Documentation on codebase of the package. Main entry-point is learn_to_normalize script. All the normalization rules are stored in submodules, this package only provides an entry point and small set of utils shared between locales:

Grammar utils

Generic utils for grammars. Recipe interacts with grammar directory through utils


Loads normalization grammars from directory of specific structure


Base class for text normalization rules.

Some functions and pynini shortcuts that are reused in grammars throughout the locales are in data_loader.py and shortcuts.py