Configuration

django-unasyncify’s configuration should be placed within your project’s pyproject.toml, inthe [tool.django-unasyncify] section.

unasync_helpers_path and unasync_helpers_import_path are both required configuration values.

paths_to_visit

Type:

list[str]

Lists paths to directories and files that should be scanned for code generation.

[tool.django_unasyncify]
paths_to_visit = [
    "src/pkg1/foo.py"
    "src/pkg2",
]

In the above configuration, we would check src/pkg1/foo.py and all .py files within src/pkg2.

By default paths_to_visit is set to ["."], which will check all .py files in your project.

While the unasyncify process on files is fairly quick, on larger projects it’s a good idea to scope things down.

unasync_helpers_path

Type:

str

unasync_helpers_path is a string to a Python file that django-unasyncify manages, that includes the definitions to the Unasync Helpers.

A file will be created at the provided path that includes decorators used by django-unasyncify. That way, code that is generated by django-unasyncify doesn’t actually need to have django-unasyncify (or its requirements).

unasync_helpers_import_path

Type:

str

The import name that would be used to import the file indicated in unasync_helpers_path. This should be a dotted path, and will be used to generate the right kind of import statements during code generation.

[tool.django-unasyncify]
unasync_helpers_path = "src/pkg1/_codegen.py"
unasync_helpers_import_path = "pkg1._codegen"

In the above configuration, we’ll end up with something like:

frok pkg1._codegen import from_codegen, generate_unasynced

After generation.

attribute_renames

Type:

dict[str, str]

In some cases you might want to rename certain identifiers based on whether it’s the sync or async variant.

For these cases you can use identifier_renames to definie a key/value mapping of these:

[tool.django-unasyncify]
identifier_renames = {
    aconnection = "connection",
    async_obj   = "sync_obj"
}

This renaming mapping will kick in during codegen for functions marked with @generate_unasynced.

It will replace usages of the identifer in attribute accesses. This is very helpful when dealing with attributes on self for an object that might manage both sync and async variations:

mark_as_used(self.async_obj)
return self.aconnection.properties

# Becomes
mark_as_used(self.sync_obj)
return self.connection.properties

This renaming is meant for attribute accesses. Async method call renaming happens automatically based on the norm of “removing the a”, and you mostly shouldn’t need this.

See Use the async/sync variant naming scheme and Handling Function Calls for details on how method renaming happens.