Merge pull request #22 from fishtown-analytics/fix/ad-creative-url-pa…

…rsing

Fix: ad creative url parsing--child links

dbt_project.yml

@@ -11,10 +11,13 @@ data-paths: ["data"]       # load CSVs from this directory with `dbt seed`
 models:
   facebook_ads:
       vars:
+
         etl:                                   #stitch or fivetran
         ads_table:                             #table
         ad_creatives_table:                    #table
         adsets_table:                          #table
         campaigns_table:                       #table
         ads_insights_table:                    #table
-        url_tag_table:                         #only for fivetran
+        ad_creatives__child_links_table:       #table -- disable if on snowflake
+
+        url_tag_table:                         #only for fivetran

macros/nested_field.sql

@@ -0,0 +1,26 @@
+{% macro nested_field(field, subfields) %}
+
+    {{ adapter_macro('facebook_ads.nested_field', field, subfields) }}
+
+{% endmacro %}
+
+
+{% macro default__nested_field(field, subfields) %}
+
+    {{field}}__{{ subfields|join('__') }}
+
+{% endmacro %}
+
+
+{% macro bigquery__nested_field(field, subfields) %}
+
+    {{field}}.{{ subfields|join('.') }}
+
+{% endmacro %}
+
+
+{% macro snowflake__nested_field(field, subfields) %}
+
+    {{field ~ "['" ~ subfields|join("']['") ~ "']" }} 
+
+{% endmacro %}

macros/stitch/base/stitch_fb_ad_creatives.sql

@@ -8,81 +8,51 @@
 {% macro default__stitch_fb_ad_creatives() %}
 
 with base as (
-
-    select
 
-        id as creative_id,
-        lower(nullif(url_tags, '')) as url_tags,
-        lower(coalesce(
-          nullif(object_story_spec__link_data__call_to_action__value__link, ''),
-          nullif(object_story_spec__video_data__call_to_action__value__link, ''),
-          nullif(object_story_spec__link_data__link, '')
-        )) as url
+    select * from {{ var('ad_creatives_table') }}
 
-    from
-    {{ var('ad_creatives_table') }}
-
-), splits as (
+),
 
-    select
+child_links as (
 
-        creative_id,
-        url,
-        {{ dbt_utils.split_part('url', "'?'", 1) }} as base_url,
-        --this is a strange thing to have to do but it's because sometimes 
-        --the URL exists on the story object and we wouldn't get the appropriate 
-        --UTM params here otherwise
-        coalesce(url_tags, {{ dbt_utils.split_part('url', "'?'", 2) }} ) as url_tags
+    select * from {{ ref('fb_ad_creatives__child_links') }}
 
-    from base
-
-)
-
-select
+),
 
-    *,
-    {{ dbt_utils.get_url_host('url') }} as url_host,
-    '/' || {{dbt_utils.get_url_path('url') }} as url_path,
-
-    {{ facebook_ads.get_url_parameter() }}
+links_joined as (
 
-from splits
-
-{% endmacro %}
-
-
-{% macro snowflake__stitch_fb_ad_creatives() %}
-
-with base as (
-
     select
-
+    
         id as creative_id,
+
         lower(coalesce(
-          nullif(object_story_spec['link_data']['call_to_action']['value']['link']::varchar, ''),
-          nullif(object_story_spec['video_data']['call_to_action']['value']['link']::varchar, ''),
-          nullif(object_story_spec['link_data']['link']::varchar, '')
-      )) as url,
+          nullif(child_link, ''),
+          nullif({{ facebook_ads.nested_field('base.object_story_spec', ['link_data', 'call_to_action', 'value', 'link']) }}, ''),
+          nullif({{ facebook_ads.nested_field('base.object_story_spec', ['video_data', 'call_to_action', 'value', 'link']) }}, ''),
+          nullif({{ facebook_ads.nested_field('base.object_story_spec', ['link_data', 'link']) }}, '')
+        )) as url,
 
-      url_tags
-
-    from {{ var('ad_creatives_table') }}
-
+        lower(coalesce(
+            nullif(url_tags, {{ dbt_utils.split_part('url', "'?'", 2) }}), '')
+        ) as url_tags
+
+    from base
+    left join child_links 
+        on base.id = child_links.creative_id
+
 ),
 
 parsed as (
 
     select
 
-        creative_id,
-        url,
+        links_joined.*,
         {{ dbt_utils.split_part('url', "'?'", 1) }} as base_url,
         {{ dbt_utils.get_url_host('url') }} as url_host,
-        '/' || {{dbt_utils.get_url_path('url') }} as url_path,
-
+        {{ dbt_utils.concat(["'/'", dbt_utils.get_url_path('url')]) }} as url_path,
         {{ facebook_ads.get_url_parameter() }}
 
-    from base 
+    from links_joined 
 
 )
 

macros/stitch/base/stitch_fb_ad_creatives__child_links.sql

@@ -0,0 +1,101 @@
+-- There are cases where ads can have multiple links attached to them.
+-- This model was created to extract all the child links and return the first 
+-- instance where a URL contains utm parameters.
+
+
+{% macro stitch_fb_ad_creatives__child_links() %}
+
+    {{ adapter_macro('facebook_ads.stitch_fb_ad_creatives__child_links') }}
+
+{% endmacro %}
+
+{% macro default__stitch_fb_ad_creatives__child_links() %}
+
+
+with base as (
+
+    select * from {{ var('ad_creatives__child_links_table') }}
+
+),
+
+child_attachment_links as (
+
+    select
+
+        _sdc_source_key_id as creative_id,
+        _sdc_batched_at,
+        link as child_link
+
+    from base
+    where child_link ilike '%utm%'
+
+),
+
+aggregated as (
+
+    select distinct
+
+        creative_id,
+        first_value(child_link) over (
+            partition by creative_id
+            order by _sdc_batched_at
+            rows between unbounded preceding and unbounded following
+        ) as child_link
+
+    from child_attachment_links
+
+)
+
+select * from aggregated
+
+{% endmacro %}
+
+
+{% macro snowflake__stitch_fb_ad_creatives__child_links() %}
+
+
+{% set fields = [
+
+    'object_story_spec',
+    'child_link',
+
+]%}
+
+with base as (
+
+    select * from {{ var('ad_creatives_table') }}
+
+),
+
+child_attachment_links as (
+
+    select
+        *,
+        attachments.value:link::varchar as child_link
+
+    from base, 
+        lateral flatten (input => object_story_spec:link_data:child_attachments) attachments
+    where child_link ilike '%utm%'
+
+),
+
+aggregated as (
+
+    select distinct
+        id as creative_id,
+
+        {% for field in fields %}
+        first_value({{ field }}) over (partition by id 
+            order by _sdc_batched_at 
+            rows between unbounded preceding and unbounded following) 
+            as {{ field }}
+        {% if not loop.last%} , {% endif %}
+    {% endfor %}
+
+    from child_attachment_links
+
+)
+
+select * from aggregated
+
+{% endmacro %}

models/router/schema.yml

@@ -6,11 +6,13 @@ models:
     tests:
     - not_null
     - unique
-- name: fb_ads_adsets
+- name: fb_ad_creatives__child_links
   columns:
-  - name: adset_id
+  - name: creative_id
     tests:
-    - not_null
+    - relationships:
+        field: creative_id
+        to: ref('fb_ad_creatives')
 - name: fb_ads
   columns:
   - name: creative_id

models/router/stitch/fb_ad_creatives__child_links.sql

@@ -0,0 +1,7 @@
+{{
+    config(
+        enabled = var('etl') == 'stitch'
+    )
+}}
+
+{{ stitch_fb_ad_creatives__child_links() }}