from fastapi import APIRouter
from datetime import datetime
from datasets import load_dataset
from sklearn.metrics import accuracy_score
import random

from .utils.evaluation import TextEvaluationRequest
from .utils.emissions import tracker, clean_emissions_data, get_space_info

from peft import PeftModel
from transformers import AutoTokenizer,AutoModelForSequenceClassification,Trainer, TrainingArguments,DataCollatorWithPadding, BitsAndBytesConfig
from datasets import Dataset
import torch
import numpy as np


router = APIRouter()

DESCRIPTION = "qwen_finetuned"
ROUTE = "/text"

@router.post(ROUTE, tags=["Text Task"], 
             description=DESCRIPTION)
async def evaluate_text(request: TextEvaluationRequest):
    """
    Evaluate text classification for climate disinformation detection.
    
    Current Model: Random Baseline
    - Makes random predictions from the label space (0-7)
    - Used as a baseline for comparison
    """
    # Get space info
    username, space_url = get_space_info()

    # Define the label mapping
    LABEL_MAPPING = {
        "0_not_relevant": 0,
        "1_not_happening": 1,
        "2_not_human": 2,
        "3_not_bad": 3,
        "4_solutions_harmful_unnecessary": 4,
        "5_science_unreliable": 5,
        "6_proponents_biased": 6,
        "7_fossil_fuels_needed": 7
    }

    # Load and prepare the dataset
    dataset = load_dataset(request.dataset_name)

    # Convert string labels to integers
    dataset = dataset.map(lambda x: {"label": LABEL_MAPPING[x["label"]]})

    # Split dataset
    train_test = dataset["train"]
    test_dataset = dataset["test"]
    
    # Start tracking emissions
    tracker.start()
    tracker.start_task("inference")

    #--------------------------------------------------------------------------------------------
    # YOUR MODEL INFERENCE CODE HERE
    # Update the code below to replace the random baseline by your model inference within the inference pass where the energy consumption and emissions are tracked.
    #--------------------------------------------------------------------------------------------   
    
    # Make random predictions (placeholder for actual model inference)
    true_labels = test_dataset["label"]
    predictions = [random.randint(0, 7) for _ in range(len(true_labels))]
    # path_adapter = 'Rcarvalo/Qwen_finetuned'
    path_adapter = 'MatthiasPicard/QwenTest3'
    path_model = "Qwen/Qwen2.5-3B-Instruct"

    bnb_config = BitsAndBytesConfig(
    load_in_8bit=True
    )
    
    base_model = AutoModelForSequenceClassification.from_pretrained(
    path_model,
    num_labels=len(LABEL_MAPPING),
    device_map="auto",
    torch_dtype=torch.bfloat16,
    quantization_config=bnb_config
    )    

    model = PeftModel.from_pretrained(base_model, path_adapter)
    model.eval()
    tokenizer = AutoTokenizer.from_pretrained(path_model)
    tokenizer.pad_token = tokenizer.eos_token  # Or any other token depending on your model
    tokenizer.pad_token_id = tokenizer.eos_token_id

    model.config.pad_token_id = tokenizer.pad_token_id
    model.config.use_cache = False
    model.config.pretraining_tp = 1
    def preprocess_function(df):
        return tokenizer(df["quote"], truncation=True)
    tokenized_test = test_dataset.map(preprocess_function, batched=True)
     
    # training_args = torch.load("training_args.bin")
    # training_args.eval_strategy='no'
     
    trainer = Trainer(
        model=model,
        tokenizer=tokenizer
    )
    
    preds = trainer.predict(tokenized_test)
    
    
    
    # Run inference
    # predictions = predict(tokenized_test)
    # print(predictions)
    predictions = np.array([np.argmax(x) for x in preds[0]])

    #--------------------------------------------------------------------------------------------
    # YOUR MODEL INFERENCE STOPS HERE
    #--------------------------------------------------------------------------------------------   

    
    # Stop tracking emissions
    emissions_data = tracker.stop_task()
    
    # Calculate accuracy
    accuracy = accuracy_score(true_labels, predictions)
    
    # Prepare results dictionary
    results = {
        "username": username,
        "space_url": space_url,
        "submission_timestamp": datetime.now().isoformat(),
        "model_description": DESCRIPTION,
        "accuracy": float(accuracy),
        "energy_consumed_wh": emissions_data.energy_consumed * 1000,
        "emissions_gco2eq": emissions_data.emissions * 1000,
        "emissions_data": clean_emissions_data(emissions_data),
        "api_route": ROUTE,
        "dataset_config": {
            "dataset_name": request.dataset_name,
            "test_size": request.test_size,
            "test_seed": request.test_seed
        }
    }
    
    return results