Dedup transactions (#277)

* initial transaction dedup concept

* first test

* working copy

* performance optimization

* docs

* code cleanup

models/staging/recommended_events/README.md

@@ -27,4 +27,14 @@ models:
           +enabled: true
 ```
 
-Not all recommended events have been implemented. If you need a specific event, please consider creating a pull request with the model that you need in the [dbt-ga4 GitHub repository](https://github.com/Velir/dbt-ga4).
+Not all recommended events have been implemented. If you need a specific event, please consider creating a pull request with the model that you need in the [dbt-ga4 GitHub repository](https://github.com/Velir/dbt-ga4).
+
+## Purchase Event Transaction Deduplication
+
+The `stg_ga4__event_purchase_deduplicated` model builds on the `sgt_ga4__event_purchase` model. It is disabled by default and thus needs to be enabled along with the `stg_ga4__event_purchase` model.
+
+The model only processes purchase events that fall within the window as defined by `static_incremental_days` and can only reliably be expected to deduplicate purchase events occurring in the same day.
+
+The model provides a highly-performant, minimum-viable product for this feature returning only data from the first purchase event with a matching `transaction_id` within the processing window.
+
+You are encouraged to copy this model to your project and customize it there should this MVP be insufficient for your needs.

models/staging/recommended_events/stg_ga4__event_purchase_deduplicated.sql

@@ -0,0 +1,27 @@
+{% if not flags.FULL_REFRESH %}
+    {% set partitions_to_query = ['current_date'] %}
+    {% for i in range(var('static_incremental_days', 1)) %}
+        {% set partitions_to_query = partitions_to_query.append('date_sub(current_date, interval ' + (i+1)|string + ' day)') %}
+    {% endfor %}
+{% endif %}
+
+{{
+  config(
+      enabled = false,
+  )
+}}
+with purch as (
+    select
+        *
+    from {{ref('stg_ga4__event_purchase')}}
+    {% if not flags.FULL_REFRESH %}
+        where event_date_dt in ({{ partitions_to_query | join(',') }})
+    {% endif %}
+    qualify row_number() over(
+        partition by transaction_id
+            order by event_timestamp
+    ) = 1
+)
+select
+    *
+from purch