When using `--empty`, `adapter.get_columns_in_relation()` returns an error

[zeev-finaloop]

Describe the bug

The adapter.get_columns_in_relation and dbt_utils.get_filtered_columns_in_relation on Snowflake are basically running the query describe table YOUR_TABLE. When running dbt build/run with the --empty flag, the generated query is describe table (select * from YOUR_TABLE where false limit 0). The latter is a broken syntax query and throws an error.

Steps to reproduce

Just build any macro with: {%- set relation_column_names = adapter.get_columns_in_relation(your_model) -%} Run dbt build --empty -d

Expected results

List of columns

Actual results

You will get something like: 001003 (42000): SQL compilation error: syntax error line 1 at position 15 unexpected '('.

Screenshots and log output

System information

The contents of your packages.yml file:

Which database are you using dbt with?

• [ ] postgres
• [ ] redshift
• [ ] bigquery
• [X] snowflake
• [ ] other (specify: ____)

The output of dbt --version:

Core:
  - installed: 1.8.1
  - latest:    1.8.1 - Up to date!

Plugins:
  - snowflake: 1.8.2 - Up to date!

Additional context

Are you interested in contributing the fix?

[dataders]

thanks for flagging @zeev-finaloop! we're actually aware of the issue with adapter.get_columns_in_relation() and --empty -- we're tracking it in https://github.com/dbt-labs/dbt-snowflake/issues/1033 (somewhat related: https://github.com/dbt-labs/dbt-adapters/issues/213).

[dataders]

I'm tempted to close this as the problem and solution exists outside of the dbt-utils package, but I'll leave it open for now so that users encountering the same issue may discover it here

[kokorin]

I found this DBT Core issue: CT-1919 Create get_column_schema_from_query macro and it looks get_column_schema_from_query macro can help with --empty.

[kokorin]

Here is a workaround which makes union_releations to work with --empty:

{%- macro union_relations(relations, column_override=none, include=[], exclude=[], source_column_name='_dbt_source_relation', where=none) -%}
    {% if invocation_args_dict.get('empty', False) %}
        {% set actual_relations = [] %}
        {% for relation in relations %}
            {% set database, schema, identifier = relation.render().split('.') %}
            {%- set actual_relation = adapter.get_relation(database, schema, identifier) -%}
            {% do actual_relations.append(actual_relation) %}      
        {% endfor %}
        {% set relations = actual_relations %}
        {% set where = 'false limit 0'%}
    {% endif %}
    {{ return(adapter.dispatch('union_relations', 'dbt_utils')(relations, column_override, include, exclude, source_column_name, where)) }}
{% endmacro %}

The keys here are relation.render() which converts relation to full table name and set where = 'false limit 0' which applies exact same filter to relations.