Column-level granularity

[gastlich]

Describe the feature

Hello guys, thank your for working on this amazing package. We are trying to use it as our only tool when it comes to DBT project validation. In order to achieve this, we need to run certain custom validation rules on the top of columns, not just nodes (models). To give you an example, we would like to forbid certain column names to be used across our project.

We hoped, that int_all_graph_resources is going to give as this answer by seeing column_name column in it, but it looks like it's set only for generic tests.

Then we thought, maybe we can find anything in the underlaying staging models like stg_nodes, but still no luck.

Describe alternatives you've considered

Lastly, we check the underlaying implementation:

dbt-project-evaluator/macros/unpack/get_node_values.sql

Lines 31 to 32 in 601263c

	node.columns.values() | list | length,
	node.columns.values() | list | selectattr('description') | list | length,

And it's clear you gather only information related to columns count, not what they are.

Additionally, there was a request last year related to validating missing column descriptions, which was automatically closed:
#199

Who will this benefit?

Anyone who is willing to start using column-level rules.

Are you interested in contributing this feature?

I gave it a go: #386