Merge pull request #2 from fishtown-analytics/feature/redshift

Feature/redshift
dbt-labs · Oct 26, 2018 · 9304fd6 · 9304fd6
2 parents e384ffd + c56f4f5
commit 9304fd6
Show file tree

Hide file tree

Showing 3 changed files with 65 additions and 5 deletions.
diff --git a/macros/cross-adapter-modeling/base/segment_web_page_views.sql b/macros/cross-adapter-modeling/base/segment_web_page_views.sql
@@ -7,7 +7,51 @@
 
 {% macro default__segment_web_page_views() %}
 
-{{ exceptions.raise_compiler_error("macro segment_web_page_views not implemented for this adapter") }}
+with source as (
+
+    select * from {{var('segment_page_views_table')}}
+
+),
+
+renamed as (
+
+    select
+
+        id as page_view_id,
+        anonymous_id,
+        user_id,
+
+        received_at as received_at_tstamp,
+        sent_at as sent_at_tstamp,        
+        timestamp as tstamp,
+
+        url as page_url,
+        split_part(split_part(split_part(url, '//', 2),'/', 1),'?',1)::varchar 
+            as page_url_host,
+        path as page_url_path,
+        title as page_title,
+        search as page_url_query,
+
+        referrer,
+        ltrim(split_part(split_part(referrer, '.com', 1), '//',2),'www.')::varchar 
+            as referrer_host,
+
+        context_campaign_source as utm_source,
+        context_campaign_medium as utm_medium,
+        context_campaign_name as utm_campaign,
+        context_campaign_term as utm_term,
+        context_campaign_content as utm_content,
+        nullif(split_part(split_part(url, 'gclid=', 2),'&', 1)::varchar,'') 
+            as gclid,
+
+        context_ip as ip,        
+        context_user_agent as user_agent
+
+    from source
+
+)
+
+select * from renamed
 
 {% endmacro %}
 

diff --git a/macros/cross-adapter-modeling/sessionization/segment_web_page_views__sessionized.sql b/macros/cross-adapter-modeling/sessionization/segment_web_page_views__sessionized.sql
@@ -70,8 +70,7 @@ lagged as (
 
         lag(tstamp) over (
             partition by anonymous_id 
-            order by page_view_number 
-            rows between unbounded preceding and unbounded following
+            order by page_view_number
             ) as previous_tstamp
 
     from numbered

diff --git a/macros/cross-adapter-modeling/sessionization/segment_web_sessions__initial.sql b/macros/cross-adapter-modeling/sessionization/segment_web_sessions__initial.sql
@@ -29,7 +29,6 @@
     'utm_medium' : 'utm_medium',
     'utm_campaign' : 'utm_campaign',
     'utm_term' : 'utm_term',
-    'mkw_id' : 'mkw_id',
     'gclid' : 'gclid',
     'page_url' : 'first_page_url',
     'page_url_host' : 'first_page_url_host',
@@ -88,8 +87,26 @@ diffs as (
 
     from agg
 
+),
+
+tiers as (
+
+    select
+
+        *,
+
+        case
+            when duration_in_s between 0 and 9 then '0s to 9s'
+            when duration_in_s between 10 and 29 then '10s to 29s'
+            when duration_in_s between 30 and 59 then '30s to 59s'
+            when duration_in_s > 59 then '60s or more'
+            else null
+        end as duration_in_s_tier
+
+    from diffs
+
 )
 
-select * from diffs
+select * from tiers
 
 {% endmacro %}