73 lines
2.3 KiB
Python
73 lines
2.3 KiB
Python
|
import json
|
|||
|
|
import os
|
|||
|
|
|
|||
|
|
import pymysql
|
|||
|
|
|
|||
|
|
base_props = {
|
|||
|
|
'name': "姓名",
|
|||
|
|
'gender': "性别",
|
|||
|
|
'birthday': "出生日期",
|
|||
|
|
'region': "国家/地区",
|
|||
|
|
'type_or_dept': "军种/部门",
|
|||
|
|
'category': "人员类别",
|
|||
|
|
'military_rank': "军衔",
|
|||
|
|
'occupation': "单位职务",
|
|||
|
|
'specialty': "专长领域",
|
|||
|
|
'contact': "联系方式",
|
|||
|
|
'social_media': "社交媒体账号",
|
|||
|
|
'update_time': "数据日期",
|
|||
|
|
'introduction': "个人简介",
|
|||
|
|
'honors': "所获荣誉",
|
|||
|
|
'education': "教育背景",
|
|||
|
|
'source': "信息来源",
|
|||
|
|
'career': "从业背景",
|
|||
|
|
'keywords': "关键词",
|
|||
|
|
'website': "个人主页",
|
|||
|
|
'achievements': "主要成就",
|
|||
|
|
'rank_up': "军衔晋升",
|
|||
|
|
'party': "政党",
|
|||
|
|
'remarks': "备注",
|
|||
|
|
'foreigners_info': "外裔情况",
|
|||
|
|
'religion': "宗教信仰",
|
|||
|
|
'other_contacts': "其他联系方式",
|
|||
|
|
'address': "个人住址"
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
result_file_path = r"E:\yuxin\nuofang-data\person\result0515"
|
|||
|
|
|
|||
|
|
if __name__ == '__main__':
|
|||
|
|
db = pymysql.connect(host='39.101.194.63', port=23306,
|
|||
|
|
user='root', passwd='passok123A', db='nfm', charset='utf8mb4')
|
|||
|
|
cursor = db.cursor()
|
|||
|
|
l_item = {}
|
|||
|
|
count = 0
|
|||
|
|
for single_json_file in os.listdir(result_file_path):
|
|||
|
|
count += 1
|
|||
|
|
result_file = open(result_file_path + "\\" + single_json_file, "r", encoding='utf-8')
|
|||
|
|
result_json = json.loads(result_file.read())
|
|||
|
|
result_file.close()
|
|||
|
|
|
|||
|
|
t_item_content = {}
|
|||
|
|
url_info = result_json['url_info']
|
|||
|
|
person_id = url_info[37:45]
|
|||
|
|
t_item_content['id'] = person_id
|
|||
|
|
base_info = result_json['base_info']
|
|||
|
|
website_info = result_json['website_info']
|
|||
|
|
|
|||
|
|
for _ in base_props:
|
|||
|
|
val = base_props[_]
|
|||
|
|
if val in base_info:
|
|||
|
|
t_item_content[_] = base_info[val]
|
|||
|
|
elif val in website_info:
|
|||
|
|
t_item_content[_] = website_info[val]
|
|||
|
|
|
|||
|
|
col_list = [_ for _ in t_item_content]
|
|||
|
|
val_list = ["'" + t_item_content[_].replace("'", "’") + "'" for _ in t_item_content]
|
|||
|
|
cols = ", ".join(col_list)
|
|||
|
|
vals = ", ".join(val_list)
|
|||
|
|
sql_insert = 'INSERT INTO m_person_page (%s) VALUES (%s)' % (cols, vals)
|
|||
|
|
print("[No. {}] {}".format(count, sql_insert))
|
|||
|
|
cursor.execute(sql_insert)
|
|||
|
|
db.commit()
|
|||
|
|
db.close()
|