Write database-specific sample profiles.yml file in dbt init

[drewbanin]

Describe the feature

Let's add a flag to dbt init which controls the sample contents of the ~/.dbt/profiles.yml file.

Example usage:

dbt init my-project --adapter=bigquery
dbt init my-project --adapter=snowflake
dbt init my-project --adapter=redshift

dbt's adapter plugins should provide their own sample profiles .yml contents. Presently, dbt writes out a Redshift-specific profile.

Additional context

Hi Felipe :)

Who will this benefit?

BQ/Snowflake/Postgres/Spark/Presto users!

[brunomurino]

I'm happy to take this, if still available @drewbanin

[brunomurino]

I'm trying to set up a try-catch with the possible adapter options, but can't seem to find a 'list' of possible adapters. I found some stuff in dbt/core/dbt/include/global_project/macros/adapters/common.sql but not sure this is it.

My idea is to create a template profiles.yml in each adapter folder and create a dictionary such that STARTER_PROFILE[sample_adapter] would get the sample from the adapter.

How does that sound?

[beckjake]

I'm trying to set up a try-catch with the possible adapter options, but can't seem to find a 'list' of possible adapters.

There is no formal list of possible adapters, because users can write their own adapter plugins.

My idea is to create a template profiles.yml in each adapter folder

In #2590, I added a get_include_paths function to dbt.adapters.factory, and the first item in the return value of get_include_paths is always the given adapter's project. Assuming it gets merged, that might make this easier. Each adapter plugin could have a sample_profiles.yml inside its include path (so alongside the dbt_project.yml), and you could use that and the --adapter argument to decide what to use.

You'd have to do something like:

factory.load_plugin(args.adapter)
path = factory.get_include_paths(args.adapter)[0]
sample_path = path / 'sample_profiles.yml'
if not sample_path.exists():
    raise RuntimeException(f'No sample profile for {args.adapter}')
shutil.copy_file(sample_path, dest)

If the chosen adapter is installed and has a sample profile, it should behave okay.

[brunomurino]

makes sense, it would indeed make it easier.

I'm trying to checkout the branch fix/adapter-macro-namespacing but can't find it (?).
If I can't check that out, then I'd just wait on your PR to bet merged

[beckjake]

@brunomurino you might have to set up fishtown-analytics/dbt as a git remote and git fetch that, I think like this:

git remote add upstream https://github.com/fishtown-analytics/dbt.git
git fetch upstream
git checkout upstream/fix/adapter-macro-namespacing

That will probably drop you into a detached HEAD state, so then you'll need to use git checkout -b to create a new local branch

[brunomurino]

Worked like a charm.

Just created the PR for this: #2594
alongside PRs for presto and spark.