Block bad dates from SQL (#4318)

* Block bad dates from SQL

* Approved exceptions

Author: tunetheweb

sql/.sqlfluff

@@ -52,14 +52,13 @@ extended_capitalisation_policy = upper
 
 [sqlfluff:rules:convention.blocked_words]
 # Regex of blocked SQL that should not be used.
-# Can be overridden with `-- noqa: CV09` for those chapters using secondary pages
+# Can be overridden with `-- noqa: CV09` for those chapters using other dates
 # TABLESAMPLE - sometimes used for testing. Shouldn't be used in production as not random.
 # sample_data - sometimes used for testing. Shouldn't be used in production.
-# Block 2022_05_12 (contains secondary pages)
-# Block 2022_06_09 (contains secondary pages)
-# Block 2022_07_01 (probably forgot to update month to June for 2022)
-# Block 2021_06_01 (probably forgot to update month to July for 2021)
-blocked_regex = (TABLESAMPLE|sample_data|2022_?05_?12|2022_?06_?09|2022_?07_?01|2021_?06_?01)
+# The Web Almanac has run in different months due to timing or delays. This
+# means it's easy to use the wrong month by copying a previous query and only
+# updating the year and not the month. Block those months to warn about this.
+blocked_regex = (TABLESAMPLE|sample_data|2025-?08-?01|2025-?06-?01|2024-?08-?01|2024-?07-?01|2022-?08-?01|2022-?07-?01|2021-?06-?01|2021-?08-?01|2020-?07-?01)
 
 [sqlfluff:rules:convention.count_rows]
 prefer_count_0 = True

sql/2020/cms/core_web_vitals_distribution.sql

@@ -36,7 +36,7 @@ ON
   IF(device = 'desktop', 'desktop', 'mobile') = client
 WHERE
   # The CrUX 202008 dataset is not available until September 8.
-  date = '2020-07-01'
+  date = '2020-07-01' -- noqa: CV09
 GROUP BY
   client,
   cms

sql/2020/cms/core_web_vitals_passing.sql

@@ -61,7 +61,7 @@ ON
   IF(device = 'desktop', 'desktop', 'mobile') = client
 WHERE
   # The CrUX 202008 dataset is not available until September 8.
-  date = '2020-07-01'
+  date = '2020-07-01' -- noqa: CV09
 GROUP BY
   client,
   cms

sql/2021/media/video_source_types.sql

@@ -11,7 +11,7 @@ FROM
   UNNEST(REGEXP_EXTRACT_ALL(body, '(?i)(<video.*?</video>)')) AS video,
   UNNEST(REGEXP_EXTRACT_ALL(video, r'(?i)type\s*=\s*["\'](video/[^\'";?]*)')) AS video_type
 WHERE
-  date = '2021-08-01' AND
+  date = '2021-08-01' AND -- noqa: CV09
   firstHtml
 GROUP BY
   client,

sql/2021/media/video_tag_and_js_player.sql

@@ -43,7 +43,7 @@ FROM (
     FROM
       `httparchive.almanac.requests`
     WHERE
-      date = '2021-08-01' AND
+      date = '2021-08-01' AND -- noqa: CV09
       type = 'script'
     GROUP BY
       client,

sql/2022/accessibility/focus_outline_0.sql

@@ -59,7 +59,7 @@ FROM (
     `httparchive.almanac.parsed_css`,
     UNNEST(getFocusStylesOutline0(css)) AS sets_outline_0
   WHERE
-    date = '2022-07-01'
+    date = '2022-07-01' -- noqa: CV09
   GROUP BY
     client,
     page

sql/2022/accessibility/focus_visible.sql

@@ -73,7 +73,7 @@ FROM (
   LEFT JOIN
     UNNEST(getSelectorParts(css).pseudo_class) AS pseudo_class
   WHERE
-    date = '2022-07-01' AND
+    date = '2022-07-01' AND -- noqa: CV09
     # Limit the size of the CSS to avoid OOM crashes.
     LENGTH(css) < 0.1 * 1024 * 1024
   GROUP BY

sql/2022/accessibility/media_query_features.sql

@@ -50,7 +50,7 @@ FROM (
   LEFT JOIN
     UNNEST(getMediaQueryFeatures(css)) AS feature
   WHERE
-    date = '2022-07-01' AND
+    date = '2022-07-01' AND -- noqa: CV09
     feature IS NOT NULL
 )
 JOIN (

sql/2022/accessibility/units_properties.sql

@@ -117,7 +117,7 @@ FROM (
       `httparchive.almanac.parsed_css`,
       UNNEST(getPropertyUnits(css)) AS unit
     WHERE
-      date = '2022-07-01' AND
+      date = '2022-07-01' AND -- noqa: CV09
       # Limit the size of the CSS to avoid OOM crashes.
       LENGTH(css) < 0.1 * 1024 * 1024
   )

sql/2022/css/all_functions.sql

@@ -62,7 +62,7 @@ FROM (
     totals
   USING (client)
   WHERE
-    date = '2022-07-01'
+    date = '2022-07-01' -- noqa: CV09
   GROUP BY
     client,
     prop