Custom script directory

By default, fal assumes that the directory where scripts are stored is the root directory of a dbt project. You can change this. Within your dbt_project.yml, you can add a custom variable that will tell fal where to look for scripts:

...
vars:
  fal-scripts-path: "my_scripts_dir"

Now, if a script (after.py) is put in my_scripts_dir directory within your dbt project, you can refer to it by name in schema.yml:

models:
  - name: some_model
    meta:
      fal:
        scripts:
            - after.py

Similarly, you can do it when using the --scripts flag:

fal run --scripts after.py

Local imports in your fal scripts

The directory containing fal scripts is temporarily added to sys.path during runtime. This means that you can import local modules in your scripts directory. Say we have the following structure in our scripts directory:

└── my_scripts_dir
    ├── after.py
    ├── random_script.py
    └── my_utils
        ├── custom_functions.py
        └── message
            └── slack.py

This will allow you to do local imports in after.py:

from my_utils.message.slack import send_slack_message
...
send_slack_message(my_message)
...

Also, within slack.py you can do relative imports:

from ..custom_functions import format_string
...
formatted = format_string(my_string)
...

Note that in this example, my_scripts_dir is not itself loaded as a module, so a from .. import wouldn't work in custom_functions.py. Instead, you can do import random_script, because everything in my_scripts_dir is available in sys.path.