# coding=utf-8
# Copyright 2025 the Fast authors and The HuggingFace Inc. team. All rights reserved.
#
# 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.
"""Fast Image processor class for TextNet."""

from typing import Optional, Union

import torch
from torchvision.transforms.v2 import functional as F

from ...image_processing_utils import BatchFeature
from ...image_processing_utils_fast import BaseImageProcessorFast, DefaultFastImageProcessorKwargs
from ...image_transforms import (
    get_resize_output_image_size,
    group_images_by_shape,
    reorder_images,
)
from ...image_utils import (
    IMAGENET_DEFAULT_MEAN,
    IMAGENET_DEFAULT_STD,
    ChannelDimension,
    ImageInput,
    PILImageResampling,
    SizeDict,
)
from ...processing_utils import Unpack
from ...utils import (
    TensorType,
    auto_docstring,
)


class TextNetFastImageProcessorKwargs(DefaultFastImageProcessorKwargs):
    """
    size_divisor (`int`, *optional*, defaults to 32):
        Ensures height and width are rounded to a multiple of this value after resizing.
    """

    size_divisor: Optional[int]


@auto_docstring
class TextNetImageProcessorFast(BaseImageProcessorFast):
    resample = PILImageResampling.BILINEAR
    image_mean = IMAGENET_DEFAULT_MEAN
    image_std = IMAGENET_DEFAULT_STD
    size = {"shortest_edge": 640}
    default_to_square = False
    crop_size = {"height": 224, "width": 224}
    do_resize = True
    do_center_crop = False
    do_rescale = True
    do_normalize = True
    do_convert_rgb = True
    size_divisor = 32
    valid_kwargs = TextNetFastImageProcessorKwargs

    def __init__(self, **kwargs: Unpack[TextNetFastImageProcessorKwargs]) -> None:
        super().__init__(**kwargs)

    @auto_docstring
    def preprocess(self, images: ImageInput, **kwargs: Unpack[TextNetFastImageProcessorKwargs]) -> BatchFeature:
        return super().preprocess(images, **kwargs)

    def resize(
        self,
        image: "torch.Tensor",
        size: SizeDict,
        interpolation: Optional["F.InterpolationMode"] = None,
        antialias: bool = True,
        size_divisor: int = 32,
        **kwargs,
    ) -> "torch.Tensor":
        if size.shortest_edge:
            new_size = get_resize_output_image_size(
                image,
                size=size.shortest_edge,
                default_to_square=False,
                input_data_format=ChannelDimension.FIRST,
            )
        else:
            raise ValueError(f"Size must contain 'shortest_edge' key. Got {size}.")
        # ensure height and width are divisible by size_divisor
        height, width = new_size
        if height % size_divisor != 0:
            height += size_divisor - (height % size_divisor)
        if width % size_divisor != 0:
            width += size_divisor - (width % size_divisor)

        return super().resize(
            image, SizeDict(height=height, width=width), interpolation=interpolation, antialias=antialias
        )

    def _preprocess(
        self,
        images: list["torch.Tensor"],
        do_resize: bool,
        size: SizeDict,
        size_divisor: int,
        interpolation: Optional["F.InterpolationMode"],
        do_center_crop: bool,
        crop_size: SizeDict,
        do_rescale: bool,
        rescale_factor: float,
        do_normalize: bool,
        image_mean: Optional[Union[float, list[float]]],
        image_std: Optional[Union[float, list[float]]],
        disable_grouping: Optional[bool],
        return_tensors: Optional[Union[str, TensorType]],
        **kwargs,
    ) -> BatchFeature:
        # Group images by size for batched resizing
        grouped_images, grouped_images_index = group_images_by_shape(images, disable_grouping=disable_grouping)
        resized_images_grouped = {}
        for shape, stacked_images in grouped_images.items():
            if do_resize:
                stacked_images = self.resize(
                    image=stacked_images, size=size, interpolation=interpolation, size_divisor=size_divisor
                )
            resized_images_grouped[shape] = stacked_images
        resized_images = reorder_images(resized_images_grouped, grouped_images_index)

        # Group images by size for further processing
        # Needed in case do_resize is False, or resize returns images with different sizes
        grouped_images, grouped_images_index = group_images_by_shape(resized_images, disable_grouping=disable_grouping)
        processed_images_grouped = {}
        for shape, stacked_images in grouped_images.items():
            if do_center_crop:
                stacked_images = self.center_crop(stacked_images, crop_size)
            # Fused rescale and normalize
            stacked_images = self.rescale_and_normalize(
                stacked_images, do_rescale, rescale_factor, do_normalize, image_mean, image_std
            )
            processed_images_grouped[shape] = stacked_images

        processed_images = reorder_images(processed_images_grouped, grouped_images_index)
        processed_images = torch.stack(processed_images, dim=0) if return_tensors else processed_images

        return BatchFeature(data={"pixel_values": processed_images}, tensor_type=return_tensors)


__all__ = ["TextNetImageProcessorFast"]