def get_student_info(id,name,idcard):
id = str(id)
if int(id[2])>3: base_url = 'http://xxjw.hnust.cn/xxjw'
else : base_url = 'http://kdjw.hnust.cn/kdjw'
# if int(id[2])>3: base_url = 'http://localhost:2167/xxjw'
# else : base_url = 'http://localhost:2169/kdjw'
opener = requests.Session()
url_img = base_url+'/verifycode.servlet'
try: opener.get(base_url,timeout=1)
except: return {'error':True,'msg':"7.服务器网络故障，可能查询的人太多了，亲再等等"}
for i in range(5):
try: img = opener.get(url_img,timeout=1).content
except : continue
code = ocr.recognize(img)
url_submit = base_url+'/xscjcx_check.jsp'
payload = {
'xsxm':name,
'xssfzh':idcard,
'yzm':code,
}
try: res = opener.post(url_submit,payload,timeout=1).text
except : continue
if 'yzm_guoq' in res : return {'error':True,'msg':"1.验证码过期"}
if 'yzm_cuowu' in res : continue
if 'notQueryXs' in res : return {'error':True,'msg':"3.未找到您输入信息的学生"}
if 'systemError' in res : return {'error':True,'msg':"4.您访问的功能出现错误"}
url_scope = base_url+'/xscjcx.jsp?yzbh=' + res[-36:-4]
try: html = opener.get(url_scope,timeout=1).text
except : continue
if not re.search('\d{10}',html) : return {'error':True,'msg':"5.登陆成功了，但是似乎没有你的信息哦"}
return htmlToJson.htmlToJson(html)
return {'error':True,'msg':"8.服务器网络故障，可能查询的人太多了，亲再等等"}
def parse(self,response):
url = response.url
html = response.body_as_unicode()
try:
js = htmlToJson.htmlToJson(html)
# import ipdb;ipdb.set_trace()
MYSCOPE_DB.save(js)
except :
pass
def parse(self, response):
url = response.url
html = response.body_as_unicode()
try:
js = htmlToJson.htmlToJson(html)
# import ipdb;ipdb.set_trace()
MYSCOPE_DB.save(js)
except:
pass
def get_student_info(id, name, idcard):
id = str(id)
if int(id[2]) > 3: base_url = 'http://xxjw.hnust.cn/xxjw'
else: base_url = 'http://kdjw.hnust.cn/kdjw'
# if int(id[2])>3: base_url = 'http://localhost:2167/xxjw'
# else : base_url = 'http://localhost:2169/kdjw'
opener = requests.Session()
url_img = base_url + '/verifycode.servlet'
try:
opener.get(base_url, timeout=1)
except:
return {'error': True, 'msg': "7.服务器网络故障，可能查询的人太多了，亲再等等"}
for i in range(5):
try:
img = opener.get(url_img, timeout=1).content
except:
continue
code = ocr.recognize(img)
url_submit = base_url + '/xscjcx_check.jsp'
payload = {
'xsxm': name,
'xssfzh': idcard,
'yzm': code,
}
try:
res = opener.post(url_submit, payload, timeout=1).text
except:
continue
if 'yzm_guoq' in res: return {'error': True, 'msg': "1.验证码过期"}
if 'yzm_cuowu' in res: continue
if 'notQueryXs' in res: return {'error': True, 'msg': "3.未找到您输入信息的学生"}
if 'systemError' in res: return {'error': True, 'msg': "4.您访问的功能出现错误"}
url_scope = base_url + '/xscjcx.jsp?yzbh=' + res[-36:-4]
try:
html = opener.get(url_scope, timeout=1).text
except:
continue
if not re.search('\d{10}', html):
return {'error': True, 'msg': "5.登陆成功了，但是似乎没有你的信息哦"}
return htmlToJson.htmlToJson(html)
return {'error': True, 'msg': "8.服务器网络故障，可能查询的人太多了，亲再等等"}
if not dic : return "{'status':false}"
# if (datetime.datetime.now()-dic['datetime']).total_seconds() > 24*3600 : return getScoreByWeb(id)
dic['from'] = 'mongodb'
dic['datetime'] = dic['datetime'].isoformat()
return json.dumps(dic,ensure_ascii=False,indent=None,encoding='UTF8')
def getScoreByWeb(id):
id=str(id)
# if id[2]=='5': url = 'http://211.67.208.67/xxjw/xscjcx.jsp?yzbh='
# else: url = 'http://211.67.208.69/kdjw/xscjcx.jsp?yzbh='
if id[2]=='5' or id[2]=='6': url = 'http://127.0.0.1:2222/xxjw/xscjcx.jsp?yzbh='
else: url = 'http://127.0.0.1:2222/kdjw/xscjcx.jsp?yzbh='
try:
html=urllib2.urlopen(url+str(id),timeout=5).read()
if len(html)<100 : return u"{'error':true,'msg':'内网服务器脱机'}"
except Exception,e:
return u"{'error':true,'msg':'服务器加载数据失败'}"
dic = htmlToJson.htmlToJson(html)
MYSCOPE_DB.save(dic)
del dic['_id']
dic['from'] = 'web'
dic['datetime'] = dic['datetime'].isoformat()
return json.dumps(dic,ensure_ascii=False,indent=None,encoding='UTF8')
if __name__ == 'main':
print s.encode('gbk','ignore')
