`i"TdZddlZejdkrddlZnddlZGddZdS)z"English Normalizer class for CLVP.N) ceZdZdZdedefdZdedefdZdedefdZdedefd Z dedefd Z dedefd Z dedefd Z dedefd Z dedefdZdedefdZdZdS)EnglishNormalizerc^ddD|_gd|_gd|_gd|_dS)Ncng|]2}tjd|dztj|df3S)z\b%s\.r)recompile IGNORECASE).0xs ~/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/transformers/models/clvp/number_normalizer.py z.EnglishNormalizer.__init__..sH   Z QqT)2= 9 91Q4 @   ))mrsmisess)mrmister)drdoctor)stsaint)cocompany)jrjunior)majmajor)gengeneral)drsdoctors)revreverend)lt lieutenant)hon honorable)sgtsergeant)captcaptain)esqesquire)ltdlimited)colcolonel)ftfort) onetwothreefourfivesixseveneightnine) teneleventwelvethirteenfourteenfifteensixteen seventeeneighteennineteen) r6r6twentythirtyfortyfiftysixtyseventyeightyninety)_abbreviationsonesteenstens)selfs r__init__zEnglishNormalizer.__init__s\     0a``      kjj rnumreturnc|dkrdS|dkr%d|t|zS|dkr |j|S|dkr|j|dz S|dkr7|j|dz|dzdkrd||dzzndzS|d kr:|j|dzd z|dzdkrd ||dzzndzS|d krB||d zd z|d zdkrd||d zzndzS|dkrB||d zdz|d zdkrd||d zzndzS|dkrB||dzdz|dzdkrd||dzzndzS|dkrB||dzdz|dzdkrd||dzzndzS|dkrB||dzdz|dzdkrd||dzzndzSdS)ax Converts numbers(`int`) to words(`str`). Please note that it only supports upto - "'nine hundred ninety-nine quadrillion, nine hundred ninety-nine trillion, nine hundred ninety-nine billion, nine hundred ninety-nine million, nine hundred ninety-nine thousand, nine hundred ninety-nine'" or `number_to_words(999_999_999_999_999_999)`. rzerozminus d-r6 hundred i@Bz thousand, iʚ;z millionlJ)z billionlI5z trillionlNZoz quadrillionznumber out of range)number_to_wordsabsrSrTrU)rVrXs rrdz!EnglishNormalizer.number_to_wordsFs !886 1WWd223s88<<< < 2XX9S> ! 2XX:cBh' ' 3YY9SBY'SVY[S[_`S`S`31E1EcBh1O1O+O+Ofhi i 4ZZ #*% 2_beh_hlm_m_mcDADjAoo4$..sTz::::SUW  = $$SI%566CF?VWCWCW4$..sY????]_a  $ $ $$$SM%9::GJ]GZ^_G_G_4$..s]/BCCCCegi  ( ( ($$S,=%=>>KNQbKbfgKgKg4$..s5F/FGGGGmoq  , , ,$$S,A%ABB !22a774//6K0KLLLL  )(rtextcT|dddS)z+ Converts unicode to ascii asciiignorezutf-8)encodedecoderVrfs rconvert_to_asciiz"EnglishNormalizer.convert_to_asciis&{{7H--44W===rmc|d}|d}t|dkr|dzS|drt|dnd}t|dkr|drt|dnd}|r#|r!|dkrdnd}|dkrdnd }|d |d |d |S|r|dkrdnd}|d |S|r|dkrdnd }|d |Sd S) zZ This method is used to expand numerical dollar values into spoken words. r .z dollarsrdollardollarscentcentsrbrcz zero dollars)groupsplitlenint)rVrnmatchpartsrsru dollar_unit cent_units r_expand_dollarsz!EnglishNormalizer._expand_dollarss*  C   u::>>:% %#(82#eAh---!$UaE!HE!H !  "u "&-ll(( K"'1**'I%,WWkkk555))L L  "&-ll(( K%gg{{3 3  ""'1**'I#eeYY/ /!>rcT|dddS)zF This method is used to remove commas from sentences. r ,r6rvreplacerVrns r_remove_commasz EnglishNormalizer._remove_commass$wwqzz!!#r***rcT|dddS)zO This method is used to expand '.' into spoken word ' point '. r rpz point rrs r_expand_decimal_pointz'EnglishNormalizer._expand_decimal_points$wwqzz!!#y111rcdddd}t|ddd}d|d zkr |d zd krd }n||dzd }|||zS) z` This method is used to expand ordinals such as '1st', '2nd' into spoken words. rndrd)r rqrrNr\r^r]th)ryrvgetrd)rVrXordinal_suffixessuffixs r_expand_ordinalz!EnglishNormalizer._expand_ordinals $66#))A,,ss#$$ s??sSyBFF%))#(D99F##C((611rcXt|d}|dkrn|dkrh|dkrdS|dkr!|dkrd||dzzS|dzdkr||dzd zS||S||S) a  This method acts as a preprocessing step for numbers between 1000 and 3000 (same as the original repository, link : https://github.com/neonbjb/tortoise-tts/blob/4003544b6ff4b68c09856e04d3eff9da26d023c2/tortoise/utils/tokenizer.py#L86) rr`i iz two thousandiz two thousand r^ra)ryrvrd)rVrnrXs r_expand_numberz EnglishNormalizer._expand_numbers !''!**oo ::#**d{{%~td &)=)=cCi)H)HHHsa++C3J77*DD++C000'',, ,rc@tjd|j|}tjdd|}tjd|j|}tjd|j|}tjd|j|}tjd|j|}|S)z This method is used to normalize numbers within a text such as converting the numbers to words, removing commas, etc. z([0-9][0-9,]+[0-9])u£([0-9,]*[0-9])z \1 poundsz\$([0-9.,]*[0-9])z([0-9]++\.[0-9]+)z[0-9]++(st|nd|rd|th)z[0-9]+)r subrr~rrrrls rnormalize_numbersz#EnglishNormalizer.normalize_numberss v,d.A4HHv)<>>v*D,@$GGv*D,FMMv-t/CTJJvi!4d;; rcL|jD]\}}tj|||}|S)z/ Expands the abbreviate words. )rRr r)rVrfregex replacements rexpand_abbreviationsz&EnglishNormalizer.expand_abbreviationss5#'"5 4 4 E;6%d33DD rcRtjtjdd|S)z. Removes multiple whitespaces z\s+rb)r rr rls rcollapse_whitespacez%EnglishNormalizer.collapse_whitespaces"vbj((#t444rc||}|}||}||}||}|dd}|S)z Converts text to ascii, numbers / number-like quantities to their spelt-out counterparts and expands abbreviations "r6)rmlowerrrrrrls r__call__zEnglishNormalizer.__call__su $$T**zz||%%d++((..''--||C$$ rN)__name__ __module__ __qualname__rWrystrrdrmr~rrrrrrrrrrrrs'k'k'kR9)39)39)9)9)9)v>S>S>>>> """"""0++++++ 2s2s2222 23 23 2 2 2 2------( c c    555555     rr)__doc__sys version_infor rrrrrrsy )( w IIIIXXXXXXXXXXr