From cf893f97a9bb20d13cdd949f284f302b67715754 Mon Sep 17 00:00:00 2001
From: Max Ostapenko <max@httparchive.org>
Date: Wed, 20 Aug 2025 23:53:13 +0000
Subject: [PATCH 1/3] fix month

---
 .../output/reports/reports_html_elements_popularity.js | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/definitions/output/reports/reports_html_elements_popularity.js b/definitions/output/reports/reports_html_elements_popularity.js
index fe5f4f94..fec88a01 100644
--- a/definitions/output/reports/reports_html_elements_popularity.js
+++ b/definitions/output/reports/reports_html_elements_popularity.js
@@ -1,5 +1,3 @@
-const pastMonth = constants.fnPastMonth(constants.currentMonth)
-
 publish('html_elements_popularity', {
   schema: 'reports',
   type: 'incremental',
@@ -18,7 +16,7 @@ try {
 ''';
 
 DELETE FROM ${ctx.self()}
-WHERE date = '${pastMonth}';
+WHERE date = '${constants.currentMonth}';
 `).query(ctx => `
 WITH pages_data AS (
   SELECT
@@ -29,7 +27,7 @@ WITH pages_data AS (
     custom_metrics.element_count
   FROM ${ctx.ref('crawl', 'pages')}
   WHERE
-    date = '${pastMonth}' ${constants.devRankFilter}
+    date = '${constants.currentMonth}' ${constants.devRankFilter}
 ),
 
 totals AS (
@@ -69,9 +67,9 @@ SELECT
       "destination": "cloud_storage",
       "config": {
         "bucket": "${constants.bucket}",
-        "name": "${constants.storagePath}${pastMonth.replaceAll('-', '_')}/htmlElementPopularity.json"
+        "name": "${constants.storagePath}${constants.currentMonth.replaceAll('-', '_')}/htmlElementPopularity.json"
       },
-      "query": "SELECT * EXCEPT(date) FROM ${ctx.self()} WHERE date = '${pastMonth}'"
+      "query": "SELECT * EXCEPT(date) FROM ${ctx.self()} WHERE date = '${constants.currentMonth}'"
     }'''
   );
 `)

From cdad0da7592ea70316bfe17e6038b6a266e67795 Mon Sep 17 00:00:00 2001
From: Max Ostapenko <max@httparchive.org>
Date: Wed, 20 Aug 2025 23:53:33 +0000
Subject: [PATCH 2/3] fix column order

---
 definitions/output/crawl/parsed_css.js | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/definitions/output/crawl/parsed_css.js b/definitions/output/crawl/parsed_css.js
index 648c2fb8..ec3abd1e 100644
--- a/definitions/output/crawl/parsed_css.js
+++ b/definitions/output/crawl/parsed_css.js
@@ -17,8 +17,8 @@ WHERE date = '${constants.currentMonth}'
 `).query(ctx => `
 SELECT
   * EXCEPT(css),
-  SAFE.PARSE_JSON(css, wide_number_mode=>'round') AS css,
-  NULL AS css_backup
+  NULL AS css_backup,
+  SAFE.PARSE_JSON(css, wide_number_mode=>'round') AS css
 FROM ${ctx.ref('crawl_staging', 'parsed_css')}
 WHERE date = '${constants.currentMonth}'
   AND client = 'desktop'
@@ -31,8 +31,8 @@ WHERE date = '${constants.currentMonth}'
 INSERT INTO ${ctx.self()}
 SELECT
   * EXCEPT(css),
-  SAFE.PARSE_JSON(css, wide_number_mode=>'round') AS css,
-  NULL AS css_backup
+  NULL AS css_backup,
+  SAFE.PARSE_JSON(css, wide_number_mode=>'round') AS css
 FROM ${ctx.ref('crawl_staging', 'parsed_css')}
 WHERE date = '${constants.currentMonth}'
   AND client = 'mobile'

From 36607a589f55baf9fbcc2fa12092ea35a578a3fd Mon Sep 17 00:00:00 2001
From: Max Ostapenko <max@httparchive.org>
Date: Wed, 20 Aug 2025 23:53:53 +0000
Subject: [PATCH 3/3] pages_per_client assert

---
 definitions/output/crawl/pages.js | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/definitions/output/crawl/pages.js b/definitions/output/crawl/pages.js
index 91fd3388..197789d2 100644
--- a/definitions/output/crawl/pages.js
+++ b/definitions/output/crawl/pages.js
@@ -30,6 +30,20 @@ ORDER BY cnt_pages DESC
 HAVING cnt_pages > 200
   `)
 
+assert('pages_per_client')
+  .tags(['crawl_complete'])
+  .query(ctx => `
+SELECT
+  client,
+  COUNT(DISTINCT page) AS cnt_pages
+FROM ${ctx.ref('crawl_staging', 'pages')}
+WHERE
+  date = '${constants.currentMonth}'
+GROUP BY
+  client
+HAVING cnt_pages < 20000000
+  `)
+
 publish('pages', {
   type: 'incremental',
   protected: true,