HTTP 2022 additional queries

sql/2022/http/h2_adoption_each_cdn_breakdown.sql

@@ -0,0 +1,29 @@
+#standardSQL
+
+# Percentage of requests using HTTP/2+ vs HTTP/1.1 broken down by whether the
+# request was served from CDN.
+
+SELECT
+  client,
+  _cdn_provider,
+  CASE
+    WHEN LOWER(protocol) = 'quic' OR LOWER(protocol) LIKE 'h3%' THEN 'HTTP/2+'
+    WHEN LOWER(protocol) = 'http/2' OR LOWER(protocol) = 'http/3' THEN 'HTTP/2+'
+    WHEN protocol IS NULL THEN 'Unknown'
+    ELSE UPPER(protocol)
+  END AS http_version_category,
+  COUNT(0) AS num_reqs,
+  SUM(COUNT(0)) OVER (PARTITION BY client, _cdn_provider) AS total_reqs,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client, _cdn_provider) AS pct_reqs
+FROM
+  `httparchive.almanac.requests`
+WHERE
+  date = '2022-06-01' AND
+  LENGTH(_cdn_provider) > 0
+GROUP BY
+  client,
+  _cdn_provider,
+  http_version_category
+ORDER BY
+  client ASC,
+  num_reqs DESC

sql/2022/http/protocol_advertised_via_altsvc.sql

@@ -0,0 +1,57 @@
+#standardSQL
+# Protocol advertised via alt-svc breakdown
+CREATE TEMPORARY FUNCTION extractHTTPHeader(HTTPheaders STRING, header STRING)
+RETURNS ARRAY<STRING> LANGUAGE js AS """
+try {
+  var headers = JSON.parse(HTTPheaders);
+
+  // Filter by header name (which is case insensitive)
+  // If multiple headers it's the same as comma separated
+  const result = [];
+  const allAltSvcHeaderValues = headers.filter(h => h.name.toLowerCase() == header.toLowerCase()).map(h => h.value);
+  for (let altsvcHeaderValue of allAltSvcHeaderValues) {
+    const splittedAltSvcHeaderValue = altsvcHeaderValue.split(",");
+    for (let altsvcToken of splittedAltSvcHeaderValue) {
+      const protocolPortToken = altsvcToken.trim().split(";")[0];
+      const protocolToken = protocolPortToken.split("=")[0];
+      result.push(protocolToken);
+    }
+  }
+  return result;
+} catch (e) {
+  return [];
+}
+""";
+
+WITH altsvcTable AS (
+  SELECT
+    client,
+    url,
+    extractHTTPHeader(response_headers, 'alt-svc') AS protocol
+  FROM
+    `httparchive.almanac.requests`
+  WHERE
+    date = '2022-06-01' AND
+    firstHtml
+)
+SELECT
+  client,
+  protocol,
+  COUNT(0) AS protocol_count,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total_advertised,
+  COUNT(0) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct_advertised
+FROM
+  (
+    SELECT
+      client,
+      url,
+      flattened_protocol AS protocol
+    FROM
+      altsvcTable, altsvcTable.protocol AS flattened_protocol
+  )
+GROUP BY
+  client,
+  protocol
+ORDER BY
+  client ASC,
+  total_advertised DESC

sql/2022/http/sites_using_link_preload.sql

@@ -0,0 +1,35 @@
+#standardSQL
+
+# Number of <link rel="preload">.
+
+CREATE TEMPORARY FUNCTION getNumLinkRelPreload(payload STRING)
+RETURNS INT LANGUAGE js AS """
+try {
+  const $ = JSON.parse(payload)
+  const almanac = JSON.parse($._almanac);
+  const link_preload_nodes = almanac['link-nodes']['nodes'].filter(link_node => link_node['rel'] === 'preload')
+  return link_preload_nodes.length;
+} catch (e) {
+  return -1;
+}
+""";
+
+SELECT
+  client,
+  percentile,
+  APPROX_QUANTILES(num_link_rel_preload, 1000)[OFFSET(percentile * 10)] AS num_percentiles
+FROM (
+  SELECT
+    _TABLE_SUFFIX AS client,
+    url AS page,
+    getNumLinkRelPreload(payload) AS num_link_rel_preload
+  FROM
+    `httparchive.pages.2022_06_01_*`
+),
+UNNEST([10, 25, 50, 75, 90, 95, 100]) AS percentile
+GROUP BY
+  client,
+  percentile
+ORDER BY
+  client,
+  percentile

sql/2022/http/sites_with_altsvc.sql

@@ -0,0 +1,16 @@
+#standardSQL
+# Fraction of sites with alt-svc header
+SELECT
+  client,
+  COUNTIF(CONTAINS_SUBSTR(response_headers, 'alt-svc')) AS sites_with_altsvc,
+  SUM(COUNT(0)) OVER (PARTITION BY client) AS total,
+  COUNTIF(CONTAINS_SUBSTR(response_headers, 'alt-svc')) / SUM(COUNT(0)) OVER (PARTITION BY client) AS pct_sites_with_altsvc
+FROM
+  `httparchive.almanac.requests`
+WHERE
+  date = '2022-06-01' AND
+  firstHtml
+GROUP BY
+  client
+ORDER BY
+  pct_sites_with_altsvc DESC