# Copyright 2021 Zilliz. 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.

import sys
from pathlib import Path
import torch
from torchvision import transforms

from towhee.types.image_utils import to_pil
from towhee.operator.base import NNOperator, OperatorFlag
from towhee.types.arg import arg, to_image_color
from towhee import register


@register(output_schema=['vec'])
class RuClip(NNOperator):
    """
    Russian CLIP multi-modal embedding operator
    """
    def __init__(self, model_name: str, modality: str):
        self.modality = modality
        self.device = "cuda" if torch.cuda.is_available() else "cpu"
        
        path = str(Path(__file__).parent)
        sys.path.append(path)
        import ruclip
        sys.path.pop()
        clip, processor = ruclip.load('ruclip-vit-base-patch32-384', device=self.device)
        templates = ['{}', 'это {}', 'на картинке {}', 'это {}, домашнее животное']
        self.predictor = ruclip.Predictor(clip, processor, self.device, bs=1, templates=templates)

    def inference_single_data(self, data):
        if self.modality == 'image':
            vec = self._inference_from_image(data)
        elif self.modality == 'text':
            vec = self._inference_from_text(data)
        else:
            raise ValueError("modality[{}] not implemented.".format(self.modality))
        return vec.detach().cpu().numpy().flatten()

    def __call__(self, data):
        if not isinstance(data, list):
            data = [data]
        else:
            data = data
        results = []
        for single_data in data:
            result = self.inference_single_data(single_data)
            results.append(result)
        if len(data) == 1:
            return results[0]
        else:
            return results

    def _inference_from_text(self, text):
        text_features =  self.predictor.get_text_latents([text])
        return text_features

    @arg(1, to_image_color('RGB'))
    def _inference_from_image(self, img):
        img = to_pil(img)
        image_features = self.predictor.get_image_latents([img])
        return image_features