# The implementation is adopted from Swin-Transformer-1D, made publicly available
# at https://github.com/meraks/Swin-Transformer-1D

import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.checkpoint as checkpoint
from torch.nn.init import trunc_normal_


def drop_path(x,
              drop_prob: float = 0.,
              training: bool = False,
              scale_by_keep: bool = True):
    """Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
    This is the same as the DropConnect impl I created for EfficientNet, etc networks, however,
    the original name is misleading as 'Drop Connect' is a different form of dropout in a separate paper...
    See discussion: https://github.com/tensorflow/tpu/issues/494#issuecomment-532968956 ... I've opted for
    changing the layer and argument names to 'drop path' rather than mix DropConnect as a layer name and use
    'survival rate' as the argument.
    """
    if drop_prob == 0. or not training:
        return x
    keep_prob = 1 - drop_prob
    shape = (x.shape[0], ) + (1, ) * (
        x.ndim - 1)  # work with diff dim tensors, not just 2D ConvNets
    random_tensor = x.new_empty(shape).bernoulli_(keep_prob)
    if keep_prob > 0.0 and scale_by_keep:
        random_tensor.div_(keep_prob)
    return x * random_tensor


class DropPath(nn.Module):
    """Drop paths (Stochastic Depth) per sample  (when applied in main path of residual blocks).
    """

    def __init__(self, drop_prob: float = 0., scale_by_keep: bool = True):
        super(DropPath, self).__init__()
        self.drop_prob = drop_prob
        self.scale_by_keep = scale_by_keep

    def forward(self, x):
        return drop_path(x, self.drop_prob, self.training, self.scale_by_keep)

    def extra_repr(self):
        return f'drop_prob={round(self.drop_prob,3):0.3f}'


class Mlp(nn.Module):

    def __init__(self,
                 in_features,
                 hidden_features=None,
                 out_features=None,
                 act_layer=nn.GELU,
                 drop=0.):
        super().__init__()
        out_features = out_features or in_features
        hidden_features = hidden_features or in_features
        self.fc1 = nn.Linear(in_features, hidden_features)
        self.act = act_layer()
        self.fc2 = nn.Linear(hidden_features, out_features)
        self.drop = nn.Dropout(drop)

    def forward(self, x):
        x = self.fc1(x)
        x = self.act(x)
        x = self.drop(x)
        x = self.fc2(x)
        x = self.drop(x)
        return x


def window_partition(x, window_size):
    """
    Args:
        x: (B, L, C)
        window_size (int): window size
    Returns:
        windows: (num_windows*B, window_size, C)
    """
    B, L, C = x.shape
    x = x.view(B, L // window_size, window_size, C)
    windows = x.permute(0, 1, 2, 3).contiguous().view(-1, window_size, C)
    return windows


def window_reverse(windows, window_size, L):
    """
    Args:
        windows: (num_windows*B, window_size, window_size, C)
        window_size (int): Window size
        L (int): sequence length
    Returns:
        x: (B, L, C)
    """
    B = int(windows.shape[0] / (L / window_size))
    x = windows.view(B, L // window_size, window_size, -1)
    x = x.permute(0, 1, 2, 3).contiguous().view(B, L, -1)
    return x


class WindowAttention_1D(nn.Module):
    r""" Window based multi-head self attention (W-MSA) module with relative position bias.
    It supports both of shifted and non-shifted window.
    Args:
        dim (int): Number of input channels.
        window_size (int): The height and width of the window.
        num_heads (int): Number of attention heads.
        qkv_bias (bool, optional):  If True, add a learnable bias to query, key, value. Default: True
        attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0
        proj_drop (float, optional): Dropout ratio of output. Default: 0.0
        pretrained_window_size (int): The height and width of the window in pre-training.
    """

    def __init__(self,
                 dim,
                 window_size,
                 num_heads,
                 qkv_bias=True,
                 attn_drop=0.,
                 proj_drop=0.,
                 pretrained_window_size=0):

        super().__init__()
        self.dim = dim
        self.window_size = window_size  # Wl
        self.pretrained_window_size = pretrained_window_size
        self.num_heads = num_heads

        self.logit_scale = nn.Parameter(
            torch.log(10 * torch.ones((num_heads, 1, 1))), requires_grad=True)

        # mlp to generate continuous relative position bias
        self.cpb_mlp = nn.Sequential(
            nn.Linear(1, 512, bias=True), nn.ReLU(inplace=True),
            nn.Linear(512, num_heads, bias=False))

        # get relative_coords_table
        relative_coords_l = torch.arange(
            -(self.window_size - 1), self.window_size, dtype=torch.float32)
        relative_coords_table = torch.stack(
            torch.meshgrid([relative_coords_l], indexing='ij')).permute(
                1, 0).contiguous().unsqueeze(0)  # 1, 2*Wl-1, 1
        if pretrained_window_size > 0:
            relative_coords_table[:, :, :] /= (pretrained_window_size - 1)
        else:
            relative_coords_table[:, :, :] /= (self.window_size - 1)
        relative_coords_table *= 8  # normalize to -8, 8
        relative_coords_table = torch.sign(relative_coords_table) * torch.log2(
            torch.abs(relative_coords_table) + 1.0) / np.log2(8)

        self.register_buffer('relative_coords_table', relative_coords_table)

        # get pair-wise relative position index for each token inside the window
        coords_l = torch.arange(self.window_size)
        coords = torch.stack(torch.meshgrid([coords_l],
                                            indexing='ij'))  # 1, Wl
        coords_flatten = torch.flatten(coords, 1)  # 1, Wl
        relative_coords = coords_flatten[:, :,
                                         None] - coords_flatten[:,
                                                                None, :]  # 1, Wl, Wl
        relative_coords = relative_coords.permute(1, 2,
                                                  0).contiguous()  # Wl, Wl, 1
        relative_coords[:, :,
                        0] += self.window_size - 1  # shift to start from 0
        relative_position_index = relative_coords.sum(-1)  # Wl, Wl
        self.register_buffer('relative_position_index',
                             relative_position_index)

        self.qkv = nn.Linear(dim, dim * 3, bias=False)
        if qkv_bias:
            self.q_bias = nn.Parameter(torch.zeros(dim))
            self.v_bias = nn.Parameter(torch.zeros(dim))
        else:
            self.q_bias = None
            self.v_bias = None
        self.attn_drop = nn.Dropout(attn_drop)
        self.proj = nn.Linear(dim, dim)
        self.proj_drop = nn.Dropout(proj_drop)
        self.softmax = nn.Softmax(dim=-1)

    def forward(self, x, mask=None):
        """
        Args:
            x: input features with shape of (num_windows*B, N, C)
            mask: (0/-inf) mask with shape of (num_windows, Wl, Wl) or None
        """
        B_, N, C = x.shape
        qkv_bias = None
        if self.q_bias is not None:
            qkv_bias = torch.cat(
                (self.q_bias,
                 torch.zeros_like(self.v_bias,
                                  requires_grad=False), self.v_bias))
        qkv = F.linear(input=x, weight=self.qkv.weight, bias=qkv_bias)
        qkv = qkv.reshape(B_, N, 3, self.num_heads, -1).permute(2, 0, 3, 1, 4)
        q, k, v = qkv[0], qkv[1], qkv[
            2]  # make torchscript happy (cannot use tensor as tuple)

        # cosine attention
        attn = (
            F.normalize(q, dim=-1) @ F.normalize(k, dim=-1).transpose(-2, -1))
        logit_scale = torch.clamp(
            self.logit_scale,
            max=torch.log(torch.tensor(1. / 0.01, device=attn.device))).exp()
        attn = attn * logit_scale

        relative_position_bias_table = self.cpb_mlp(
            self.relative_coords_table).view(-1, self.num_heads)
        relative_position_bias = relative_position_bias_table[
            self.relative_position_index.view(-1)].view(
                self.window_size, self.window_size, -1)  # Wl,l,nH
        relative_position_bias = relative_position_bias.permute(
            2, 0, 1).contiguous()  # nH, Wl, Wl
        relative_position_bias = 16 * torch.sigmoid(relative_position_bias)
        attn = attn + relative_position_bias.unsqueeze(0)

        if mask is not None:
            nW = mask.shape[0]
            attn = attn.view(B_ // nW, nW, self.num_heads, N,
                             N) + mask.unsqueeze(1).unsqueeze(0)
            attn = attn.view(-1, self.num_heads, N, N)
            attn = self.softmax(attn)
        else:
            attn = self.softmax(attn)

        attn = self.attn_drop(attn)

        x = (attn @ v).transpose(1, 2).reshape(B_, N, C)
        x = self.proj(x)
        x = self.proj_drop(x)
        return x


def compute_mask(L, window_size, shift_size):
    Lp = int(np.ceil(L / window_size)) * window_size
    img_mask = torch.zeros((1, Lp, 1))  # 1 Lp 1
    pad_size = int(Lp - L)
    if (pad_size == 0) or (pad_size + shift_size == window_size):
        segs = (slice(-window_size), slice(-window_size, -shift_size),
                slice(-shift_size, None))
    elif pad_size + shift_size > window_size:
        seg1 = int(window_size * 2 - L + shift_size)
        segs = (slice(-seg1), slice(-seg1, -window_size),
                slice(-window_size, -shift_size), slice(-shift_size, None))
    elif pad_size + shift_size < window_size:
        seg1 = int(window_size * 2 - L + shift_size)
        segs = (slice(-window_size), slice(-window_size, -seg1),
                slice(-seg1, -shift_size), slice(-shift_size, None))
    cnt = 0
    for d in segs:
        img_mask[:, d, :] = cnt
        cnt += 1
    mask_windows = window_partition(img_mask, window_size)  # nW, ws, 1
    mask_windows = mask_windows.squeeze(-1)  # nW, ws
    attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2)
    attn_mask = attn_mask.masked_fill(attn_mask != 0,
                                      float(-100.0)).masked_fill(
                                          attn_mask == 0, float(0.0))
    return attn_mask


class SwinTransformerBlock_1D(nn.Module):
    r""" Swin Transformer Block.
    Args:
        dim (int): Number of input channels.
        num_heads (int): Number of attention heads.
        window_size (int): Window size.
        shift_size (int): Shift size for SW-MSA.
        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
        qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
        drop (float, optional): Dropout rate. Default: 0.0
        attn_drop (float, optional): Attention dropout rate. Default: 0.0
        drop_path (float, optional): Stochastic depth rate. Default: 0.0
        act_layer (nn.Module, optional): Activation layer. Default: nn.GELU
        norm_layer (nn.Module, optional): Normalization layer.  Default: nn.LayerNorm
        pretrained_window_size (int): Window size in pre-training.
    """

    def __init__(self,
                 dim,
                 num_heads,
                 window_size=7,
                 shift_size=0,
                 mlp_ratio=4.,
                 qkv_bias=True,
                 drop=0.,
                 attn_drop=0.,
                 drop_path=0.,
                 act_layer=nn.GELU,
                 norm_layer=nn.LayerNorm,
                 pretrained_window_size=0):
        super().__init__()
        self.dim = dim
        self.num_heads = num_heads
        self.window_size = window_size
        self.shift_size = shift_size
        self.mlp_ratio = mlp_ratio

        assert 0 <= self.shift_size < self.window_size, 'shift_size must in 0-window_size'

        self.norm1 = norm_layer(dim)
        self.attn = WindowAttention_1D(
            dim,
            window_size=self.window_size,
            num_heads=num_heads,
            qkv_bias=qkv_bias,
            attn_drop=attn_drop,
            proj_drop=drop,
            pretrained_window_size=pretrained_window_size)

        self.drop_path = DropPath(
            drop_path) if drop_path > 0. else nn.Identity()
        self.norm2 = norm_layer(dim)
        mlp_hidden_dim = int(dim * mlp_ratio)
        self.mlp = Mlp(
            in_features=dim,
            hidden_features=mlp_hidden_dim,
            act_layer=act_layer,
            drop=drop)

    def forward(self, x):
        B, L, C = x.shape

        attn_mask = compute_mask(L, self.window_size,
                                 self.shift_size).to(x.device)

        shortcut = x
        # x = x.view(B, L, C)

        # padding x
        pad_r = (self.window_size - L % self.window_size) % self.window_size
        x = F.pad(x, (0, 0, 0, pad_r))
        _, Lp, _ = x.shape

        # cyclic shift
        if self.shift_size > 0:
            shifted_x = torch.roll(x, shifts=(-self.shift_size), dims=(1))
        else:
            shifted_x = x

        # partition windows
        x_windows = window_partition(shifted_x,
                                     self.window_size)  # nW*B, window_size, C
        x_windows = x_windows.view(-1, self.window_size,
                                   C)  # nW*B, window_siz, C

        # W-MSA/SW-MSA
        attn_windows = self.attn(
            x_windows, mask=attn_mask)  # nW*B, window_size, C

        # merge windows
        attn_windows = attn_windows.view(-1, self.window_size, C)
        shifted_x = window_reverse(attn_windows, self.window_size,
                                   Lp)  # B L' C

        # reverse cyclic shift
        if self.shift_size > 0:
            x = torch.roll(shifted_x, shifts=(self.shift_size), dims=(1))
        else:
            x = shifted_x
        x = x.view(B, Lp, C)
        # reverse padding x
        x = x[:, :L, :].contiguous()
        x = shortcut + self.drop_path(self.norm1(x))

        # FFN
        x = x + self.drop_path(self.norm2(self.mlp(x)))

        return x


class PatchMerging(nn.Module):
    """ Patch Merging Layer
    Args:
        dim (int): Number of input channels.
        norm_layer (nn.Module, optional): Normalization layer.  Default: nn.LayerNorm
    """

    def __init__(self, dim, norm_layer=nn.LayerNorm):
        super().__init__()
        self.dim = dim
        # self.reduction = nn.Linear(2 * dim, dim, bias=False)
        # self.norm = norm_layer(2 * dim)

    def forward(self, x):
        """ Forward function.
        Args:
            x: Input feature, tensor size (B, L, C).
        """
        B, L, C = x.shape
        x = F.pad(x, (0, 0, 0, L % 2))

        x0 = x[:, 0::2, :]  # B L/2 C
        x1 = x[:, 1::2, :]  # B L/2 C

        x = torch.maximum(x0, x1)

        return x


class BasicLayer(nn.Module):
    """ A basic Swin Transformer layer for one stage.
    Args:
        dim (int): Number of input channels.
        depth (int): Number of blocks.
        num_heads (int): Number of attention heads.
        window_size (int): Local window size.
        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
        qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
        drop (float, optional): Dropout rate. Default: 0.0
        attn_drop (float, optional): Attention dropout rate. Default: 0.0
        drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0
        norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
        downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None
        use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.
        pretrained_window_size (int): Local window size in pre-training.
    """

    def __init__(self,
                 dim,
                 depth,
                 num_heads,
                 window_size,
                 mlp_ratio=4.,
                 qkv_bias=True,
                 drop=0.,
                 attn_drop=0.,
                 drop_path=0.,
                 norm_layer=nn.LayerNorm,
                 downsample=None,
                 use_checkpoint=False,
                 pretrained_window_size=0):

        super().__init__()
        self.dim = dim
        self.depth = depth
        self.use_checkpoint = use_checkpoint

        # build blocks
        self.blocks = nn.ModuleList([
            SwinTransformerBlock_1D(
                dim=dim,
                num_heads=num_heads,
                window_size=window_size,
                shift_size=0 if (i % 2 == 0) else window_size // 2,
                mlp_ratio=mlp_ratio,
                qkv_bias=qkv_bias,
                drop=drop,
                attn_drop=attn_drop,
                drop_path=drop_path[i]
                if isinstance(drop_path, list) else drop_path,
                norm_layer=norm_layer,
                pretrained_window_size=pretrained_window_size)
            for i in range(depth)
        ])

        # patch merging layer
        if downsample is not None:
            self.downsample = downsample(dim=dim, norm_layer=norm_layer)
        else:
            self.downsample = None

    def forward(self, x):
        for blk in self.blocks:
            if self.use_checkpoint:
                x = checkpoint.checkpoint(blk, x)
            else:
                x = blk(x)

        proposal = x
        if self.downsample is not None:
            x = self.downsample(x)
        return x, proposal

    def _init_respostnorm(self):
        for blk in self.blocks:
            nn.init.constant_(blk.norm1.bias, 0)
            nn.init.constant_(blk.norm1.weight, 0)
            nn.init.constant_(blk.norm2.bias, 0)
            nn.init.constant_(blk.norm2.weight, 0)


class PatchEmbed1D(nn.Module):
    """ Video to Patch Embedding.
    Args:
        patch_size (int): Patch token size. Default: 4.
        in_chans (int): Number of input video channels. Default: 3.
        embed_dim (int): Number of linear projection output channels. Default: 96.
        norm_layer (nn.Module, optional): Normalization layer. Default: None
    """

    def __init__(self,
                 patch_size=4,
                 in_chans=32,
                 embed_dim=128,
                 norm_layer=None):
        super().__init__()
        self.patch_size = patch_size

        self.in_chans = in_chans
        self.embed_dim = embed_dim

        self.proj = nn.Conv1d(
            in_chans, embed_dim, kernel_size=patch_size, stride=patch_size)

        if norm_layer is not None:
            self.norm = norm_layer(embed_dim)
        else:
            self.norm = None

    def forward(self, x):
        """Forward function."""
        # padding
        _, _, L = x.size()
        pad_r = (self.patch_size - L % self.patch_size) % self.patch_size
        x = F.pad(x, (0, pad_r))
        x = self.proj(x)  # B C Wl
        if self.norm is not None:
            # Wl = x.size(2)
            x = x.transpose(1, 2)
            x = self.norm(x)
            # x = x.transpose(1, 2).view(-1, self.embed_dim, Wl)

        return x


class SwinTransformerV2_1D(nn.Module):

    def __init__(self,
                 patch_size=4,
                 in_chans=32,
                 embed_dim=96,
                 depths=[2, 2, 6, 2],
                 num_heads=[3, 6, 12, 24],
                 window_size=[7, 7, 7, 7],
                 mlp_ratio=4.,
                 qkv_bias=True,
                 drop_rate=0.,
                 attn_drop_rate=0.,
                 drop_path_rate=0.1,
                 norm_layer=nn.LayerNorm,
                 patch_norm=True,
                 use_checkpoint=False,
                 pretrained_window_sizes=[0, 0, 0, 0],
                 **kwargs):
        super().__init__()

        self.num_layers = len(depths)
        self.embed_dim = embed_dim
        self.patch_norm = patch_norm
        self.num_features = int(embed_dim * 2**(self.num_layers - 1))
        self.mlp_ratio = mlp_ratio

        # split image into non-overlapping patches
        self.patch_embed = PatchEmbed1D(
            patch_size=patch_size,
            in_chans=in_chans,
            embed_dim=embed_dim,
            norm_layer=norm_layer if self.patch_norm else None)

        self.pos_drop = nn.Dropout(p=drop_rate)
        # stochastic depth
        dpr = [
            x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))
        ]  # stochastic depth decay rule

        # build layers
        self.layers = nn.ModuleList()
        for i_layer in range(self.num_layers):
            layer = BasicLayer(
                dim=embed_dim,
                depth=depths[i_layer],
                num_heads=num_heads[i_layer],
                window_size=window_size[i_layer],
                mlp_ratio=self.mlp_ratio,
                qkv_bias=qkv_bias,
                drop=drop_rate,
                attn_drop=attn_drop_rate,
                drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])],
                norm_layer=norm_layer,
                downsample=PatchMerging if
                (i_layer < self.num_layers - 1) else None,
                use_checkpoint=use_checkpoint,
                pretrained_window_size=pretrained_window_sizes[i_layer])
            self.layers.append(layer)

        self.apply(self._init_weights)
        for bly in self.layers:
            bly._init_respostnorm()

    def _init_weights(self, m):
        if isinstance(m, nn.Linear):
            trunc_normal_(m.weight, std=.02)
            if isinstance(m, nn.Linear) and m.bias is not None:
                nn.init.constant_(m.bias, 0)
        elif isinstance(m, nn.LayerNorm):
            nn.init.constant_(m.bias, 0)
            nn.init.constant_(m.weight, 1.0)

    @torch.jit.ignore
    def no_weight_decay(self):
        return {'absolute_pos_embed'}

    @torch.jit.ignore
    def no_weight_decay_keywords(self):
        return {'cpb_mlp', 'logit_scale', 'relative_position_bias_table'}

    def forward_features(self, x):
        x = self.patch_embed(x)
        x = self.pos_drop(x)

        proposals = list()
        for layer in self.layers:
            x, proposal = layer(x)
            proposals.append(proposal)

        return proposals

    def forward(self, x):
        return self.forward_features(x)