U ,-e@sdZddlZddlmZmZmZddlmZddlm Z ddl m Z ddl m Z mZmZmZdd lmZd d lmZGd d d e ZdS)zq Processor class for InstructBLIP. Largely copy of Blip2Processor with addition of a tokenizer for the Q-Former. N)ListOptionalUnion) BatchFeature) ImageInput)ProcessorMixin)PaddingStrategyPreTokenizedInput TextInputTruncationStrategy) TensorType) AutoTokenizercseZdZdZddgZdZdZfddZdee e e e e e e fe e e eefe e eefeeeeeee e e e e e e ee eefed d dZddZddZeddZfddZeddZZS)InstructBlipProcessora Constructs an InstructBLIP processor which wraps a BLIP image processor and a LLaMa/T5 tokenizer into a single processor. [`InstructBlipProcessor`] offers all the functionalities of [`BlipImageProcessor`] and [`AutoTokenizer`]. See the docstring of [`~BlipProcessor.__call__`] and [`~BlipProcessor.decode`] for more information. Args: image_processor (`BlipImageProcessor`): An instance of [`BlipImageProcessor`]. The image processor is a required input. tokenizer (`AutoTokenizer`): An instance of ['PreTrainedTokenizer`]. The tokenizer is a required input. qformer_tokenizer (`AutoTokenizer`): An instance of ['PreTrainedTokenizer`]. The Q-Former tokenizer is a required input. image_processor tokenizerZBlipImageProcessorrcst||||_dSN)super__init__qformer_tokenizer)selfrrr __class__y/var/www/html/Darija-Ai-Train/env/lib/python3.8/site-packages/transformers/models/instructblip/processing_instructblip.pyr2szInstructBlipProcessor.__init__NTFr)imagestextadd_special_tokenspadding truncation max_lengthstridepad_to_multiple_ofreturn_attention_maskreturn_overflowing_tokensreturn_special_tokens_maskreturn_offsets_mappingreturn_token_type_ids return_lengthverbosereturn_tensorsreturncKs|dkr|dkrtdt}|dk r|jf|||||||| | | | | |||d|}|||jf|||||||| | | | | |||d|}|d|d<|d|d<|dk r|j||d}|||S) a  This method uses [`BlipImageProcessor.__call__`] method to prepare image(s) for the model, and [`BertTokenizerFast.__call__`] to prepare text for the model. Please refer to the docstring of the above two methods for more information. Nz,You have to specify at least images or text.)rrrr r!r"r#r$r%r&r'r(r)r*r+Z input_idsZqformer_input_idsZattention_maskZqformer_attention_mask)r+) ValueErrorrrupdaterpopr)rrrrrr r!r"r#r$r%r&r'r(r)r*r+kwargsencodingZ text_encodingZqformer_text_encodingZimage_encodingrrr__call__8sb  zInstructBlipProcessor.__call__cOs|jj||S)z This method forwards all its arguments to PreTrainedTokenizer's [`~PreTrainedTokenizer.batch_decode`]. Please refer to the docstring of this method for more information. )r batch_decoderargsr0rrrr3sz"InstructBlipProcessor.batch_decodecOs|jj||S)z This method forwards all its arguments to PreTrainedTokenizer's [`~PreTrainedTokenizer.decode`]. Please refer to the docstring of this method for more information. )rdecoder4rrrr6szInstructBlipProcessor.decodecCs"|jj}|jj}tt||Sr)rmodel_input_namesrlistdictfromkeys)rZtokenizer_input_namesZimage_processor_input_namesrrrr7sz'InstructBlipProcessor.model_input_namesc sTtj|rtd|dtj|ddtj|d}|j|tj|f|S)NzProvided path (z#) should be a directory, not a fileT)exist_okr) ospathisfiler-makedirsjoinrsave_pretrainedr)rZsave_directoryr0Zqformer_tokenizer_pathrrrrAs   z%InstructBlipProcessor.save_pretrainedcKs.tj|dd}|j|f|}||||S)Nr)Z subfolder)rfrom_pretrainedZ_get_arguments_from_pretrainedappend)clsZpretrained_model_name_or_pathr0rr5rrrrBs z%InstructBlipProcessor.from_pretrained)NNTFNNrNNFFFFFTN)__name__ __module__ __qualname____doc__ attributesZimage_processor_classZtokenizer_classrrrr r rboolstrr r rintr rr2r3r6propertyr7rA classmethodrB __classcell__rrrrrs^    O  r)rHr<typingrrrZimage_processing_utilsrZ image_utilsrZprocessing_utilsrZtokenization_utils_baser r r r utilsr autorrrrrrs