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 getLinkAddress(text=""):
reg = r"(通讯地址)(：|:)?( )*([\u4e00-\u9fa5]+)"
list = extrRegEntity(text, reg, index=[3], keywords=["通讯地址"])
_list = nlpUtils.get_loc_list(arrUtils.listToString(list))
if len(_list) > 0:
return list
return []
def getNativeHouseHold(text=""):
reg = r"(籍贯)(：|:)?( )*([\u4e00-\u9fa5]+)"
list = extrRegEntity(text, reg, index=[3], keywords=["籍贯"])
_list = nlpUtils.get_loc_list(arrUtils.listToString(list))
if len(_list) > 0:
return list
return []
def test_person_name(self):
res = nlp.get_per_list(
'各处室、直属单位：经研究决定：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；陶尚同志任省职业介绍中心（省就业培训中心）代理科科长。'
)
print(res)
def test_loc_name(self):
res = nlp.get_loc_list('上海市松江区九亭镇虬泾路118号')
print(res)
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()
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()