# -*- coding: utf-8 -*-
from transformers import BertTokenizer, BertForSequenceClassification, Trainer, TrainingArguments, EarlyStoppingCallback
import pandas as pd
from sklearn.model_selection import train_test_split
import torch
import re
from sklearn.metrics import precision_score, recall_score, f1_score, accuracy_score, confusion_matrix
import matplotlib.pyplot as plt

import seaborn as sns
import warnings
warnings.filterwarnings('ignore')
# 检查设备
print("Available devices:", "GPU" if torch.cuda.is_available() else "Only CPU")

# 加载数据
df = pd.read_csv("data/sms_classify.csv")
texts = df["text"].tolist()
labels = df["label"].tolist()

# 在数据加载后添加文本清洗函数 --0320
def clean_text(text):
    # 移除多余的空格
    text = ' '.join(text.split())
    # 移除特殊字符
    text = re.sub(r'[【】]', '', text)
    return text

texts = [clean_text(text) for text in texts]


# 标签映射
label2id = {label: idx for idx, label in enumerate(set(labels))}
id2label = {idx: label for label, idx in label2id.items()}
labels = [label2id[label] for label in labels]

# 划分数据集
train_texts, val_texts, train_labels, val_labels = train_test_split(texts, labels, test_size=0.2)

# 加载分词器和模型（使用BERT中文模型）
tokenizer = BertTokenizer.from_pretrained("bert-base-chinese")
model = BertForSequenceClassification.from_pretrained(
    "bert-base-chinese",
    num_labels=len(label2id),
    id2label=id2label,
    label2id=label2id
)


# 数据编码
train_encodings = tokenizer(train_texts, truncation=True, padding=True, max_length=64)
val_encodings = tokenizer(val_texts, truncation=True, padding=True, max_length=64)

# 转换为PyTorch Dataset
class SMSDataset(torch.utils.data.Dataset):
    def __init__(self, encodings, labels):
        self.encodings = encodings
        self.labels = labels

    def __getitem__(self, idx):
        item = {key: torch.tensor(val[idx]) for key, val in self.encodings.items()}
        item["labels"] = torch.tensor(self.labels[idx])
        return item

    def __len__(self):
        return len(self.labels)

train_dataset = SMSDataset(train_encodings, train_labels)
val_dataset = SMSDataset(val_encodings, val_labels)

# 训练参数（CPU优化）定义训练参数（批次大小、学习率、保存策略等）
training_args = TrainingArguments(
    output_dir="./models/classifier",
    num_train_epochs=3,
    per_device_train_batch_size=8,
    per_device_eval_batch_size=8,
    logging_dir="./logs",
    evaluation_strategy="epoch",
    save_strategy="epoch",
    load_best_model_at_end=True,
    no_cuda=True,  # 强制使用CPU
    # 新增参数
    metric_for_best_model="accuracy",
    greater_is_better=True,
    learning_rate=2e-5,
    warmup_ratio=0.1,  # 使用warmup_ratio替代warmup_steps
    weight_decay=0.01,
)


# 添加评估指标计算  --0320
def compute_metrics(pred):
    labels = pred.label_ids
    preds = pred.predictions.argmax(-1)
    precision = precision_score(labels, preds, average='weighted')
    recall = recall_score(labels, preds, average='weighted')
    f1 = f1_score(labels, preds, average='weighted')
    acc = accuracy_score(labels, preds)
    return {
        'accuracy': acc,
        'f1': f1,
        'precision': precision,
        'recall': recall
    }

# 训练器
# 更新Trainer
trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=train_dataset,
    eval_dataset=val_dataset,
    compute_metrics=compute_metrics,
    callbacks=[EarlyStoppingCallback(early_stopping_patience=2)]  # 添加早停回调
)


# 开始训练
trainer.train()

# 在训练后添加模型评估
eval_results = trainer.evaluate()
print("评估结果:", eval_results)

# 保存混淆矩阵
predictions = trainer.predict(val_dataset)
preds = predictions.predictions.argmax(-1)
cm = confusion_matrix(val_labels, preds)
plt.figure(figsize=(10,8))
sns.heatmap(cm, annot=True, fmt='d', xticklabels=list(id2label.values()), yticklabels=list(id2label.values()))
plt.title('混淆矩阵')
plt.savefig('./models/classifier/confusion_matrix.png')

# 保存模型和分词器
model_path = "./models/classifier"
model.save_pretrained(model_path)
tokenizer.save_pretrained(model_path)