jaga-translate / language_directions.py
gburgardt's picture
Upload folder using huggingface_hub
22d9697 verified
from helsinki_models import helsinki_models, get_clearly_formatted_langauge_directions
from iso639_wrapper import get_name_from_iso_code
from collections import OrderedDict
from utils import convert_keys_to_lowercase, match_in_keys, match_in_values
def get_all_source_languages():
"""
Returns a human-readable `dict source_languages_names:codes`
based on the available models.
"""
source_languages = {}
language_directions = get_clearly_formatted_langauge_directions()
for direction in language_directions:
source_lang_code = direction.split('-')[0]
source_language_name = get_name_from_iso_code(source_lang_code)
if source_language_name:
source_languages[source_language_name] = source_lang_code
source_languages = OrderedDict(sorted(source_languages.items()))
return source_languages
def get_target_languages(source_language_code):
"""
Returns a human-readable `dict of target languages names to codes`
based on the available models and the source language passed.
"""
target_languages = {}
language_directions = get_clearly_formatted_langauge_directions()
for direction in language_directions:
if direction.startswith(f"{source_language_code}-"):
target_language = direction.split('-')[1] # Extracting the last part as the target language
target_language_name = get_name_from_iso_code(target_language)
if target_language_name:
target_languages[target_language_name] = target_language
return OrderedDict(sorted(target_languages.items())), source_language_code
# Example usage:
# all_source_languages = get_all_source_languages()
# print("All Source Languages:", all_source_languages)
# source_language_code = "pa" # Replace with your desired source language
# target_languages = get_target_languages(source_language_code)
# print(f"Target Languages for {source_language_code}:", target_languages)