processing_sailvl.py

# coding=utf-8
# Copyright 2025 The Moonshot Team and HuggingFace Inc. team. All rights reserved.
#
# The code is based on the Qwen2VL processor (qwen2_vl/processing_qwen2_vl.py), but modified for KimiVL.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Processor class for KimiVL.
"""

from typing import List, Union

from transformers.feature_extraction_utils import BatchFeature
from transformers.image_utils import ImageInput
from transformers.processing_utils import ProcessingKwargs, ProcessorMixin, Unpack, _validate_images_text_input_order
from transformers.tokenization_utils_base import PreTokenizedInput, TextInput
from transformers.utils import logging
import torch


logger = logging.get_logger(__name__)


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


class SAILVLProcessor(ProcessorMixin):
    # r"""
    # Constructs a KimiVL processor which wraps a KimiVL image processor and a tokenizer into a single processor.
    # [`KimiVLProcessor`] offers all the functionalities of [`KimiVLImageProcessor`] and [`TikTokenTokenizer`]. See the
    # [`~KimiVLProcessor.__call__`] and [`~KimiVLProcessor.decode`] for more information.
    # Args:
    #     image_processor ([`KimiVLImageProcessor`], *optional*):
    #         The image processor is a required input.
    #     tokenizer ([`TikTokenTokenizer`], *optional*):
    #         The tokenizer is a required input.
    #     chat_template (`str`, *optional*): A Jinja template which will be used to convert lists of messages
    #         in a chat into a tokenizable string.
    # """

    attributes = ["image_processor", "tokenizer"]
    valid_kwargs = [ "chat_template"]
    image_processor_class = "AutoImageProcessor"
    tokenizer_class = "AutoTokenizer"

    def __init__(
        self,
        image_processor=None,
        tokenizer=None,
        chat_template=None,
        **kwargs,
    ):
        self.image_token = "<image>"
        self.num_image_token = 256
        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: Unpack[SAILVLProcessorKwargs],
    ) -> BatchFeature:
        """
        Main method to prepare for the model one or several sequences(s) and image(s). This method forwards the `text`
        and `kwargs` arguments to TikTokenTokenizer's [`~TikTokenTokenizer.__call__`] if `text` is not `None` to encode
        the text. To prepare the image(s), this method forwards the `images` and `kwrags` arguments to
        CLIPImageProcessor's [`~CLIPImageProcessor.__call__`] if `images` is not `None`. Please refer to the docstring
        of the above two methods for more information.
        Args:
            images (`PIL.Image.Image`, `np.ndarray`, `torch.Tensor`, `List[PIL.Image.Image]`, `List[np.ndarray]`, `List[torch.Tensor]`):
                The image or batch of images to be prepared. Each image can be a PIL image, NumPy array or PyTorch
                tensor. Both channels-first and channels-last formats are supported.
            text (`str`, `List[str]`, `List[List[str]]`):
                The sequence or batch of sequences to be encoded. Each sequence can be a string or a list of strings
                (pretokenized string). If the sequences are provided as list of strings (pretokenized), you must set
                `is_split_into_words=True` (to lift the ambiguity with a batch of sequences).
            return_tensors (`str` or [`~utils.TensorType`], *optional*):
                If set, will return tensors of a particular framework. Acceptable values are:
                - `'tf'`: Return TensorFlow `tf.constant` objects.
                - `'pt'`: Return PyTorch `torch.Tensor` objects.
                - `'np'`: Return NumPy `np.ndarray` objects.
                - `'jax'`: Return JAX `jnp.ndarray` objects.
        Returns:
            [`BatchFeature`]: A [`BatchFeature`] with the following fields:
            - **input_ids** -- List of token ids to be fed to a model. Returned when `text` is not `None`.
            - **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when
              `return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if `text` is not
              `None`).
            - **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`.
        """
        if images is None and text is None:
            raise ValueError("You have to specify at least one of `images` or `text`.")

        # check if images and text inputs are reversed for BC
        images, text = _validate_images_text_input_order(images, text)

        output_kwargs = self._merge_kwargs(
            SAILVLProcessorKwargs,
            tokenizer_init_kwargs=self.tokenizer.init_kwargs,
            **kwargs,
        )
        # print(output_kwargs)
        if images is not None:
            image_inputs = self.image_processor(images, **output_kwargs["images_kwargs"])

        else:
            image_inputs = {}

        if isinstance(text, str):
            text = [text]
        elif not isinstance(text, list) and not isinstance(text[0], str):
            raise ValueError("Invalid input text. Please provide a string, or a list of strings")

        query = text[0]
        if images is not None:
            for num_patches in image_inputs['num_patches_list']:
                image_tokens = '<img>' + '<IMG_CONTEXT>' * self.num_image_token * num_patches + '</img>'
                query = query.replace('<image>', image_tokens, 1)
            image_inputs.pop('num_patches_list')

        model_inputs = self.tokenizer(query, return_tensors='pt')
        input_ids = model_inputs['input_ids'].cuda()
        attention_mask = model_inputs['attention_mask'].cuda()
        stop_word = '<|im_end|>'
        eos_token_id = self.tokenizer.convert_tokens_to_ids(stop_word)


        text_inputs = {'input_ids': input_ids, 'attention_mask': attention_mask, 'eos_token_id': eos_token_id}
        return BatchFeature(data={**text_inputs, **image_inputs})

    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__ = ["SAILVLProcessorKwargs"]