# Copyright 2017 The TensorFlow 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.
# ==============================================================================
"""Tensorflow layers with added variables for parameter masking.

Branched from tensorflow/contrib/layers/python/layers/layers.py
"""
# pylint: disable=missing-docstring
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import six

from tensorflow.contrib.framework.python.ops import add_arg_scope
from tensorflow.contrib.framework.python.ops import variables
from tensorflow.contrib.layers.python.layers import initializers
from tensorflow.contrib.layers.python.layers import utils
from tensorflow.contrib.model_pruning.python.layers import core_layers as core
from tensorflow.python.framework import ops
from tensorflow.python.ops import init_ops
from tensorflow.python.ops import nn
from tensorflow.python.ops import variable_scope
from tensorflow.python.ops import variables as tf_variables


def _model_variable_getter(getter,
                           name,
                           shape=None,
                           dtype=None,
                           initializer=None,
                           regularizer=None,
                           trainable=True,
                           collections=None,
                           caching_device=None,
                           partitioner=None,
                           rename=None,
                           use_resource=None,
                           **_):
  """Getter that uses model_variable for compatibility with core layers."""
  short_name = name.split('/')[-1]
  if rename and short_name in rename:
    name_components = name.split('/')
    name_components[-1] = rename[short_name]
    name = '/'.join(name_components)
  return variables.model_variable(
      name,
      shape=shape,
      dtype=dtype,
      initializer=initializer,
      regularizer=regularizer,
      collections=collections,
      trainable=trainable,
      caching_device=caching_device,
      partitioner=partitioner,
      custom_getter=getter,
      use_resource=use_resource)


def _build_variable_getter(rename=None):
  """Build a model variable getter that respects scope getter and renames."""

  # VariableScope will nest the getters
  def layer_variable_getter(getter, *args, **kwargs):
    kwargs['rename'] = rename
    return _model_variable_getter(getter, *args, **kwargs)

  return layer_variable_getter


def _add_variable_to_collections(variable, collections_set, collections_name):
  """Adds variable (or all its parts) to all collections with that name."""
  collections = utils.get_variable_collections(collections_set,
                                               collections_name) or []
  variables_list = [variable]
  if isinstance(variable, tf_variables.PartitionedVariable):
    variables_list = [v for v in variable]
  for collection in collections:
    for var in variables_list:
      if var not in ops.get_collection(collection):
        ops.add_to_collection(collection, var)


@add_arg_scope
def masked_convolution(inputs,
                       num_outputs,
                       kernel_size,
                       stride=1,
                       padding='SAME',
                       data_format=None,
                       rate=1,
                       activation_fn=nn.relu,
                       normalizer_fn=None,
                       normalizer_params=None,
                       weights_initializer=initializers.xavier_initializer(),
                       weights_regularizer=None,
                       biases_initializer=init_ops.zeros_initializer(),
                       biases_regularizer=None,
                       reuse=None,
                       variables_collections=None,
                       outputs_collections=None,
                       trainable=True,
                       scope=None):
  """Adds an 2D convolution followed by an optional batch_norm layer.
  The layer creates a mask variable on top of the weight variable. The input to
  the convolution operation is the elementwise multiplication of the mask
  variable and the weigh

  It is required that 1 <= N <= 3.

  `convolution` creates a variable called `weights`, representing the
  convolutional kernel, that is convolved (actually cross-correlated) with the
  `inputs` to produce a `Tensor` of activations. If a `normalizer_fn` is
  provided (such as `batch_norm`), it is then applied. Otherwise, if
  `normalizer_fn` is None and a `biases_initializer` is provided then a `biases`
  variable would be created and added the activations. Finally, if
  `activation_fn` is not `None`, it is applied to the activations as well.

  Performs atrous convolution with input stride/dilation rate equal to `rate`
  if a value > 1 for any dimension of `rate` is specified.  In this case
  `stride` values != 1 are not supported.

  Args:
    inputs: A Tensor of rank N+2 of shape
      `[batch_size] + input_spatial_shape + [in_channels]` if data_format does
      not start with "NC" (default), or
      `[batch_size, in_channels] + input_spatial_shape` if data_format starts
      with "NC".
    num_outputs: Integer, the number of output filters.
    kernel_size: A sequence of N positive integers specifying the spatial
      dimensions of the filters.  Can be a single integer to specify the same
      value for all spatial dimensions.
    stride: A sequence of N positive integers specifying the stride at which to
      compute output.  Can be a single integer to specify the same value for all
      spatial dimensions.  Specifying any `stride` value != 1 is incompatible
      with specifying any `rate` value != 1.
    padding: One of `"VALID"` or `"SAME"`.
    data_format: A string or None.  Specifies whether the channel dimension of
      the `input` and output is the last dimension (default, or if `data_format`
      does not start with "NC"), or the second dimension (if `data_format`
      starts with "NC").  For N=1, the valid values are "NWC" (default) and
      "NCW".  For N=2, the valid values are "NHWC" (default) and "NCHW".
      For N=3, the valid values are "NDHWC" (default) and "NCDHW".
    rate: A sequence of N positive integers specifying the dilation rate to use
      for atrous convolution.  Can be a single integer to specify the same
      value for all spatial dimensions.  Specifying any `rate` value != 1 is
      incompatible with specifying any `stride` value != 1.
    activation_fn: Activation function. The default value is a ReLU function.
      Explicitly set it to None to skip it and maintain a linear activation.
    normalizer_fn: Normalization function to use instead of `biases`. If
      `normalizer_fn` is provided then `biases_initializer` and
      `biases_regularizer` are ignored and `biases` are not created nor added.
      default set to None for no normalizer function
    normalizer_params: Normalization function parameters.
    weights_initializer: An initializer for the weights.
    weights_regularizer: Optional regularizer for the weights.
    biases_initializer: An initializer for the biases. If None skip biases.
    biases_regularizer: Optional regularizer for the biases.
    reuse: Whether or not the layer and its variables should be reused. To be
      able to reuse the layer scope must be given.
    variables_collections: Optional list of collections for all the variables or
      a dictionary containing a different list of collection per variable.
    outputs_collections: Collection to add the outputs.
    trainable: If `True` also add variables to the graph collection
      `GraphKeys.TRAINABLE_VARIABLES` (see tf.Variable).
    scope: Optional scope for `variable_scope`.

  Returns:
    A tensor representing the output of the operation.

  Raises:
    ValueError: If `data_format` is invalid.
    ValueError: Both 'rate' and `stride` are not uniformly 1.
  """
  if data_format not in [None, 'NWC', 'NCW', 'NHWC', 'NCHW', 'NDHWC', 'NCDHW']:
    raise ValueError('Invalid data_format: %r' % (data_format,))

  layer_variable_getter = _build_variable_getter({
      'bias': 'biases',
      'kernel': 'weights'
  })

  with variable_scope.variable_scope(
      scope, 'Conv', [inputs], reuse=reuse,
      custom_getter=layer_variable_getter) as sc:
    inputs = ops.convert_to_tensor(inputs)
    input_rank = inputs.get_shape().ndims

    if input_rank == 3:
      raise ValueError('Sparse Convolution not supported for input with rank',
                       input_rank)
    elif input_rank == 4:
      layer_class = core.MaskedConv2D
    elif input_rank == 5:
      raise ValueError('Sparse Convolution not supported for input with rank',
                       input_rank)
    else:
      raise ValueError('Sparse Convolution not supported for input with rank',
                       input_rank)

    if data_format is None or data_format == 'NHWC':
      df = 'channels_last'
    elif data_format == 'NCHW':
      df = 'channels_first'
    else:
      raise ValueError('Unsupported data format', data_format)

    layer = layer_class(
        filters=num_outputs,
        kernel_size=kernel_size,
        strides=stride,
        padding=padding,
        data_format=df,
        dilation_rate=rate,
        activation=None,
        use_bias=not normalizer_fn and biases_initializer,
        kernel_initializer=weights_initializer,
        bias_initializer=biases_initializer,
        kernel_regularizer=weights_regularizer,
        bias_regularizer=biases_regularizer,
        activity_regularizer=None,
        trainable=trainable,
        name=sc.name,
        dtype=inputs.dtype.base_dtype,
        _scope=sc,
        _reuse=reuse)
    outputs = layer.apply(inputs)

    # Add variables to collections.
    _add_variable_to_collections(layer.kernel, variables_collections, 'weights')
    if layer.use_bias:
      _add_variable_to_collections(layer.bias, variables_collections, 'biases')

    if normalizer_fn is not None:
      normalizer_params = normalizer_params or {}
      outputs = normalizer_fn(outputs, **normalizer_params)

    if activation_fn is not None:
      outputs = activation_fn(outputs)
    return utils.collect_named_outputs(outputs_collections,
                                       sc.original_name_scope, outputs)


masked_conv2d = masked_convolution


@add_arg_scope
def masked_fully_connected(
    inputs,
    num_outputs,
    activation_fn=nn.relu,
    normalizer_fn=None,
    normalizer_params=None,
    weights_initializer=initializers.xavier_initializer(),
    weights_regularizer=None,
    biases_initializer=init_ops.zeros_initializer(),
    biases_regularizer=None,
    reuse=None,
    variables_collections=None,
    outputs_collections=None,
    trainable=True,
    scope=None):
  """Adds a sparse fully connected layer. The weight matrix is masked.

  `fully_connected` creates a variable called `weights`, representing a fully
  connected weight matrix, which is multiplied by the `inputs` to produce a
  `Tensor` of hidden units. If a `normalizer_fn` is provided (such as
  `batch_norm`), it is then applied. Otherwise, if `normalizer_fn` is
  None and a `biases_initializer` is provided then a `biases` variable would be
  created and added the hidden units. Finally, if `activation_fn` is not `None`,
  it is applied to the hidden units as well.

  Note: that if `inputs` have a rank greater than 2, then `inputs` is flattened
  prior to the initial matrix multiply by `weights`.

  Args:
    inputs: A tensor of at least rank 2 and static value for the last dimension;
      i.e. `[batch_size, depth]`, `[None, None, None, channels]`.
    num_outputs: Integer or long, the number of output units in the layer.
    activation_fn: Activation function. The default value is a ReLU function.
      Explicitly set it to None to skip it and maintain a linear activation.
    normalizer_fn: Normalization function to use instead of `biases`. If
      `normalizer_fn` is provided then `biases_initializer` and
      `biases_regularizer` are ignored and `biases` are not created nor added.
      default set to None for no normalizer function
    normalizer_params: Normalization function parameters.
    weights_initializer: An initializer for the weights.
    weights_regularizer: Optional regularizer for the weights.
    biases_initializer: An initializer for the biases. If None skip biases.
    biases_regularizer: Optional regularizer for the biases.
    reuse: Whether or not the layer and its variables should be reused. To be
      able to reuse the layer scope must be given.
    variables_collections: Optional list of collections for all the variables or
      a dictionary containing a different list of collections per variable.
    outputs_collections: Collection to add the outputs.
    trainable: If `True` also add variables to the graph collection
      `GraphKeys.TRAINABLE_VARIABLES` (see tf.Variable).
    scope: Optional scope for variable_scope.

  Returns:
     The tensor variable representing the result of the series of operations.

  Raises:
    ValueError: If x has rank less than 2 or if its last dimension is not set.
  """
  if not isinstance(num_outputs, six.integer_types):
    raise ValueError('num_outputs should be int or long, got %s.' %
                     (num_outputs,))

  layer_variable_getter = _build_variable_getter({
      'bias': 'biases',
      'kernel': 'weights'
  })

  with variable_scope.variable_scope(
      scope,
      'fully_connected', [inputs],
      reuse=reuse,
      custom_getter=layer_variable_getter) as sc:
    inputs = ops.convert_to_tensor(inputs)
    layer = core.MaskedFullyConnected(
        units=num_outputs,
        activation=None,
        use_bias=not normalizer_fn and biases_initializer,
        kernel_initializer=weights_initializer,
        bias_initializer=biases_initializer,
        kernel_regularizer=weights_regularizer,
        bias_regularizer=biases_regularizer,
        activity_regularizer=None,
        trainable=trainable,
        name=sc.name,
        dtype=inputs.dtype.base_dtype,
        _scope=sc,
        _reuse=reuse)
    outputs = layer.apply(inputs)

    # Add variables to collections.
    _add_variable_to_collections(layer.kernel, variables_collections, 'weights')
    if layer.bias is not None:
      _add_variable_to_collections(layer.bias, variables_collections, 'biases')

    # Apply normalizer function / layer.
    if normalizer_fn is not None:
      if not normalizer_params:
        normalizer_params = {}
      outputs = normalizer_fn(outputs, **normalizer_params)

    if activation_fn is not None:
      outputs = activation_fn(outputs)

    return utils.collect_named_outputs(outputs_collections,
                                       sc.original_name_scope, outputs)