# Copyright 2024-2025 The Alibaba Wan Team Authors. All rights reserved.
import torch

try:
    import flash_attn_interface

    def is_hopper_gpu():
        if not torch.cuda.is_available():
            return False
        device_name = torch.cuda.get_device_name(0).lower()
        return "h100" in device_name or "hopper" in device_name
    FLASH_ATTN_3_AVAILABLE = is_hopper_gpu()
except ModuleNotFoundError:
    FLASH_ATTN_3_AVAILABLE = False

try:
    import flash_attn
    FLASH_ATTN_2_AVAILABLE = True
except ModuleNotFoundError:
    FLASH_ATTN_2_AVAILABLE = False

# FLASH_ATTN_3_AVAILABLE = False

import warnings

__all__ = [
    'flash_attention',
    'attention',
]


def flash_attention(
    q,
    k,
    v,
    q_lens=None,
    k_lens=None,
    dropout_p=0.,
    softmax_scale=None,
    q_scale=None,
    causal=False,
    window_size=(-1, -1),
    deterministic=False,
    dtype=torch.bfloat16,
    version=None,
):
    """
    q:              [B, Lq, Nq, C1].
    k:              [B, Lk, Nk, C1].
    v:              [B, Lk, Nk, C2]. Nq must be divisible by Nk.
    q_lens:         [B].
    k_lens:         [B].
    dropout_p:      float. Dropout probability.
    softmax_scale:  float. The scaling of QK^T before applying softmax.
    causal:         bool. Whether to apply causal attention mask.
    window_size:    (left right). If not (-1, -1), apply sliding window local attention.
    deterministic:  bool. If True, slightly slower and uses more memory.
    dtype:          torch.dtype. Apply when dtype of q/k/v is not float16/bfloat16.
    """
    half_dtypes = (torch.float16, torch.bfloat16)
    assert dtype in half_dtypes
    assert q.device.type == 'cuda' and q.size(-1) <= 256

    # params
    b, lq, lk, out_dtype = q.size(0), q.size(1), k.size(1), q.dtype

    def half(x):
        return x if x.dtype in half_dtypes else x.to(dtype)

    # preprocess query
    if q_lens is None:
        q = half(q.flatten(0, 1))
        q_lens = torch.tensor(
            [lq] * b, dtype=torch.int32).to(
                device=q.device, non_blocking=True)
    else:
        q = half(torch.cat([u[:v] for u, v in zip(q, q_lens)]))

    # preprocess key, value
    if k_lens is None:
        k = half(k.flatten(0, 1))
        v = half(v.flatten(0, 1))
        k_lens = torch.tensor(
            [lk] * b, dtype=torch.int32).to(
                device=k.device, non_blocking=True)
    else:
        k = half(torch.cat([u[:v] for u, v in zip(k, k_lens)]))
        v = half(torch.cat([u[:v] for u, v in zip(v, k_lens)]))

    q = q.to(v.dtype)
    k = k.to(v.dtype)

    if q_scale is not None:
        q = q * q_scale

    if version is not None and version == 3 and not FLASH_ATTN_3_AVAILABLE:
        warnings.warn(
            'Flash attention 3 is not available, use flash attention 2 instead.'
        )

    # apply attention
    if (version is None or version == 3) and FLASH_ATTN_3_AVAILABLE:
        # Note: dropout_p, window_size are not supported in FA3 now.
        x = flash_attn_interface.flash_attn_varlen_func(
            q=q,
            k=k,
            v=v,
            cu_seqlens_q=torch.cat([q_lens.new_zeros([1]), q_lens]).cumsum(
                0, dtype=torch.int32).to(q.device, non_blocking=True),
            cu_seqlens_k=torch.cat([k_lens.new_zeros([1]), k_lens]).cumsum(
                0, dtype=torch.int32).to(q.device, non_blocking=True),
            max_seqlen_q=lq,
            max_seqlen_k=lk,
            softmax_scale=softmax_scale,
            causal=causal,
            deterministic=deterministic)[0].unflatten(0, (b, lq))
    else:
        assert FLASH_ATTN_2_AVAILABLE
        x = flash_attn.flash_attn_varlen_func(
            q=q,
            k=k,
            v=v,
            cu_seqlens_q=torch.cat([q_lens.new_zeros([1]), q_lens]).cumsum(
                0, dtype=torch.int32).to(q.device, non_blocking=True),
            cu_seqlens_k=torch.cat([k_lens.new_zeros([1]), k_lens]).cumsum(
                0, dtype=torch.int32).to(q.device, non_blocking=True),
            max_seqlen_q=lq,
            max_seqlen_k=lk,
            dropout_p=dropout_p,
            softmax_scale=softmax_scale,
            causal=causal,
            window_size=window_size,
            deterministic=deterministic).unflatten(0, (b, lq))

    # output
    return x.type(out_dtype)


def attention(
    q,
    k,
    v,
    q_lens=None,
    k_lens=None,
    dropout_p=0.,
    softmax_scale=None,
    q_scale=None,
    causal=False,
    window_size=(-1, -1),
    deterministic=False,
    dtype=torch.bfloat16,
    fa_version=None,
):
    if FLASH_ATTN_2_AVAILABLE or FLASH_ATTN_3_AVAILABLE:
        return flash_attention(
            q=q,
            k=k,
            v=v,
            q_lens=q_lens,
            k_lens=k_lens,
            dropout_p=dropout_p,
            softmax_scale=softmax_scale,
            q_scale=q_scale,
            causal=causal,
            window_size=window_size,
            deterministic=deterministic,
            dtype=dtype,
            version=fa_version,
        )
    else:
        if window_size != (-1, -1):
            raise RuntimeError(
                "PyTorch SDPA fallback does not support Wan sliding windows"
            )
        out_dtype = q.dtype
        batch_size, max_q_len = q.shape[:2]
        outputs = []
        for batch_index in range(batch_size):
            query_len = (
                max_q_len
                if q_lens is None
                else int(q_lens[batch_index].item())
            )
            key_len = (
                k.shape[1]
                if k_lens is None
                else int(k_lens[batch_index].item())
            )
            query = q[batch_index : batch_index + 1, :query_len]
            key = k[batch_index : batch_index + 1, :key_len]
            value = v[batch_index : batch_index + 1, :key_len]
            query = query.transpose(1, 2).to(dtype)
            key = key.transpose(1, 2).to(dtype)
            value = value.transpose(1, 2).to(dtype)
            if q_scale is not None:
                query = query * q_scale
            if query.shape[1] != key.shape[1]:
                if query.shape[1] % key.shape[1]:
                    raise ValueError(
                        "query heads must be divisible by key/value heads"
                    )
                repeats = query.shape[1] // key.shape[1]
                key = key.repeat_interleave(repeats, dim=1)
                value = value.repeat_interleave(repeats, dim=1)
            output = torch.nn.functional.scaled_dot_product_attention(
                query,
                key,
                value,
                dropout_p=dropout_p,
                is_causal=causal,
                scale=softmax_scale,
            ).transpose(1, 2)
            if query_len < max_q_len:
                output = torch.nn.functional.pad(
                    output, (0, 0, 0, 0, 0, max_q_len - query_len)
                )
            outputs.append(output)
        return torch.cat(outputs, dim=0).contiguous().to(out_dtype)