Variable "DBT_PACKAGE_GA4__DATASETS" and more than one dataset might be used now.

Author: Sergey Sergey

dbt_project.yml

@@ -25,12 +25,13 @@ clean-targets:         # directories to be removed by `dbt clean`
   - "dbt_packages"
 
 # vars:
-#   DBT_PACKAGE_GA4__PROJECT: "holybricks"
-#   DBT_PACKAGE_GA4__DATASET: "analytics_202294062"
+#   DBT_PACKAGE_GA4__PROJECT: "commonwealth-356813"
+#   DBT_PACKAGE_GA4__DATASET: "analytics_375681484"
+#   DBT_PACKAGE_GA4__DATASETS: "analytics_375681484,analytics_247024976"
 #   DBT_PACKAGE_GA4__INTERVAL: 14
 #   DBT_PACKAGE_GA4__INTERVAL_INCREMENTAL: 4
-#   DBT_PACKAGE_GA4__STREAM_ID: ['2059071580', '2771133919']
-#   DBT_PACKAGE_GA4__TIME_ZONE: 'America/Los_Angeles'
+#   DBT_PACKAGE_GA4__STREAM_ID: "['2059071580', '5200866205', '5200058214']"
+#   DBT_PACKAGE_GA4__TIME_ZONE: "America/Los_Angeles"
 
 
 # Configuring models

models/integration/src_ga4__events.sql

@@ -8,13 +8,42 @@
 
 
 WITH raw AS (
+
+{%- set datasets = env_var('DBT_PACKAGE_GA4__DATASETS', '') -%}
+
+{%- if datasets|length == 0 %}
+
     SELECT
         events._TABLE_SUFFIX AS TABLE_SUFFIX,
         events.*
     FROM
         {{ source('dbt_package_ga4', 'events') }} AS events
     WHERE
         events.stream_id IN UNNEST({{ env_var('DBT_PACKAGE_GA4__STREAM_ID') }})
+
+{%- else %}
+
+    {%- set datasets = datasets.split(',') -%}
+
+    {%- for dataset in datasets %}
+
+    SELECT
+        events._TABLE_SUFFIX AS TABLE_SUFFIX,
+        events.*
+    FROM
+        `{{ env_var('DBT_PACKAGE_GA4__PROJECT') }}`.`{{ dataset }}`.`events_*` AS events
+    WHERE
+        events.stream_id IN UNNEST({{ env_var('DBT_PACKAGE_GA4__STREAM_ID') }})
+
+        {%- if not loop.last %}
+
+    UNION ALL
+
+        {%- endif %}
+
+    {%- endfor %}
+
+{%- endif %}
 )
 
 SELECT * FROM raw