def process_reading(reading):
hiragana = common.to_hiragana(reading)
hiragana = hiragana.split(".")[0].strip("-")
generated_readings = {hiragana}
def generate_extra_readings(hiragana):
# handle sokuon
if hiragana[-1] in "きくちつ":
yield hiragana[:-1] + "っ"
# handle rendaku
if hiragana[0] in "かきくけこさしすせそたちつてとはひふへほ":
yield chr(ord(hiragana[0]) + 1) + hiragana[1:]
# handle handakuten for ha row
if hiragana[0] in "はひふへほ":
yield chr(ord(hiragana[0]) + 2) + hiragana[1:]
# handle homophones for dzi, dzu
if hiragana[0] in "ちつ":
yield {"ち": "じ", "つ": "ず"}[hiragana[0]] + hiragana[1:]
old = set()
while generated_readings != old:
old = generated_readings
generated_readings = generated_readings | {new_reading
for old_reading in generated_readings
for new_reading in generate_extra_readings(old_reading)}
return generated_readings
def process_furigana(kanji, kana):
if (common.to_hiragana(kanji) != kana
and len(kanji) != 0
and all(common.is_kanji(k) or ord('０') <= ord(k) <= ord('９') for k in kanji)):
return split_reading(kanji, kana, return_score=True)
else:
return [(kanji, None)], sum(NO_RUBY_PENALTY for k in kanji
if common.is_kanji(k))
def finalize_furigana(l, return_score=False):
def process_furigana(kanji, kana):
if (common.to_hiragana(kanji) != kana
and len(kanji) != 0
and all(common.is_kanji(k) or ord('０') <= ord(k) <= ord('９') for k in kanji)):
return split_reading(kanji, kana, return_score=True)
else:
return [(kanji, None)], sum(NO_RUBY_PENALTY for k in kanji
if common.is_kanji(k))
nested_furigana, scores = zip(*(process_furigana(kanji, common.to_hiragana(kana))
for kanji, kana in l))
furigana = [pair
for nested in nested_furigana
for pair in nested]
total_score = sum(scores) + KANA_MISMATCH_PENALTY * \
Levenshtein.distance(common.to_hiragana("".join(kana for _, kana in l)),
common.to_hiragana("".join(kana if kana is not None else kanji
for kanji, kana in furigana)))
return (furigana, total_score) if return_score else furigana
