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

# Configuration Dataclass (equivalent to GPTConfig in nanoGPT)
class MVTConfig:
    vocab_size = 5000  # V: Set by custom tokenizer
    block_size = 256   # T_ctx: Context length
    n_layer = 8        # N_layer: Number of decoder blocks
    n_head = 8         # N_head: Number of attention heads
    n_embd = 512       # D_embd: Embedding dimension
    batch_size = 16    # B: Batch size
    dropout = 0.1
    bias = False       # Optional bias for linear layers

# Initializing device setup
device = 'cuda' if torch.cuda.is_available() else 'cpu'

# --- 1. Causal Self-Attention Mechanism ---
class CausalSelfAttention(nn.Module):
    def __init__(self, config):
        super().__init__()
        assert config.n_embd % config.n_head == 0
        self.c_attn = nn.Linear(config.n_embd, 3 * config.n_embd, bias=config.bias)
        self.c_proj = nn.Linear(config.n_embd, config.n_embd, bias=config.bias)
        self.attn_dropout = nn.Dropout(config.dropout)
        self.resid_dropout = nn.Dropout(config.dropout)
        self.n_head = config.n_head
        self.n_embd = config.n_embd
        self.dropout = config.dropout
        self.block_size = config.block_size
        self.register_buffer("mask", torch.tril(torch.ones(config.block_size, config.block_size))
                             .view(1, 1, config.block_size, config.block_size))
        nn.init.normal_(self.c_proj.weight, mean=0.0, std=0.02 / math.sqrt(2 * config.n_layer))

    def forward(self, x):
        B, T, C = x.size()
        q, k, v = self.c_attn(x).split(self.n_embd, dim=2)
        k = k.view(B, T, self.n_head, C // self.n_head).transpose(1, 2)
        q = q.view(B, T, self.n_head, C // self.n_head).transpose(1, 2)
        v = v.view(B, T, self.n_head, C // self.n_head).transpose(1, 2)
        att = (q @ k.transpose(-2, -1)) * (1.0 / math.sqrt(k.size(-1)))
        att = att.masked_fill(self.mask[:, :, :T, :T] == 0, float('-inf'))
        att = F.softmax(att, dim=-1)
        att = self.attn_dropout(att)
        y = att @ v
        y = y.transpose(1, 2).contiguous().view(B, T, C)
        y = self.resid_dropout(self.c_proj(y))
        return y

# --- 2. Feed-Forward Network (MLP) ---
class MLP(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.c_fc = nn.Linear(config.n_embd, 4 * config.n_embd, bias=config.bias)
        self.gelu = nn.GELU()
        self.c_proj = nn.Linear(4 * config.n_embd, config.n_embd, bias=config.bias)
        self.dropout = nn.Dropout(config.dropout)
        nn.init.normal_(self.c_proj.weight, mean=0.0, std=0.02 / math.sqrt(2 * config.n_layer))

    def forward(self, x):
        x = self.c_fc(x)
        x = self.gelu(x)
        x = self.c_proj(x)
        x = self.dropout(x)
        return x

# --- 3. Transformer Block ---
class Block(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.ln_1 = nn.LayerNorm(config.n_embd, bias=config.bias)
        self.attn = CausalSelfAttention(config)
        self.ln_2 = nn.LayerNorm(config.n_embd, bias=config.bias)
        self.mlp = MLP(config)

    def forward(self, x):
        x = x + self.attn(self.ln_1(x))
        x = x + self.mlp(self.ln_2(x))
        return x

# --- 4. The MinimalGPT Model ---
class MinimalGPT(nn.Module):
    def __init__(self, config):
        super().__init__()
        # Store config parameters as instance attributes for TorchScript compatibility
        self.vocab_size = config.vocab_size
        self.block_size = config.block_size
        self.n_layer = config.n_layer
        self.n_head = config.n_head
        self.n_embd = config.n_embd
        self.dropout = config.dropout
        self.bias = config.bias

        self.transformer = nn.ModuleDict(dict(
            wte=nn.Embedding(self.vocab_size, self.n_embd),
            wpe=nn.Embedding(self.block_size, self.n_embd),
            drop=nn.Dropout(self.dropout),
            h=nn.ModuleList([Block(config) for _ in range(self.n_layer)]),
            ln_f=nn.LayerNorm(self.n_embd, bias=self.bias),
        ))
        self.lm_head = nn.Linear(self.n_embd, self.vocab_size, bias=False)
        self.transformer.wte.weight = self.lm_head.weight
        print(f"Minimal GPT Model initialized: {sum(p.numel() for p in self.parameters())/1e6:.2f}M parameters")

    def forward(self, idx, targets=None):
        B, T = idx.size()
        assert T <= self.block_size, f"Input sequence length {T} exceeds block size {self.block_size}"
        pos = torch.arange(0, T, dtype=torch.long, device=idx.device)
        tok_emb = self.transformer.wte(idx)
        pos_emb = self.transformer.wpe(pos)
        x = self.transformer.drop(tok_emb + pos_emb)
        for block in self.transformer.h:
            x = block(x)
        x = self.transformer.ln_f(x)
        logits = self.lm_head(x)
        loss = None
        if targets is not None:
            loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1), ignore_index=-1)
        else: # Return a dummy loss tensor if targets is None for TorchScript compatibility
            loss = torch.tensor(0.0, device=idx.device)
        return logits, loss