o }oi-@sddlZddlZddlmZmZmZddlmZddlm Z ddl m Z ddl m Z ddlmZee dZdd lmZd5d edefddZd6d eedeedefddZdedefddZd7dedeeeefdefddZd8dededefddZd9ded edefd!d"Z   # $  $  $ $  $d:d%ed&ed'ed(ed ed)ed*ed+ed,edeed-edeeeeffd.d/Z   #  $ $  0  $d;d%ed&ed'ed*ed+ed,edeed1ed2eed-efd3d4ZdS)<N)OptionalTupleUnion)SacreBLEUScore) ROUGEScore)word_error_rate_detail)logging)LogMode)bleurouge) DictConfig. configreturnc Cstg}|D].\}}|r|||n|}t|tr*|t||||d|q||d|q||S)a] Flatten a DictConfig object into a string of parameter names and their values. Args: config (DictConfig): The input DictConfig object. parent_key (str): The parent key for nested configurations. sep (str): Separator between keys. Returns: str: Flattened string of parameter names and their values. )sepjoin=)items isinstancer extendflatten_dict_configsplitappendr)r parent_keyrrrkvnew_keyr_/home/ubuntu/.local/lib/python3.10/site-packages/nemo/collections/asr/parts/utils/eval_utils.pyr"s   r exclude_keyscsL|sdSd}t||d|}rfdd|D}ddd|D}|S)z Flatten a DictConfig object into a string of hydra overrides for commandline, for example: >>> config = OmegaConf.create({"foo": {"bar": 1, "baz": 2}}) >>> get_hydra_override_from_config(config) "++foo.bar=1 ++foo.baz=2" r r)rcs&g|]tfddDsqS)csg|] }|ddkqS)rr)r).0yxrr Dsz=get_hydra_override_from_config...)any)r"r!r$r r&Ds&z2get_hydra_override_from_config.. cSsg|]}d|qS)z++r)r"r%rrr r&Es)rrr)rr!r overrides param_strrr(r get_hydra_override_from_config8sr,textcCstdd|}|S)zS Remove spaces before punctuations, e.g. "hello , world" -> "hello, world" z(\w)\s+([.,;!?])z\1\2)resub)r-resultrrr strip_spaces_before_punctuationsIsr1 punctuationscCs,|s dddD}|D]}||d}q |S)z+ Remove punctuations from a string cSg|]}|qSrrr"charrrr r&Vz'remove_punctuations..z!"#$%&()*+,-./:;<=>?@[\]^_`{|}~r )replace)r-r2 punctuationrrr remove_punctuationsQs r9Ten_str num_to_wordscCsdddD}dddD}dddD}|}|}|D]}||d}q|D]}||d }q*|D]}||d }q5|rS|d krKt|d d }ntjd tjdd | }|S)zY Remove unauthorized characters in a string, lower it and remove unneeded spaces cSr3rrr4rrr r&ar6zclean_label..u7/?*",.:=?_{|}~¨«·»¡¿„…‧‹›≪≫!:;ː→cSr3rrr4rrr r&br6u$`¨´‘’“”`ʻ‘’“"‘”cSr3rrr4rrr r&cr6u‘’ʻ‘’‘r r)'r:langidzyCurrently support basic num_to_words in English only. Please use Text Normalization to convert other languages! Skipping!mode) striplowerr7convert_num_to_wordsrwarningr ONCErr)r;r<r?replace_with_spacereplace_with_blankreplace_with_aposiretrrr clean_label]s(rLr?c Cs|dkregd}|}|}d}g}|D]H}|rXt|}|rW|d}||} || t|d}|sUd} |ddd}|D]} | | d7} qB|| d7}||s"q||d7}q|}|Stjdtj d |S) z} Convert digits to corresponding words. Note this is a naive approach and could be replaced with text normalization. r:) zeroonetwothreefourfivesixseveneightniner Nr)zoCurrently support basic num_to_words in English only. Please use Text Normalization to convert other languages!r@) rBrisdigitintrclearrrEr rF) r;r?r<wordsout_strnum_wordwordnumdigit digit_wordnum_strelerrr rDys>    rD pred_textF pred_manifestgt_text_attr_namepred_text_attr_nameclean_groundtruth_textuse_ceroutput_filenameignore_capitalizationignore_punctuationstrip_punc_spacec ! Csg} g} g} |r dnd}t|d}|D]}|}|sqt|}||vrFd|vr/d}n td|ddd|fWdS||}||}|rZt||d}|rit|| d }t|| d }n | rst |}t |}|r}| }| }t |g|g|d \}}}}}|||<||d <||d <||d <||d<| || || |qWdn1swYt | | |d \}}}}}|s|}n|}t|d}| D]}t |||d|qWdn1swYdt| d |||d |d |d|i} || |fS)a  Calculate wer, inserion, deletion and substitution rate based on groundtruth text and pred_text_attr_name (pred_text) We use WER in function name as a convention, but Error Rate (ER) currently support Word Error Rate (WER) and Character Error Rate (CER) cerwerrr-ground-truth text attribute z= is not present in manifest! Cannot calculate WER. Returning!Nr>r2) hypotheses referencesrjtokensins_ratedel_ratesub_ratewrsamples)open readlinesrBjsonloadsrinforLr9r1rCrrdumpwriteflushlen)!rfrgrhrir?rjrkrlrmr2rnr{hypsrefs eval_metricfplinesamplehyprefrprvrwrxry total_wer total_tokenstotal_ins_ratetotal_del_ratetotal_sub_rateoutput_manifest_w_werfout total_resrrr cal_write_wers              .     rr metric metric_argsc  Cs6g} g} g} |tvrtd|dt|r!t|di|nt|} t|d}|D]}|}|s9q0t|}||vrcd|vrId}n t d|d|ddd|fWdS||}||}|r~t ||d}t ||d}n | rt |}t |}|r| }| }|d kr| |g|gg }n| || }|||<| || || |q0Wdn1swY|d krd d | D} | | |  }|s|}n|}t|d }| D]}t|||d |qWdn 1s wYdt| ||i}|||fS)Nzmetric z& is not supported! Please choose from rqr-rrz. is not present in manifest! Cannot calculate z . Returning!rsr cSsg|]}|gqSrr)r"rrrr r&8sz)cal_write_text_metric..rzrr{r)TEXT_METRICS_MAPPING ValueErrorkeysr|r}rBr~rrrr9r1rCitemrrrrr)rfrgrhrkrlrmr2rrrnr{rrmetric_calculatorrrrrrscore total_scorerrrrrr cal_write_text_metricsr            (     r)r rr)NN)N)Tr:)r:) Nr-reFr:FNFFNF) Nr-reNFFNr NF)r~r.typingrrrtorchmetrics.textrtorchmetrics.text.rouger nemo.collections.asr.metrics.werr nemo.utilsrnemo.utils.nemo_loggingr r omegaconfr strrlistr,r1r9boolrLrDdictrrrrrr s       $ %    _