Refactor: Use toyaml method for source generation about dbt-codegen HOT 3 CLOSED

@clrcrl I took a pass at this today, but due to how pyyaml's dump style, the format differs substantially (despite being functionally identical).

Original Test Output:

version: 2

models:
  - name: data__a_relation
    description: ""
    columns:
      - name: col_a
        description: ""

      - name: col_b
        description: ""

Test Output Using toyaml():

version: 2
models:
- name: data__a_relation
  description: ''
  columns:
  - name: col_a
    description: ''
  - name: col_b
    description: ''

The key differences:

• A list inside a dict is no longer indented.
• Blank lines between top-level keys and columns.
• Single quotes for empty strings instead of double quotes.

This bothers me because it departs from the fishtown style. If dbt core replaced pyyaml with the drop-in replacement ruamel we could add some code there that would set the indentation (IMO the biggest issue with the change), but I haven't yet found a fix for the blank lines or single quotes.

What are your thoughts here? Should we stick with the list of strings originally? I like the spirit of this issue because it's a more maintainable way, but I also don't like introducing something that messes with what seems like the default dbt yaml style.

from dbt-codegen.

Ah dang, that's a shame that we didn't hook it up to a yaml-dumper that is in our favorite Fishtown format. IIRC (from other projects) formatting YAML output is not a quick and easy task.

I think given the current constraints of the library, I'm going to close this issue, and leave the "append new lines" method in place! I'm excited to see some other developments in the space (e.g. dbt-sugar and dbt-helper) which might be better solutions that the current package-based approach!

from dbt-codegen.

(also TY for taking a look at this!!)

from dbt-codegen.