processing_ax_ve.py

from typing import List, Union, Optional
from transformers.feature_extraction_utils import BatchFeature
from transformers.image_utils import ImageInput
from transformers.tokenization_utils_base import PreTokenizedInput, TextInput
from transformers.processing_utils import ProcessingKwargs, ProcessorMixin, ImagesKwargs


class AXVEImagesKwargs(ImagesKwargs):
    min_pixels: Optional[int]
    max_pixels: Optional[int]
    patch_size: Optional[int]
    merge_size: Optional[int]


class AXVEProcessorKwargs(ProcessingKwargs, total=False):
    images_kwargs: AXVEImagesKwargs
    _defaults = {
        "text_kwargs": {
            "padding": False,
        }
    }


class AXVEProcessor(ProcessorMixin):
    attributes = ["image_processor", "tokenizer"]
    image_processor_class = "AutoImageProcessor"
    tokenizer_class = "AutoTokenizer"

    def __init__(
        self,
        image_processor=None,
        tokenizer=None,
        chat_template=None,
        **kwargs
    ):        
        self.image_token = "<|image_pad|>" if not hasattr(tokenizer, "image_token") else tokenizer.image_token

        self.image_token_id = (
            tokenizer.image_token_id
            if getattr(tokenizer, "image_token_id", None)
            else tokenizer.convert_tokens_to_ids(self.image_token)
        )
        super().__init__(image_processor, tokenizer, chat_template=chat_template)
        
    def __call__(
        self,
        images: ImageInput = None,
        text: Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]] = None,
        **kwargs
    ) -> BatchFeature:
        output_kwargs = self._merge_kwargs(
            AXVEProcessorKwargs,
            tokenizer_init_kwargs=self.tokenizer.init_kwargs,
            **kwargs,
        )        
        if images is not None:
            image_inputs = self.image_processor(images=images, **output_kwargs["images_kwargs"])
            image_grid_hw = image_inputs["image_grid_hw"]
        else:
            image_inputs = {}
            image_grid_hw = None

        if not isinstance(text, list):
            text = [text]

        text = text.copy()  # below lines change text in-place
        if image_grid_hw is not None:
            merge_length = self.image_processor.merge_size**2
            index = 0
            for i in range(len(text)):
                while self.image_token in text[i]:
                    num_image_tokens = image_grid_hw[index].prod() // merge_length
                    text[i] = text[i].replace(self.image_token, "<|placeholder|>" * num_image_tokens, 1)
                    index += 1
                text[i] = text[i].replace("<|placeholder|>", self.image_token)

        return_tensors = output_kwargs["text_kwargs"].pop("return_tensors", None)
        # return_mm_token_type_ids = output_kwargs["text_kwargs"].pop("return_mm_token_type_ids", None)
        text_inputs = self.tokenizer(text, **output_kwargs["text_kwargs"])        
        text_inputs.pop('token_type_ids', None)

        return BatchFeature(data={**text_inputs, **image_inputs}, tensor_type=return_tensors)
        
    def batch_decode(self, *args, **kwargs):
        """
        This method forwards all its arguments to LlamaTokenizerFast's [`~PreTrainedTokenizer.batch_decode`]. Please
        refer to the docstring of this method for more information.
        """
        return self.tokenizer.batch_decode(*args, **kwargs)

    def decode(self, *args, **kwargs):
        """
        This method forwards all its arguments to LlamaTokenizerFast's [`~PreTrainedTokenizer.decode`]. Please refer to
        the docstring of this method for more information.
        """
        return self.tokenizer.decode(*args, **kwargs)

    @property
    def model_input_names(self):
        tokenizer_input_names = self.tokenizer.model_input_names
        image_processor_input_names = self.image_processor.model_input_names
        return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names))

__all__ = ["AXVEProcessor"]