app.py

import sqlite3
import joblib
import jieba
import os
import numpy as np
from flask import Flask, jsonify, request, send_from_directory
from flask_cors import CORS
from collections import Counter
from datetime import datetime, timedelta

# === 配置 ===
app = Flask(__name__, static_folder='static')
CORS(app)

DB_NAME = 'database.db'
MODEL_PATH = 'models/sentiment_nb.pkl'

# === 1. 加载模型 ===
model = None
if os.path.exists(MODEL_PATH):
    try:
        model = joblib.load(MODEL_PATH)
        print("✅ 后端: 情感模型加载成功")
    except Exception as e:
        print(f"❌ 模型加载失败: {e}")
else:
    print("⚠️ 警告: 未找到模型文件，预测功能将不可用")


def get_db_connection():
    conn = sqlite3.connect(DB_NAME)
    conn.row_factory = sqlite3.Row
    return conn


@app.route('/')
def index():
    return send_from_directory('.', 'index.html')


# === 4. API: 仪表盘数据 (已修改为真实数据统计) ===
@app.route('/api/dashboard')
def get_dashboard_data():
    try:
        conn = get_db_connection()
        cursor = conn.cursor()

        # A. 总数
        cursor.execute("SELECT COUNT(*) FROM posts")
        total_posts = cursor.fetchone()[0]

        # B. 来源
        cursor.execute("SELECT source, COUNT(*) FROM posts GROUP BY source")
        sources = dict(cursor.fetchall())

        # C. 情感分布
        cursor.execute("SELECT sentiment_label, COUNT(*) FROM posts GROUP BY sentiment_label")
        sentiment_counts = {1: 0, 0: 0, -1: 0}
        for row in cursor.fetchall():
            sentiment_counts[row[0]] = row[1]

        distribution = [sentiment_counts.get(1, 0), sentiment_counts.get(0, 0), sentiment_counts.get(-1, 0)]

        # D. 24小时趋势 (核心修改：统计真实数据库中的时间分布)
        # 获取所有帖子的时间和情感
        cursor.execute("SELECT sentiment_label, publish_time FROM posts")
        rows = cursor.fetchall()

        # 初始化 0点到23点的桶
        hours_data = {i: {'pos': 0, 'neg': 0} for i in range(24)}

        for row in rows:
            try:
                # 数据库存的时间格式通常是 "2023-xx-xx 14:30:00"
                # 我们截取字符串解析出小时 (Hour)
                time_str = str(row[1])
                # 解析时间对象
                dt = datetime.strptime(time_str.split('.')[0], "%Y-%m-%d %H:%M:%S")
                h = dt.hour

                # 归类
                if row[0] == 1:  # 积极
                    hours_data[h]['pos'] += 1
                elif row[0] == -1:  # 消极
                    hours_data[h]['neg'] += 1
            except Exception as e:
                continue  # 忽略解析错误的时间

        # 整理成前端图表需要的数组格式 (按0点-23点排序)
        trend_data = {
            'labels': [f"{i}:00" for i in range(24)],
            'pos': [hours_data[i]['pos'] for i in range(24)],
            'neg': [hours_data[i]['neg'] for i in range(24)]
        }

        conn.close()

        return jsonify({
            'total': total_posts,
            'sources': sources,
            'distribution': distribution,
            'trend': trend_data
        })

    except Exception as e:
        print(f"Dashboard Error: {e}")
        return jsonify({'error': str(e)}), 500


@app.route('/api/analyze', methods=['POST'])
def analyze_text():
    data = request.json
    text = data.get('text', '')

    if not text or not model:
        return jsonify({'error': '无输入或模型未加载'}), 400

    words = " ".join(jieba.cut(text))
    label_idx = model.predict([words])[0]
    proba = model.predict_proba([words])[0]

    label_map = {1: '积极 (Positive)', 0: '中立 (Neutral)', -1: '消极 (Negative)'}
    keywords = [w for w in jieba.cut(text) if len(w) > 1][:5]

    return jsonify({
        'label': label_map.get(label_idx, '未知'),
        'probs': {'neg': float(proba[0]), 'neu': float(proba[1]), 'pos': float(proba[2])},
        'keywords': keywords
    })


if __name__ == '__main__':
    print("🚀 舆情系统后端已启动...")
    app.run(debug=True, port=5000)