math_dup_check/math_dup_app.py

from gevent import monkey; monkey.patch_all()
from flask import Flask, request, jsonify
import requests
from gevent.pywsgi import WSGIServer

import config
from hnsw_logic import Hnsw_Logic
from school_dup_logic import School_Dup_Logic

app = Flask(__name__)

# 日志采集初始化
math_dup_logger = config.LogConfig(config.math_dup_path, "math_dup").get_log()
# hnsw_logic初始化
hnsw_logic = Hnsw_Logic(math_dup_logger)
# 校本题库查重逻辑判断初始化
school_dup_logic = School_Dup_Logic(math_dup_logger)

# 云题库查重
@app.route('/repeat', methods=['GET', 'POST'])
def repeat_retrieve():
    if request.method == 'POST':
        # 获取post数据
        topics_dict = request.get_json()
        # 接收日志采集
        math_dup_logger.info(config.log_msg.format(id="云题库查重",
                                                   type="repeat接收",
                                                   message=topics_dict))
        if isinstance(topics_dict, dict):
            retrieve_list = topics_dict["topics"]
            callback_url = topics_dict["callback_url"]
        elif isinstance(topics_dict, list):
            retrieve_list = topics_dict
            callback_url = None
        else:
            return jsonify("请输入正确格式数据")
        if not retrieve_list:
            return jsonify("请输入查重数据")
        # 调用清洗分词函数和句向量计算函数
        res_list = hnsw_logic.logic_process(retrieve_list, hnsw_index=0)
        res_dict = {retrieve_list[i]["topic_id"]:ele for i,ele in enumerate(res_list) if len(ele)>0}
        if callback_url is not None:
            requests.post(callback_url, json=res_dict, timeout=10)
        # 返回日志采集
        math_dup_logger.info(config.log_msg.format(id="云题库查重",
                                                   type="repeat返回",
                                                   message=res_dict))
        return jsonify(res_dict)

# 校本题库查重
@app.route('/chc', methods=['GET', 'POST'])
def chc_retrieve():
    if request.method == 'POST':
        # 获取post数据
        topics_dict = request.get_json()
        if not topics_dict:
            return jsonify("请输入查重数据")
        # 接收日志采集
        math_dup_logger.info(config.log_msg.format(id="校本题库查重",
                                                   type="chc接收",
                                                   message=topics_dict))
        # chc转移异步查重
        requests.post(r"http://localhost:8858/chc/transfer", json=topics_dict, timeout=10)
        return jsonify({"errcode": 0, "errmsg": "OK"})

# 校本题库查重
@app.route('/chc/process', methods=['GET', 'POST'])
def chc_process():
    if request.method == 'POST':
        # 获取post数据
        topics_dict = request.get_json()
        res_dict = school_dup_logic(hnsw_logic.logic_process, topics_dict["topics"])
        # 将结果post给callback_url
        requests.post(topics_dict["callback_url"], json=res_dict, timeout=10)
        # 获取chc查重试卷id
        post_doc_id = topics_dict["callback_url"].split('/')[-1].split('\\')[-1]
        # 返回日志采集
        math_dup_logger.info(config.log_msg.format(
                                        id="校本题库查重",
                                        type="chc返回-{}查重完毕".format(post_doc_id),
                                        message=res_dict))
        return jsonify("")

# 云题库数据更新
@app.route('/cloud/update', methods=['GET', 'POST'])
def cloud_update():
    if request.method == 'POST':
        # 获取post数据
        update_dict = request.get_json()
        # 接收日志采集
        math_dup_logger.info(config.log_msg.format(id="云题库数据更新",
                                                   type="cloud/update接收",
                                                   message=update_dict))
        if not update_dict:
            return jsonify("请输入更新数据")
        # 更新HNSW模型
        hnsw_logic.update(update_dict["id"], hnsw_index=0)

        return jsonify("数据更新完毕")


if __name__ == '__main__':
    # app.run(host='0.0.0.0',port='8855')
    server = WSGIServer(('0.0.0.0', 8855), app)
    server.serve_forever()