o jf@sHddlZddlZd dedefddZdefdd Zd edefd d ZdS)NFTnumreturncst|}tt|dkrtdd|vrtd|rdnd}|r"d}|r3|r(dnd d }|r0d nd }n|r7d ndd}|rD|rAdnd}nd}|rJdnd} dd} t|}g} |ddkrd| |dn |ddkrq| |dd|vr|dd\} n|dd} tr^fddtt dd D} g}t | D]\}}t|dkr|dq|dkrt|d!kr||| |dqg}| d"}t t |D]^\}}|d#kr|r|dq|dkr|t|q|dkr|d$kr|dd#kr||dq|dkr&|d%kr&||||dq|t|||dq| |}|dkrE||q||| |dq| | | dn| d| r~| |d!| d&fd'd(| Dd&| S))uConvert numerical arabic numbers (0->9) to chinese hanzi numbers (〇 -> 九) Args: num (str): arabic number to convert big (bool, optional): use financial characters. Defaults to False. simp (bool, optional): use simplified characters instead of tradictional characters. Defaults to True. o (bool, optional): use 〇 for 'zero'. Defaults to False. twoalt (bool, optional): use 两/兩 for 'two' when appropriate. Defaults to False. Raises: ValueError: if number is more than 1e48 ValueError: if 'e' exposent in number Returns: str: converted number as hanzi characters gsȡ1Iznumber out of rangeez$scientific notation is not supportedu 正负点u 正負點Fu零壹贰叁肆伍陆柒捌玖u零壹貳參肆伍陸柒捌玖u 拾佰仟u贰u貳u〇一二三四五六七八九u零一二三四五六七八九u 十百千u两u兩u二u!万亿兆京垓秭穰沟涧正载u!萬億兆京垓秭穰溝澗正載cSsdddtt|DS)Ncss|]\}}|VqdSN).0kgrr^/home/kuhnn/.local/lib/python3.10/site-packages/TTS/tts/utils/text/chinese_mandarin/numbers.py 3sz1_num2chinese....)join itertoolsgroupbyreversed)lrrr 3sz_num2chinese..r+-.z+-Ncs"g|] }t|dd|qS)r)max)ri)integerrr ?s"z _num2chinese..r012rc3s|] }t|VqdSr)int)rch)c_basicrr r dsz_num2chinese..)strabsfloat ValueErrorappendlstripsplitr!rangelen enumeratezfillrstripr )rbigsimpotwoaltndc_symbolc_unit1c_twoaltc_unit2revuniqresult remaindersplitted intresultnuunitulistncr"ustrr)r#rr _num2chinese s|         "    rCcCs|}t|S)zfunction to apply in a match, transform all numbers in a match by chinese characters Args: match (re.Match): numbers regex matches Returns: str: replaced characters for the numbers )grouprC)match match_strrrr _number_replacehs rGtextcCstdt|}|S)zReplace all arabic numbers in a text by their equivalent in chinese characters (simplified) Args: text (str): input text to transform Returns: str: output text z[0-9]+)resubrG)rHrrr %replace_numbers_to_characters_in_textus rK)FTFF)rrIr$rCrGrKrrrr s \