def getPersonName(text=""):
list = RegUtils.getPersonName(text)
if len(list) > 0:
return list[0]
NlpUtils.get_per_list(text)
if len(list) > 0:
return list[0]
return ""
def testGetJobs3(self):
list = reg.getJobsCurr(self.str_text12)
print(list)
self.assertIn("中共十九届中央委员", list)
self.assertIn("国务委员", list)
self.assertIn("公安部部长", list)
self.assertIn("党委书记", list)
self.assertIn("总警监", list)
self.assertIn("国务院党组成员", list)
def testResume(self):
dict = {}
dict['volk'] = reg.getVolk(self.str_text)
dict['edu'] = reg.getEducations(self.str_text)
dict['alias'] = reg.getAliasName(self.str_text)
dict['job'] = reg.getJobsCurr(self.str_text)
dict['majors'] = reg.getMajors(self.str_text)
dict['schools'] = reg.getSchools(self.str_text)
dict['location'] = reg.getLocation(self.str_text)
print(dict)
def testGetLocation(self):
list = reg.getLocation(self.str_text7)
print(list)
self.assertIn('浙江龙游', list)
def testBirthday(self):
list = reg.getBirthday(
"犯罪嫌疑人王海涛，男，汉族，1978年2月10日出生，身份证号码120109197802105034，户籍地：天津市滨海新区大港太平镇六间房村。"
)
print(list)
self.assertIn("1978年2月10日", list)
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 testPolitics2(self):
list = reg.getPolitics("政治面貌 中共党员,中共预备党员 入党时间 1996-11-22 民族 汉族 ")
print(list)
def testMobilePhone(self):
list = reg.getMobilePhone(self.str_text16)
print(list)
self.assertIn("18515061589", list)
def testFirstDateInJob(self):
list = reg.getFirstDateInJob(self.str_text16)
print(list)
self.assertIn("2011年08月", 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 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 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 testDisappointment(self):
list = reg.getDisappointment(self.str_text15)
print(list)
def testGetLocation3(self):
list = reg.getLocation(
'张任德（在逃人员编号：T8432009999992015090170），男，汉族，1963年12月15日生，户籍地：广西自治区防城港市港口区企沙镇赤沙村赤东组28号'
)
print(list)
self.assertIn('广西自治区防城港市港口区企沙镇赤沙村赤东组28号', list)
def testGetLocation2(self):
list = reg.getLocation('户籍地：天津市滨海新区大港太平镇六间房村。')
print(list)
self.assertIn('天津市滨海新区大港太平镇六间房村', list)
def testHouseHold(self):
list = reg.getRegisterHouseHold(self.str_text16)
print(list)
self.assertIn("河南省濮阳县", list)
def testNativeHouseHold(self):
list = reg.getNativeHouseHold(self.str_text16)
print(list)
self.assertIn("河南省濮阳县", list)
def testPersonName2(self):
list = reg.getPersonName("姓名：逯长松 性别： 男 照片 ")
print(list)
self.assertIn("逯长松", list)
def testLinkAddress(self):
list = reg.getLinkAddress(self.str_text16)
print(list)
self.assertIn("朝阳区康营小区", list)
def testPersonName3(self):
list = reg.getPersonName("犯罪嫌疑人王海涛")
print(list)
self.assertIn("王海涛", list)
def testGetDateStage(self):
list = reg.getDateStage("2004/9 2008/6 江汉大学 计算机科学与技术 本科 学士学位 是")
print(list)
self.assertIn("2004/9 2008/6", list)
def testPersonName4(self):
list = reg.getPersonName('陈春友，男，汉族，1978年3月7日出生')
print(list)
self.assertIn('陈春友', 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)
def testPersonName5(self):
list = reg.getPersonName(
'金彩霞，绰号“金婉婉”，女，汉族，1998年5月14日出生，广东口音，户籍地：广东省徐闻县迈陈镇金宅村65号，身份证号码：440825199805143468。'
)
print(list)
self.assertIn('金彩霞', list)
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 testPersonName6(self):
list = reg.getPersonName(
'张任德（在逃人员编号：T8432009999992015090170），男，汉族，1963年12月15日生，户籍地：广西自治区防城港市港口区企沙镇赤沙村赤东组28号'
)
print(list)
self.assertIn('张任德', list)
def firstInStrList(_list=[]):
if len(_list) > 0:
return _list[0]
return ""
if __name__ == "__main__":
resume_dict = {}
dict_result = npt.analy_html(html_text, UrlUtils.parseUrl(html_url))
arr_resume = dict_result['resume']
arr_resume_dict = []
for resume in arr_resume:
resume['name'] = getPersonName(resume['text'])
resume['alias'] = firstInStrList(RegUtils.getAliasName(resume['text']))
resume['location'] = firstInStrList(
RegUtils.getLocation(resume['text']))
resume['nid'] = firstInStrList(RegUtils.getIdNbr(resume['text']))
resume['event_descr'] = firstInStrList(
RegUtils.getEventDescrs(resume['text']))
resume['birthday'] = firstInStrList(
RegUtils.getBirthday(resume['text']))
resume['volk'] = firstInStrList(RegUtils.getVolk(resume['text']))
resume['gender'] = firstInStrList(RegUtils.getGender(resume['text']))
if resume['name'] != '' or resume['nid'] != '':
arr_resume_dict.append(resume)
arr_resume_dict = npt.clean_image_url(arr_resume_dict)
print(arr_resume_dict)
def testPolitics(self):
list = reg.getPolitics(self.str_text16)
print(list)
self.assertIn("党员", list)
import myconfig.conf as config
import nbtest.utils.ArrayUtils as arrUtils
import nbtest.utils.RegUtils as regUtils
import nbtest.utils.NlpUtils as nlpUtils
if __name__ == "__main__":
db = pymysql.connect(host=config.databaseip,
user=config.databaseuser,
password=config.databasepasswd,
database=config.databasename)
cursor = db.cursor()
dict_result = npt.analy_html(html_text)
resume_text = arrUtils.listToString(dict_result['resume'])
dict_result['reason'] = arrUtils.listToString(dict_result['reason'])
dict_result['result'] = arrUtils.listToString(dict_result['result'])
dict_resume = regUtils.analyToResumeDict(resume_text)
dict_result['name'] = arrUtils.listToString(
nlpUtils.get_per_list(resume_text))
dict_result.update(dict_resume)
cursor.execute(
"insert into aml_cnnnews_crawl (name, volk, education, job, major, school, location, punish_reason , punish_result) VALUES ('%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s') "
% (dict_result['name'], dict_result['volk'], dict_result['edu'],
dict_result['job'], dict_result['majors'], dict_result['schools'],
dict_result['location'], dict_result['reason'],
dict_result['result']))
db.commit()
db.close()
def testMarriage(self):
list = reg.getMarriage(self.str_text16)
print(list)
self.assertIn("未婚", list)