PiUdZddlZddlZddlmZddlmZmZddlm Z m Z m Z ddl m Z e dZe ed<d ed ed d fd ZGdd ejZGddeZGddeZGddeZdS)z5Manages words count mismatches for the espeak backendN)Logger)ListTuple) TypeAliasLiteralUnion) Separator)warnignore WordMismatchmodeloggerreturnBaseWordsMismatchc tttd} |||S#t$r<t d|dd|dwxYw)aOReturns a word count mismatch processor according to `mode` The `mode` can be one of the following: - `ignore` to ignore words mismatches - `warn` to display a warning on each mismatched utterance - `remove` to remove any utterance containing a words mismatch Raises a RuntimeError if the `mode` is unknown. )r r removezmode z invalid, must be in z, N)IgnoreWarnRemoveKeyError RuntimeErrorjoinkeys)r r processorss |/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/phonemizer/backend/espeak/words_mismatch.pyget_words_mismatch_processorrsJ z$'''  MD M Mtyy9J9J/K/K M M   s (AA.c HeZdZdZejdZdefdZe efde e de e ej fde efdZde eeeeffd Zd ed efd Zde e fd Zde e defdZejde e de e fdZdS)rz4The base class of all word count mismatch processorsz\s+rc0||_g|_g|_dSN)_logger _count_txt _count_phn)selfrs r__init__zBaseWordsMismatch.__init__:s textwordseprc|ttjstjfd|DS)z;Return the number of words contained in each line of `text`c g|]@}tdtj|DAS)cg|]}||Sr+).0ws r z=BaseWordsMismatch._count_words...IsAAAqqAAAAr%)lenresplitstrip)r,liner's rr.z2BaseWordsMismatch._count_words..HsT AABHWdjjll;;AAA B Br%) isinstancer0Patternescape)clsr&r's `r _count_wordszBaseWordsMismatch._count_words?sS '2:.. )i((G r%c*t|jt|jkr9tdt|jdt|jdt t |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=c0g|]\}\}}||k|||fSr+r+)r,ntps rr.z7BaseWordsMismatch._mismatched_lines..Ys7#!VaAvv1Ivvr%)r/r!r"r enumeratezip)r#s r_mismatched_linesz#BaseWordsMismatch._mismatched_linesLs t  3t#7#7 7 71"%do"6"611do..1122 2  c$/4?;; < < r% nmismatchnlinescp|r3|jdt||z ddz||dSdS)z$Logs a high level undetailed warningz1words count mismatch on %s%% of the lines (%s/%s)dN)r warninground)r#rArBs r_resumezBaseWordsMismatch._resume^s^  G L Ci&(!,,s2Iv G G G G G G Gr%c:|||_dS)z-Stores the number of words in each input lineN)r8r!r#r&s r count_textzBaseWordsMismatch.count_textes++D11r% separatorcF|||j|_dS)z.Stores the number of words in each output lineN)r8wordr")r#r&rLs rcount_phonemizedz"BaseWordsMismatch.count_phonemizedis++D).AAr%cdS)zDetects and process word count misatches according to the mode This method is called at the very end of phonemization, during post-processing. Nr+rJs rprocesszBaseWordsMismatch.processmsr%N)__name__ __module__ __qualname____doc__r0compile _RE_SPACESrr$ classmethodrstrrr5intr8rr@rHrKr rOabcabstractmethodrQr+r%rrr6sx>>F##Jv /9  s) 3 ?+ >B#Y   [ 4c3m(<#=$GGcGGGG2tCy2222BT#YB9BBBB  DI $s)      r%c:eZdZdZdeedeefdZdS)rzIgnores word count mismatchesr&rc|t|t||Sr)rHr/r@rJs rrQzIgnore.processzs5 S//1122CII>>> r%NrRrSrTrUrrYrQr+r%rrrwsD''DI$s)r%rc:eZdZdZdeedeefdZdS)rz Warns on every mismatch detectedr&rc|}|D]&\}}}|jd|dz||'|t |t ||S)Nz>words count mismatch on line %s (expected %s words but get %s))r@r rFrHr/)r#r&mismatchnumntxtnphns rrQz Warn.processs~))++' % %OCt L 1at % % % % S]]CII... r%Nr_r+r%rrrsD** DI $s)      r%rc:eZdZdZdeedeefdZdS)rz6Removes any utterance containing a word count mismatchr&rcd|D}|t|t||jd|D]}d||<|S)Ncg|] }|d S)rr+)r,r3s rr.z"Remove.process..sAAADGAAAr%zremoving the mismatched lines)r@rHr/r rF)r#r&rcindexs rrQzRemove.processswAA(>(>(@(@AAA S]]CII... <===  EDKK r%Nr_r+r%rrrsD@@DI$s)r%r)rUr[r0loggingrtypingrrtyping_extensionsrrrphonemizer.separatorr r __annotations__rABCrrrrr+r%rrrsv<;; 7777777777******""23 i333|VH[0> > > > > > > > B                     r%