poc_qlm/采集任务清单.py

# -*- coding: utf-8 -*-
"""
Created on 2024-10-18 
---------
@summary:  
---------
@author: Dzr
"""
from pymongo import MongoClient
import pandas as pd


to_db = MongoClient('192.168.3.182', 27017)
coll = to_db['31zg_poc']['keyword_company']

suffix_lst = '重机,挖掘机,装载机,泵送,桥泵车,搅拌车,拖泵,搅拌站,车载泵,搅拌车,重起,汽车起重机,履带起重机,桩机,旋挖钻,大旋挖,中旋挖,小旋挖,路机,铣刨机,摊铺机,平地机,压路机,沥青站'.split(',')
suffix_set = set(suffix_lst)

f = '/Users/dongzhaorui/Desktop/qlm数据采集.xlsx'
df = pd.read_excel(f)
df.fillna('',  inplace=True)

company_lst = []
for _, i in df.iterrows():
    items = i.to_dict()
    s_company = str(items['集团名称']).strip()
    s_sub_company = str(items['二级局名称']).strip()

    if s_company and s_company not in company_lst:
        company_lst.append(s_company)

    if s_sub_company and s_sub_company not in company_lst:
        company_lst.append(s_sub_company)

data = []
for suffix in suffix_set:
    print(suffix)

    for company in company_lst:
        data.append({'s_suffix': suffix, 's_company': company, 's_keyword': f'{company}+{suffix}'})

    if len(data) == 100:
        coll.insert_many(data, ordered=False)
        data = []

if len(data) > 0:
    coll.insert_many(data, ordered=False)

print('1234')