o i@sUdZddlZddlZddlmZddlmZmZddlm Z m Z m Z ddl m Z e dZe ed<d ed ed d fd dZGdd d ejZGdddeZGdddeZGdddeZdS)z5Manages words count mismatches for the espeak backendN)Logger)ListTuple) TypeAliasLiteralUnion) Separator)warnignore WordMismatchmodeloggerreturnBaseWordsMismatchc CsJtttd}z|||WSty$td|dd|dw)aOReturns a word count mismatch processor according to `mode` The `mode` can be one of the following: - `ignore` to ignore words mismatches - `warn` to display a warning on each mismatched utterance - `remove` to remove any utterance containing a words mismatch Raises a RuntimeError if the `mode` is unknown. )r r removezmode z invalid, must be in z, N)IgnoreWarnRemoveKeyError RuntimeErrorjoinkeys)r r processorsr\/home/ubuntu/.local/lib/python3.10/site-packages/phonemizer/backend/espeak/words_mismatch.pyget_words_mismatch_processors  rc @seZdZdZedZdefddZe efde e de e ej fde efd d Zde eeeeffd d Zd edefddZde e fddZde e defddZejde e de e fddZdS)rz4The base class of all word count mismatch processorsz\s+r cCs||_g|_g|_dSN)_logger _count_txt _count_phn)selfr rrr__init__:s zBaseWordsMismatch.__init__textwordseprcs(ttjs tfdd|DS)z;Return the number of words contained in each line of `text`c s*g|]}tddt|DqS)cSsg|]}|r|qSrr).0wrrr Iz=BaseWordsMismatch._count_words...)lenresplitstripr$liner#rrr&Hsz2BaseWordsMismatch._count_words..) isinstancer)Patternescape)clsr"r#rr.r _count_words?s   zBaseWordsMismatch._count_wordscCsPt|jt|jkrtdt|jdt|jddtt|j|jDS)zReturns a list of (num_line, nwords_input, nwords_output) Consider only the lines where nwords_input != nwords_output. Raises a RuntimeError if input and output do not have the same number of lines. zBnumber of lines in input and output must be equal, we have: input=z , output=cSs&g|]\}\}}||kr|||fqSrr)r$ntprrrr&Ys  z7BaseWordsMismatch._mismatched_lines..)r(rrr enumeratezip)r rrr_mismatched_linesLsz#BaseWordsMismatch._mismatched_lines nmismatchnlinescCs,|r|jdt||dd||dSdS)z$Logs a high level undetailed warningz1words count mismatch on %s%% of the lines (%s/%s)dN)rwarninground)r r:r;rrr_resume^s zBaseWordsMismatch._resumecCs|||_dS)z-Stores the number of words in each input lineN)r3rr r"rrr count_texteszBaseWordsMismatch.count_text separatorcCs|||j|_dS)z.Stores the number of words in each output lineN)r3wordr)r r"rCrrrcount_phonemizedisz"BaseWordsMismatch.count_phonemizedcCsdS)zDetects and process word count misatches according to the mode This method is called at the very end of phonemization, during post-processing. NrrArrrprocessmszBaseWordsMismatch.processN)__name__ __module__ __qualname____doc__r)compile _RE_SPACESrr! classmethodrstrrr0intr3rr9r@rBrrEabcabstractmethodrFrrrrr6s&     c@*eZdZdZdeedeefddZdS)rzIgnores word count mismatchesr"rcCs|t|t||Sr)r@r(r9rArrrrFzszIgnore.processNrGrHrIrJrrNrFrrrrrwrc@rR)rz Warns on every mismatch detectedr"rcCsF|}|D]\}}}|jd|d||q|t|t||S)Nz>words count mismatch on line %s (expected %s words but get %s))r9rr>r@r()r r"mismatchnumntxtnphnrrrrFs z Warn.processNrSrrrrrrTrc@rR)rz6Removes any utterance containing a word count mismatchr"rcCsHdd|D}|t|t||jd|D]}d||<q|S)NcSsg|]}|dqS)rrr,rrrr&r'z"Remove.process..zremoving the mismatched lines)r9r@r(rr>)r r"rVindexrrrrFs   zRemove.processNrSrrrrrrTr)rJrPr)loggingrtypingrrtyping_extensionsrrrphonemizer.separatorrr __annotations__rABCrrrrrrrrs  A