question-repeat/app.py

from flask import Flask, request, jsonify
from flask_cors import CORS
from duplicate_checker import QuestionDuplicateChecker

app = Flask(__name__)
CORS(app) # 启用跨域支持
# 初始化查重器（全局单例，避免重复加载索引）
checker = QuestionDuplicateChecker()

@app.route('/api/check_duplicate', methods=['POST'])
def check_duplicate():
    """
    题目查重 API 接口 (提交前预检模式)
    参数: stem, options, answer, solution
    """
    data = request.get_json()
    print(f"📥 收到查重请求 (内容比对): {data}")
    if not data:
        return jsonify({"code": -1, "message": "Missing content"}), 400

    # 提取内容字段
    question_data = {
        "stem": data.get('stem', ''),
        "options": data.get('options', ''),
        "answer": data.get('answer', ''),
        "solution": data.get('solution', '')
    }

    if not question_data["stem"]:
        return jsonify({"code": -1, "message": "stem is required"}), 400

    # 确保索引已加载（多进程下避免空索引）
    checker.ensure_index_loaded()
    # 执行基于内容的查重
    result = checker.check_duplicate_by_content(question_data)

    # 增加详细日志
    top_score = result["top_similar"][0]["similarity"] if result.get("top_similar") else "N/A"
    print(f"🔍 查重决策详情: status={result.get('status')}, "
          f"is_duplicate={result.get('is_duplicate')}, "
          f"max_score={top_score}, "
          f"gpt_checked={result.get('gpt_checked', False)}")

    if result.get("status") == "error":
        return jsonify({"code": -1, "message": result.get("message")}), 500

    if result.get("is_duplicate"):
        item = result["top_similar"][0]
        gpt_info = " (经 GPT-4o 深度核验)" if result.get("gpt_checked") else ""
        return jsonify({
            "code": -1,
            "result": {
                "repeatIdList": [{
                    "questionsId": item["id"],
                    "repeatMsg": f"相似度: {item['similarity']}{gpt_info}。相似点: {item['similar_point']}"
                }]
            }
        })
    else:
        return jsonify({"code": 0, "result": "ok"})

@app.route('/api/sync', methods=['POST'])
def sync_index():
    """手动触发全量同步接口"""
    print("🔄 收到同步索引请求")
    try:
        checker.ensure_index_loaded()
        started = checker.sync_all_from_db()
        if started:
            return jsonify({"code": 0, "result": "Sync completed"})
        return jsonify({"code": 0, "result": "Sync already running"})
    except Exception as e:
        return jsonify({"code": -1, "message": str(e)}), 500

@app.route('/api/confirm_repeat', methods=['POST'])
def confirm_repeat():
    """
    人工确认查重结果接口
    参数: questionId, isRepeat (0: 无相似, 1: 有重复)
    """
    data = request.get_json()
    print(f"📥 收到确认结果请求: {data}")
    if not data:
        return jsonify({"code": -1, "message": "Missing JSON body"}), 400
    
    question_id = data.get('questionId')
    is_repeat = data.get('isRepeat')

    if question_id is None or is_repeat is None:
        return jsonify({"code": -1, "message": "Missing questionId or isRepeat"}), 400

    try:
        checker.ensure_index_loaded()
        success = checker.confirm_repeat(int(question_id), int(is_repeat))
        if success:
            return jsonify({"code": 0, "result": "ok"})
        else:
            return jsonify({"code": -1, "message": "Failed to update"}), 500
    except Exception as e:
        return jsonify({"code": -1, "message": str(e)}), 500

@app.route('/api/question_info', methods=['GET'])
def get_question_info():
    """
    查询题目在向量库中的信息
    参数: questionId
    """
    question_id = request.args.get('questionId')
    if not question_id:
        return jsonify({"code": -1, "message": "Missing questionId"}), 400
    
    try:
        checker.ensure_index_loaded()
        result = checker.get_question_data(int(question_id))
        return jsonify({
            "code": 0,
            "result": result
        })
    except ValueError:
        return jsonify({"code": -1, "message": "Invalid questionId format"}), 400
    except Exception as e:
        return jsonify({"code": -1, "message": str(e)}), 500

@app.route('/api/index_info', methods=['GET'])
def get_index_info():
    """查看当前索引文件路径及条数"""
    checker.ensure_index_loaded()
    index_count = int(checker.index.ntotal) if checker.index else 0
    return jsonify({
        "code": 0,
        "result": {
            "index_path": checker.index_path,
            "metadata_path": checker.metadata_path,
            "index_count": index_count,
            "metadata_count": len(checker.metadata)
        }
    })

if __name__ == '__main__':
    # 启动服务，默认 5000 端口
    app.run(host='0.0.0.0', port=8888, debug=False)