text_to_vector/utils/title_ner.py

# coding:utf-8
import re

from a2s.a2s_client import a2s_execute
from a2s.tools import json_serialize, json_deserialize
from loguru import logger
from config import daili
from utils.request_fun import top_t


def start(data: dict):
    # 本次不使用SSL，所以channel是不安全的
    result = {}
    try:
        retry = 5
        for r in range(retry):
            bytes_data = json_serialize(data)
            result = a2s_execute(daili, 'title_ner', 60, bytes_data)
            if result is None:
                continue
            result = json_deserialize(result)
            return result
    except Exception as e:
        logger.info(str(e))
    return result


def title_topic_merge(text):
    """
    标题信息抽取,合并多标的物查询
    """
    tet = re.sub(r'[^\w\s]', '', text)
    print(tet)
    input_text = {"text": tet}
    res = start(input_text)
    topic_res = ''
    flag = ''
    if res:
        res_list = res.get('result',[])
        for i in res_list:
            target = i.get('TARGET', [])
            topic_res = ''.join([topic[0] for topic in target])
    if topic_res in ['建设']:
        topic_res = ''
    return topic_res,flag


def title_topic_process(text,):
    """
    标题信息抽取
    """
    input_text = {"text": re.sub(r'[^\w\s]', '', text).replace('定点','')}
    pattern = r'项目'
    count_re = len(re.findall(pattern, text))
    res = start(input_text)
    topic_res = ''
    flag = ''
    if res:
        res_list = res.get('result',[])
        for i in res_list:
            target = i.get('TARGET', [])
            if count_re >=2:
                topic_res = ''.join([topic[0] for topic in target])
            else:
                for j in target:
                    topic_res = j[0]
                    flag = 'ner'
    if topic_res in ['建设']:
        topic_res = ''
    if not topic_res:
        topic_res = text
        flag = 'title'
    return topic_res,flag


def topic_trace(title,projectname):
    """
    主干词抽取
    """
    if '采购意向' in projectname:
        return title
    if ('采购意向' in title or '...' in title) and projectname:
        title_topic, flag = title_topic_process(projectname)
    else:
        title_topic, flag = title_topic_process(title)
        if title_topic == title and projectname:
            title_topic, flag = title_topic_process(projectname)
    if not title_topic:
        title_topic = top_t(title)
    if not title_topic:
        title_topic = top_t(projectname)
    if not title_topic:
        title_topic = title
    title_topic = re.sub(r'[^\w\s]', '', title_topic)
    return title_topic

if __name__ == '__main__':
    data = " 广州公司-(珠海)智慧能源-显示屏-2312(急)变更公告"
    r = start({"text": data})
    print(topic_trace(data, data))
    print(r)