From 03036dba1e33f06078feeb01f9b0008790160f0d Mon Sep 17 00:00:00 2001
From: Simo Tumelius <simo.tumelius@gmail.com>
Date: Tue, 17 May 2022 18:46:35 +0300
Subject: [PATCH] Improve profiler performance in Snowflake by reducing table
 scans using a single CTE from which column profiles are calculated (#48)

Co-authored-by: Simo Tumelius <simo@datamie.fi>
---
 .gitignore             |  1 +
 macros/get_profile.sql | 10 ++++++++--
 2 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/.gitignore b/.gitignore
index 4811d0b..8f3a9dc 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,5 +1,6 @@
 target/
 dbt_modules/
+dbt_packages/
 logs/
 
 venv/
diff --git a/macros/get_profile.sql b/macros/get_profile.sql
index 2b31fcc..ef46a3b 100644
--- a/macros/get_profile.sql
+++ b/macros/get_profile.sql
@@ -50,7 +50,13 @@
   {{ log("Column data types: " ~ data_type_map, info=False) }}
 
   {% set profile_sql %}
-    with column_profiles as (
+    with source_data as (
+      select
+        *
+      from {{ relation }}
+    ),
+
+    column_profiles as (
       {% for column_name in profile_column_names %}
         {% set data_type = data_type_map.get(column_name.lower(), "") %}
         select 
@@ -88,7 +94,7 @@
           {%- endif %}
           cast(current_timestamp as {{ dbt_profiler.type_string() }}) as profiled_at,
           {{ loop.index }} as _column_position
-        from {{ relation }}
+        from source_data
 
         {% if not loop.last %}union all{% endif %}
       {% endfor %}