Using the schema compiler to generate Python api in one file

[zhengliuX]

Hi, when I usin flatc to generate python api from the schema, there are a lot of python files generated(the structure is a little complex), is there a way that I can generate only one file containing all the APIs?

[Advait251206]

As far as I know, flatc does not currently provide an option to generate all Python bindings into a single file.

The Python code generator typically mirrors the schema structure, generating separate modules for tables, structs, enums, unions, and namespaces. This keeps the generated code organized and helps avoid naming conflicts, especially for larger schemas.

If distributing a large number of generated files is inconvenient, a few possible workarounds are:

• Package the generated files as a Python package and import from a common entry point.
• Create your own wrapper module that re-exports the generated classes and functions.
• Add a build step that concatenates or bundles the generated files, though this would be a custom solution rather than a supported flatc feature.
• Consider generating code into a dedicated package directory and distributing that package instead of individual files.

For example, a simple wrapper module could look like:

from myschema.Foo import Foo
from myschema.Bar import Bar
from myschema.MyEnum import MyEnum

__all__ = ["Foo", "Bar", "MyEnum"]

which gives consumers a single import location even though the generated code remains split across multiple files.

I don't believe there is currently a flatc flag equivalent to "generate a single Python file," but I'd be interested to hear if maintainers know of an existing option or have considered supporting one.