Implement review

Author: judahrand

README.md

@@ -706,15 +706,15 @@ This macro returns the sql required to build a date spine. The spine will includ
 }}
 ```
 
-#### dedupe ([source](macros/sql/dedupe.sql))
-This macro returns the sql required to remove deduplicate rows from a model or source.
+#### deduplicate ([source](macros/sql/deduplicate.sql))
+This macro returns the sql required to remove duplicate rows from a model or source.
 
 **Usage:**
 
 ```
-{{ dbt_utils.dedupe(
-    source('my_source', 'my_table'),
-    "user_id, cast(timestamp as day)",
+{{ dbt_utils.deduplicate(
+    relation=source('my_source', 'my_table'),
+    group_by="user_id, cast(timestamp as day)",
     order_by="timestamp desc"
    )
 }}

integration_tests/data/sql/data_dedupe.csv integration_tests/data/sql/data_deduplicate.csv

@@ -146,4 +146,4 @@ models:
   - name: test_dedupe
     tests:
       - dbt_utils.equality:
-          compare_model: ref('data_dedupe_expected')
+          compare_model: ref('data_deduplicate_expected')

integration_tests/data/sql/data_dedupe_expected.csv integration_tests/data/sql/data_deduplicate_expected.csv

@@ -0,0 +1,7 @@
+with deduped as (
+
+    {{ dbt_utils.dededuplicate(ref('data_deduplicate'), group_by='user_id', order_by='version desc') | indent }}
+
+)
+
+select * from deduped

integration_tests/models/sql/schema.yml

@@ -0,0 +1,46 @@
+{%- macro deduplicate(relation, group_by, order_by=none) -%}
+    {{ return(adapter.dispatch('deduplicate', 'dbt_utils')(relation, group_by, order_by=order_by)) }}
+{% endmacro %}
+
+{%- macro default__deduplicate(relation, group_by, order_by=none) -%}
+
+    select
+        {{ dbt_utils.star(relation, relation_alias='deduped') | indent }}
+    from (
+        select
+            _inner.*,
+            row_number() over (
+                partition by {{ group_by }}
+                {% if order_by is not none -%}
+                order by {{ order_by }}
+                {%- endif %}
+            ) as rn
+        from {{ relation }} as _inner
+    ) as deduped
+    where deduped.rn = 1
+
+{%- endmacro -%}
+
+{#
+--  It is more performant to deduplicate using `array_agg` with a limit
+--  clause in BigQuery:
+--  https://github.com/dbt-labs/dbt-utils/issues/335#issuecomment-788157572
+#}
+{%- macro bigquery__deduplicate(relation, group_by, order_by=none) -%}
+
+    select
+        {{ dbt_utils.star(relation, relation_alias='deduped') | indent }}
+    from (
+        select
+            array_agg (
+                original
+                {% if order_by is not none-%}
+                order by {{ order_by }}
+                {%- endif %}
+                limit 1
+            )[offset(0)] as deduped
+        from {{ relation }} as original
+        group by {{ group_by }}
+    )
+
+{%- endmacro -%}