fix 2022/privacy

HTTPArchive · May 19, 2024 · ce5dea3 · ce5dea3
1 parent 85c91d6
commit ce5dea3
Show file tree

Hide file tree

Showing 18 changed files with 71 additions and 53 deletions.
diff --git a/sql/2022/privacy/most_common_client_hints.sql b/sql/2022/privacy/most_common_client_hints.sql
@@ -29,7 +29,7 @@ meta_tags AS (
       JSON_VALUE(payload, '$._almanac') AS metrics
     FROM
       `httparchive.pages.2022_06_01_*`
-    ),
+  ),
     UNNEST(JSON_QUERY_ARRAY(metrics, '$.meta-nodes.nodes')) meta_node
   WHERE
     JSON_VALUE(meta_node, '$.http-equiv') IS NOT NULL

diff --git a/sql/2022/privacy/most_common_permissions_policy_directives.sql b/sql/2022/privacy/most_common_permissions_policy_directives.sql
@@ -62,7 +62,7 @@ meta_tags AS (
       JSON_VALUE(payload, '$._almanac') AS metrics
     FROM
       `httparchive.pages.2022_06_01_*`
-    ),
+  ),
     UNNEST(JSON_QUERY_ARRAY(metrics, '$.meta-nodes.nodes')) meta_node
   WHERE
     JSON_VALUE(meta_node, '$.http-equiv') IS NOT NULL
@@ -92,8 +92,10 @@ merged_policy AS (
     meta_tags
   USING (client, page)
   WHERE
-    (header_name IN ('feature-policy', 'permissions-policy') OR
-      tag_name IN ('feature-policy', 'permissions-policy')) AND
+    (
+      header_name IN ('feature-policy', 'permissions-policy') OR
+      tag_name IN ('feature-policy', 'permissions-policy')
+    ) AND
     header_value IS NOT NULL AND
     tag_value IS NOT NULL
 )

diff --git a/sql/2022/privacy/most_common_purposes_for_iab_tcf_v2.sql b/sql/2022/privacy/most_common_purposes_for_iab_tcf_v2.sql
@@ -4,8 +4,10 @@
 # https://stackoverflow.com/a/65054751/7391782
 # Warning: fails if there are colons in the keys/values, but these are not expected
 
-CREATE TEMPORARY FUNCTION ExtractKeyValuePairs(input STRING) RETURNS ARRAY < STRUCT < key STRING,
-  value STRING > > AS (
+CREATE TEMPORARY FUNCTION ExtractKeyValuePairs(input STRING) RETURNS ARRAY < STRUCT <
+  key STRING,
+  value STRING
+> > AS (
   (
     SELECT
       ARRAY(
@@ -41,8 +43,8 @@ WITH pages_iab_tcf_v2 AS (
 SELECT
   client,
   field,
-  result.key AS key,
-  result.value AS value,
+  result.key,
+  result.value,
   COUNT(0) AS number_of_websites_with_purpose
 FROM
   (

diff --git a/sql/2022/privacy/nb_sites_with_cname_tracking.sql b/sql/2022/privacy/nb_sites_with_cname_tracking.sql
@@ -25,7 +25,7 @@ WITH websites_using_cname_tracking AS (
 
 totals AS (
   SELECT
-    _TABLE_SUFFIX AS _TABLE_SUFFIX,
+    _TABLE_SUFFIX,
     count(0) AS total_pages
   FROM
     `httparchive.summary_pages.2022_06_01_*`

diff --git a/sql/2022/privacy/nb_sites_with_cname_tracking_per_public_suffix.sql b/sql/2022/privacy/nb_sites_with_cname_tracking_per_public_suffix.sql
@@ -10,7 +10,7 @@ WITH websites_using_cname_tracking AS (
 
 totals AS (
   SELECT
-    _TABLE_SUFFIX AS _TABLE_SUFFIX,
+    _TABLE_SUFFIX,
     count(0) AS total_pages
   FROM
     `httparchive.summary_pages.2022_06_01_*`

diff --git a/sql/2022/privacy/nb_sites_with_cname_tracking_per_rank.sql b/sql/2022/privacy/nb_sites_with_cname_tracking_per_rank.sql
@@ -8,7 +8,7 @@ WITH websites_using_cname_tracking AS (
 
 totals AS (
   SELECT
-    _TABLE_SUFFIX AS _TABLE_SUFFIX,
+    _TABLE_SUFFIX,
     rank_grouping,
     count(0) AS total_pages
   FROM

diff --git a/sql/2022/privacy/number_of_websites_using_each_cmp.sql b/sql/2022/privacy/number_of_websites_using_each_cmp.sql
@@ -16,7 +16,7 @@ WITH totals AS (
 SELECT
   _TABLE_SUFFIX AS client,
   app,
-  total_websites AS total_websites,
+  total_websites,
   COUNT(DISTINCT url) AS number_of_websites,
   COUNT(DISTINCT url) / total_websites AS percent_of_websites
 FROM

diff --git a/sql/2022/privacy/number_of_websites_using_each_fingerprinting.sql b/sql/2022/privacy/number_of_websites_using_each_fingerprinting.sql
@@ -16,7 +16,7 @@ WITH totals AS (
 SELECT
   _TABLE_SUFFIX AS client,
   app,
-  total_websites AS total_websites,
+  total_websites,
   COUNT(DISTINCT url) AS number_of_websites,
   COUNT(DISTINCT url) / total_websites AS percent_of_websites
 FROM

diff --git a/sql/2022/privacy/number_of_websites_using_each_geolocation.sql b/sql/2022/privacy/number_of_websites_using_each_geolocation.sql
@@ -16,7 +16,7 @@ WITH totals AS (
 SELECT
   _TABLE_SUFFIX AS client,
   app,
-  total_websites AS total_websites,
+  total_websites,
   COUNT(DISTINCT url) AS number_of_websites,
   COUNT(DISTINCT url) / total_websites AS percent_of_websites
 FROM

diff --git a/sql/2022/privacy/number_of_websites_using_each_retargeting.sql b/sql/2022/privacy/number_of_websites_using_each_retargeting.sql
@@ -16,7 +16,7 @@ WITH totals AS (
 SELECT
   _TABLE_SUFFIX AS client,
   app,
-  total_websites AS total_websites,
+  total_websites,
   COUNT(DISTINCT url) AS number_of_websites,
   COUNT(DISTINCT url) / total_websites AS percent_of_websites
 FROM

diff --git a/sql/2022/privacy/number_of_websites_with_client_hints.sql b/sql/2022/privacy/number_of_websites_with_client_hints.sql
@@ -29,7 +29,7 @@ meta_tags AS (
       JSON_VALUE(payload, '$._almanac') AS metrics
     FROM
       `httparchive.pages.2022_06_01_*`
-    ),
+  ),
     UNNEST(JSON_QUERY_ARRAY(metrics, '$.meta-nodes.nodes')) meta_node
   WHERE
     JSON_VALUE(meta_node, '$.http-equiv') IS NOT NULL

diff --git a/sql/2022/privacy/number_of_websites_with_floc_opt_out.sql b/sql/2022/privacy/number_of_websites_with_floc_opt_out.sql
@@ -29,7 +29,7 @@ meta_tags AS (
       JSON_VALUE(payload, '$._almanac') AS metrics
     FROM
       `httparchive.pages.2022_06_01_*`
-    ),
+  ),
     UNNEST(JSON_QUERY_ARRAY(metrics, '$.meta-nodes.nodes')) meta_node
   WHERE
     JSON_VALUE(meta_node, '$.http-equiv') IS NOT NULL

diff --git a/sql/2022/privacy/number_of_websites_with_mediadevices_blink_usage.sql b/sql/2022/privacy/number_of_websites_with_mediadevices_blink_usage.sql
@@ -6,7 +6,7 @@ SELECT DISTINCT
   feature,
   num_urls,
   total_urls,
-  pct_urls AS pct_urls
+  pct_urls
 FROM
   `httparchive.blink_features.usage`
 WHERE

diff --git a/sql/2022/privacy/number_of_websites_with_nb_trackers.sql b/sql/2022/privacy/number_of_websites_with_nb_trackers.sql
@@ -50,7 +50,7 @@ FROM (
   GROUP BY
     client,
     page
-  )
+)
 JOIN
   totals
 USING (client)
@@ -92,7 +92,7 @@ FROM (
   GROUP BY
     client,
     page
-  )
+)
 JOIN
   totals
 USING (client)

diff --git a/sql/2022/privacy/number_of_websites_with_origin_trial_from_token.sql b/sql/2022/privacy/number_of_websites_with_origin_trial_from_token.sql
@@ -183,7 +183,7 @@ meta_tags AS (
       JSON_VALUE(payload, '$._almanac') AS metrics
     FROM
       `httparchive.pages.2022_06_01_*`
-    ),
+  ),
     UNNEST(JSON_QUERY_ARRAY(metrics, '$.meta-nodes.nodes')) meta_node
   WHERE
     JSON_VALUE(meta_node, '$.http-equiv') IS NOT NULL

diff --git a/sql/2022/privacy/number_of_websites_with_permissions_policy.sql b/sql/2022/privacy/number_of_websites_with_permissions_policy.sql
@@ -40,7 +40,7 @@ meta_tags AS (
       JSON_VALUE(payload, '$._almanac') AS metrics
     FROM
       `httparchive.pages.2022_06_01_*`
-    ),
+  ),
     UNNEST(JSON_QUERY_ARRAY(metrics, '$.meta-nodes.nodes')) meta_node
   WHERE
     JSON_VALUE(meta_node, '$.http-equiv') IS NOT NULL
@@ -59,22 +59,28 @@ FROM (
   SELECT
     client,
     rank_grouping,
-    COUNT(DISTINCT IF(
-      header_name = 'feature-policy' OR
-      tag_name = 'feature-policy',
-      page, NULL)
+    COUNT(
+      DISTINCT IF(
+        header_name = 'feature-policy' OR
+        tag_name = 'feature-policy',
+        page, NULL
+      )
     ) AS number_of_websites_with_feature_policy,
-    COUNT(DISTINCT IF(
-      header_name = 'permissions-policy' OR
-      tag_name = 'permissions-policy',
-      page, NULL)
+    COUNT(
+      DISTINCT IF(
+        header_name = 'permissions-policy' OR
+        tag_name = 'permissions-policy',
+        page, NULL
+      )
     ) AS number_of_websites_with_permissions_policy,
-    COUNT(DISTINCT IF(
-      header_name = 'feature-policy' OR
-      tag_name = 'feature-policy' OR
-      header_name = 'permissions-policy' OR
-      tag_name = 'permissions-policy',
-      page, NULL)
+    COUNT(
+      DISTINCT IF(
+        header_name = 'feature-policy' OR
+        tag_name = 'feature-policy' OR
+        header_name = 'permissions-policy' OR
+        tag_name = 'permissions-policy',
+        page, NULL
+      )
     ) AS number_of_websites_with_any_policy,
     COUNT(DISTINCT page) AS number_of_websites
   FROM

diff --git a/sql/2022/privacy/number_of_websites_with_permissions_policy_blink_usage.sql b/sql/2022/privacy/number_of_websites_with_permissions_policy_blink_usage.sql
@@ -6,7 +6,7 @@ SELECT DISTINCT
   feature,
   num_urls,
   total_urls,
-  pct_urls AS pct_urls
+  pct_urls
 FROM
   `httparchive.blink_features.usage`
 WHERE

diff --git a/sql/2022/privacy/number_of_websites_with_referrerpolicy.sql b/sql/2022/privacy/number_of_websites_with_referrerpolicy.sql
@@ -49,28 +49,36 @@ FROM (
   SELECT
     client,
     COUNT(DISTINCT IF(
-        entire_document_policy_meta IS NOT NULL,
-        page, NULL)) AS number_of_websites_with_entire_document_policy_meta,
+      entire_document_policy_meta IS NOT NULL,
+      page, NULL
+    )) AS number_of_websites_with_entire_document_policy_meta,
     COUNT(DISTINCT IF(
-        entire_document_policy_header IS NOT NULL,
-        page, NULL)) AS number_of_websites_with_entire_document_policy_header,
-    COUNT(DISTINCT IF(
-      entire_document_policy_meta IS NOT NULL OR
       entire_document_policy_header IS NOT NULL,
-      page, NULL)
+      page, NULL
+    )) AS number_of_websites_with_entire_document_policy_header,
+    COUNT(
+      DISTINCT IF(
+        entire_document_policy_meta IS NOT NULL OR
+        entire_document_policy_header IS NOT NULL,
+        page, NULL
+      )
     ) AS number_of_websites_with_entire_document_policy,
     COUNT(DISTINCT IF(
-        ARRAY_LENGTH(individual_requests) > 0,
-        page, NULL)) AS number_of_websites_with_any_individual_requests,
+      ARRAY_LENGTH(individual_requests) > 0,
+      page, NULL
+    )) AS number_of_websites_with_any_individual_requests,
     COUNT(DISTINCT IF(
-        ARRAY_LENGTH(link_relations) > 0,
-        page, NULL)) AS number_of_websites_with_any_link_relations,
-    COUNT(DISTINCT IF(
-      entire_document_policy_meta IS NOT NULL OR
-      entire_document_policy_header IS NOT NULL OR
-      ARRAY_LENGTH(individual_requests) > 0 OR
       ARRAY_LENGTH(link_relations) > 0,
-      page, NULL)
+      page, NULL
+    )) AS number_of_websites_with_any_link_relations,
+    COUNT(
+      DISTINCT IF(
+        entire_document_policy_meta IS NOT NULL OR
+        entire_document_policy_header IS NOT NULL OR
+        ARRAY_LENGTH(individual_requests) > 0 OR
+        ARRAY_LENGTH(link_relations) > 0,
+        page, NULL
+      )
     ) AS number_of_websites_with_any_referrer_policy,
     COUNT(DISTINCT page) AS number_of_websites
   FROM