get_models seems broken in 1.8

[sdebruyn]

Steps to reproduce

1. Install dbt-core 1.8.0b2 with a compatible adapter
2. Create a new analysis and paste {{ log('repro: ' ~ codegen.get_models(directory='somedir')) }}
3. Run dbt run -s your_analysis

Expected results

A log with your models

Actual results

Compilation error 'dict object' has no attribute 'nodes'

System information

packages:
  - package: dbt-labs/codegen
    version: 0.12.1

Which database are you using dbt with?

• postgres
• redshift
• bigquery
• snowflake
• other (specify: ____________)

The output of dbt --version:

Core:
  - installed: 1.8.0-b2
  - latest:    1.7.11   - Ahead of latest version!

Plugins:
  - postgres:    1.8.0b1 - Ahead of latest version!
  - timescaledb: 1.8.0b2 - Up to date!

The operating system you're using: macOS

The output of python --version: Python 3.11.8

[gwenwindflower]

thanks @sdebruyn -- we're getting ready to cut a new release next week so we'll dig into this and make sure it gets fixed before we do that! thanks for the heads up!

[gwenwindflower]

hey @sdebruyn were you wrapping that in an if execute tag? wondering if it's just this -- there aren't nodes in the graph in the parsing stage that would typically apply to an analysis file.

[mrhorvath]

Could the example in the readme (https://github.com/dbt-labs/dbt-codegen?tab=readme-ov-file#generate_model_yaml-source) that uses codegen.get_models be updated to include 'if execute'? As the documentation is written now, it results in the error described in this issue.

[SuaveCharlie]

Could the example in the readme (https://github.com/dbt-labs/dbt-codegen?tab=readme-ov-file#generate_model_yaml-source) that uses codegen.get_models be updated to include 'if execute'? As the documentation is written now, it results in the error described in this issue.

Echoing this, I've got it running in an analysis file by wrapping it in an if execute block, but it then complained that I needed to add -- depends on x which voids the point of using get_models(directory='some_dir')

# analyses/codegen/app_installs.sql

-- depends_on: {{ ref('mobile_app_installs') }}

{% if execute %}
    {% set models_to_generate = codegen.get_models(directory='app_installs') %}
    {{ codegen.generate_model_yaml(
        model_names = models_to_generate,
        upstream_descriptions = true,
        include_data_types = true
    ) }}
{% endif %}

---

Scratch that, the above doesn't work; it adds duplicate entries to my various yaml files

[stephen-elves-mayden]

This still seems to be broken in 0.13.1 and the work around, doesn't seem to.....work?

[stephen-elves-mayden]

thanks @sdebruyn -- we're getting ready to cut a new release next week so we'll dig into this and make sure it gets fixed before we do that! thanks for the heads up!

I'm guessing this didn't happen?