o ?؂i@s$dZddlZddlmZmZddlmZmZddlm Z m Z m Z m Z ddl mZGdddeZeGd d d ZeGd d d ZGd ddeZidddddddddddddddddd d!d"d#d$ddddddddddddddd d"d$d%Zd&ed'efd(d)ZdS)*z| Base classes for transcription validators. All validators inherit from BaseValidator and produce ValidationResult objects. N)ABCabstractmethod) dataclassfield)ListOptionalDictAny)Enumc@s eZdZdZdZdZdZdZdS)ValidatorStatuszStatus of a validator.not_initializedreadyerrordisabledN)__name__ __module__ __qualname____doc__NOT_INITIALIZEDREADYERRORDISABLEDrr4/home/ubuntu/maya3_transcribe/src/validators/base.pyr s r c@sLeZdZUdZeed<eed<eed<dZeeed<e defdd Z dS) WordAlignmentz3Represents alignment information for a single word.word start_timeend_timeN confidencereturncCs |j|jS)N)rrselfrrrdurations zWordAlignment.duration) rrrrstr__annotations__floatrrpropertyr"rrrrrs rc@seZdZUdZeed<eed<eed<dZeeed<dZ eeed<e e dZ e eed <dZeeed <dZeeed <dZeeed <dZeeed <dZeeeefed<edeefddZdeeeffddZdS)ValidationResultz:Result from a single validator for a single audio segment.validator_name audio_pathsuccessN error_message transcription)default_factoryword_alignmentsoverall_confidencealignment_scoreprocessing_time_secaudio_duration_sec raw_outputrcCs2|jr|jrt|j}|jdkr||jSdS)z3Calculate speaking rate if transcription available.rN)r,r2lensplit)r! word_countrrrwords_per_second<s   z!ValidationResult.words_per_secondc Cs<|j|j|j|j|jdd|jD|j|j|j|j |j d S)z(Convert to dictionary for serialization.cSs&g|]}|j|j|j|j|jdqS)rrrrr"r8).0warrr Msz,ValidationResult.to_dict.. r(r)r*r+r,r.r/r0r1r2r7r<r rrrto_dictEs zValidationResult.to_dict)rrrrr#r$boolr+rr,rlistr.rrr/r%r0r1r2r3rr r&r7r=rrrrr'!s r'c @seZdZUdZdZeed<dZeed<ddefdd Z e d efd d Z e ddede eded e fddZ ddeede eeded ee fddZd efddZddZd eeeffddZd S) BaseValidatora Abstract base class for all transcription validators. Each validator should: 1. Initialize model/resources in __init__ or setup() 2. Implement validate() to process a single audio file 3. Return ValidationResult with relevant metrics base_validatornamezBase validator class descriptionTenabledcKs||_tj|_||_d|_dS)z Initialize the validator. Args: enabled: Whether this validator is active **kwargs: Validator-specific configuration N)rDr rstatusconfig_model)r!rDkwargsrrr__init__ls zBaseValidator.__init__rcCdS)z Initialize models and resources. Called lazily before first validation. Returns: True if setup successful, False otherwise Nrr rrrsetupys zBaseValidator.setupNter)reference_textlanguagecCrJ)am Validate/process a single audio file. Args: audio_path: Path to audio file reference_text: Optional reference transcription for alignment language: Language code (te=Telugu, hi=Hindi, kn=Kannada, etc.) Returns: ValidationResult with transcription/alignment data Nr)r!r)rMrNrrrvalidateszBaseValidator.validate audio_pathsreference_textsc CsFg}|p dgt|}t||D]\}}||||}||q|S)z Validate multiple audio files. Default implementation calls validate() for each file. Override for batch-optimized processing. N)r4ziprOappend) r!rPrQrNresults ref_textsr)ref_textresultrrrvalidate_batchs   zBaseValidator.validate_batchc Cs||jtjkr8z|}|rtjntj|_Wn"ty7}ztj|_td|jd|WYd}~dSd}~ww|jtjkS)z&Ensure validator is set up before use.[z] Setup failed: NF) rEr rrKrr ExceptionprintrB)r!r*errr ensure_setups  zBaseValidator.ensure_setupcCs d|_dS)z&Release resources. Override if needed.N)rGr rrrcleanups zBaseValidator.cleanupcCs|j|j|j|jj|jdS)zGet validator information.)rBrCrDrErF)rBrCrDrEvaluerFr rrrget_infos zBaseValidator.get_info)T)NrL)rrrrrBr#r$rCr>rIrrKrr'rOrrXr]r^rr r`rrrrr@_sB        r@telugurLhindihikannadakntamilta malayalammlbengalibngujaratigumarathimrpunjabipaodiaorenglishen)rmrorqrsrurNrcCst||S)z(Convert language name to ISO 639-1 code.)LANGUAGE_CODESgetlower)rNrrrnormalize_language_codesry)rosabcrr dataclassesrrtypingrrrr enumr r rr'r@rvr#ryrrrrsl  =i