o wih @s(ddlZddlmZddlmZmZddlmZddlm Z ddl m Z m Z m Z mZddlmZmZddlmZmZmZmZmZe d d hBZe d d hBZe d dhBZe ddhBZe ddhBZe ddhBZGdddeZdee e fdee e ffddZ!e eedededee ej"ffddZ#dS)N)MonoCut)CutMixedCut)ifnone)registered_prompt_format_fn) BOOL_FALSE BOOL_TRUE PNC_FALSEPNC_TRUE)ModalityPromptFormatter) CANARY2_BOCTX CANARY_BOS CANARY_EOSCANARY_SPECIAL_TOKENIZERCanaryTokenizeritnz<|itn|>noitn <|noitn|> timestampz <|timestamp|> notimestamp<|notimestamp|>diarizez <|diarize|> nodiarize <|nodiarize|>c seZdZdZdZdededeje dddd d ejejej e e Bej e e Bej eeBej eeBd d d ededejid ededejid iZdedededeeffdd ZZS)Canary2PromptFormattercanary2 assistantuserz|decodercontext|zA|emotion||source_lang||target_lang||pnc||itn||timestamp||diarize|<|emo:undefined|>z<|emo:neutral|>z <|emo:angry|>z <|emo:happy|>z <|emo:sad|>)decodercontextemotion source_lang target_langpncrrr)templateslots user_partialr z|text|textprompt_templateexpected_slots slot_valuesreturncst|}tj|||dS)N)r)r*r+)%map_manifest_values_to_special_tokenssuper encode_turn)selfr)r*r+ __class__d/home/ubuntu/sommelier/.venv/lib/python3.10/site-packages/nemo/collections/common/prompts/canary2.pyr/]sz"Canary2PromptFormatter.encode_turn)__name__ __module__ __qualname__NAME OUTPUT_ROLEr rr Text TextLiteralr r ITN_TRUE ITN_FALSETIMESTAMP_TRUETIMESTAMP_FALSE DIARIZE_TRUE DIARIZE_FALSErTEMPLATEstrdictlistintr/ __classcell__r3r3r1r4r-s:      *-rr+r,cCs|}d}dD]"}||vr*||}dr|ds*||}d|d||<d}qdD]-}d|d}d|d}||vrZ||||fvrZ||dd d d |fvrT|n|||<d}q-|rgtj|vrgt|tj<|S) NF)r"r#z<|z|>T)r$rrrz<|noyes1Truetrue)copy startswithendswithr PROMPT_LANGUAGE_SLOTr)r+any_special_token_presentkvval true_token false_tokenr3r3r4r-js$$  " r-cutpromptc sttrjttstddddh}|tj}|r.tdjd|ddd d d d d d}fdd|D}t ||j <| D]\}}|jvrWj|n|||<qIt d|dg}d ddjD} |t dd| |j tjdjjdid||} t|jtr|jj} n|jt} | dksJd| dd| ksJd| d| ddd| d<| S)z Prepend and append control tokens to the token sequence as per Canary 2.0 format. The prompt format syntax is defined in :class:`Canary2PromptFormatter` z_Expected input audio to have a single channel (required MonoCut/MixedCut, but we received: cut=)r"r#zWe found cut with ID z% that is missing the following keys: zNPlease ensure that every utterance in the input manifests contains these keys.rrrrz<|pnc|>)r r!rrrr$csi|]}|j|qSr3)custom).0slotrVr3r4 szcanary2..r)roler& css |] }|jdur|jVqdS)N)r()r[sr3r3r4 szcanary2..rr(rzGInvalid tokenizer: tokenizer.token_to_id('{CANARY_EOS}') returned {eos} answer_idszr?r@rArrDrCr-Tensorrr3r3r3r4s*         "=&