import torch
import torch.nn as nn
import torch.nn.functional as F
import math

class RotaryPositionalEmbedding(nn.Module):
    def __init__(self, dim, max_seq_len=2048, base=10000):
        super().__init__()
        inv_freq = 1.0 / (base ** (torch.arange(0, dim, 2).float() / dim))
        self.register_buffer('inv_freq', inv_freq)
        self.max_seq_len = max_seq_len
        
    def forward(self, seq_len, device):
        t = torch.arange(seq_len, device=device).type_as(self.inv_freq)
        freqs = torch.einsum('i,j->ij', t, self.inv_freq)
        emb = torch.cat((freqs, freqs), dim=-1)
        return emb.cos(), emb.sin()

def apply_rotary_pos_emb(q, k, cos, sin):
    def rotate_half(x):
        x1, x2 = x.chunk(2, dim=-1)
        return torch.cat((-x2, x1), dim=-1)
    q_embed = (q * cos) + (rotate_half(q) * sin)
    k_embed = (k * cos) + (rotate_half(k) * sin)
    return q_embed, k_embed

class MultiHeadSelfAttention(nn.Module):
    def __init__(self, d_model, n_heads, dropout=0.1, max_seq_len=2048):
        super().__init__()
        assert d_model % n_heads == 0
        self.d_model = d_model
        self.n_heads = n_heads
        self.d_k = d_model // n_heads
        self.q_linear = nn.Linear(d_model, d_model, bias=False)
        self.k_linear = nn.Linear(d_model, d_model, bias=False)
        self.v_linear = nn.Linear(d_model, d_model, bias=False)
        self.out_linear = nn.Linear(d_model, d_model, bias=False)
        self.dropout = nn.Dropout(dropout)
        self.attn_dropout = nn.Dropout(dropout)
        self.rope = RotaryPositionalEmbedding(self.d_k, max_seq_len)
        self.flash = hasattr(torch.nn.functional, 'scaled_dot_product_attention')
        
    def forward(self, x, mask=None):
        batch_size, seq_len, d_model = x.size()
        Q = self.q_linear(x).view(batch_size, seq_len, self.n_heads, self.d_k).transpose(1, 2)
        K = self.k_linear(x).view(batch_size, seq_len, self.n_heads, self.d_k).transpose(1, 2)
        V = self.v_linear(x).view(batch_size, seq_len, self.n_heads, self.d_k).transpose(1, 2)
        
        cos, sin = self.rope(seq_len, x.device)
        cos = cos[None, None, :, :]
        sin = sin[None, None, :, :]
        Q, K = apply_rotary_pos_emb(Q, K, cos, sin)
        
        if self.flash and mask is None:
            context = F.scaled_dot_product_attention(Q, K, V, attn_mask=None, dropout_p=0.0, is_causal=True)
        else:
            scores = torch.matmul(Q, K.transpose(-2, -1)) / math.sqrt(self.d_k)
            if mask is not None:
                scores = scores.masked_fill(mask == 0, float('-inf'))
            attn_weights = F.softmax(scores, dim=-1)
            attn_weights = self.attn_dropout(attn_weights)
            context = torch.matmul(attn_weights, V)
        
        context = context.transpose(1, 2).contiguous().view(batch_size, seq_len, d_model)
        output = self.out_linear(context)
        return self.dropout(output)

class FeedForward(nn.Module):
    def __init__(self, d_model, d_ff, dropout=0.1):
        super().__init__()
        self.linear1 = nn.Linear(d_model, d_ff)
        self.linear2 = nn.Linear(d_ff, d_model)
        self.dropout = nn.Dropout(dropout)
        
    def forward(self, x):
        return self.linear2(self.dropout(F.gelu(self.linear1(x))))

class RMSNorm(nn.Module):
    def __init__(self, dim, eps=1e-6):
        super().__init__()
        self.eps = eps
        self.weight = nn.Parameter(torch.ones(dim))
        
    def forward(self, x):
        norm = torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + self.eps)
        return x * norm * self.weight

class TransformerBlock(nn.Module):
    def __init__(self, d_model, n_heads, d_ff, dropout=0.1, max_seq_len=2048, use_swiglu=False):
        super().__init__()
        self.attention = MultiHeadSelfAttention(d_model, n_heads, dropout, max_seq_len)
        self.feed_forward = FeedForward(d_model, d_ff, dropout)
        self.norm1 = RMSNorm(d_model)
        self.norm2 = RMSNorm(d_model)
        self.dropout = nn.Dropout(dropout)
        
    def forward(self, x, mask=None):
        x = x + self.attention(self.norm1(x), mask)
        x = x + self.feed_forward(self.norm2(x))
        return x

class MTPMiniModel(nn.Module):
    def __init__(self, vocab_size, d_model=512, n_layers=8, n_heads=8, 
                 d_ff=2048, max_seq_len=512, dropout=0.2, use_swiglu=False):
        super().__init__()
        self.vocab_size = vocab_size
        self.d_model = d_model
        self.max_seq_len = max_seq_len
        self.token_embedding = nn.Embedding(vocab_size, d_model)
        self.dropout = nn.Dropout(dropout)
        self.blocks = nn.ModuleList([
            TransformerBlock(d_model, n_heads, d_ff, dropout, max_seq_len, use_swiglu)
            for _ in range(n_layers)
        ])
        self.norm_f = RMSNorm(d_model)
        self.lm_head = nn.Linear(d_model, vocab_size, bias=False)
        self.lm_head.weight = self.token_embedding.weight
        self.apply(self._init_weights)
        
    def _init_weights(self, module):
        if isinstance(module, nn.Linear):
            torch.nn.init.normal_(module.weight, mean=0.0, std=0.02)
            if module.bias is not None:
                torch.nn.init.zeros_(module.bias)
        elif isinstance(module, nn.Embedding):
            torch.nn.init.normal_(module.weight, mean=0.0, std=0.02)
    
    def forward(self, input_ids, targets=None, use_eos_weight=False):
        batch_size, seq_len = input_ids.size()
        mask = torch.tril(torch.ones(seq_len, seq_len, device=input_ids.device)).view(1, 1, seq_len, seq_len)
        x = self.dropout(self.token_embedding(input_ids))
        for block in self.blocks:
            x = block(x, mask)
        x = self.norm_f(x)
        logits = self.lm_head(x)
        
        loss = None
        if targets is not None:
            if use_eos_weight:
                weights = torch.ones(self.vocab_size, device=logits.device)
                weights[3] = 2.0 
                loss = F.cross_entropy(logits.view(-1, self.vocab_size), targets.view(-1), weight=weights, label_smoothing=0.1)
            else:
                loss = F.cross_entropy(logits.view(-1, self.vocab_size), targets.view(-1), label_smoothing=0.1)
        return logits, loss

    def count_parameters(self):
        return sum(p.numel() for p in self.parameters() if p.requires_grad)