def test_fail_bost(self, testdata):
login = Login(self.driver)
login.login(testdata["username"],testdata["password"])
time.sleep(3)
real = login.get_title()
hope = testdata["hope"]
self.assertEqual(real, hope)
def DoTasker():
lstUserData = Data.laodUserData()
emails = []
for d in lstUserData:
login = Login(d[0], d[1])
skey, reason = login.autologin()
if skey == False:
# 签到失败,删除用户信息,发送邮件提醒
email = Data.getUserEmailData(d[0])
if email != None and len(email) > 0:
emails.append(email)
Data.removeUserData(d[0])
Data.removeUserEmailData(d[0])
sign = Sign(d[0], skey)
if sign.autosign() == False:
pass
# time.sleep(1)
# print emails
if len(emails) > 0:
email = libs.mail.Message(Config.smtp_user, emails, Config.mail_notify)
try:
conn = libs.mail.Connection(Config.smtp_server, 25, Config.smtp_user, Config.smtp_pass)
conn.send_message(email)
except:
pass
def main ():
#make sure to shown tacskIcon right when use pyinstaller to make exe file you can commont it
ctypes.windll.shell32.SetCurrentProcessExplicitAppUserModelID("BatterayDetection")
me = singleton.SingleInstance()
app = QtGui.QApplication(sys.argv)
splash = QSplashScreen(QPixmap(":/Start/Resource/start.png"), Qt.WindowStaysOnTopHint)
splash.setDisabled(True) # Disables widget input events, or when mouse click splash will be hide
splash.show()
splash.showMessage("Start Application ......", Qt.AlignBottom, Qt.black)
app.processEvents()
start = time()
while time() - start < 1:
sleep(0.1)
app.processEvents()
login = Login()
#login.show()
splash.finish(login)
if (login.exec_() == QtGui.QDialog.Accepted) :
main = MainWindow()
main.show()
sys.exit(app.exec_())
def change_password(self): #pylint:disable=missing-docstring,wrong-import-order,trailing-whitespace, multiple-imports
try:
a = Login(user=self.user, password=self.pswd) #pylint:disable=invalid-name
session = a.get_session()
change_password_html = session.get(CHANGE_PASSWORD_PAGE)
soup_data = change_password_html.text
soup = BeautifulSoup(soup_data, "html.parser")
state = soup.find(id="__VIEWSTATE")
viewstate = state["value"]
payload = {
"__EVENTTARGET": "",
"__EVENTARGUMENT": "",
" __VIEWSTATE": viewstate,
"txtOldPassword": self.pswd,
"txtNewPassword": self.new_pswd,
"txtConfirmPassword": self.new_pswd,
"btnUpdate": "Change Password",
"hdPasswordLength": 6,
"hdRollid": 3
}
session.post(CHANGE_PASSWORD_PAGE, data=payload, verify=False)
except: #pylint:disable=bare-except
traceback.print_exc(file=sys.stdout)
def DoTasker():
lstUserData = Data.laodUserData()
emails = []
for d in lstUserData:
login = Login(d[0], d[1])
skey, reason = login.autologin()
if skey == False:
#签到失败,删除用户信息,发送邮件提醒
email = Data.getUserEmailData(d[0])
if email != None and len(email) > 0:
emails.append(email)
Data.removeUserData(d[0])
Data.removeUserEmailData(d[0])
sign = Sign(d[0], skey)
if sign.autosign() == False:
pass
#time.sleep(1)
#print emails
if len(emails) > 0:
email = libs.mail.Message(Config.smtp_user, emails, Config.mail_notify)
try:
conn = libs.mail.Connection(Config.smtp_server, 25,
Config.smtp_user, Config.smtp_pass)
conn.send_message(email)
except:
pass
def LoginActivity(email, password):
firebase = FB.Config(Settings.settings)
user = FB.UserLogin(email=email, password=password, cfg=firebase)
try:
session["user"] = user.auth.get_account_info(
user.user["idToken"])["users"]
except Exception as e:
print(e)
def callback():
login = Login()
username = usuario.get()
password = contrasena.get()
if login.verificar(username, password) == "True":
message.configure(text="Bienvenido")
from AltasBajasGUI import AltasGUI
else:
message.configure(
text="Nombre de usuario o contraseña incorrectos")
def loginProcess(self):
self.id = ui.log_idInput.text()
password = ui.log_passInput.text()
self.login = Login(self.conn)
self.login.checkResult(self.id, password)
if self.login.loginResult == 1:
ui.stackedWidget.setCurrentIndex(1)
self.showList()
elif self.login.loginResult == 2:
self.login_alarm("아이디 혹은 비밀번호를 다시 입력해주세요")
def instBot():
work_done = False
while not work_done:
user = UserDetails()
selenium = SeleniumDetails()
selenium.getDriver().get("https://www.instagram.com")
login = Login(user, selenium)
if login.main():
profile = Profile(user, selenium)
if profile.main():
work_done = True
selenium.getDriver().quit()
def __init__(self, email, password):
self.headers = {
'Referer': 'https://github.com/',
'User-Agent':
'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/57.0.2987.133 Safari/537.36',
'Host': 'github.com'
}
self.email = email
self.password = password
self.foller_url = 'https://github.com/{name}?tab=followers'
self.login = Login()
self.session = self.login.login(email, password)
self.set = set()
def get_categories_and_save_to_db():
login = Login()
login.login()
a_elements = driver.find_elements_by_xpath(
"//table[@class='categories']//child::a")
categories = []
for a_element in a_elements:
link_href = a_element.get_attribute('href')
category_id = int((link_href.split('?')[-1]).split("=")[-1])
category_name = a_element.get_attribute('text')
new_category = Category(category_name, category_id)
categories.append(new_category)
database.insert_category(category_id, category_name)
return categories
def login():
form = LoginForm(request.form)
#get username / password
username = form.username.data
password = form.password.data
accounts = root.child('accounts').get()
list = []
for accountid in accounts:
eachaccount = accounts[accountid]
if eachaccount['account type'] == 'iuser' or eachaccount[
'account type'] == 'idoctor' or eachaccount[
'account type'] == 'iinstructor':
account = Login(eachaccount['account type'],
eachaccount['username'], eachaccount['password'])
account.set_registerid(accountid)
list.append(account)
#---------------------------------------------------------
#--------------------------------------------------------
if request.method == 'POST' and form.validate():
username_list = []
password_list = []
while True:
for result in list:
username_list.append(result.get_username())
password_list.append(result.get_password())
print(username_list)
print(password_list)
break
if username in username_list and password in password_list:
if username_list.index(username) == password_list.index(password):
session['logged_in'] = True
return redirect(url_for('view_Booking_Page'))
else:
error = 'Invalid login'
flash(error, 'danger')
return render_template('login.html', form=form)
# print("Username:"******"Password:",account.get_password())
# if username == account.get_username() and password == account.get_password():
# session['logged_in'] = True
# return redirect(url_for('view_Booking_Page'))
# else:
# error = 'Invalid login'
# flash(error,'danger')
# return render_template('login.html',form=form)
return render_template('login.html', form=form)
def login():
while True:
os.system("cls")
print("[ LOGIN ]")
username = input("ENTER USERNAME >> ")
password = getpass.getpass("ENTER PASSWORD >> ")
user = Login(username, password)
if user.valid_user():
print("Login Successful!")
time.sleep(1)
break
else:
print("Login Unsuccessful, try again..")
time.sleep(1)
continue
def run_login(self, widget):
# Seleccionar un Usuario o llama a crear uno nuevo.
from Login import Login
login = Login()
login.connect("crear_usuario", self.crear_usuario)
login.connect("load_usuario", self.run_menu)
login.run()
return False
def test_search_006(self):
"""测试搜索搜索功能"""
Login(self.wd).test_login002("13253728975", "123456w")
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[2]/div[4]')
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[1]/div[1]/div')
time.sleep(2)
Base(self.wd).sendKeys(
By.CSS_SELECTOR,
'#app > div:nth-child(2) > div.search-input > div > div > div > input[type=text]',
'朝花夕拾')
time.sleep(2)
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[1]/div/span[2]')
all = WebDriverWait(
self.wd, 30,
1).until(lambda x: x.find_elements(By.CSS_SELECTOR, '.item-title'))
text = all[1].text
print(text)
try:
self.wd.find_element_by_xpath("//*[contains(text(),'朝花夕拾')]")
except:
return False
def test_search_007(self):
"""测试阅读页面是否存在"""
Login(self.wd).test_login002("13253728975", "123456w")
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[2]/div[4]')
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[1]/div[1]/div')
print('点击输入框')
time.sleep(2)
# Base.(self.wd).sendKeys(By.CSS_SELECTOR,'#app > div:nth-child(2) > div.search-input > div > div > div > input[type=text]')
Base(self.wd).sendKeys(
By.CSS_SELECTOR,
'#app > div:nth-child(2) > div.search-input > div > div > div > input[type=text]',
'朝花夕拾')
print('输入')
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[1]/div/span[2]')
time.sleep(2)
Base(self.wd).ClickElement(
By.XPATH,
'//*[@id="app"]/div[2]/div[2]/div/div[1]/div/ul[1]/div/li/div/div[2]'
)
time.sleep(2)
text = Base(self.wd).text(By.CSS_SELECTOR, '.author')
assert text == '鲁迅'
def test_search_009(self):
"""测试--搜索历史--"""
Login(self.wd).test_login002("13253728975", "123456w")
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[2]/div[4]')
time.sleep(1)
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[1]/div[1]/div')
print('点击输入框')
time.sleep(2)
Base(self.wd).sendKeys(By.CSS_SELECTOR, '.input-container>input',
'朝花夕拾')
print('输入')
time.sleep(1)
# 点击搜索
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[1]/div/span[2]')
print('sou')
time.sleep(2)
Base(self.wd).ClickElement(
By.CSS_SELECTOR,
'#app div:nth-child(2) > div.search-input > div > span.cancel')
time.sleep(1)
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[1]/div[1]/div')
# time.sleep(4)
all = WebDriverWait(self.wd, 30, 1).until(lambda x: x.find_elements(
By.CSS_SELECTOR, 'div.search-history > ul > li:nth-child(1) > span'
))
all[0].click()
time.sleep(1)
text = Base(self.wd).text(By.CSS_SELECTOR, '.author')
assert text == '鲁迅'
def test_search_005(self):
"""测试搜索取消功能"""
Login(self.wd).test_login002("13253728975", "123456w")
Base(self.wd).ClickElement(By.XPATH,
'//*[@id="app"]/div[2]/div[2]/div[4]')
time.sleep(1)
Base(self.wd).ClickElement(
By.CSS_SELECTOR,
'#app>div:nth-child(2)>div:nth-child(1)>div.discover-search>div')
print(5)
time.sleep(2)
Base(self.wd).sendKeys(
By.CSS_SELECTOR,
'#app > div:nth-child(2) > div.search-input > div > div > div > input[type=text]',
'朝花夕拾')
time.sleep(2)
Base(self.wd).ClickElement(
By.CSS_SELECTOR,
'#app>div:nth-child(2)>div.search-input>div>span.cancel')
print(5)
text = Base(self.wd).text(
By.CSS_SELECTOR,
'#app > div:nth-child(2) > div:nth-child(1) > div.van-list > div:nth-child(1) > div:nth-child(1) > div.list-title > span'
)
assert text == '根据《零点超控》推荐'
class Solution:
newUserId = 1110
print("1.LOGIN \n 2.SIGNUP")
opt = int(input("enter the option"))
if (opt == 1):
user = "******"
userName = input("enter the username")
userType = input("enter the usertype:(Admin/Student)")
password = input("enter the password")
oldUser = Login(user, userName, userType, password)
else:
user = "******"
userName = newUserId + 1
userType = input("enter the usertype(STUDENT/ADMIN):")
password = input("set the password")
newUser = Login(user, userName, userType, password)
def __init__(self, *args, **kwargs):
# __init__ function for class Tk
tk.Tk.__init__(self, *args, **kwargs)
# creating a container and setting the layout, which can change the frames based on user activity
container = tk.Frame(self)
container.pack(side="top", fill="both", expand=True)
container.grid_rowconfigure(0, weight=1)
container.grid_columnconfigure(0, weight=1)
# Below mentioned are the list of frames which are called upon with the button press, we are sending the
# sending the value of new window as string so that there would be new window populated basing on the user event
self.frames = {
"Login":
Login(parent=container, controller=self),
"HomePage":
HomePage(parent=container, controller=self),
"RegisteringProducts":
RegisteringProducts(parent=container, controller=self),
"IndividualBilling":
IndividualBilling(parent=container, controller=self),
"ShippingProducts":
ShippingProducts(parent=container, controller=self)
}
# Aligning the layout of all available frame windows, row=0, column=0 implies they all be on same position
self.frames["Login"].grid(row=0, column=0, sticky="nsew")
self.frames["HomePage"].grid(row=0, column=0, sticky="nsew")
self.frames["RegisteringProducts"].grid(row=0, column=0, sticky="nsew")
self.frames["IndividualBilling"].grid(row=0, column=0, sticky="nsew")
self.frames["ShippingProducts"].grid(row=0, column=0, sticky="nsew")
# First window to be shown when user opens app
self.show_frame("Login")
def __init__(self):
"""
初始化相关参数
"""
self.login = Login()
self.login.login()
self.headers = self.login.getHeaders()
self.data = self.login.getData()
self.session = self.login.getSession()
# self.session.encoding = 'utf8'
# print self.session.
# data link list
self.havefinished_list = [] # 已经处理的数据连接
self.waitting_list = [] # 未处理的数据连接
self.record_topic_data = dict()
self.record_topic_link_data = dict()
def configure_game():
global root
update_window_size(window, 800, 450)
if root is not None:
root.destroy()
root = tk.Frame()
Login(root, auth, configure_menu)
def login(self, username, password=Tzjconfig.password, url=Tzjconfig.login_url):
'''
登录，获取token，这里需填写用户名，password默认是test1111，url有默认值
:param username:
:param password:
:param url:
:return:
'''
ll = Login()
backdata = ll.tzj_login(url=url, username=username, password=password)
if backdata.has_key('token'):
self.token = backdata['token']
return self.token
else:
print U'无法获取token'
return self.token
def test_view__all_classes(self):
a = App(Login(DjangoInterface()), UserEdits(), CourseEdit())
result = a.command("LoggedInUser")
# Error cases
self.assertEqual("Not enrolled in any classes", result)
# Success
self.assertEqual("List of classes: ", result)
def show_go_screen():
fondo_menu = pygame.transform.scale(pygame.image.load("recursos/background.png"), [800, 600])
screen.blit(fondo_menu, [0, 0])
draw_text(screen, "Wing Gundam Zero", 65, WIDTH // 2, HEIGHT // 4)
draw_text(screen, "Press Esc to return main menu", 20, WIDTH // 2, HEIGHT * 3/4)
draw_text(screen, "Press Enter to play", 20, WIDTH // 2, HEIGHT // 2)
pygame.display.flip()
waiting = True
while waiting:
clock.tick(60)
for event in pygame.event.get():
if event.type == pygame.QUIT:
pygame.mixer_music.stop()
pygame.quit()
if event.type == pygame.KEYUP:
if event.key == pygame.K_RETURN:
waiting = False
if event.key == pygame.K_ESCAPE:
pygame.mixer_music.stop()
login = Login()
menu = Game(documento_usuario, nombre, nombre_usuario, correo, fecha_nacimiento, pais, login)
menu.menu_principal()
pygame.quit()
sys.exit()
class Teste:
l = Login()
while true:
try:
l.CallMenu(l.Validation())
except ValueError:
print("Login Inválido")
def test_ViewDatabase(self):
a = App(Login(DjangoInterface()), UserEdits(), CourseEdit())
result = a.command("LoggedInUser")
# Error cases
self.assertEqual("Illegal permissions to do this activity", result)
self.assertEqual("Error while connecting to the database", result)
# Success
self.assertEqual("Data gathered", result)
def test_login(self):
driver = self.driver
driver.maximize_window()
driver.get(self.base_url)
"""登录"""
Login().user_login(driver)
"""退出"""
Logout().user_logout(driver)
def main():
# Basic response to begin conversation
#first perform a login
obj = Reminder()
mailobj = Mail()
global counter
global name
global username
log = Login()
name, username = log.logger()
#print("USERNAME="******" ".join(re.findall("[a-zA-Z0-9]+", str(name)))
if name != '':
counter = 0
if counter == 1: #Add name to database yet to implement
print("Welcome to ChatterBox. What is your name?")
name = input()
print("Hello " + name + "\n I'm Alice")
if counter == 0:
print('Hello ' + name + ". Welcome back. Here are todays reminders...")
#os.system('play --no-show-progress --null --channels 1 synth %s sine %f' % (1, 500))
time.sleep(0.5)
obj.RemindersToday(username)
print("Here are your emails")
time.sleep(2)
mailobj.LatestMails()
time.sleep(1)
print("How can I help you?")
#reception of inputs, for each input, perform processing to understand what the input requires.
Statement = input()
Statement = Statement.lower()
# print(Statement)
reply = Statement
counter = 0
global cond
cond = 0
#start giving replies
while process(Statement) != -1:
#preprocessing done
Statement = input("How can I help you?\n")
def test_Add_project(self):
"""新建长租产品"""
driver = self.driver
driver.maximize_window()
driver.get(self.base_url)
Login().user_login(driver)
Add_Project().Add_project(driver)
sleep(10)
def logout(self):
"""Log out and show the login screen."""
if self.user is not None:
self.startDb.logEntry(self.user[0], 2, None, self.cursor)
self.connection.commit()
self.user = None
self.setWindowTitle(self.getVersion())
widget = Login(self, self)
self.changeDisplay(widget)
def test_send__TA_email(self):
a = App(Login(DjangoInterface()), UserEdits(), CourseEdit())
result = a.command("From To Subject Body TA UpdatedUser")
# Error cases
self.assertEqual("Illegal permissions to do this activity", result)
self.assertEqual("TA(s) do not exist", result)
self.assertEqual("Illegal email subject", result)
self.assertEqual("Illegal email body", result)
# Success
self.assertEqual("Email sent to TA(s) successfully", result)
def __init__(self):
http_client.HTTPConnection.debuglevel = 0
try:
Logging.info("geting cookies")
self.requests = requests.Session()
self.requests.cookies = cookielib.LWPCookieJar("zhihu_api/cookies")
self.cookies = self.requests.cookies
self.requests.cookies.load(ignore_discard=True)
Login.islogin()
except NotLogin:
Login.login()
except:
Logging.error(u"找不到cookie")
reload(sys)
sys.setdefaultencoding("utf8")
Logging.info("Default encoding: " + sys.getdefaultencoding())
self.proxies = {"http": "http://127.0.0.1:8080", "https": "http://127.0.0.1:8080"}
def login(s): l = Login() l.username = USERNAME l.password = PASSWORD l.platform = PLATFORM l.version = VERSION l.protocol = PROTOCOL s.send(str(l))
def __init__(self): """ 初始化相关参数 """ self.login = Login() self.login.login() self.headers = self.login.getHeaders() self.data = self.login.getData() self.session = self.login.getSession() # self.session.encoding = 'utf8' # print self.session. # data link list self.havefinished_list = [] # 已经处理的数据连接 self.waitting_list = [] # 未处理的数据连接 self.record_topic_data = dict() self.record_topic_link_data = dict()
def nsuite():
n = unittest.TestSuite()
n.addTest(Signup.SignUp("test"))
n.addTest(General.general())
n.addTest(Lists.lists())
n.addTest(Views.views())
n.addTest(Subtasks.subtasks())
n.addTest(Files.files())
n.addTest(Recurrence.recurrence())
n.addTest(Positioning.positioning())
n.addTest(DefaultFolder.default_folder())
n.addTest(Sharing.sharing())
n.addTest(ListSharing.list_sharing())
n.addTest(Login.login())
return n
def main():
SEND_MSG = ("[自动回复]这里是机器人,稍后回复您的消息,"
"抱歉.(望勿频繁调戏机器人...)")
bot = Login()
LOG.info('请扫描二维码.')
print(bot.get_QRcode())
os.remove('QRcode.png')
Is_login = False
while not Is_login:
time.sleep(2)
res = bot.is_login().split(',')[-2]
LOG.info(res)
Is_login = True if '登录成功' in res else False
LOG.info('获取ptwebqq...')
bot.get_ptwebqq()
LOG.info('获取vfwebqq...')
bot.get_vfwebqq()
LOG.info('获取psessionid...')
bot.get_psessionid()
LOG.info('等待消息...')
STOP = False
is_open = True
while not STOP:
time.sleep(1)
try:
msg = bot.poll()
msg_content, from_uin, msg_type = hand_msg(msg[0])
LOG.info('{0} 发来一条消息: {1}'.format(from_uin, msg_content.encode('utf-8')))
if (is_open is True) and ('STOP' not in msg_content):
msg = "{0}[{1}]".format(SEND_MSG, random.randint(0, 10))
send_status = bot.send_msg(msg, from_uin, msg_type)
LOG.info('回复 {0}: {1}'.format(from_uin, send_status))
elif 'STOP' in msg_content:
LOG.info('CLOSE...')
bot.send_msg('CLOSED', from_uin, msg_type)
is_open = False
elif 'START' in msg_content:
LOG.info('STARTED')
bot.send_msg('OPENED', from_uin, msg_type)
is_open = True
except KeyboardInterrupt:
LOG.info('See You...')
STOP = True
except:
LOG.error(sys.exc_info())
# -*-coding: UTF-8-*-
import time,re
from HotTopic import HotTopic
from Login import Login
login = Login('yourUserName','yourPWD')
login.login()
#每天的话题
''''
topic = HotTopic(login.gsid)
topic.hot_topic()
'''
#
''''''
#搜话题
topic = HotTopic(login.gsid)
for eachline in file('key.txt','r'):
print eachline
print type(eachline)
topic.linkProducer(eachline.strip())
def testUsuarioCorrecto(self):
login = Login()
self.assertEqual("True", login.verificar("juanPerez", "cartmancuco"))
def testUsuaioIncorrecto(self):
login = Login()
self.assertEqual(
"False", login.verificar("juanitoMartinez", "noExisto"))
#-*-coding=utf-8
import sys,time
sys.path.append(r'/home/irene/Documents/python/Class')
sys.path.append(r'/home/irene/Documents/python/Class/System')
from Login import Login
from User import User
from selenium import webdriver
driver = webdriver.Firefox()
#~~~~~~~~~~~~~~~~~~~~~~~~登录~~~~~~~~~~~~~~~~~~~~~~~~~
#Login(ip,username,password)
l = Login("192.168.88.183","admin","admin",driver)
l.test_login()
#~~~~~~~~~~~~~~~~~~~~~~~~登录成功~~~~~~~~~~~~~~~~~~~~~
#~~~~~~~~~~~~~~~~~~~~~~~~注册用户~~~~~~~~~~~~~~~~~~~~~
#User(driver)
#AddUser(name,password,email,telephone,privilege)
#privilege---1:系统权限；2:上级监控；3:备份恢复；4:查看报表
u = User(driver)
#u.AddUser("kkkkkk3","dingjia123","*****@*****.**","123456789","14")
#time.sleep(5)
#u.DeleteUser("kkkkkk3",'n')
#u.LockUser("dingjia",'y')
#u.UnlockUser("dingjia",'y')
u.ModifidUser('dingjia','','','4')
#~~~~~~~~~~~~~~~~~~~~~~~~注册成功~~~~~~~~~~~~~~~~~~~~~
#~~~~~~~~~~~~~~~~~~~~~~~~客户端访问~~~~~~~~~~~~~~~~~~~
#VisitedCient(user,clients),client多客户端由;号分开
#u.VisitedCient(u,"dingjia","localhost.localdomain_192.168.88.215;rhel5_mysql_216")
class SpiderTopicData(object):
"""
获取知乎数据
"""
def __init__(self):
"""
初始化相关参数
"""
self.login = Login()
self.login.login()
self.headers = self.login.getHeaders()
self.data = self.login.getData()
self.session = self.login.getSession()
# self.session.encoding = 'utf8'
# print self.session.
# data link list
self.havefinished_list = [] # 已经处理的数据连接
self.waitting_list = [] # 未处理的数据连接
self.record_topic_data = dict()
self.record_topic_link_data = dict()
def setRootTopic(self,root_topic_id='19778317'):
self.waitting_list.append(root_topic_id)
def relogin(self):
"""
短线重连
:return:
"""
self.login.login()
self.headers = self.login.getHeaders()
self.data = self.login.getData()
self.session = self.login.getSession()
def getLinkTopic(self,link_url='https://www.zhihu.com/topic/19778317/organize/entire?parent=19778317'):
"""
获取连接下的数据
:param link_url:
:return: 数据 text
"""
# self.session
try:
res = self.session.post(link_url,data=self.data,headers=self.headers)
time.sleep(random.randint(4,12))
except:
self.relogin()
res = self.session.post(link_url,data=self.data,headers=self.headers)
time.sleep(random.randint(4,12))
topic = json.loads(res.text)
# topic = eval(res.text)
cur_topic = topic['msg'][0]
sub_topics = topic['msg'][1]
# for test
parent_topic_name = cur_topic[1].encode('utf8')
parent_topic_id = cur_topic[2]
sub_topics_name = []
sub_topics_id = []
for sub in sub_topics:
sub_topics_id.append(sub[0][2])
sub_topics_name.append(sub[0][1].encode('utf8'))
result = dict()
result["parent_topic_id"] = parent_topic_id
result["parent_topic_name"] = parent_topic_name
result["sub_topics_name"] = sub_topics_name
result["sub_topics_id"] = sub_topics_id
return result
def recordData(self,result):
"""
处理爬取的topic 数据，并记录
:param result:
:return:
"""
sub_topics_id = result['sub_topics_id']
sub_topics_name = result['sub_topics_name']
parent_topic_name = result['parent_topic_name']
parent_topic_id = result['parent_topic_id']
existed = False
child_topic_id = ''
if parent_topic_id not in self.record_topic_data:
self.record_topic_data[parent_topic_id] = parent_topic_name
for sub_id,sub_name in zip(sub_topics_id,sub_topics_name):
if sub_name == str("加载更多"):
existed = True
child_topic_id = sub_id
continue
if sub_id not in self.record_topic_data:
self.record_topic_data[sub_id] = sub_name
if sub_id not in self.record_topic_link_data:
self.record_topic_link_data[sub_id] = []
self.record_topic_link_data[sub_id].append(parent_topic_id)
return existed,parent_topic_id,child_topic_id
def getSubTopic(self,parent_topic_id='19778317',child_topic_id=''):
"""
爬取知乎某一话题下的所有子话题，仅爬取话题下一层（即仅爬取当前话题的孩子话题，孙子不管）
:param parent_topic_id: 当前话题 ID
:param child_topic_id: 子话题 ID应对显示不全时
:return:
"""
state = True
sub_topics_id = []
while state:
url_link = 'https://www.zhihu.com/topic/{0}/organize/entire'.format(parent_topic_id)
if child_topic_id:
url_link += "?child={}&parent={}".format(child_topic_id, parent_topic_id)
result = self.getLinkTopic(url_link)
sub_topics_id.extend(result['sub_topics_id'])
state,parent_topic_id,child_topic_id = self.recordData(result)
return sub_topics_id
def getAllTopic(self):
"""
爬取队列所有话题及子话题
:return:
"""
while self.waitting_list:
topic_id = self.waitting_list.pop(0)
if topic_id in self.havefinished_list:
continue
sub_topics = self.getSubTopic(parent_topic_id=topic_id)
self.waitting_list.extend((list(set(sub_topics))))
self.havefinished_list.append(topic_id)
print "当前一获取topic number:{}\t 当前已遍历 topic number: {}"\
.format(len(self.havefinished_list)+len(self.waitting_list),len(self.havefinished_list))
self.writeResulttoFile()
def writeResulttoFile(self,topic_file=config.TopicFilePath+"zhihu_topic.json",topic_link_file = config.TopicFilePath+'zhihu_topic_link.json'):
with codecs.open(topic_file,'w',encoding='utf8') as topic_fp:
json.dump(self.record_topic_data,topic_fp,ensure_ascii=False,encoding='utf8')
with codecs.open(topic_link_file,'w',encoding='utf8') as topic_link_fp:
json.dump(self.record_topic_link_data,topic_link_fp,ensure_ascii=False,encoding='utf8')
