o )iM* @sddlmZddlmZmZmZmZmZmZm Z ddl Z ddl m Z m Z mZmZer2ddlmZmZGddde ZGdd d e ZGd d d e ZeeeeefZ d ed eefddZd ed eefddZedeeddZedeeddZGddde eeefZeeefZ Gddde Z   d6dee deee d eedeed ef ddZ!Gdd d e Z" d7d!e j#deed e"fd"d#Z$eee"d$fZ% Gd%d&d&e Z&eee"d$fZ' ee%e&fZ( ed'eed(Z)ed)eed(Z* d7d*e)d+ee*d,ee+eefd ee)e*ffd-d.Z, d7d/ee)d0eee*d,eeee+eefe+eeffd eee)e*ffd1d2Z-d3eee)e*fd ee.e)ee*ffd4d5Z/dS)8)Iterable) TYPE_CHECKINGAnyGenericLiteralOptionalUnioncastN) NotRequired TypedDictTypeIsTypeVar)MultiModalDataDictMultiModalInputsc@sLeZdZUdZeed< eded< eeeefed< eeed<dS) TextPromptzSchema for a text prompt.promptrmulti_modal_datamm_processor_kwargs cache_saltN) __name__ __module__ __qualname____doc__str__annotations__r dictrrr\/home/app/PaddleOCR-VL-test/.venv_paddleocr/lib/python3.10/site-packages/vllm/inputs/data.pyr s   rc@sbeZdZUdZeeed< eeeed< eded< eee e fed< ee ed<dS) TokensPromptzSchema for a tokenized prompt.prompt_token_idstoken_type_idsrrrrN) rrrrlistintrr rrrrrrrr's    rc@s*eZdZUdZejed< eeed<dS) EmbedsPromptz2Schema for a prompt provided via token embeddings. prompt_embedsrN) rrrrtorchTensorrr rrrrrr#Ds   r#rreturncCst|to d|vo d|vSNrr$ isinstancerrrrris_tokens_promptir,cCst|to d|vo d|vSr(r)r+rrris_embeds_promptnr-r._T1_coT)bounddefault covariant_T2_coc@s:eZdZUdZeed<eeed<ee e e fed<dS)ExplicitEncoderDecoderPromptaw Represents an encoder/decoder model input prompt, comprising an explicit encoder prompt and a decoder prompt. The encoder and decoder prompts, respectively, may be formatted according to any of the [`SingletonPrompt`][vllm.inputs.data.SingletonPrompt] schemas, and are not required to have the same schema. Only the encoder prompt may have multi-modal data. mm_processor_kwargs should be at the top-level, and should not be set in the encoder/decoder prompts, since they are agnostic to the encoder/decoder. Note that an [`ExplicitEncoderDecoderPrompt`][vllm.inputs.data.ExplicitEncoderDecoderPrompt] may not be used as an input to a decoder-only model, and that the `encoder_prompt` and `decoder_prompt` fields of this data structure themselves must be [`SingletonPrompt`][vllm.inputs.data.SingletonPrompt] instances. encoder_promptdecoder_promptrN) rrrrr/rrr3r rrrrrrrr4~s  r4c@sZeZdZUdZeded< eeed< eeeed< ee ed< ee ed<dS) TokenInputszRepresents token-based inputs.tokentyperr rrN) rrrrrrr!r"r rrrrrr7s     r7rr rcCs@td|d}|dur||d<|dur||d<|dur||d<|S)zQConstruct [`TokenInputs`][vllm.inputs.data.TokenInputs] from optional values.r8)r9rNrr r)r7)rr rrinputsrrr token_inputss r;c@s8eZdZUdZeded< ejed< ee ed<dS) EmbedsInputsz#Represents embeddings-based inputs.embedsr9r$rN) rrrrrrr%r&r rrrrrr<s    r<r$cCs td|d}|dur||d<|S)zSConstruct [`EmbedsInputs`][vllm.inputs.data.EmbedsInputs] from optional values.r=)r9r$Nr)r<)r$rr:rrr embeds_inputss r>rc@s4eZdZUdZeedfed< eedfed<dS)EncoderDecoderInputsz The inputs in [`LLMEngine`][vllm.engine.llm_engine.LLMEngine] before they are passed to the model executor. This specifies the required data for encoder-decoder models. rencoderdecoderN)rrrrrr7rrrrrr?s r?_T1)r0r1_T2r5r6rcCs|duri}t|||dS)Nr5r6r)r4rDrrrbuild_explicit_enc_dec_promptsrE enc_prompts dec_promptscsRdur ttttfittrfddt||DSddt||DS)a Zip encoder and decoder prompts together into a list of [`ExplicitEncoderDecoderPrompt`][vllm.inputs.data.ExplicitEncoderDecoderPrompt] instances. ``mm_processor_kwargs`` may also be provided; if a dict is passed, the same dictionary will be used for every encoder/decoder prompt. If an iterable is provided, it will be zipped with the encoder/decoder prompts. Nc s*g|]\}}t||ttttfqSr)rEr rrr).0r5r6rrr 6sz'zip_enc_dec_prompts..cSsg|] \}}}t|||qSr)rE)rHr5r6Zmm_proc_kwargsrrrrJ>s )r rrrr*zip)rFrGrrrIrzip_enc_dec_prompts$s   rLenc_dec_promptscCsdd|DS)NcSsg|] }|d|dfqS)r5r6r)rHZenc_dec_promptrrrrJIs z)to_enc_dec_tuple_list..r)rMrrrto_enc_dec_tuple_listFsrN)NNN)N)0collections.abcrtypingrrrrrrr r%Ztyping_extensionsr r r r Zvllm.multimodal.inputsrrrrr#rZSingletonPromptr,r.r/r3r4Z PromptTyper7r!r"r;r<r&r>ZDecoderOnlyInputsr?ZSingletonInputsZProcessorInputsrBrCrrErLtuplerNrrrrs $            "