Rework filter bots macro

snowplow · Oct 17, 2024 · 16be271 · 16be271
1 parent 1223a5f
commit 16be271
Show file tree

Hide file tree

Showing 5 changed files with 16 additions and 21 deletions.
diff --git a/integration_tests/models/source/bigquery/snowplow_unified_events_stg.sql b/integration_tests/models/source/bigquery/snowplow_unified_events_stg.sql
@@ -494,7 +494,7 @@ You may obtain a copy of the Snowplow Personal and Academic License Version 1.0
       select as struct JSON_EXTRACT_scalar(json_array,'$.category') as category,
                     JSON_EXTRACT_scalar(json_array,'$.primary_impact') as primary_impact,
                     JSON_EXTRACT_scalar(json_array,'$.reason') as reason,
-                    JSON_EXTRACT_scalar(json_array,'$.spider_or_robot') as spider_or_robot
+                    cast(JSON_EXTRACT_scalar(json_array,'$.spider_or_robot') as boolean) as spider_or_robot
 
       from unnest(contexts_com_iab_snowplow_spiders_and_robots_1_0_0) as json_array
       ) as contexts_com_iab_snowplow_spiders_and_robots_1_0_0

diff --git a/macros/filter_bots.sql b/macros/filter_bots.sql
@@ -10,17 +10,32 @@ You may obtain a copy of the Snowplow Personal and Academic License Version 1.0
 {%- endmacro -%}
 
 {% macro default__filter_bots(table_alias = none) %}
+  {% if var('snowplow__enable_iab', false) %}
+    {# additional logic in case the result is null due to server anonymization #}
+    and coalesce(iab__spider_or_robot, False ) = False
+  {% endif %}
   and {% if table_alias %}{{table_alias~'.'}}{% endif %}useragent not similar to '%(bot|crawl|slurp|spider|archiv|spinn|sniff|seo|audit|survey|pingdom|worm|capture|(browser|screen)shots|analyz|index|thumb|check|facebook|PingdomBot|PhantomJS|YandexBot|Twitterbot|a_archiver|facebookexternalhit|Bingbot|BingPreview|Googlebot|Baiduspider|360(Spider|User-agent)|semalt)%'
 {% endmacro %}
 
 {% macro bigquery__filter_bots(table_alias = none) %}
+  {% if var('snowplow__enable_iab', false) %}
+    and coalesce(iab__spider_or_robot, False ) = False
+  {% endif %}
   and not regexp_contains({% if table_alias %}{{table_alias~'.'}}{% endif %}useragent, '(bot|crawl|slurp|spider|archiv|spinn|sniff|seo|audit|survey|pingdom|worm|capture|(browser|screen)shots|analyz|index|thumb|check|facebook|PingdomBot|PhantomJS|YandexBot|Twitterbot|a_archiver|facebookexternalhit|Bingbot|BingPreview|Googlebot|Baiduspider|360(Spider|User-agent)|semalt)')
 {% endmacro %}
 
 {% macro spark__filter_bots(table_alias = none) %}
+  {% if var('snowplow__enable_iab', false) %}
+   {# had to add different syntax as the coalesce based one resulted in a Spark error #}
+    (not iab__spider_or_robot = True 
+     or iab__spider_or_robot is null)
+  {% endif %}
   and not rlike({% if table_alias %}{{table_alias~'.'}}{% endif %}useragent, '.*(bot|crawl|slurp|spider|archiv|spinn|sniff|seo|audit|survey|pingdom|worm|capture|(browser|screen)shots|analyz|index|thumb|check|facebook|PingdomBot|PhantomJS|YandexBot|Twitterbot|a_archiver|facebookexternalhit|Bingbot|BingPreview|Googlebot|Baiduspider|360(Spider|User-agent)|semalt).*')
 {% endmacro %}
 
 {% macro snowflake__filter_bots(table_alias = none) %}
+  {% if var('snowplow__enable_iab', false) %}
+    and coalesce(iab__spider_or_robot, False ) = False
+  {% endif %}
   and not rlike({% if table_alias %}{{table_alias~'.'}}{% endif %}useragent, '.*(bot|crawl|slurp|spider|archiv|spinn|sniff|seo|audit|survey|pingdom|worm|capture|(browser|screen)shots|analyz|index|thumb|check|facebook|PingdomBot|PhantomJS|YandexBot|Twitterbot|a_archiver|facebookexternalhit|Bingbot|BingPreview|Googlebot|Baiduspider|360(Spider|User-agent)|semalt).*')
 {% endmacro %}
diff --git a/...l_modules/core_web_vitals/scratch/bigquery/snowplow_unified_web_vital_events_this_run.sql b/...l_modules/core_web_vitals/scratch/bigquery/snowplow_unified_web_vital_events_this_run.sql
@@ -70,14 +70,6 @@ with prep as (
 
   -- exclude bot traffic
 
-  {% if var('snowplow__enable_iab', false) %}
-    and not {{ snowplow_utils.get_field(column_name = 'contexts_com_iab_snowplow_spiders_and_robots_1_0_0',
-                              field_name = 'spider_or_robot',
-                              table_alias = 'e',
-                              type = 'boolean',
-                              array_index = 0)}} = True
-  {% endif %}
-
   {{ snowplow_unified.filter_bots() }}
 
 )

diff --git a/...modules/core_web_vitals/scratch/databricks/snowplow_unified_web_vital_events_this_run.sql b/...modules/core_web_vitals/scratch/databricks/snowplow_unified_web_vital_events_this_run.sql
@@ -70,14 +70,6 @@ with prep as (
 
   -- exclude bot traffic
 
-  {% if var('snowplow__enable_iab', false) %}
-    and not {{ snowplow_utils.get_field(column_name = 'contexts_com_iab_snowplow_spiders_and_robots_1',
-                              field_name = 'spider_or_robot',
-                              table_alias = 'e',
-                              type = 'boolean',
-                              array_index = 0)}} = True
-  {% endif %}
-
   {{ snowplow_unified.filter_bots() }}
 
 )

diff --git a/...al_modules/core_web_vitals/scratch/default/snowplow_unified_web_vital_events_this_run.sql b/...al_modules/core_web_vitals/scratch/default/snowplow_unified_web_vital_events_this_run.sql
@@ -69,10 +69,6 @@ with prep as (
 
   -- exclude bot traffic
 
-  {% if var('snowplow__enable_iab', false) %}
-    and not e.iab__spider_or_robot = True
-  {% endif %}
-
   {{ snowplow_unified.filter_bots() }}
 
 )