def testAppoint2(self): list = reg.getDisappointment("免去吕国范的河南省水利厅副厅长,河南省人民政府移民工作领导小组办公室主任职务。") print(list) self.assertIn("河南省水利厅副厅长,河南省人民政府移民工作领导小组办公室主任", list)
def testDisappointment(self): list = reg.getDisappointment(self.str_text15) print(list)
def testAppoint1(self): list = reg.getDisappointment("免去王红(女)的河南省公共资源交易中心主任职务。") print(list) self.assertIn("河南省公共资源交易中心主任", 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()