Merge pull request #275 from kyleburke-meq/main

Snowflake: Add expression parameter to columns

integration_tests/models/plugins/snowflake/people_expression.sql

@@ -0,0 +1,4 @@
+SELECT
+ {{ dbt_utils.star(from=ref('people')) }},
+  split_part(email, '@', 2) as email_domain
+FROM {{ ref('people') }}

integration_tests/models/plugins/snowflake/snowflake_external.yml

@@ -173,4 +173,31 @@ sources:
                 - id
                 - first_name
                 - last_name
-                - email_alias
+                - email_alias
+
+        # test for column expression
+      - name: people_json_expression
+        external: *json-people
+        columns:
+          - name: id
+            data_type: int
+          - name: first_name
+            data_type: varchar(64)
+          - name: last_name
+            data_type: varchar(64)
+          - name: email
+            data_type: varchar(64)
+          - name: email_domain
+            data_type: varchar(64)
+            alias: EMAIL_DOMAIN
+            quote: true
+            expression: split_part(value:email::VARCHAR, '@', 2)
+        tests:
+          - dbt_utils.equality:
+              compare_model: ref('people_expression')
+              compare_columns:
+                - id
+                - first_name
+                - last_name
+                - email
+                - email_domain

macros/plugins/snowflake/create_external_table.sql

@@ -26,14 +26,25 @@
         {%- endfor -%}{%- endif -%}
         {%- if not infer_schema -%}
             {%- for column in columns %}
-                {%- set column_alias = column.alias if column.alias else column.name %}
-                {%- set column_alias_quoted = adapter.quote(column_alias) if column.quote else column_alias %}
                 {%- set column_quoted = adapter.quote(column.name) if column.quote else column.name %}
+                {%- set column_alias -%}
+                    {%- if 'alias' in column and column.quote -%}
+                        {{adapter.quote(column.alias)}}
+                    {%- elif 'alias' in column -%}
+                        {{column.alias}}
+                    {%- else -%}
+                        {{column_quoted}}
+                    {%- endif -%}
+                {%- endset %}
                 {%- set col_expression -%}
-                    {%- set col_id = 'value:c' ~ loop.index if is_csv else 'value:' ~ column_quoted -%}
-                    (case when is_null_value({{col_id}}) or lower({{col_id}}) = 'null' then null else {{col_id}} end)
+                    {%- if column.expression -%}
+                        {{column.expression}}
+                    {%- else -%}
+                        {%- set col_id = 'value:c' ~ loop.index if is_csv else 'value:' ~ column_alias -%}
+                        (case when is_null_value({{col_id}}) or lower({{col_id}}) = 'null' then null else {{col_id}} end)
+                    {%- endif -%}
                 {%- endset %}
-                {{column_alias_quoted}} {{column.data_type}} as ({{col_expression}}::{{column.data_type}})
+                {{column_alias}} {{column.data_type}} as ({{col_expression}}::{{column.data_type}})
                 {{- ',' if not loop.last -}}
             {% endfor %}
         {% else %}

sample_sources/snowflake.yml

@@ -32,6 +32,18 @@ sources:
           - name: etl_tstamp
             data_type: timestamp
             description: "Timestamp event began ETL"
+          - name: etl timestamp
+            # Use double-quoted identifiers for name and identifier
+            quote: true
+            # Specifying alias lets us rename etl timestamp to "etl_timestamp"
+            alias: etl_timestamp
+            data_type: timestamp
+            description: "Timestamp event began ETL with a double quoted identifier"
+          - name: etl_date
+            data_type: date
+            description: "Date event began ETL"
+            # Expressions can manipulate the variant value prior to casting to data_type.
+            expression: TRY_TO_DATE(VALUE:etl_tstamp::VARCHAR, 'YYYYMMDD')
           - name: contexts
             data_type: variant
             description: "Contexts attached to event by Tracker"