import inspect
import math
from typing import Callable, List, Optional, Tuple, Union

import torch
import torch.nn.functional as F
from torch import nn

from diffusers.models.attention_processor import Attention, AttnProcessor2_0

try:
    import flash_attn_interface
    IS_FLASH3_AVAILABLE = True
except Exception as e:
    print(f"flash_attn3 load fail: {e}")
    IS_FLASH3_AVAILABLE = False


class CogVideoXFlashAttn3ControlnetXsProcessor:
    def __init__(self):
        pass

    def __call__(
        self,
        attn: Attention,
        hidden_states: torch.Tensor,
        encoder_hidden_states = None,
        attention_mask: Optional[torch.Tensor] = None,
        image_rotary_emb: Optional[torch.Tensor] = None,
    ) -> torch.Tensor:

        batch_size, sequence_length, _ = hidden_states.shape

        if attention_mask is not None:
            attention_mask = attn.prepare_attention_mask(attention_mask, sequence_length, batch_size)
            attention_mask = attention_mask.view(batch_size, attn.heads, -1, attention_mask.shape[-1])

        query = attn.to_q(hidden_states)
        key = attn.to_k(hidden_states)
        value = attn.to_v(hidden_states)

        inner_dim = key.shape[-1]
        head_dim = inner_dim // attn.heads

        query = query.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
        key = key.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
        value = value.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)

        if attn.norm_q is not None:
            query = attn.norm_q(query)
        if attn.norm_k is not None:
            key = attn.norm_k(key)

        # Apply RoPE if needed
        if image_rotary_emb is not None:
            from diffusers.models.embeddings import apply_rotary_emb

            query = apply_rotary_emb(query, image_rotary_emb)
            if not attn.is_cross_attention:
                key = apply_rotary_emb(key, image_rotary_emb)

        hidden_states = flash_attn_interface.flash_attn_func(
            query.transpose(1, 2), key.transpose(1, 2), value.transpose(1, 2)
        )

        hidden_states = hidden_states[0]

        hidden_states = hidden_states.reshape(batch_size, -1, attn.heads * head_dim)

        # linear proj
        hidden_states = attn.to_out[0](hidden_states)
        # dropout
        hidden_states = attn.to_out[1](hidden_states)

        return hidden_states

class CogVideoXFlashAttn3Processor:
    def __init__(self):
        pass

    def __call__(
        self,
        attn: Attention,
        hidden_states: torch.Tensor,
        encoder_hidden_states: torch.Tensor,
        attention_mask: Optional[torch.Tensor] = None,
        image_rotary_emb: Optional[torch.Tensor] = None,
    ) -> torch.Tensor:
        text_seq_length = encoder_hidden_states.size(1)

        hidden_states = torch.cat([encoder_hidden_states, hidden_states], dim=1)

        batch_size, sequence_length, _ = (
            hidden_states.shape if encoder_hidden_states is None else encoder_hidden_states.shape
        )

        if attention_mask is not None:
            attention_mask = attn.prepare_attention_mask(attention_mask, sequence_length, batch_size)
            attention_mask = attention_mask.view(batch_size, attn.heads, -1, attention_mask.shape[-1])

        query = attn.to_q(hidden_states)
        key = attn.to_k(hidden_states)
        value = attn.to_v(hidden_states)

        inner_dim = key.shape[-1]
        head_dim = inner_dim // attn.heads

        query = query.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
        key = key.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)
        value = value.view(batch_size, -1, attn.heads, head_dim).transpose(1, 2)

        if attn.norm_q is not None:
            query = attn.norm_q(query)
        if attn.norm_k is not None:
            key = attn.norm_k(key)

        # Apply RoPE if needed
        if image_rotary_emb is not None:
            from diffusers.models.embeddings import apply_rotary_emb

            query[:, :, text_seq_length:] = apply_rotary_emb(query[:, :, text_seq_length:], image_rotary_emb)
            if not attn.is_cross_attention:
                key[:, :, text_seq_length:] = apply_rotary_emb(key[:, :, text_seq_length:], image_rotary_emb)

        hidden_states = flash_attn_interface.flash_attn_func(
            query.transpose(1, 2), key.transpose(1, 2), value.transpose(1, 2)
        )

        hidden_states = hidden_states[0]

        hidden_states = hidden_states.reshape(batch_size, -1, attn.heads * head_dim)

        # linear proj
        hidden_states = attn.to_out[0](hidden_states)
        # dropout
        hidden_states = attn.to_out[1](hidden_states)

        encoder_hidden_states, hidden_states = hidden_states.split(
            [text_seq_length, hidden_states.size(1) - text_seq_length], dim=1
        )
        return hidden_states, encoder_hidden_states