# Copyright (c) 2021  PaddlePaddle Authors. 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 copy
import logging

import paddle
import paddle.fluid.contrib.slim.quantization as Q
from paddle.fluid.contrib.slim.quantization import AbsmaxQuantizer
from paddle.fluid.contrib.slim.quantization import HistQuantizer
from paddle.fluid.contrib.slim.quantization import KLQuantizer
from paddle.fluid.contrib.slim.quantization import PerChannelAbsmaxQuantizer
from ...common import get_logger

_logger = get_logger(__name__, level=logging.INFO)

__all__ = [
    'PTQ',
    'AbsmaxQuantizer',
    'HistQuantizer',
    'KLQuantizer',
    'PerChannelAbsmaxQuantizer',
]


class PTQ(object):
    """
    Static post training quantization.
    """

    def __init__(self,
                 activation_quantizer=Q.KLQuantizer(),
                 weight_quantizer=Q.PerChannelAbsmaxQuantizer()):
        """
        Args:
            activation_quantizer(Quantizer): The quantizer method for activation.
                Default: KLQuantizer.
            weight_quantizer(Quantizer): The quantizer method for weight.
                Default: PerChannelAbsmaxQuantizer.
        """
        assert isinstance(activation_quantizer, tuple(Q.SUPPORT_ACT_QUANTIZERS))
        assert isinstance(weight_quantizer, tuple(Q.SUPPORT_WT_QUANTIZERS))

        quant_config = Q.PTQConfig(
            activation_quantizer=activation_quantizer,
            weight_quantizer=weight_quantizer)

        self.ptq = Q.ImperativePTQ(quant_config=quant_config)

    def quantize(self, model, inplace=False):
        """
        Quantize the input model.

        Args:
            model(paddle.nn.Layer): The model to be quantized.
            inplace(bool): Whether apply quantization to the input model.
                           Default: False.
        Returns:
            quantized_model(paddle.nn.Layer): The quantized model.
        """
        assert isinstance(model, paddle.nn.Layer), \
            "The model must be the instance of paddle.nn.Layer."

        return self.ptq.quantize(model=model, inplace=inplace)

    def save_quantized_model(self, model, path, input_spec=None):
        """
        Save the quantized inference model.

        Args:
            model (Layer): The model to be saved.
            path (str): The path prefix to save model. The format is 
                ``dirname/file_prefix`` or ``file_prefix``.
            input_spec (list[InputSpec|Tensor], optional): Describes the input
                of the saved model's forward method, which can be described by
                InputSpec or example Tensor. If None, all input variables of 
                the original Layer's forward method would be the inputs of
                the saved model. Default: None.

        Returns:
            None
        """
        assert isinstance(model, paddle.nn.Layer), \
            "The model must be the instance of paddle.nn.Layer."

        training = model.training
        if training:
            model.eval()

        self.ptq.save_quantized_model(
            model=model, path=path, input_spec=input_spec)

        if training:
            model.train()