[CT-1282] [Bug] Cannot reference a Source containing a dot (.) in node selection

[rossserven]

Is this a new bug in dbt-core?

• I believe this is a new bug in dbt-core
• I have searched the existing issues, and I could not find an existing issue for this bug

Current Behavior

Trying to reference a Source through node selection and I’m running into an issue because our sources contain periods. e.g.:

sources:
  - name: raw.test
    database: 'RAW'
    schema: 'TESTING'
    tables:
      - name: TEST_TABLE

so this doesn’t work: source:raw.test.TEST_TABLE
however if it’s defined as:

sources:
  - name: test
    database: 'FIVETRAN_RAW'
    schema: 'TESTING'
    tables:
      - name: TEST_TABLE

this works: source:test.TEST_TABLE

Expected Behavior

sources:
  - name: raw.test
    database: 'RAW'
    schema: 'TESTING'
    tables:
      - name: TEST_TABLE

this should work (or optionally enclosed with quotes): source:raw.test.TEST_TABLE

Steps To Reproduce

1. define a Source with a dot/period
2. Attempt to reference that source in node selection

Relevant log output

[0m15:38:00 No nodes selected!
[0m15:38:00 'source:raw.test.TEST_TABLE' does not match any nodes

Environment

- OS:
- Python:
- dbt:

Which database adapter are you using with dbt?

snowflake

Additional Context

No response

[lostmygithubaccount]

hi @rossserven, thanks for opening this issue. I think it's related to #1269, tagging @jtcohen6 for his input -- it looks like this is how dbt has been for a long time so I'm going to classify this as an enhancement rather than a bug.

@jtcohen6 should we close as a duplicate in favor of #1269, or is this different? I don't see a lot of discussion about sources in that issue

[jtcohen6]

The ability to use dots in some resource names is something that's, at best, semi-supported within dbt today. We don't (to my knowledge) document that you can definitely do this, but last year a community member did add tests for models with dots in their names (#3247) to ensure that we didn't accidentally regress on this capability in future versions.

I do see the connection to #1269, insofar as the dot-in-model-name approach (dynamically stripped to form schema.alias combos) was following this pattern, and originally motivated by the lack of namespacing support within dbt: https://discourse.getdbt.com/t/extracting-schema-and-model-names-from-the-filename/575

We simply must add namespacing support next year. #1269 is the oldest and narrowest version of that issue, but the way I want to see us doing this is by:

• continuing to require unique model names within one project
• making it much, much easier to deploy multiple projects together, including lineage across projects (Better mechanisms for cross-project lineage #5244)

All that being said: @rossserven Could I ask what your purpose is behind naming a source raw.test, instead of (e.g.) raw__test? Is it essential to your use case that the name contain dots? Or is this just something you'd expected to work, and we should more clearly document that it doesn't?

[rossserven]

@jtcohen6 there was no requirement/purpose for raw. we just used that notation and at the time didn't see or read any limitations around it. we can work to move to a new standard without the dot since this is now known.

[lostmygithubaccount]

I've opened a docs issue to track potential improvements, going to close this one out @rossserven