o [iA2 @sdZddlZddlZddlZddlmZmZmZmZddl m Z m Z ddl m Z ddl mZddlmZmZmZe Gd d d Ze Gd d d ZGd ddZ  d@dedeededefddZedkrLddlZejddZejdddejddddejdddd d!ejd"d#d$d%d!ejd&d'd(ejd)d'd(eZeej ej ej!ej"d*Z#e#j$ej%ej&ej!d+Z'e(d,d-e(d.e(d-e(d/ej%e(d0ej!e(d1e'j)d2d3e(e'j*+D]O\Z,Z-e-j.rd4nd5Z/e(e/d6e,d7e-j.r7e-j0re(d8e-j0dd9d:e-j1r)e(d;e2e-j1d<e-j3r6e(d=e-j3d>qe(d?e-j4qe#5e'ge#6dSdS)Az Validator Runner ================ Coordinates running multiple validators on audio segments. Supports toggling individual validators and collecting aggregated results. N)ListDictAnyOptional) dataclassfield)datetime)Path) BaseValidatorValidationResultnormalize_language_codec@seZdZUdZdZeed<dZeed<dZe ed<dZ e ed<d Z e ed <dZ eed <e ed Zee efed <e ed Zee efed<dS) RunnerConfigz(Configuration for the validation runner.Tenable_indicmfaenable_indic_conformertelanguageautodevice./validation_results output_dir save_resultsdefault_factoryindicmfa_optionsindic_conformer_optionsN)__name__ __module__ __qualname____doc__rbool__annotations__rrstrrrrrdictrrrrr$r$6/home/ubuntu/maya3_transcribe/src/validators/runner.pyrs       rc@sjeZdZUdZeed<eeed<eed<eedZ e ee fed<dZ e ed<d e eeffd d Zd S) AggregatedResultz?Aggregated results from all validators for a single audio file. audio_pathreference_textrrresultstotal_processing_time_secreturncCs(|j|j|j|jdd|jDdS)NcSi|] \}}||qSr$to_dict).0nameresultr$r$r% :z,AggregatedResult.to_dict..)r'r(rr+r))r'r(rr+r)itemsselfr$r$r%r/4szAggregatedResult.to_dictN)rrrrr"r!rrr#r)rr r+floatrr/r$r$r$r%r&'s   r&c@seZdZdZ     d#dededed ed ef d d Zd dZdeee ffddZ   d$dede ede ede e ede f ddZ    d%de ede e ede ede e ede ede e f ddZ d&de e de edefdd Zd!d"ZdS)'ValidatorRunnerai Runs multiple validators on audio files. Example: ```python runner = ValidatorRunner( enable_indicmfa=True, enable_indic_conformer=True, language="te" ) result = runner.validate( audio_path="audio.flac", reference_text="reference transcription", language="te" ) ``` TrrrrrrrrcKs:td|||||ddd|D|_i|_d|_dS)a| Initialize the validator runner. Args: enable_indicmfa: Enable IndicMFA validator (forced alignment) enable_indic_conformer: Enable IndicConformer validator (ASR) language: Default language code device: Device for ML models ("cuda", "cpu", "auto") output_dir: Directory to save results )rrrrrcSs i|] \}}|dr||qS)_options)endswith)r0kvr$r$r%r3ns z,ValidatorRunner.__init__..FNr$)rr5config validators _initialized)r7rrrrrkwargsr$r$r%__init__Us zValidatorRunner.__init__c Cs|jrdStd|jjrAzddlm}|dddi|jj|jd<tdWnty@}z td |WYd}~nd}~ww|jj r|zdd l m }|dd|jj d |jj |jd <td Wnty{}z td|WYd}~nd}~wwtdt|jdd|_dS)z"Initialize all enabled validators.Nz#[Runner] Initializing validators...r )IndicMFAValidatorenabledTindicmfau%[Runner] ✓ IndicMFA validator addedu#[Runner] ✗ IndicMFA unavailable: )IndicConformerValidator)rDrindic_conformeru+[Runner] ✓ IndicConformer validator addedu)[Runner] ✗ IndicConformer unavailable: [Runner] z validators initializedr$)r@printr>rindicmfa_validatorrCrr? ImportErrorrindic_conformer_validatorrFrrlen)r7rCerFr$r$r%_init_validatorsts@     z ValidatorRunner._init_validatorsr,cCs|dd|jDS)z%Get information about all validators.cSr-r$)get_info)r0r1 validatorr$r$r%r3r4z6ValidatorRunner.get_validator_info..)rOr?r5r6r$r$r%get_validator_infosz"ValidatorRunner.get_validator_infoNr'r(r?c Cs@|t|p |jj}t}t|||d}|pt|j}|D]t} | |jvr2t d| dq"|j| } | j s;q"t d| dz*| j |||d} | |j | <| j rVdnd} | jp\d} t d | d | d | d d Wq"ty}zt d| d|t| |dt|d|j | <WYd}~q"d}~wwt||_|S)a Run validation on a single audio file. Args: audio_path: Path to audio file reference_text: Reference transcription (required for MFA) language: Language code (defaults to config) validators: Specific validators to run (None = all enabled) Returns: AggregatedResult with results from all validators r'r(rz%[Runner] Warning: Unknown validator ''z[Runner] Running ...✓✗r*rH : .2fsu [Runner] ✗ z failed: F)validator_namer'success error_messageN)rOr r>rtimer&listr?keysrIrDvalidater)r]processing_time_sec Exceptionr r"r+)r7r'r(rr?lang start_timer2validators_to_runr1rQvalidation_resultstatus proc_timerNr$r$r%rbsN    " zValidatorRunner.validate audio_pathsreference_textsprogress_callbackc Csg}|p dgt|}t|}td|dtt||D]0\} \} } |r.|| d|ntd| dd|dt| j|j| | ||d} || q|S) a Run validation on multiple audio files. Args: audio_paths: List of audio file paths reference_texts: List of reference transcriptions language: Language code validators: Specific validators to run progress_callback: Called with (current, total) for progress Returns: List of AggregatedResults Nz[Runner] Processing z files...r z[Runner] File /rY)r'r(rr?)rMrI enumeratezipr r1rbappend) r7rkrlrr?rmr) ref_textstotalir'ref_textr2r$r$r%validate_batchs $ zValidatorRunner.validate_batchr)filenamecCstj|jjdd|durtd}d|d}tj|jj|}t |jj t |j dt|dd |Dd }t|d d d }tj||dddWdn1sZwYtd||S)z Save validation results to JSON file. Args: results: List of AggregatedResults filename: Optional filename (auto-generated if None) Returns: Path to saved file T)exist_okNz %Y%m%d_%H%M%Svalidation_results_z.json)rr?cSsg|]}|qSr$r.)r0rr$r$r% -sz0ValidatorRunner.save_results..) timestampr> results_countr)wzutf-8)encodingF) ensure_asciiindentzunner] Results saved to: )osmakedirsr>rrnowstrftimepathjoin isoformatrr`r?rarMopenjsondumprI)r7r)rwr| output_path output_datafr$r$r%rs"     zValidatorRunner.save_resultscCs$|jD]}|qtddS)z&Release resources from all validators.z[Runner] Cleanup completeN)r?valuescleanuprI)r7rQr$r$r%r6s  zValidatorRunner.cleanup)TTrrr)NNN)NNNN)N)rrrrr r"rBrOrrrRrrr&rbcallablervrrr$r$r$r%r9Asx #   H   . 'r9rr'r(rr,cKs.tdd|i|}|j|||d}||S)a4 Convenience function to run all validators on a single file. Args: audio_path: Path to audio file reference_text: Reference transcription language: Language code **kwargs: Additional options for ValidatorRunner Returns: AggregatedResult rrSNr$)r9rbr)r'r(rrArunnerr2r$r$r%run_all_validators=sr__main__zRun transcription validators) descriptionzPath to audio file)helpz --referencez-rzReference transcriptionz --languagez-lz Language code)defaultrz--outputz-orzOutput directoryz --disable-mfa store_true)actionz--disable-conformer)rrrrrS z<============================================================zValidation SummaryzAudio: z Language: z Total time: rZr[rVrWrX:z Transcription: drUz Word alignments: z wordsz Confidence: z.3fz Error: )Nr)7rrrr_typingrrrr dataclassesrrrpathlibr baser r r rr&r9r"rrargparseArgumentParserparser add_argument parse_argsargs disable_mfadisable_conformerroutputrrbr' referencer2rIr+r)r5r1vrr]ri transcriptionword_alignmentsrMoverall_confidencer^rrr$r$r$r%s