From 2ecf22f808246e379bf55e0aeb4c992105ddd48b Mon Sep 17 00:00:00 2001
From: Guillaume Wagner <guillaume.luc.wagner@gmail.com>
Date: Wed, 21 Aug 2024 11:33:29 -0400
Subject: [PATCH] Added support for structs and modes (repeated and required)

---
 .../bigquery/create_external_table.sql        | 16 +++++++-
 sample_sources/bigquery.yml                   | 38 +++++++++++++++++++
 2 files changed, 52 insertions(+), 2 deletions(-)

diff --git a/macros/plugins/bigquery/create_external_table.sql b/macros/plugins/bigquery/create_external_table.sql
index fe7a5ac8..5e4716e6 100644
--- a/macros/plugins/bigquery/create_external_table.sql
+++ b/macros/plugins/bigquery/create_external_table.sql
@@ -1,3 +1,16 @@
+{%- macro create_struct_type_definition(fields) -%}
+    STRUCT<
+    {%for field in fields%}{{get_column_definition(field)}}{{- ',' if not loop.last -}}{%- endfor %}
+    >
+{%- endmacro -%}
+
+{%- macro get_column_definition(column) -%}
+    {%- set column_quoted = adapter.quote(column.name) if column.quote else column.name %}
+    {%- set column_type = create_struct_type_definition(column.fields) if column.data_type == 'STRUCT' else column.data_type %}
+    {%- set column_type = 'ARRAY<' ~ column_type ~ '>' if column.mode == 'REPEATED' else column_type %}
+    {{column_quoted}} {{column_type}} {{- ' NOT NULL' if column.mode == 'REQUIRED'}}
+{%- endmacro -%}
+
 {% macro bigquery__create_external_table(source_node) %}
     {%- set columns = source_node.columns.values() -%}
     {%- set external = source_node.external -%}
@@ -19,8 +32,7 @@
     create or replace external table {{source(source_node.source_name, source_node.name)}}
         {%- if columns -%}(
             {% for column in columns %}
-                {%- set column_quoted = adapter.quote(column.name) if column.quote else column.name %}
-                {{column_quoted}} {{column.data_type}} {{- ',' if not loop.last -}}
+                {{ get_column_definition(column) }} {{- ',' if not loop.last -}}
             {%- endfor -%}
         )
         {% endif %}
diff --git a/sample_sources/bigquery.yml b/sample_sources/bigquery.yml
index cfe18cde..9d86a60e 100644
--- a/sample_sources/bigquery.yml
+++ b/sample_sources/bigquery.yml
@@ -26,6 +26,7 @@ sources:
           - name: app_id
             data_type: varchar(255)
             description: "Application ID"
+            mode: REQUIRED
           - name: domain_sessionidx
             data_type: int
             description: "A visit / session index"
@@ -58,3 +59,40 @@ sources:
               - 'gs://bucket_a/path/*'
               - 'gs://bucket_b/path/*'
               - 'gs://bucket_c/more/specific/path/file.csv'
+
+
+  - name: users
+    schema: users
+    tables:
+      - name: users
+        external:
+          location: gs://my-prod-bucket/users.jsonl
+          options:
+            format: JSON
+        columns:
+        - name: id
+          data_type: INTEGER
+          mode: REQUIRED
+          description: >
+            User ID from the prod database
+        - name: name
+          data_type: STRING
+          description: "User full name"
+        - name: addresses
+          data_type: STRUCT
+          mode: REPEATED
+          description: A list of addresses where the user has lived
+          fields:
+            - name: city
+              data_type: STRING
+              description: "City name"
+            - name: address
+              data_type: STRING
+              description: "Street address"
+            - name: postal_code
+              data_type: STRING
+              description: "Postal code"
+            - name: country
+              data_type: STRING
+              description: "Country name"
+          
\ No newline at end of file