
+FROM (
+  SELECT
+    client,
+    page,
+    IF(NET.HOST(url) IN (
+      SELECT domain FROM `httparchive.almanac.third_parties` WHERE date = '2021-07-01' AND category != 'hosting'
+    ), 'third party', 'first party') AS host,
+    getHeader(JSON_EXTRACT(payload, '$.response.headers'), 'SourceMap') IS NOT NULL AS has_sourcemap_header
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2021-07-01' AND
+    type = 'script')
+JOIN (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    COUNT(0) AS total_pages
+  FROM
+    `httparchive.summary_pages.2021_07_01_*`
+  GROUP BY
+    client)
+USING
+  (client)
+GROUP BY
+  client,
+  host,
+  has_sourcemap_header
+ORDER BY
+  client,
+  host,
+  has_sourcemap_header
diff --git a/sql/2021/javascript/unused_js_bytes_distribution.sql b/sql/2021/javascript/unused_js_bytes_distribution.sql
new file mode 100644
index 00000000000..62872729613
--- /dev/null
+++ b/sql/2021/javascript/unused_js_bytes_distribution.sql
@@ -0,0 +1,12 @@
+#standardSQL
+# Distribution of unused JS request bytes per page
+SELECT
+  percentile,
+  APPROX_QUANTILES(CAST(JSON_EXTRACT_SCALAR(report, "$.audits.unused-javascript.details.overallSavingsBytes") AS INT64) / 1024, 1000)[OFFSET(percentile * 10)] AS js_kilobytes
+FROM
+  `httparchive.lighthouse.2021_07_01_mobile`,
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+GROUP BY
+  percentile
+ORDER BY
+  percentile
diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
index db9cf094386..a4463e2c5c7 100644
--- a/sql/2021/javascript/web_components_specs.sql
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -10,21 +10,30 @@ SELECT
 FROM (
   SELECT
     _TABLE_SUFFIX AS client,
-    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+    (CASE 
+     WHEN ARRAY_LENGTH(JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
-        '$.web_component_specs.custom_elements') AS INT64) AS custom_elements,
-    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$.web_component_specs.custom_elements'))) > 1 THEN 1
+    ELSE 0 
+    END) AS custom_elements,
+     (CASE 
+     WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
-        '$.web_component_specs.shadow_roots') AS INT64) AS shadow_roots,
-    CAST(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$.web_component_specs.shadow_roots'))) > 1 THEN 1
+    ELSE 0 
+    END) AS shadow_roots,
+    (CASE 
+     WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
           '$._javascript'),
-        '$.web_component_specs.template') AS INT64) AS template
+        '$.web_component_specs.template'))) > 1 THEN 1
+    ELSE 0 
+    END) AS template
   FROM
-    `httparchive.sample_data.pages_*` ),
+    `httparchive.pages.2021_09_01_*` ),
   UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
   percentile,
   client
 ORDER BY
   percentile,
-  client
+  client
\ No newline at end of file
diff --git a/sql/2021/pwa/README.md b/sql/2021/pwa/README.md
index 134094ad1aa..669d9ba6664 100644
--- a/sql/2021/pwa/README.md
+++ b/sql/2021/pwa/README.md
@@ -8,3 +8,11 @@
 
   Analysts: if helpful, you can use this README to give additional info about the queries.
 -->
+
+Resources:
+
+- [Chapter issue](https://github.com/HTTPArchive/almanac.httparchive.org/issues/2153)
+- [Planning doc](https://docs.google.com/document/d/1grNpHfQgBZow7fTw9UNZk8E4eWdQV97ZyenPmaye1JU/)
+- [Results sheet](https://docs.google.com/spreadsheets/d/16AkIdDBBkCR5Kgb7kyfYvnNLQBu23Vsh7MUSFHW9RtA/)
+- [2019 chapter](https://almanac.httparchive.org/en/2019/pwa)
+- [2020 chapter](https://almanac.httparchive.org/en/2020/pwa)
diff --git a/sql/2021/pwa/assetlink_usage.sql b/sql/2021/pwa/assetlink_usage.sql
new file mode 100644
index 00000000000..784ece26f23
--- /dev/null
+++ b/sql/2021/pwa/assetlink_usage.sql
@@ -0,0 +1,61 @@
+#standardSQL
+# assetlink usage
+
+SELECT
+  'PWA sites' AS type,
+  _TABLE_SUFFIX AS client,
+  COUNT(0) AS freq,
+  total,
+  COUNT(0) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.manifests') != "[]" AND
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true" AND
+  JSON_EXTRACT(payload, '$._pwa.manifests') != "[]" AND
+  JSON_EXTRACT_SCALAR(JSON_VALUE(payload, "$._well-known"), "$['/.well-known/assetlinks.json'].found") = 'true'
+GROUP BY
+  client,
+  total
+UNION ALL
+SELECT
+  'All sites' AS type,
+  _TABLE_SUFFIX AS client,
+  COUNT(0) AS freq,
+  total,
+  COUNT(0) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT_SCALAR(JSON_VALUE(payload, "$._well-known"), "$['/.well-known/assetlinks.json'].found") = 'true'
+GROUP BY
+  client,
+  total
+ORDER BY
+  type DESC,
+  freq / total DESC,
+  client
diff --git a/sql/2021/pwa/install_events.sql b/sql/2021/pwa/install_events.sql
new file mode 100644
index 00000000000..edd578ceff4
--- /dev/null
+++ b/sql/2021/pwa/install_events.sql
@@ -0,0 +1,69 @@
+#standardSQL
+# SW install events
+
+CREATE TEMPORARY FUNCTION getInstallEvents(payload STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var payloadJSON = JSON.parse(payload);
+
+  /* YouTube iFrames account for a lot of these, so we exclude them */
+  /* Cannot use filter as it is a complex object and not a straight array */
+  function filterYouTube(info) {
+
+    var objectKeys = Object.keys(info);
+    objectKeys = objectKeys.trim().split(',');
+    for(var i = 0; i < objectKeys.length; i++) {
+        if(objectKeys[i].toLowerCase().includes('youtube')) {
+            delete info[objectKeys[i]];
+        }
+    }
+    return info;
+  }
+
+  var windowEventListenersInfo = Object.values(filterYouTube(payloadJSON.windowEventListenersInfo)).flat();
+  var windowPropertiesInfo = Object.values(filterYouTube(payloadJSON.windowPropertiesInfo)).flat()
+
+  return [...new Set([...windowEventListenersInfo ,...windowPropertiesInfo])];
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  install_event,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getInstallEvents(JSON_EXTRACT(payload, '$._pwa'))) AS install_event
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      -- This condition filters out tests that might have broken when running the 'pwa' metric
+      -- as even pages without any pwa capabilities will have a _pwa object with empty fields
+      JSON_EXTRACT(payload, '$._pwa') != "[]"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  (
+    JSON_EXTRACT(payload, '$._pwa.windowEventListenersInfo') != "[]" OR
+    JSON_EXTRACT(payload, '$._pwa.windowPropertiesInfo') != "[]"
+  ) AND
+  install_event != '' AND
+  install_event != '[]'
+GROUP BY
+  client,
+  total,
+  install_event
+ORDER BY
+  freq / total DESC,
+  client
diff --git a/sql/2021/pwa/lighthouse_pwa_audits.sql b/sql/2021/pwa/lighthouse_pwa_audits.sql
new file mode 100644
index 00000000000..f68354926fc
--- /dev/null
+++ b/sql/2021/pwa/lighthouse_pwa_audits.sql
@@ -0,0 +1,67 @@
+#standardSQL
+# Get summary of all lighthouse PWA audits, for both PWA pages and all pages
+# Note scores, weightings, groups and descriptions may be off in mixed months when new versions of Lighthouse roles out
+
+CREATE TEMPORARY FUNCTION getAudits(auditRefs STRING, audits STRING)
+RETURNS ARRAY<STRUCT<id STRING, weight INT64, audit_group STRING, title STRING, description STRING, score INT64>> LANGUAGE js AS '''
+var auditrefs = JSON.parse(auditRefs);
+var audits = JSON.parse(audits);
+var results = [];
+for (auditref of auditrefs) {
+  results.push({
+    id: auditref.id,
+    weight: auditref.weight,
+    audit_group: auditref.group,
+    description: audits[auditref.id].description,
+    score: audits[auditref.id].score
+  });
+}
+return results;
+''';
+
+SELECT
+  'PWA Sites' AS type,
+  audits.id AS id,
+  COUNTIF(audits.score > 0) AS num_pages,
+  COUNT(0) AS total,
+  COUNTIF(audits.score IS NOT NULL) AS total_applicable,
+  SAFE_DIVIDE(COUNTIF(audits.score > 0), COUNTIF(audits.score IS NOT NULL)) AS pct,
+  APPROX_QUANTILES(audits.weight, 100)[OFFSET(50)] AS median_weight,
+  MAX(audits.audit_group) AS audit_group,
+  MAX(audits.description) AS description
+FROM
+  `httparchive.lighthouse.2021_07_01_mobile`,
+  UNNEST(getAudits(JSON_EXTRACT(report, '$.categories.pwa.auditRefs'), JSON_EXTRACT(report, '$.audits'))) AS audits
+JOIN
+  (
+    SELECT
+      url
+    FROM
+      `httparchive.pages.2021_07_01_mobile`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true" AND
+      JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+  )
+USING (url)
+GROUP BY
+  audits.id
+UNION ALL
+SELECT
+  'ALL Sites' AS type,
+  audits.id AS id,
+  COUNTIF(audits.score > 0) AS num_pages,
+  COUNT(0) AS total,
+  COUNTIF(audits.score IS NOT NULL) AS total_applicable,
+  SAFE_DIVIDE(COUNTIF(audits.score > 0), COUNTIF(audits.score IS NOT NULL)) AS pct,
+  APPROX_QUANTILES(audits.weight, 100)[OFFSET(50)] AS median_weight,
+  MAX(audits.audit_group) AS audit_group,
+  MAX(audits.description) AS description
+FROM
+  `httparchive.lighthouse.2021_07_01_mobile`,
+  UNNEST(getAudits(JSON_EXTRACT(report, '$.categories.pwa.auditRefs'), JSON_EXTRACT(report, '$.audits'))) AS audits
+GROUP BY
+  audits.id
+ORDER BY
+  type DESC,
+  median_weight DESC,
+  id
diff --git a/sql/2021/pwa/lighthouse_pwa_score.sql b/sql/2021/pwa/lighthouse_pwa_score.sql
new file mode 100644
index 00000000000..a57997e05d5
--- /dev/null
+++ b/sql/2021/pwa/lighthouse_pwa_score.sql
@@ -0,0 +1,48 @@
+#standardSQL
+# Percentiles of lighthouse pwa score
+# This metric comes from Lighthouse only and is only available in mobile in HTTP Archive dataset
+SELECT
+  '2021_07_01' AS date,
+  'PWA Sites' AS type,
+  percentile,
+  APPROX_QUANTILES(score, 1000)[OFFSET(percentile * 10)] * 100 AS score
+FROM (
+  SELECT
+    url,
+    CAST(JSON_EXTRACT(report, '$.categories.pwa.score') AS NUMERIC) AS score
+  FROM
+    `httparchive.lighthouse.2021_07_01_mobile`)
+JOIN
+  (
+    SELECT
+      url
+    FROM
+      `httparchive.pages.2021_07_01_mobile`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true" AND
+      JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+  )
+USING (url),
+  UNNEST([10, 25, 50, 75, 90]) AS percentile
+GROUP BY
+  date,
+  percentile
+UNION ALL
+SELECT
+  '2021_07_01' AS date,
+  'All Sites' AS type,
+  percentile,
+  APPROX_QUANTILES(score, 1000)[OFFSET(percentile * 10)] * 100 AS score
+FROM (
+  SELECT
+    CAST(JSON_EXTRACT(report, '$.categories.pwa.score') AS NUMERIC) AS score
+  FROM
+    `httparchive.lighthouse.2021_07_01_mobile`),
+  UNNEST([10, 25, 50, 75, 90]) AS percentile
+GROUP BY
+  date,
+  percentile
+ORDER BY
+  date,
+  type DESC,
+  percentile
diff --git a/sql/2021/pwa/manifests_and_service_workers.sql b/sql/2021/pwa/manifests_and_service_workers.sql
new file mode 100644
index 00000000000..afd882eaf70
--- /dev/null
+++ b/sql/2021/pwa/manifests_and_service_workers.sql
@@ -0,0 +1,21 @@
+#standardSQL
+# Counting Manifests and Service Workers
+SELECT
+  client,
+  SAFE_DIVIDE(SUM(ServiceWorker), SUM(COUNT(0)) OVER (PARTITION BY client)) AS ServiceWorkers,
+  SAFE_DIVIDE(SUM(manifests), SUM(COUNT(0)) OVER (PARTITION BY client)) AS Manifests,
+  SAFE_DIVIDE(COUNTIF(ServiceWorker > 0 OR manifests > 0), SUM(COUNT(0)) OVER (PARTITION BY client)) AS Either,
+  SAFE_DIVIDE(COUNTIF(ServiceWorker > 0 AND manifests > 0), SUM(COUNT(0)) OVER (PARTITION BY client)) AS Both,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    IF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true", 1, 0) AS ServiceWorker,
+    IF(JSON_EXTRACT(payload, '$._pwa.manifests') != "[]", 1, 0) AS manifests
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  )
+GROUP BY
+  client
+ORDER BY
+  client
diff --git a/sql/2021/pwa/manifests_not_json_parsable.sql b/sql/2021/pwa/manifests_not_json_parsable.sql
new file mode 100644
index 00000000000..3e28d0c1dc7
--- /dev/null
+++ b/sql/2021/pwa/manifests_not_json_parsable.sql
@@ -0,0 +1,52 @@
+#standardSQL
+# Manifests that are not JSON parsable for service worker pages and all pages
+
+CREATE TEMP FUNCTION canParseManifest(manifest STRING)
+RETURNS BOOLEAN LANGUAGE js AS '''
+try {
+  var manifestJSON = Object.values(JSON.parse(manifest))[0];
+  if (typeof manifestJSON === 'string' && manifestJSON.trim() != '') {
+    return false;
+  }
+  if (typeof manifestJSON === 'string' && manifestJSON.trim() === '') {
+    return null;
+  }
+  return true;
+} catch {
+  return false;
+}
+''';
+
+SELECT
+  'PWA Pages' AS type,
+  _TABLE_SUFFIX AS client,
+  canParseManifest(JSON_EXTRACT(payload, '$._pwa.manifests')) AS can_parse,
+  COUNT(0) AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.manifests') != "[]" AND
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+GROUP BY
+  client,
+  can_parse
+UNION ALL
+SELECT
+  'All Pages' AS type,
+  _TABLE_SUFFIX AS client,
+  canParseManifest(JSON_EXTRACT(payload, '$._pwa.manifests')) AS can_parse,
+  COUNT(0) AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+GROUP BY
+  client,
+  can_parse
+ORDER BY
+  type DESC,
+  freq / total DESC,
+  can_parse,
+  client
diff --git a/sql/2021/pwa/manifests_preferring_native_apps.sql b/sql/2021/pwa/manifests_preferring_native_apps.sql
new file mode 100644
index 00000000000..5b9ba1eba4b
--- /dev/null
+++ b/sql/2021/pwa/manifests_preferring_native_apps.sql
@@ -0,0 +1,46 @@
+#standardSQL
+# % manifests preferring native apps for service worker pages and all pages
+
+CREATE TEMP FUNCTION prefersNative(manifest STRING)
+RETURNS BOOLEAN LANGUAGE js AS '''
+try {
+  var $ = Object.values(JSON.parse(manifest))[0];
+  return $.prefer_related_applications == true && $.related_applications.length > 0;
+} catch (e) {
+  return null;
+}
+''';
+
+SELECT
+  'PWA Pages' AS type,
+  _TABLE_SUFFIX AS client,
+  prefersNative(JSON_EXTRACT(payload, '$._pwa.manifests')) AS prefersNative,
+  COUNT(0) AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.manifests') != "[]" AND
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+GROUP BY
+  client,
+  prefersNative
+UNION ALL
+SELECT
+  'All Pages' AS type,
+  _TABLE_SUFFIX AS client,
+  prefersNative(JSON_EXTRACT(payload, '$._pwa.manifests')) AS prefersNative,
+  COUNT(0) AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+GROUP BY
+  client,
+  prefersNative
+ORDER BY
+  type DESC,
+  freq / total DESC,
+  prefersNative,
+  client
diff --git a/sql/2021/pwa/popular_pwa_libraries.sql b/sql/2021/pwa/popular_pwa_libraries.sql
new file mode 100644
index 00000000000..d43c92ae920
--- /dev/null
+++ b/sql/2021/pwa/popular_pwa_libraries.sql
@@ -0,0 +1,100 @@
+#standardSQL
+# Popular PWA script
+CREATE TEMPORARY FUNCTION getSWLibraries(importScriptsInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  /* 'importScriptsInfo' returns an array of script that might import other script
+     The final array of script comes from the combination of both */
+  var ObjKeys = Object.keys(JSON.parse(importScriptsInfo));
+  var ObjValues = Object.values(JSON.parse(importScriptsInfo));
+  var script = ObjKeys.concat(ObjValues);
+  /* Replacing spaces and commas */
+  for (var i = 0; i < script.length; i++) {
+      script[i] = script[i].toString().trim().replace(/'/g, "");
+  }
+
+  /* Creating a Set to eliminate duplicates and transforming back to an array to respect the function signature */
+  return Array.from(new Set(script));
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  client,
+  COUNTIF(workbox > 0) / total AS workbox,
+  COUNTIF(sw_toolbox > 0) / total AS sw_toolbox,
+  COUNTIF(firebase > 0) / total AS firebase,
+  COUNTIF(OneSignalSDK > 0) / total AS OneSignalSDK,
+  COUNTIF(najva > 0) / total AS najva,
+  COUNTIF(upush > 0) / total AS upush,
+  COUNTIF(cache_polyfill > 0) / total AS cache_polyfill,
+  COUNTIF(analytics_helper > 0) / total AS analytics_helper,
+  COUNTIF(recaptcha > 0) / total AS recaptcha,
+  COUNTIF(pwabuilder > 0) / total AS pwabuilder,
+  COUNTIF(pushprofit > 0) / total AS pushprofit,
+  COUNTIF(sendpulse > 0) / total AS sendpulse,
+  COUNTIF(quora > 0) / total AS quora,
+  COUNTIF(none_of_the_above > 0) / total AS none_of_the_above,
+  COUNTIF(importscripts > 0) / total AS uses_importscript,
+  total
+FROM
+  (
+    SELECT
+      _TABLE_SUFFIX AS client,
+      url,
+      COUNT(0) AS importscripts,
+      COUNTIF(LOWER(script) LIKE '%workbox%') AS workbox,
+      COUNTIF(LOWER(script) LIKE '%sw-toolbox%') AS sw_toolbox,
+      COUNTIF(LOWER(script) LIKE '%firebase%') AS firebase,
+      COUNTIF(LOWER(script) LIKE '%onesignalsdk%') AS OneSignalSDK,
+      COUNTIF(LOWER(script) LIKE '%najva%') AS najva,
+      COUNTIF(LOWER(script) LIKE '%upush%') AS upush,
+      COUNTIF(LOWER(script) LIKE '%cache-polyfill%') AS cache_polyfill,
+      COUNTIF(LOWER(script) LIKE '%analytics-helper%') AS analytics_helper,
+      COUNTIF(LOWER(script) LIKE '%recaptcha%') AS recaptcha,
+      COUNTIF(LOWER(script) LIKE '%pwabuilder%') AS pwabuilder,
+      COUNTIF(LOWER(script) LIKE '%pushprofit%') AS pushprofit,
+      COUNTIF(LOWER(script) LIKE '%sendpulse%') AS sendpulse,
+      COUNTIF(LOWER(script) LIKE '%quore%') AS quora,
+      COUNTIF(LOWER(script) NOT LIKE '%workbox%' AND
+        LOWER(script) NOT LIKE '%sw-toolbox%' AND
+        LOWER(script) NOT LIKE '%firebase%' AND
+        LOWER(script) NOT LIKE '%onesignalsdk%' AND
+        LOWER(script) NOT LIKE '%najva%' AND
+        LOWER(script) NOT LIKE '%upush%' AND
+        LOWER(script) NOT LIKE '%cache-polyfill.js%' AND
+        LOWER(script) NOT LIKE '%analytics-helper.js%' AND
+        LOWER(script) NOT LIKE '%recaptcha%' AND
+        LOWER(script) NOT LIKE '%pwabuilder%' AND
+        LOWER(script) NOT LIKE '%pushprofit%' AND
+        LOWER(script) NOT LIKE '%sendpulse%' AND
+        LOWER(script) NOT LIKE '%quora%') AS none_of_the_above
+    FROM
+      `httparchive.pages.2021_07_01_*`,
+      UNNEST(getSWLibraries(JSON_EXTRACT(payload, '$._pwa.importScriptsInfo'))) AS script
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.importScriptsInfo') != "[]" AND
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX,
+      url
+  )
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX AS client,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      client
+  )
+USING (client)
+GROUP BY
+  client,
+  total
+ORDER BY
+  client
diff --git a/sql/2021/pwa/popular_pwa_libraries_helper.sql b/sql/2021/pwa/popular_pwa_libraries_helper.sql
new file mode 100644
index 00000000000..f6a18a66cb4
--- /dev/null
+++ b/sql/2021/pwa/popular_pwa_libraries_helper.sql
@@ -0,0 +1,48 @@
+#standardSQL
+# Use this sql to find popular library imports for popular_pwa_libraries.sql
+# And also other importscripts used in service workers
+CREATE TEMPORARY FUNCTION getSWLibraries(importScriptsInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  /* 'importScriptsInfo' returns an array of libraries that might import other libraries
+     The final array of libraries comes from the combination of both */
+  var ObjKeys = Object.keys(JSON.parse(importScriptsInfo));
+  var ObjValues = Object.values(JSON.parse(importScriptsInfo));
+  var libraries = ObjKeys.concat(ObjValues);
+  /* Replacing spaces and commas */
+  for (var i = 0; i < libraries.length; i++) {
+      libraries[i] = libraries[i].toString().trim().replace(/'/g, "");
+  }
+
+  /* Creating a Set to eliminate duplicates and transforming back to an array to respect the function signature */
+  return Array.from(new Set(libraries));
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  script,
+  COUNT(DISTINCT url) AS freq
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getSWLibraries(JSON_EXTRACT(payload, '$._pwa.importScriptsInfo'))) AS script
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.importScriptsInfo') != "[]" AND
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true" AND
+  LOWER(script) NOT LIKE '%workbox%' AND
+  LOWER(script) NOT LIKE '%sw-toolbox%' AND
+  LOWER(script) NOT LIKE '%firebase%' AND
+  LOWER(script) NOT LIKE '%onesignalsdk%' AND
+  LOWER(script) NOT LIKE '%najva%' AND
+  LOWER(script) NOT LIKE '%upush%' AND
+  LOWER(script) NOT LIKE '%cache-polyfill.js%' AND
+  LOWER(script) NOT LIKE '%analytics-helper.js%' AND
+  LOWER(script) NOT LIKE '%recaptcha%' AND
+  LOWER(script) NOT LIKE '%pwabuilder%'
+GROUP BY
+  _TABLE_SUFFIX,
+  script
+ORDER BY
+  freq DESC
diff --git a/sql/2021/pwa/pwa_blink_usage.sql b/sql/2021/pwa/pwa_blink_usage.sql
new file mode 100644
index 00000000000..c6af5b77d77
--- /dev/null
+++ b/sql/2021/pwa/pwa_blink_usage.sql
@@ -0,0 +1,21 @@
+#standardSQL
+# PWA features tracked in blink_features.usage
+SELECT DISTINCT
+  client,
+  feature,
+  num_urls,
+  total_urls,
+  num_urls / total_urls AS pct
+FROM
+  `httparchive.blink_features.usage`
+WHERE
+  yyyymmdd = '20210701' AND
+  (
+    feature LIKE '%ServiceWorker%' OR
+    feature LIKE '%BackgroundSync%' OR
+    feature LIKE '%GetInstalledRelatedApps%'
+  )
+ORDER BY
+  num_urls DESC,
+  client,
+  feature
diff --git a/sql/2021/pwa/pwa_notification_acceptance_rates.sql b/sql/2021/pwa/pwa_notification_acceptance_rates.sql
new file mode 100644
index 00000000000..e570e2372b1
--- /dev/null
+++ b/sql/2021/pwa/pwa_notification_acceptance_rates.sql
@@ -0,0 +1,47 @@
+#standardSQL
+# 13_16b: Web Push Notification CRUX stats (10th / 25th / 50th / 75th / 90th / 100th percentile) for PWA origins
+# Based on https://github.com/HTTPArchive/almanac.httparchive.org/blob/main/sql/2020/ecommerce/webpushstats_ecommsites.sql
+
+SELECT
+  date,
+  client,
+  percentile,
+  APPROX_QUANTILES(notification_permission_accept, 1000 RESPECT NULLS)[OFFSET(percentile * 10)] AS notification_permission_accept,
+  APPROX_QUANTILES(notification_permission_deny, 1000 RESPECT NULLS)[OFFSET(percentile * 10)] AS notification_permission_deny,
+  APPROX_QUANTILES(notification_permission_ignore, 1000 RESPECT NULLS)[OFFSET(percentile * 10)] AS notification_permission_ignore,
+  APPROX_QUANTILES(notification_permission_dismiss, 1000 RESPECT NULLS)[OFFSET(percentile * 10)] AS notification_permission_dismiss
+FROM
+  `chrome-ux-report.materialized.metrics_summary`
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX AS client,
+      RTRIM(url, "/") AS origin,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX,
+      url
+  )
+USING
+  (origin),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+WHERE
+  date IN ('2021-07-01') AND
+  (
+    notification_permission_accept IS NOT NULL OR
+    notification_permission_deny IS NOT NULL OR
+    notification_permission_ignore IS NOT NULL OR
+    notification_permission_dismiss IS NOT NULL
+  )
+GROUP BY
+  date,
+  client,
+  percentile
+ORDER BY
+  date,
+  client,
+  percentile
diff --git a/sql/2021/pwa/sw_adoption_over_time.sql b/sql/2021/pwa/sw_adoption_over_time.sql
new file mode 100644
index 00000000000..a5279652502
--- /dev/null
+++ b/sql/2021/pwa/sw_adoption_over_time.sql
@@ -0,0 +1,15 @@
+#standardSQL
+# SW adoption over time
+SELECT DISTINCT
+  REGEXP_REPLACE(yyyymmdd, r'(\d{4})(\d{2})(\d{2})', r'\1-\2\-3') AS date,
+  client,
+  num_urls AS freq,
+  total_urls AS total,
+  pct_urls AS pct
+FROM
+  `httparchive.blink_features.usage`
+WHERE
+  feature = 'ServiceWorkerControlledPage'
+ORDER BY
+  date DESC,
+  client
diff --git a/sql/2021/pwa/sw_adoption_over_time_ranking.sql b/sql/2021/pwa/sw_adoption_over_time_ranking.sql
new file mode 100644
index 00000000000..7babb7ffc4c
--- /dev/null
+++ b/sql/2021/pwa/sw_adoption_over_time_ranking.sql
@@ -0,0 +1,57 @@
+#standardSQL
+# SW adoption over time, including ranking
+SELECT
+  REGEXP_REPLACE(yyyymmdd, r'(\d{4})(\d{2})(\d{2})', r'\1-\2\-3') AS date,
+  client,
+  rank_grouping,
+  CASE
+    WHEN rank_grouping = 10000000 THEN 'all'
+    ELSE FORMAT("%'d", rank_grouping)
+  END AS ranking,
+  COUNT(0) AS freq,
+  total,
+  COUNT(0) / total AS pct
+FROM
+  (
+    SELECT DISTINCT
+      FORMAT_DATE('%Y%m%d', yyyymmdd) AS yyyymmdd,
+      client,
+      url,
+      rank
+    FROM
+      `httparchive.blink_features.features`
+    WHERE
+      feature = 'ServiceWorkerControlledPage' AND
+      yyyymmdd >= '2021-05-01'
+  ),
+  UNNEST([1000, 10000, 100000, 1000000, 10000000]) AS rank_grouping
+JOIN (
+  SELECT
+    REPLACE(SUBSTR(_TABLE_SUFFIX, 0, 10), '_', '') AS yyyymmdd,
+    SUBSTR(_TABLE_SUFFIX, 12) AS client,
+    rank_grouping,
+    COUNT(0) AS total
+  FROM
+    `httparchive.summary_pages.*`,
+    UNNEST([1000, 10000, 100000, 1000000, 10000000]) AS rank_grouping
+  WHERE
+    _TABLE_SUFFIX > '2021_05_01' AND
+    rank <= rank_grouping
+  GROUP BY
+    yyyymmdd,
+    rank_grouping,
+    client
+)
+USING
+  (yyyymmdd, client, rank_grouping)
+WHERE
+  rank <= rank_grouping
+GROUP BY
+  yyyymmdd,
+  client,
+  total,
+  rank_grouping
+ORDER BY
+  date DESC,
+  rank_grouping,
+  client
diff --git a/sql/2021/pwa/sw_events.sql b/sql/2021/pwa/sw_events.sql
new file mode 100644
index 00000000000..d691d8aedab
--- /dev/null
+++ b/sql/2021/pwa/sw_events.sql
@@ -0,0 +1,46 @@
+#standardSQL
+# SW events
+
+CREATE TEMPORARY FUNCTION getSWEvents(payload STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var payloadJSON = JSON.parse(payload);
+  var swEventListenersInfo = (Object.values(payloadJSON.swEventListenersInfo)).flat();
+  var swPropertiesInfo = (Object.values(payloadJSON.swPropertiesInfo)).flat();
+  return [...new Set([...swEventListenersInfo ,...swPropertiesInfo])];
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  event,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getSWEvents(JSON_EXTRACT(payload, '$._pwa'))) AS event
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+GROUP BY
+  client,
+  total,
+  event
+ORDER BY
+  freq / total DESC,
+  client
diff --git a/sql/2021/pwa/sw_methods.sql b/sql/2021/pwa/sw_methods.sql
new file mode 100644
index 00000000000..bd17c522a62
--- /dev/null
+++ b/sql/2021/pwa/sw_methods.sql
@@ -0,0 +1,44 @@
+#standardSQL
+# SW methods
+CREATE TEMPORARY FUNCTION getSWMethods(swMethodsInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var swMethods = JSON.parse(swMethodsInfo);
+  return Array.from(new Set(Object.values(swMethods).flat()));
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  sw_method,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getSWMethods(JSON_EXTRACT(payload, '$._pwa.swMethodsInfo'))) AS sw_method
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true" AND
+  JSON_EXTRACT(payload, '$._pwa.swMethodsInfo') != "[]"
+GROUP BY
+  client,
+  total,
+  sw_method
+ORDER BY
+  freq / total DESC,
+  client
diff --git a/sql/2021/pwa/sw_objects.sql b/sql/2021/pwa/sw_objects.sql
new file mode 100644
index 00000000000..0bedc3fa969
--- /dev/null
+++ b/sql/2021/pwa/sw_objects.sql
@@ -0,0 +1,47 @@
+#standardSQL
+# SW objects
+CREATE TEMPORARY FUNCTION getSWObjects(swObjectsInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var swObjects = Object.values(JSON.parse(swObjectsInfo));
+  if (typeof swObjects != 'string') {
+    swObjects = swObjects.toString();
+  }
+  swObjects = swObjects.trim().split(',');
+  return Array.from(new Set(swObjects));
+} catch (e) {
+  return [];
+}
+''';
+SELECT
+  _TABLE_SUFFIX AS client,
+  sw_object,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getSWObjects(JSON_EXTRACT(payload, '$._pwa.swObjectsInfo'))) AS sw_object
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true" AND
+  JSON_EXTRACT(payload, '$._pwa.swObjectsInfo') != "[]"
+GROUP BY
+  client,
+  total,
+  sw_object
+ORDER BY
+  freq / total DESC,
+  client
diff --git a/sql/2021/pwa/sw_objects_name_only.sql b/sql/2021/pwa/sw_objects_name_only.sql
new file mode 100644
index 00000000000..ac5bdb9f644
--- /dev/null
+++ b/sql/2021/pwa/sw_objects_name_only.sql
@@ -0,0 +1,47 @@
+#standardSQL
+# SW objects
+CREATE TEMPORARY FUNCTION getSWObjects(swObjectsInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var swObjects = Object.values(JSON.parse(swObjectsInfo));
+  if (typeof swObjects != 'string') {
+    swObjects = swObjects.toString();
+  }
+  swObjects = swObjects.trim().split(',').map(x => x.split('.')[0]);
+  return Array.from(new Set(swObjects));
+} catch (e) {
+  return [];
+}
+''';
+SELECT
+  _TABLE_SUFFIX AS client,
+  sw_object,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getSWObjects(JSON_EXTRACT(payload, '$._pwa.swObjectsInfo'))) AS sw_object
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true" AND
+  JSON_EXTRACT(payload, '$._pwa.swObjectsInfo') != "[]"
+GROUP BY
+  client,
+  total,
+  sw_object
+ORDER BY
+  freq / total DESC,
+  client
diff --git a/sql/2021/pwa/sw_registration_properties.sql b/sql/2021/pwa/sw_registration_properties.sql
new file mode 100644
index 00000000000..ee614ccb0dd
--- /dev/null
+++ b/sql/2021/pwa/sw_registration_properties.sql
@@ -0,0 +1,48 @@
+#standardSQL
+# SW registration properties
+CREATE TEMPORARY FUNCTION getSWRegistrationProperties(swRegistrationPropertiesInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var swRegistrationProperties = Object.values(JSON.parse(swRegistrationPropertiesInfo));
+  if (typeof swRegistrationProperties != 'string') {
+    swRegistrationProperties = swRegistrationProperties.toString();
+  }
+  swRegistrationProperties = swRegistrationProperties.trim().split(',');
+  return Array.from(new Set(swRegistrationProperties));
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  sw_registration_properties,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getSWRegistrationProperties(JSON_EXTRACT(payload, '$._pwa.swRegistrationPropertiesInfo'))) AS sw_registration_properties
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true" AND
+  JSON_EXTRACT(payload, '$._pwa.swRegistrationPropertiesInfo') != "[]"
+GROUP BY
+  client,
+  total,
+  sw_registration_properties
+ORDER BY
+  freq / total DESC,
+  client
diff --git a/sql/2021/pwa/sw_registration_properties_name_only.sql b/sql/2021/pwa/sw_registration_properties_name_only.sql
new file mode 100644
index 00000000000..326a4eddbc2
--- /dev/null
+++ b/sql/2021/pwa/sw_registration_properties_name_only.sql
@@ -0,0 +1,48 @@
+#standardSQL
+# SW registration properties
+CREATE TEMPORARY FUNCTION getSWRegistrationProperties(swRegistrationPropertiesInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var swRegistrationProperties = Object.values(JSON.parse(swRegistrationPropertiesInfo));
+  if (typeof swRegistrationProperties != 'string') {
+    swRegistrationProperties = swRegistrationProperties.toString();
+  }
+  swRegistrationProperties = swRegistrationProperties.trim().split(',').map(x => x.split('.')[0]);
+  return Array.from(new Set(swRegistrationProperties));
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  sw_registration_properties,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getSWRegistrationProperties(JSON_EXTRACT(payload, '$._pwa.swRegistrationPropertiesInfo'))) AS sw_registration_properties
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true" AND
+  JSON_EXTRACT(payload, '$._pwa.swRegistrationPropertiesInfo') != "[]"
+GROUP BY
+  client,
+  total,
+  sw_registration_properties
+ORDER BY
+  freq / total DESC,
+  client
diff --git a/sql/2021/pwa/top_manifest_categories.sql b/sql/2021/pwa/top_manifest_categories.sql
new file mode 100644
index 00000000000..9653ffdd719
--- /dev/null
+++ b/sql/2021/pwa/top_manifest_categories.sql
@@ -0,0 +1,86 @@
+#standardSQL
+# Top manifest categories
+CREATE TEMPORARY FUNCTION getCategories(manifest STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var $ = Object.values(JSON.parse(manifest))[0];
+  var categories = $.categories;
+  if (typeof categories == 'string') {
+    return [categories];
+  }
+  return categories;
+} catch (e) {
+  return null;
+}
+''';
+
+WITH totals AS (
+  SELECT
+    _TABLE_SUFFIX,
+    COUNT(0) AS total,
+    COUNTIF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true") AS pwa_total
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  WHERE
+    JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+  GROUP BY
+    _TABLE_SUFFIX
+),
+
+manifests_categories AS (
+  SELECT
+    'All Sites' AS type,
+    _TABLE_SUFFIX AS client,
+    category,
+    COUNT(DISTINCT url) AS freq,
+    total,
+    COUNT(DISTINCT url) / total AS pct,
+    COUNTIF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true") AS pwa_freq,
+    pwa_total,
+    COUNTIF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true") / pwa_total AS pwa_pct
+  FROM
+    `httparchive.pages.2021_07_01_*`,
+    UNNEST(getCategories(JSON_EXTRACT(payload, '$._pwa.manifests'))) AS category
+  JOIN
+    totals
+  USING (_TABLE_SUFFIX)
+  WHERE
+    JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+  GROUP BY
+    client,
+    category,
+    total,
+    pwa_total
+  HAVING
+    category IS NOT NULL
+  ORDER BY
+    type DESC,
+    freq / total DESC,
+    category,
+    client
+)
+
+SELECT
+  'PWA Sites' AS type,
+  client,
+  category,
+  pwa_freq AS freq,
+  pwa_total AS total,
+  pwa_pct AS pct
+FROM
+  manifests_categories
+UNION ALL
+SELECT
+  'All Sites' AS type,
+  client,
+  category,
+  freq,
+  total,
+  pct
+FROM
+  manifests_categories
+ORDER BY
+  type DESC,
+  pct DESC,
+  category,
+  client
diff --git a/sql/2021/pwa/top_manifest_display_values.sql b/sql/2021/pwa/top_manifest_display_values.sql
new file mode 100644
index 00000000000..1dade49983d
--- /dev/null
+++ b/sql/2021/pwa/top_manifest_display_values.sql
@@ -0,0 +1,56 @@
+#standardSQL
+# Top most used display values in manifest files
+
+CREATE TEMP FUNCTION getDisplay(manifest STRING) RETURNS STRING LANGUAGE js AS '''
+try {
+  var $ = Object.values(JSON.parse(manifest))[0];
+  if (!('display' in $)) {
+    return '(not set)';
+  }
+  return $.display;
+} catch {
+  return '(not set)'
+}
+''';
+
+SELECT
+  "PWA Sites" AS type,
+  _TABLE_SUFFIX AS client,
+  getDisplay(JSON_EXTRACT(payload, '$._pwa.manifests')) AS display,
+  COUNT(0) AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.manifests') != "[]" AND
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+GROUP BY
+  client,
+  display -- noqa: PRS
+QUALIFY
+  display IS NOT NULL AND
+  freq > 100
+UNION ALL
+SELECT
+  "All Sites" AS type,
+  _TABLE_SUFFIX AS client,
+  getDisplay(JSON_EXTRACT(payload, '$._pwa.manifests')) AS display,
+  COUNT(0) AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+GROUP BY
+  client,
+  display -- noqa: PRS
+QUALIFY
+  display IS NOT NULL AND
+  freq > 100
+ORDER BY
+  type DESC,
+  freq / total DESC,
+  display,
+  client
diff --git a/sql/2021/pwa/top_manifest_icon_sizes.sql b/sql/2021/pwa/top_manifest_icon_sizes.sql
new file mode 100644
index 00000000000..9b628603b19
--- /dev/null
+++ b/sql/2021/pwa/top_manifest_icon_sizes.sql
@@ -0,0 +1,88 @@
+#standardSQL
+# Top manifest icon sizes
+CREATE TEMPORARY FUNCTION getIconSizes(manifest STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var $ = Object.values(JSON.parse(manifest))[0];
+  return $.icons.map(icon => icon.sizes);
+} catch (e) {
+  return null;
+}
+''';
+
+WITH totals AS (
+  SELECT
+    _TABLE_SUFFIX,
+    COUNT(0) AS total,
+    COUNTIF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true") AS pwa_total
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  WHERE
+    JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+  GROUP BY
+    _TABLE_SUFFIX
+),
+
+manifests_icon_sizes AS (
+  SELECT
+    'All Sites' AS type,
+    _TABLE_SUFFIX AS client,
+    size,
+    COUNT(DISTINCT url) AS freq,
+    total,
+    COUNT(DISTINCT url) / total AS pct,
+    COUNTIF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true") AS pwa_freq,
+    pwa_total,
+    COUNTIF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true") / pwa_total AS pwa_pct
+  FROM
+    `httparchive.pages.2021_07_01_*`,
+    UNNEST(getIconSizes(JSON_EXTRACT(payload, '$._pwa.manifests'))) AS size
+  JOIN
+    totals
+  USING (_TABLE_SUFFIX)
+  WHERE
+    JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+  GROUP BY
+    client,
+    size,
+    total,
+    pwa_total
+  HAVING
+    size IS NOT NULL
+  ORDER BY
+    type DESC,
+    freq / total DESC,
+    size,
+    client
+)
+
+SELECT
+  'PWA Sites' AS type,
+  client,
+  size,
+  pwa_freq AS freq,
+  pwa_total AS total,
+  pwa_pct AS pct
+FROM
+  manifests_icon_sizes
+WHERE
+  size IS NOT NULL AND
+  freq > 100
+UNION ALL
+SELECT
+  'All Sites' AS type,
+  client,
+  size,
+  freq,
+  total,
+  pct
+FROM
+  manifests_icon_sizes
+WHERE
+  size IS NOT NULL AND
+  freq > 100
+ORDER BY
+  type DESC,
+  pct DESC,
+  size,
+  client
diff --git a/sql/2021/pwa/top_manifest_orientations.sql b/sql/2021/pwa/top_manifest_orientations.sql
new file mode 100644
index 00000000000..74a82d6ad3d
--- /dev/null
+++ b/sql/2021/pwa/top_manifest_orientations.sql
@@ -0,0 +1,52 @@
+#standardSQL
+# Top manifest orientations
+
+CREATE TEMP FUNCTION getOrientation(manifest STRING) RETURNS STRING LANGUAGE js AS '''
+try {
+  var $ = Object.values(JSON.parse(manifest))[0];
+  if (!('orientation' in $)) {
+    return '(not set)';
+  }
+  return $.orientation;
+} catch {
+  return '(not set)'
+}
+''';
+
+SELECT
+  "PWA Sites" AS type,
+  _TABLE_SUFFIX AS client,
+  getOrientation(JSON_EXTRACT(payload, '$._pwa.manifests')) AS orientation,
+  COUNT(0) AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.manifests') != "[]" AND
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+GROUP BY
+  type,
+  client,
+  orientation
+UNION ALL
+SELECT
+  "All Sites" AS type,
+  _TABLE_SUFFIX AS client,
+  getOrientation(JSON_EXTRACT(payload, '$._pwa.manifests')) AS orientation,
+  COUNT(0) AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+GROUP BY
+  type,
+  client,
+  orientation
+ORDER BY
+  type DESC,
+  freq / total DESC,
+  orientation,
+  client
diff --git a/sql/2021/pwa/top_manifest_properties.sql b/sql/2021/pwa/top_manifest_properties.sql
new file mode 100644
index 00000000000..4b38eb79d68
--- /dev/null
+++ b/sql/2021/pwa/top_manifest_properties.sql
@@ -0,0 +1,92 @@
+#standardSQL
+# Top manifest properties
+
+CREATE TEMP FUNCTION getManifestProps(manifest STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var manifestJSON = Object.values(JSON.parse(manifest))[0];
+  if (typeof manifestJSON === 'string') {
+    return null;
+  }
+  return Object.keys(manifestJSON);
+} catch {
+  return null
+}
+''';
+
+WITH totals AS (
+  SELECT
+    _TABLE_SUFFIX,
+    COUNT(0) AS total,
+    COUNTIF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true") AS pwa_total
+  FROM
+    `httparchive.pages.2021_07_01_*`
+  WHERE
+    JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+  GROUP BY
+    _TABLE_SUFFIX
+),
+
+manifests_properties AS (
+  SELECT
+    'All Sites' AS type,
+    _TABLE_SUFFIX AS client,
+    property,
+    COUNT(DISTINCT url) AS freq,
+    total,
+    COUNT(DISTINCT url) / total AS pct,
+    COUNTIF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true") AS pwa_freq,
+    pwa_total,
+    COUNTIF(JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true") / pwa_total AS pwa_pct
+  FROM
+    `httparchive.pages.2021_07_01_*`,
+    UNNEST(getManifestProps(JSON_EXTRACT(payload, '$._pwa.manifests'))) AS property
+  JOIN
+    totals
+  USING (_TABLE_SUFFIX)
+  WHERE
+    JSON_EXTRACT(payload, '$._pwa.manifests') != "[]"
+  GROUP BY
+    client,
+    property,
+    total,
+    pwa_total
+  HAVING
+    property IS NOT NULL
+  ORDER BY
+    type DESC,
+    freq / total DESC,
+    property,
+    client
+)
+
+SELECT
+  'PWA Sites' AS type,
+  client,
+  property,
+  pwa_freq AS freq,
+  pwa_total AS total,
+  pwa_pct AS pct
+FROM
+  manifests_properties
+WHERE
+  property IS NOT NULL AND
+  freq > 100
+UNION ALL
+SELECT
+  'All Sites' AS type,
+  client,
+  property,
+  freq,
+  total,
+  pct
+FROM
+  manifests_properties
+WHERE
+  property IS NOT NULL AND
+  freq > 100
+ORDER BY
+  type DESC,
+  pct DESC,
+  property,
+  client
diff --git a/sql/2021/pwa/workbox_methods.sql b/sql/2021/pwa/workbox_methods.sql
new file mode 100644
index 00000000000..ff3955ffa39
--- /dev/null
+++ b/sql/2021/pwa/workbox_methods.sql
@@ -0,0 +1,57 @@
+#standardSQL
+#Workbox methods
+CREATE TEMPORARY FUNCTION getWorkboxMethods(workboxInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var workboxPackageMethods = Object.values(JSON.parse(workboxInfo));
+  if (typeof workboxPackageMethods == 'string') {
+    workboxPackageMethods = [workboxPackageMethods];
+  }
+  var workboxMethods = [];
+  /* Replacing spaces and commas */
+  for (var i = 0; i < workboxPackageMethods.length; i++) {
+      var workboxItems = workboxPackageMethods[i].toString().trim().split(',');
+      for(var j = 0; j < workboxItems.length; j++) {
+        if(workboxItems[j].indexOf(':') == -1) {
+          workboxMethods.push(workboxItems[j].trim());
+        }
+      }
+  }
+  return Array.from(new Set(workboxMethods));
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  workbox_method,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getWorkboxMethods(JSON_EXTRACT(payload, '$._pwa.workboxInfo'))) AS workbox_method
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.workboxInfo') != "[]" AND
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+GROUP BY
+  client,
+  workbox_method,
+  total
+ORDER BY
+  pct DESC,
+  client
diff --git a/sql/2021/pwa/workbox_packages.sql b/sql/2021/pwa/workbox_packages.sql
new file mode 100644
index 00000000000..8d079376232
--- /dev/null
+++ b/sql/2021/pwa/workbox_packages.sql
@@ -0,0 +1,60 @@
+#standardSQL
+#Workbox package and methods
+CREATE TEMPORARY FUNCTION getWorkboxPackages(workboxInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var workboxPackageMethods = Object.values(JSON.parse(workboxInfo));
+  if (typeof workboxPackageMethods == 'string') {
+    workboxPackageMethods = [workboxPackageMethods];
+  }
+  var workboxPackages = [];
+  /* Replacing spaces and commas */
+  for (var i = 0; i < workboxPackageMethods.length; i++) {
+      var workboxItems = workboxPackageMethods[i].toString().trim().split(',');
+      for(var j = 0; j < workboxItems.length; j++) {
+        var workboxItem = workboxItems[j];
+        var firstColonIndex = workboxItem.indexOf(':');
+        if(firstColonIndex > -1) {
+          var workboxPackage = workboxItem.trim().substring(0, workboxItem.indexOf(':', firstColonIndex + 1));
+          workboxPackages.push(workboxPackage);
+        }
+      }
+  }
+  return Array.from(new Set(workboxPackages));
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  workbox_package,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getWorkboxPackages(JSON_EXTRACT(payload, '$._pwa.workboxInfo'))) AS workbox_package
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.workboxInfo') != "[]" AND
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+GROUP BY
+  client,
+  workbox_package,
+  total
+ORDER BY
+  pct DESC,
+  client
diff --git a/sql/2021/pwa/workbox_usage.sql b/sql/2021/pwa/workbox_usage.sql
new file mode 100644
index 00000000000..f4e6472c506
--- /dev/null
+++ b/sql/2021/pwa/workbox_usage.sql
@@ -0,0 +1,18 @@
+#standardSQL
+# Workbox usage
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  COUNTIF(JSON_EXTRACT(payload, '$._pwa.workboxInfo') != "[]") AS freq,
+  SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS total,
+  COUNTIF(JSON_EXTRACT(payload, '$._pwa.workboxInfo') != "[]") / SUM(COUNT(0)) OVER (PARTITION BY _TABLE_SUFFIX) AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.manifests') != "[]" AND
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+GROUP BY
+  client
+ORDER BY
+  freq / total DESC,
+  client
diff --git a/sql/2021/pwa/workbox_versions.sql b/sql/2021/pwa/workbox_versions.sql
new file mode 100644
index 00000000000..b79c904f205
--- /dev/null
+++ b/sql/2021/pwa/workbox_versions.sql
@@ -0,0 +1,60 @@
+#standardSQL
+#Workbox versions
+CREATE TEMPORARY FUNCTION getWorkboxVersions(workboxInfo STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS '''
+try {
+  var workboxPackageMethods = Object.values(JSON.parse(workboxInfo));
+  if (typeof workboxPackageMethods == 'string') {
+    workboxPackageMethods = [workboxPackageMethods];
+  }
+  var workboxVersions = [];
+  /* Replacing spaces and commas */
+  for (var i = 0; i < workboxPackageMethods.length; i++) {
+      var workboxItems = workboxPackageMethods[i].toString().trim().split(',');
+      for(var j = 0; j < workboxItems.length; j++) {
+        var workboxItem = workboxItems[j];
+        var firstColonIndex = workboxItem.indexOf(':');
+        if(firstColonIndex > -1) {
+          var workboxVersion = workboxItem.trim().substring(workboxItem.indexOf(':', firstColonIndex + 1) + 1);
+          workboxVersions.push(workboxVersion);
+        }
+      }
+  }
+  return Array.from(new Set(workboxVersions));
+} catch (e) {
+  return [];
+}
+''';
+
+SELECT
+  _TABLE_SUFFIX AS client,
+  workbox_version,
+  COUNT(DISTINCT url) AS freq,
+  total,
+  COUNT(DISTINCT url) / total AS pct
+FROM
+  `httparchive.pages.2021_07_01_*`,
+  UNNEST(getWorkboxVersions(JSON_EXTRACT(payload, '$._pwa.workboxInfo'))) AS workbox_version
+JOIN
+  (
+    SELECT
+      _TABLE_SUFFIX,
+      COUNT(0) AS total
+    FROM
+      `httparchive.pages.2021_07_01_*`
+    WHERE
+      JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+    GROUP BY
+      _TABLE_SUFFIX
+  )
+USING (_TABLE_SUFFIX)
+WHERE
+  JSON_EXTRACT(payload, '$._pwa.workboxInfo') != "[]" AND
+  JSON_EXTRACT(payload, '$._pwa.serviceWorkerHeuristic') = "true"
+GROUP BY
+  _TABLE_SUFFIX,
+  workbox_version,
+  total
+ORDER BY
+  pct DESC,
+  client
diff --git a/sql/2021/third-parties/README.md b/sql/2021/third-parties/README.md
deleted file mode 100644
index fd33b80fdb8..00000000000
--- a/sql/2021/third-parties/README.md
+++ /dev/null
@@ -1,10 +0,0 @@
-# 2021 Third Parties queries
-
-<!--
-  This directory contains all of the 2021 Third Parties chapter queries.
-
-  Each query should have a corresponding `metric_name.sql` file.
-  Note that readers are linked to this directory, so try to make the SQL file names descriptive for easy browsing.
-
-  Analysts: if helpful, you can use this README to give additional info about the queries.
--->
diff --git a/src/content/en/2021/third-parties.md b/src/content/en/2021/third-parties.md
deleted file mode 100644
index 4eb1dff3c5e..00000000000
--- a/src/content/en/2021/third-parties.md
+++ /dev/null
@@ -1,20 +0,0 @@
----
-#See https://github.com/HTTPArchive/almanac.httparchive.org/wiki/Authors'-Guide#metadata-to-add-at-the-top-of-your-chapters
-title: Third Parties
-description: TODO
-authors: []
-reviewers: []
-analysts: []
-editors: []
-translators: []
-results: https://docs.google.com/spreadsheets/d/1tf4RMF8SYr6he9tbqt61yuFJ_QK-F-i7XPxaPkpKSDI/
-featured_quote: TODO
-featured_stat_1: TODO
-featured_stat_label_1: TODO
-featured_stat_2: TODO
-featured_stat_label_2: TODO
-featured_stat_3: TODO
-featured_stat_label_3: TODO
----
-
-## TODO
diff --git a/src/package-lock.json b/src/package-lock.json
index f4962128c73..3ad760bab0c 100644
--- a/src/package-lock.json
+++ b/src/package-lock.json
@@ -13,8 +13,8 @@
         "fs-extra": "10.0.0",
         "jsdom": "17.0.0",
         "node-fetch": "3.0.0",
-        "node-watch": "0.7.1",
-        "prettier": "2.4.0",
+        "node-watch": "0.7.2",
+        "prettier": "2.4.1",
         "rainbow-code": "2.1.7",
         "recursive-readdir": "2.2.2",
         "run-script-os": "1.1.6",
@@ -738,9 +738,9 @@
       }
     },
     "node_modules/node-watch": {
-      "version": "0.7.1",
-      "resolved": "https://registry.npmjs.org/node-watch/-/node-watch-0.7.1.tgz",
-      "integrity": "sha512-UWblPYuZYrkCQCW5PxAwYSxaELNBLUckrTBBk8xr1/bUgyOkYYTsUcV4e3ytcazFEOyiRyiUrsG37pu6I0I05g==",
+      "version": "0.7.2",
+      "resolved": "https://registry.npmjs.org/node-watch/-/node-watch-0.7.2.tgz",
+      "integrity": "sha512-g53VjSARRv1JdST0LZRIg8RiuLr1TaBbVPsVvxh0/0Ymvi0xYUjDuoqQQAWtHJQUXhiShowPT/aXKNeHBcyQsw==",
       "dev": true,
       "engines": {
         "node": ">=6"
@@ -827,9 +827,9 @@
       }
     },
     "node_modules/prettier": {
-      "version": "2.4.0",
-      "resolved": "https://registry.npmjs.org/prettier/-/prettier-2.4.0.tgz",
-      "integrity": "sha512-DsEPLY1dE5HF3BxCRBmD4uYZ+5DCbvatnolqTqcxEgKVZnL2kUfyu7b8pPQ5+hTBkdhU9SLUmK0/pHb07RE4WQ==",
+      "version": "2.4.1",
+      "resolved": "https://registry.npmjs.org/prettier/-/prettier-2.4.1.tgz",
+      "integrity": "sha512-9fbDAXSBcc6Bs1mZrDYb3XKzDLm4EXXL9sC1LqKP5rZkT6KRr/rf9amVUcODVXgguK/isJz0d0hP72WeaKWsvA==",
       "dev": true,
       "bin": {
         "prettier": "bin-prettier.js"
@@ -1793,9 +1793,9 @@
       }
     },
     "node-watch": {
-      "version": "0.7.1",
-      "resolved": "https://registry.npmjs.org/node-watch/-/node-watch-0.7.1.tgz",
-      "integrity": "sha512-UWblPYuZYrkCQCW5PxAwYSxaELNBLUckrTBBk8xr1/bUgyOkYYTsUcV4e3ytcazFEOyiRyiUrsG37pu6I0I05g==",
+      "version": "0.7.2",
+      "resolved": "https://registry.npmjs.org/node-watch/-/node-watch-0.7.2.tgz",
+      "integrity": "sha512-g53VjSARRv1JdST0LZRIg8RiuLr1TaBbVPsVvxh0/0Ymvi0xYUjDuoqQQAWtHJQUXhiShowPT/aXKNeHBcyQsw==",
       "dev": true
     },
     "nwsapi": {
@@ -1861,9 +1861,9 @@
       "dev": true
     },
     "prettier": {
-      "version": "2.4.0",
-      "resolved": "https://registry.npmjs.org/prettier/-/prettier-2.4.0.tgz",
-      "integrity": "sha512-DsEPLY1dE5HF3BxCRBmD4uYZ+5DCbvatnolqTqcxEgKVZnL2kUfyu7b8pPQ5+hTBkdhU9SLUmK0/pHb07RE4WQ==",
+      "version": "2.4.1",
+      "resolved": "https://registry.npmjs.org/prettier/-/prettier-2.4.1.tgz",
+      "integrity": "sha512-9fbDAXSBcc6Bs1mZrDYb3XKzDLm4EXXL9sC1LqKP5rZkT6KRr/rf9amVUcODVXgguK/isJz0d0hP72WeaKWsvA==",
       "dev": true
     },
     "psl": {
diff --git a/src/package.json b/src/package.json
index 3df0a11b5b0..76a2e16463b 100644
--- a/src/package.json
+++ b/src/package.json
@@ -37,8 +37,8 @@
     "fs-extra": "10.0.0",
     "jsdom": "17.0.0",
     "node-fetch": "3.0.0",
-    "node-watch": "0.7.1",
-    "prettier": "2.4.0",
+    "node-watch": "0.7.2",
+    "prettier": "2.4.1",
     "rainbow-code": "2.1.7",
     "recursive-readdir": "2.2.2",
     "run-script-os": "1.1.6",
diff --git a/src/requirements.txt b/src/requirements.txt
index df8c69a6c14..7c0bfb6e227 100644
--- a/src/requirements.txt
+++ b/src/requirements.txt
@@ -4,4 +4,4 @@ gunicorn==20.1.0
 pytest==6.2.5
 pytest-watch==4.2.0
 pytest-cov==2.12.1
-sqlfluff==0.6.5
+sqlfluff==0.6.6

From a42124bd2e53b1ea284a5e0c1b000655758deaa1 Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Tue, 19 Oct 2021 03:47:54 +0530
Subject: [PATCH 20/23] lint fixes

---
 sql/2021/javascript/web_components_specs.sql | 56 ++++++++++----------
 1 file changed, 28 insertions(+), 28 deletions(-)

diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
index 917ebafeef0..993ad2bf2d5 100644
--- a/sql/2021/javascript/web_components_specs.sql
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -1,39 +1,39 @@
 SELECT
   client,
   percentile,
-  APPROX_QUANTILES(custom_elements, 1000)[
-    OFFSET(percentile * 10)] AS custom_elements,
-  APPROX_QUANTILES(shadow_roots, 1000)[
-    OFFSET(percentile * 10)] AS shadow_roots,
-  APPROX_QUANTILES(template, 1000)[
-    OFFSET(percentile * 10)] AS template
-FROM (
-  SELECT
+  APPROX_QUANTILES(custom_elements,
+  1000)[OFFSET(percentile * 10)] AS custom_elements,
+  APPROX_QUANTILES(shadow_roots,
+  1000)[OFFSET(percentile * 10)] AS shadow_roots,
+  APPROX_QUANTILES(template,
+  1000)[OFFSET(percentile * 10)] AS template 
+FROM
+  (SELECT
     _TABLE_SUFFIX AS client,
-    (CASE 
-     WHEN ARRAY_LENGTH(JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-          '$._javascript'),
-        '$.web_component_specs.custom_elements'))) > 1 THEN 1
-    ELSE 0 
+    (CASE       
+        WHEN ARRAY_LENGTH(JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+        '$.web_component_specs.custom_elements'))) > 1 THEN 1     
+        ELSE 0      
     END) AS custom_elements,
-     (CASE 
-     WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-          '$._javascript'),
-        '$.web_component_specs.shadow_roots'))) > 1 THEN 1
-    ELSE 0 
+    (CASE       
+        WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+        '$.web_component_specs.shadow_roots'))) > 1 THEN 1     
+        ELSE 0      
     END) AS shadow_roots,
-    (CASE 
-     WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-          '$._javascript'),
-        '$.web_component_specs.template'))) > 1 THEN 1
-    ELSE 0 
-    END) AS template
+    (CASE       
+        WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+        '$._javascript'),
+        '$.web_component_specs.template'))) > 1 THEN 1     
+        ELSE 0      
+    END) AS template   
   FROM
     `httparchive.pages.2021_09_01_*` ),
-  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
+UNNEST([10, 25, 50, 75, 90, 100]) AS percentile 
 GROUP BY
   percentile,
-  client
+  client 
 ORDER BY
-  percentile,
-  client
+    percentile,
+    client
\ No newline at end of file

From 6025b6c35331eef83fbe273b170e7450d904b3d6 Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Tue, 19 Oct 2021 03:52:48 +0530
Subject: [PATCH 21/23] lint fix -2

---
 sql/2021/javascript/web_components_specs.sql | 78 +++++++++++---------
 1 file changed, 45 insertions(+), 33 deletions(-)

diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
index 993ad2bf2d5..c6b2475f649 100644
--- a/sql/2021/javascript/web_components_specs.sql
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -1,39 +1,51 @@
 SELECT
-  client,
-  percentile,
-  APPROX_QUANTILES(custom_elements,
-  1000)[OFFSET(percentile * 10)] AS custom_elements,
-  APPROX_QUANTILES(shadow_roots,
-  1000)[OFFSET(percentile * 10)] AS shadow_roots,
-  APPROX_QUANTILES(template,
-  1000)[OFFSET(percentile * 10)] AS template 
+    client,
+    percentile,
+    APPROX_QUANTILES(custom_elements,
+        1000)[OFFSET(percentile * 10)] AS custom_elements,
+    APPROX_QUANTILES(shadow_roots,
+        1000)[OFFSET(percentile * 10)] AS shadow_roots,
+    APPROX_QUANTILES(template,
+        1000)[OFFSET(percentile * 10)] AS template
 FROM
-  (SELECT
-    _TABLE_SUFFIX AS client,
-    (CASE       
-        WHEN ARRAY_LENGTH(JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-        '$._javascript'),
-        '$.web_component_specs.custom_elements'))) > 1 THEN 1     
-        ELSE 0      
-    END) AS custom_elements,
-    (CASE       
-        WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-        '$._javascript'),
-        '$.web_component_specs.shadow_roots'))) > 1 THEN 1     
-        ELSE 0      
-    END) AS shadow_roots,
-    (CASE       
-        WHEN ARRAY_LENGTH( JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-        '$._javascript'),
-        '$.web_component_specs.template'))) > 1 THEN 1     
-        ELSE 0      
-    END) AS template   
-  FROM
+    (SELECT
+        _table_suffix AS client,
+        (CASE
+                 WHEN
+            ARRAY_LENGTH(
+                JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+                                                                       '$._javascript'
+                                                                  ),
+                                                                       '$.web_component_specs.custom_elements'
+                                                                  ))) > 1 THEN 1
+                 ELSE 0
+             END) AS custom_elements,
+        (CASE
+                 WHEN
+            ARRAY_LENGTH(
+                JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+                                                                       '$._javascript'
+                                                                  ),
+                                                                       '$.web_component_specs.shadow_roots'
+                                                                  ))) > 1 THEN 1
+                 ELSE 0
+             END) AS shadow_roots,
+        (CASE
+                 WHEN
+            ARRAY_LENGTH(
+                JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
+                                                                       '$._javascript'
+                                                                  ),
+                                                                       '$.web_component_specs.template'
+                                                                  ))) > 1 THEN 1
+                 ELSE 0
+             END) AS template
+    FROM
     `httparchive.pages.2021_09_01_*` ),
-UNNEST([10, 25, 50, 75, 90, 100]) AS percentile 
+    UNNEST([10, 25, 50, 75, 90, 100])
 GROUP BY
-  percentile,
-  client 
+    percentile,
+    client
 ORDER BY
     percentile,
-    client
\ No newline at end of file
+    client

From 11028b8a19aeeda3306d2260f45184b3f07ada0e Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Wed, 20 Oct 2021 12:55:49 +0530
Subject: [PATCH 22/23] add review updates

---
 .../most_requested_content_type.sql           | 30 +++++----
 sql/2021/javascript/web_components_specs.sql  | 66 +++++--------------
 2 files changed, 37 insertions(+), 59 deletions(-)

diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index 19b7cb9eb22..ce263c9aabe 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -1,14 +1,22 @@
 SELECT
-    client,
-    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%image%') as images,
-    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%font%') as font, 
-    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%css%') as css, 
-    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%javascript%') as javascript, 
-    (select count(*) from `httparchive.almanac.requests` WHERE date = '2021-07-01' and resp_content_type like '%json%') as json,
-    count(*) as total
-  FROM
-    `httparchive.almanac.requests`
+  client,
+  CASE
+    WHEN resp_content_type LIKE '%image%' THEN 'images'
+    WHEN resp_content_type LIKE '%font%' THEN 'font'
+    WHEN resp_content_type LIKE '%css%' THEN 'css'
+    WHEN resp_content_type LIKE '%javascript%' THEN 'javascript'
+    WHEN resp_content_type LIKE '%json%' THEN 'json'
+    ELSE 'other'
+  END AS content_type,
+  COUNT(0) AS count,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct
+FROM
+  `httparchive.almanac.requests`
+WHERE
+  date = '2021-07-01'
 GROUP BY
-  client
+  client,
+  content_type
 ORDER BY
-  client
+  pct DESC
\ No newline at end of file
diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
index c6b2475f649..f264ed97260 100644
--- a/sql/2021/javascript/web_components_specs.sql
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -1,51 +1,21 @@
 SELECT
-    client,
-    percentile,
-    APPROX_QUANTILES(custom_elements,
-        1000)[OFFSET(percentile * 10)] AS custom_elements,
-    APPROX_QUANTILES(shadow_roots,
-        1000)[OFFSET(percentile * 10)] AS shadow_roots,
-    APPROX_QUANTILES(template,
-        1000)[OFFSET(percentile * 10)] AS template
-FROM
-    (SELECT
-        _table_suffix AS client,
-        (CASE
-                 WHEN
-            ARRAY_LENGTH(
-                JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-                                                                       '$._javascript'
-                                                                  ),
-                                                                       '$.web_component_specs.custom_elements'
-                                                                  ))) > 1 THEN 1
-                 ELSE 0
-             END) AS custom_elements,
-        (CASE
-                 WHEN
-            ARRAY_LENGTH(
-                JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-                                                                       '$._javascript'
-                                                                  ),
-                                                                       '$.web_component_specs.shadow_roots'
-                                                                  ))) > 1 THEN 1
-                 ELSE 0
-             END) AS shadow_roots,
-        (CASE
-                 WHEN
-            ARRAY_LENGTH(
-                JSON_QUERY_ARRAY(JSON_EXTRACT(JSON_EXTRACT_SCALAR(payload,
-                                                                       '$._javascript'
-                                                                  ),
-                                                                       '$.web_component_specs.template'
-                                                                  ))) > 1 THEN 1
-                 ELSE 0
-             END) AS template
-    FROM
-    `httparchive.pages.2021_09_01_*` ),
-    UNNEST([10, 25, 50, 75, 90, 100])
+  client,
+  percentile,
+  APPROX_QUANTILES(custom_elements, 1000)[OFFSET(percentile * 10)] AS custom_elements,
+  APPROX_QUANTILES(shadow_roots, 1000)[OFFSET(percentile * 10)] AS shadow_roots,
+  APPROX_QUANTILES(template, 1000)[OFFSET(percentile * 10)] AS template
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    ARRAY_LENGTH(JSON_EXTRACT_ARRAY(JSON_EXTRACT_SCALAR(payload, '$._javascript'), '$.web_component_specs.custom_elements')) AS custom_elements,
+    ARRAY_LENGTH(JSON_EXTRACT_ARRAY(JSON_EXTRACT_SCALAR(payload, '$._javascript'), '$.web_component_specs.shadow_roots')) AS shadow_roots,
+    ARRAY_LENGTH(JSON_EXTRACT_ARRAY(JSON_EXTRACT_SCALAR(payload, '$._javascript'), '$.web_component_specs.template')) AS template
+  FROM
+    `httparchive.pages.2021_09_01_*`),
+  UNNEST([10, 25, 50, 75, 90, 100]) AS percentile
 GROUP BY
-    percentile,
-    client
+  percentile,
+  client
 ORDER BY
-    percentile,
-    client
+  percentile,
+  client
\ No newline at end of file

From d036f20a2d07a339acdc50a40c8ae1b7b8f6276c Mon Sep 17 00:00:00 2001
From: Nishu Goel <nishu0505@gmail.com>
Date: Wed, 20 Oct 2021 13:00:13 +0530
Subject: [PATCH 23/23] add trailing newline

---
 sql/2021/javascript/most_requested_content_type.sql | 2 +-
 sql/2021/javascript/web_components_specs.sql        | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/sql/2021/javascript/most_requested_content_type.sql b/sql/2021/javascript/most_requested_content_type.sql
index ce263c9aabe..573059a6279 100644
--- a/sql/2021/javascript/most_requested_content_type.sql
+++ b/sql/2021/javascript/most_requested_content_type.sql
@@ -19,4 +19,4 @@ GROUP BY
   client,
   content_type
 ORDER BY
-  pct DESC
\ No newline at end of file
+  pct DESC
diff --git a/sql/2021/javascript/web_components_specs.sql b/sql/2021/javascript/web_components_specs.sql
index f264ed97260..8dff570db5b 100644
--- a/sql/2021/javascript/web_components_specs.sql
+++ b/sql/2021/javascript/web_components_specs.sql
@@ -18,4 +18,4 @@ GROUP BY
   client
 ORDER BY
   percentile,
-  client
\ No newline at end of file
+  client