import torch
import gradio as gr
import torch.nn as nn

import pickle
import time

# Load precomputed text or read from file if it doesn't exist
try:
    # Try to load precomputed text and encoded tensor
    with open('/kaggle/input/precomputed-stories/precomputed_text.pkl', 'rb') as f:
        text = pickle.load(f)
    data = torch.load('/kaggle/input/precomputed-stories/precomputed_data.pt')
    print("Loaded precomputed data.")

except FileNotFoundError:
    # If the precomputed data doesn't exist, read the text and encode it
    print("Precomputed data not found. Reading and processing the text file...")
    
    start_time = time.time()

    # Read text file
    with open('/kaggle/input/long-discord/messages.txt', 'r', encoding='utf-8') as f:
        text = f.read()

    # Save precomputed text
    with open('precomputed_text.pkl', 'wb') as f:
        pickle.dump(text, f)

    # Encode text
    chars = sorted(set(text))
    vocab_size = len(chars)
    string_to_int = {ch: i for i, ch in enumerate(chars)}
    encode = lambda s: [string_to_int[c] for c in s]

    encoded_text = encode(text)
    
    # Convert to tensor
    data = torch.tensor(encoded_text, dtype=torch.long)

    # Save the tensor for future use
    torch.save(data, 'precomputed_data.pt')

    end_time = time.time()
    print(f"Processed and saved data in {end_time - start_time:.4f} seconds.")

# Split data for training and validation
n = int(0.8 * len(data))
train_data = data[:n]
val_data = data[n:]

print("Data is ready for model initialization.")





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

chars = sorted(set(text))
vocab_size = len(chars)
n_embd = 384
n_head = 4
n_layer = 4
block_size = 128
dropout = 0.2
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

string_to_int = { ch:i for i,ch in enumerate(chars) }
int_to_string = { i:ch for i,ch in enumerate(chars) }
encode = lambda s: [string_to_int[c] for c in s]
decode = lambda l: ''.join([int_to_string[i] for i in l])

class FeedFoward(nn.Module):
    """ a simple linear layer followed by a non-linearity """

    def __init__(self, n_embd):
        super().__init__()
        self.net = nn.Sequential(
            nn.Linear(n_embd, 4 * n_embd),
            nn.ReLU(),
            nn.Linear(4 * n_embd, n_embd),
            nn.Dropout(dropout),
        )

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


class Head(nn.Module):
    """ one head of self-attention """

    def __init__(self, head_size):
        super().__init__()
        self.key = nn.Linear(n_embd, head_size, bias=False)
        self.query = nn.Linear(n_embd, head_size, bias=False)
        self.value = nn.Linear(n_embd, head_size, bias=False)
        self.register_buffer('tril', torch.tril(torch.ones(block_size, block_size)))

        self.dropout = nn.Dropout(dropout)

    def forward(self, x):
        # input of size (batch, time-step, channels)
        # output of size (batch, time-step, head size)
        B,T,C = x.shape
        k = self.key(x)   # (B,T,hs)
        q = self.query(x) # (B,T,hs)
        # compute attention scores ("affinities")
        wei = q @ k.transpose(-2,-1) * k.shape[-1]**-0.5 # (B, T, hs) @ (B, hs, T) -> (B, T, T)
        wei = wei.masked_fill(self.tril[:T, :T] == 0, float('-inf')) # (B, T, T)
        wei = F.softmax(wei, dim=-1) # (B, T, T)
        wei = self.dropout(wei)
        # perform the weighted aggregation of the values
        v = self.value(x) # (B,T,hs)
        out = wei @ v # (B, T, T) @ (B, T, hs) -> (B, T, hs)
        return out

# [1, 0, 0]
# [1, 0.6, 0]
# [1, 0.6, 0.4]

class MultiHeadAttention(nn.Module):
    """ multiple heads of self-attention in parallel """

    def __init__(self, num_heads, head_size):
        super().__init__()
        self.heads = nn.ModuleList([Head(head_size) for _ in range(num_heads)])
        self.proj = nn.Linear(head_size * num_heads, n_embd)
        self.dropout = nn.Dropout(dropout)

    def forward(self, x):
        out = torch.cat([h(x) for h in self.heads], dim=-1) # (B, T, F) -> (B, T, [h1, h1, h1, h1, h2, h2, h2, h2, h3, h3, h3, h3])
        out = self.dropout(self.proj(out))
        return out

class Block(nn.Module):
    """ Transformer block: communication followed by computation """

    def __init__(self, n_embd, n_head):
        # n_embd: embedding dimension, n_head: the number of heads we'd like
        super().__init__()
        head_size = n_embd // n_head
        self.sa = MultiHeadAttention(n_head, head_size)
        self.ffwd = FeedFoward(n_embd)
        self.ln1 = nn.LayerNorm(n_embd)
        self.ln2 = nn.LayerNorm(n_embd)

    def forward(self, x):
        y = self.sa(x)
        x = self.ln1(x + y)
        y = self.ffwd(x)
        x = self.ln2(x + y)
        return x

    
class GPTLanguageModel(nn.Module):
    def __init__(self, vocab_size):
        super().__init__()
        self.token_embedding_table = nn.Embedding(vocab_size, n_embd)
        self.position_embedding_table = nn.Embedding(block_size, n_embd)
        self.blocks = nn.Sequential(*[Block(n_embd, n_head=n_head) for _ in range(n_layer)])
        self.ln_f = nn.LayerNorm(n_embd) # final layer norm
        self.lm_head = nn.Linear(n_embd, vocab_size)


        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, index, targets=None):
        B, T = index.shape


        # idx and targets are both (B,T) tensor of integers
        tok_emb = self.token_embedding_table(index) # (B,T,C)
        pos_emb = self.position_embedding_table(torch.arange(T, device=device)) # (T,C)
        x = tok_emb + pos_emb # (B,T,C)
        x = self.blocks(x) # (B,T,C)
        x = self.ln_f(x) # (B,T,C)
        logits = self.lm_head(x) # (B,T,vocab_size)

        if targets is None:
            loss = None
        else:
            B, T, C = logits.shape
            logits = logits.view(B*T, C)
            targets = targets.view(B*T)
            loss = F.cross_entropy(logits, targets)

        return logits, loss

    def generate(self, index, max_new_tokens):
        # index is (B, T) array of indices in the current context
        for _ in range(max_new_tokens):
            # crop idx to the last block_size tokens
            index_cond = index[:, -block_size:]
            # get the predictions
            logits, loss = self.forward(index_cond)
            # focus only on the last time step
            logits = logits[:, -1, :] # becomes (B, C)
            # apply softmax to get probabilities
            probs = F.softmax(logits, dim=-1) # (B, C)
            # sample from the distribution
            index_next = torch.multinomial(probs, num_samples=1) # (B, 1)
            # append sampled index to the running sequence
            index = torch.cat((index, index_next), dim=1) # (B, T+1)
        return index
model = GPTLanguageModel(vocab_size)
model.load_state_dict(torch.load( "/kaggle/input/longtext/transformers/default/1/longtext.pth", weights_only=True, map_location=device)["modelState"])
model.to(device)
print('loaded successfully!')


prompt = 'i look around and the world is strange'
context = torch.tensor(encode(prompt), dtype=torch.long, device=device)
generated_chars = decode(model.generate(context.unsqueeze(0), max_new_tokens=200)[0].tolist())
print(generated_chars)



if __name__ == "__main__":
    gr.Interface(fn=main, inputs=[gr.Textbox(label='Starting context'), gr.Number(label="Maximum output tokens")], outputs=[gr.Textbox(label="Response:")], title="mattGPT", article="I TELL STORIES").launch()