Merge pull request #72 from davidwitk/fix-dbt-models

Fix dbt and IMDb models

models/duckdb/stg_imdb.sql

@@ -10,13 +10,13 @@ base as (
         extracted_at :: timestamp as extracted_at,
         title,
         first_value(title) over (partition by movie_id order by extracted_at :: timestamp) as title_first,
-        rank,
+        rank :: int as rank,
         case
             when rating_20230119_2215 is not null then rating_20230119_2215
             else split_part(rating, chr(160), 1) :: double
         end as rating,
-        rating_count,
-        year,
+        rating_count :: int as rating_count,
+        year :: int as year,
         link
     from source
 

models/postgres/admin/fct_dbt_invocations.sql

@@ -0,0 +1 @@
+select * from {{ ref('stg_dbt_invocations') }}

models/postgres/admin/fct_dbt_model_executions.sql

@@ -0,0 +1 @@
+select * from {{ ref('stg_dbt_model_executions') }}

models/postgres/staging/dbt/_dbt.yml

@@ -0,0 +1,7 @@
+sources:
+  - name: raw_dbt
+
+    tables:
+      - name: invocations
+      - name: model_executions
+

models/postgres/staging/dbt/stg_dbt_invocations.sql

@@ -0,0 +1,31 @@
+with source as (
+    select * from {{ source('raw_dbt', 'invocations') }}
+),
+
+base as (
+
+    select
+        command_invocation_id,
+        dbt_version,
+        project_name,
+        run_started_at,
+        dbt_command,
+        full_refresh_flag,
+        target_profile_name,
+        target_name,
+        target_schema,
+        target_threads,
+        dbt_cloud_project_id,
+        dbt_cloud_job_id,
+        dbt_cloud_run_id,
+        dbt_cloud_run_reason_category,
+        dbt_cloud_run_reason,
+        env_vars,
+        dbt_vars,
+        invocation_args,
+        dbt_custom_envs
+    from source
+
+)
+
+select * from base

models/postgres/staging/dbt/stg_dbt_model_executions.sql

@@ -0,0 +1,29 @@
+with source as (
+    select * from {{ source('raw_dbt', 'model_executions') }}
+),
+
+base as (
+
+    select
+        command_invocation_id || node_id as model_execution_id,
+        command_invocation_id,
+        node_id,
+        run_started_at,
+        was_full_refresh,
+        split_part(thread_id, '-', 2) as thread_id,
+        status,
+        compile_started_at,
+        query_completed_at,
+        total_node_runtime,
+        rows_affected
+        as materialization,
+        schema
+        as name,
+        alias,
+        message,
+        adapter_response
+    from source
+
+)
+
+select * from base