Split Braxen by language

Kaggle version, based on this

from pathlib import Path

BRAXEN_DICT = Path("/kaggle/input/download-and-fix-braxen/braxen/dict/braxen-sv.tsv")

data = {}

with open(BRAXEN_DICT) as bf:
    for line in bf:
        if not "\t" in line:
            continue
        parts = line.split("\t")
        lang = parts[3]
        word = parts[0]
        phon = parts[1]
        if lang == "unk" or lang == "":
            continue
        if not lang in data:
            data[lang] = {}
        if not word in data[lang]:
            data[lang][word] = []
        data[lang][word].append(phon)

def clean_phones(phone_string):
    phones = phone_string.split(" ")
    phones = [p for p in phones if not p in ["~", "|", ".", "-"]]
    phones = [p[1:] if p and p[0:1] in [",", "'", '"'] else p for p in phones]
    return " ".join(phones)

for lang in data:
    out_path = Path(f"braxen-{lang}.txt")
    with open(out_path, "w") as of:
        for word in sorted(data[lang].keys()):
            for prons in set(data[lang][word]):
                phones = clean_phones(prons)
                of.write(f"{word}\t{phones}\n")