def testAppoint3(self): list = reg.getAppointment("任命党培红(女)为河南省民政厅副厅长(试用期一年);") print(list) self.assertIn("河南省民政厅副厅长", list)
def testAppointmentJob2(self): list = reg.getAppointment('提名梁中基为中国林业集团有限公司副总经理') print(list) self.assertIn('中国林业集团有限公司副总经理', list)
def testAppointmentJob3(self): list = reg.getAppointment('卢镱逢同志任省专家和留学人员服务中心(省留学人员创业服务中心)七级职员') print(list) self.assertIn('省专家和留学人员服务中心(省留学人员创业服务中心)七级职员', list)
def testAppointmentJob1(self): list = reg.getAppointment('杨思雷同志任省职业介绍中心(省就业培训中心)职介科科长') print(list) self.assertIn('省职业介绍中心(省就业培训中心)职介科科长', list)
def testGetAppointment3(self): list = reg.getAppointment(self.str_text7) print(list) self.assertIn("国家民族事务委员会主任", list)
def testGetAppointment2(self): list = reg.getAppointment(self.str_text12) print(list) self.assertEqual(0, len(list))
def testGetAppointment(self): list = reg.getAppointment(self.str_text14) print(list) self.assertIn("教育部部长", list)
import myconfig.conf as config import re import pymysql if __name__ == "__main__": # db = pymysql.connect(host=config.databaseip, user=config.databaseuser, password=config.databasepasswd, database=config.databasename) # cursor = db.cursor() arr_person_resume = [] dict_result = npt.analy_html(html_text) for appoint_texts in dict_result['appoint']: for appoint_text in re.split(r";|;|。", appoint_texts): dict_resume = {} dict_resume['appoint'] = arrUtils.listToString(regUtils.getAppointment(appoint_text)) dict_resume['disappoint'] = arrUtils.listToString(regUtils.getDisappointment(appoint_text)) dict_resume['name'] = arrUtils.listToString(arrUtils.flatListToStr(nlpUtils.get_per_list(appoint_text)), ',') if dict_resume['appoint'] == '' and dict_resume['disappoint'] == '': continue arr_person_resume.append(dict_resume) print(dict_resume) # cursor.execute("insert into aml_cnnnews_crawl (name, appointment , disappointment) VALUES ('%s', '%s', '%s') " % # (dict_resume['name'], dict_resume['appoint'], dict_resume['disappoint'])) print(dict_result) # db.commit() # db.close()