def TestCnWeiboFetcher(user, pwd):
uid = 10029
# uid = 10057
# uid = 10111
# uid = 10145
# uid = 10211
# uid = 10318
# uid = 10361
# uid = 10392
page = 1
f_weibos = './file/weibos-%s.txt' % (uid)
f_follows = './file/follows-%s.txt' % (uid)
f_fans = './file/fans-%s.txt' % (uid)
f_infos = './file/infos-%s.txt' % (uid)
#-------------------------------------------
start = time.time()
fetcher = CnWeiboFetcher(username=user, password=pwd)
print 'test for check cookie weibo.com'
print fetcher.check_cookie()
print 'test for check user exist...'
print fetcher.check_user(uid)
print 'test for fetch weibo...'
url = 'http://weibo.cn/%s?page=%s' % (uid, page)
html = fetcher.fetch(url)
with codecs.open(f_weibos, 'w', 'utf-8') as f:
f.write(html)
print 'test for fetch follows...'
url = 'http://weibo.cn/%s/follow?page=%s' % (uid, page)
html = fetcher.fetch(url)
with codecs.open(f_follows, 'w', 'utf-8') as f:
f.write(html)
print 'test for fetch fans...'
url = 'http://weibo.cn/%s/fans?page=%s' % (uid, page)
html = fetcher.fetch(url)
with codecs.open(f_fans, 'w', 'utf-8') as f:
f.write(html)
print 'test for fetch infos...'
url = 'http://weibo.cn/%s/info' % uid
html = fetcher.fetch(url)
with codecs.open(f_infos, 'w', 'utf-8') as f:
f.write(html)
cost_time = int(time.time() - start)
print 'finished: # connections: %s, cost time: %s' % (
fetcher.n_connections, cost_time)
def TestCnWeiboFetcher(user, pwd):
uid = 10029
# uid = 10057
# uid = 10111
# uid = 10145
# uid = 10211
# uid = 10318
# uid = 10361
# uid = 10392
page = 1
f_weibos = "./file/weibos-%s.txt" % (uid)
f_follows = "./file/follows-%s.txt" % (uid)
f_fans = "./file/fans-%s.txt" % (uid)
f_infos = "./file/infos-%s.txt" % (uid)
# -------------------------------------------
start = time.time()
fetcher = CnWeiboFetcher(username=user, password=pwd)
print "test for check cookie weibo.com"
print fetcher.check_cookie()
print "test for check user exist..."
print fetcher.check_user(uid)
print "test for fetch weibo..."
url = "http://weibo.cn/%s?page=%s" % (uid, page)
html = fetcher.fetch(url)
with codecs.open(f_weibos, "w", "utf-8") as f:
f.write(html)
print "test for fetch follows..."
url = "http://weibo.cn/%s/follow?page=%s" % (uid, page)
html = fetcher.fetch(url)
with codecs.open(f_follows, "w", "utf-8") as f:
f.write(html)
print "test for fetch fans..."
url = "http://weibo.cn/%s/fans?page=%s" % (uid, page)
html = fetcher.fetch(url)
with codecs.open(f_fans, "w", "utf-8") as f:
f.write(html)
print "test for fetch infos..."
url = "http://weibo.cn/%s/info" % uid
html = fetcher.fetch(url)
with codecs.open(f_infos, "w", "utf-8") as f:
f.write(html)
cost_time = int(time.time() - start)
print "finished: # connections: %s, cost time: %s" % (fetcher.n_connections, cost_time)
def TestCnWeiboCrawler(user, pwd):
uid = 10029
# uid = 10057
# uid = 10111
# uid = 10145
# uid = 10211
# uid = 10318
# uid = 10361
# uid = 10392
store_path = './file/'
fetcher = CnWeiboFetcher(username=user, password=pwd)
start = time.time()
login_ok = fetcher.check_cookie()
if not login_ok:
print 'login failed.'
sys.exit()
fetcher.n_connections = 0
print 'crawl follows'
crawler = CnWeiboCrawler(fetcher, store_path, uid=uid)
crawler.crawl_follows()
fetcher.n_connections = 0
print 'crawl fans'
crawler = CnWeiboCrawler(fetcher, store_path, uid=uid)
crawler.crawl_fans()
cost_time = int(time.time() - start)
print 'finished: # connections: %s, cost time: %s' % (
fetcher.n_connections, cost_time)
def TestCnWeiboCrawler(user, pwd):
uid = 10029
# uid = 10057
# uid = 10111
# uid = 10145
# uid = 10211
# uid = 10318
# uid = 10361
# uid = 10392
store_path = './file/'
fetcher = CnWeiboFetcher(username=user, password=pwd)
start = time.time()
login_ok = fetcher.check_cookie()
if not login_ok:
print 'login failed.'
sys.exit()
fetcher.n_connections = 0
print 'crawl follows'
crawler = CnWeiboCrawler(fetcher, store_path, uid=uid)
crawler.crawl_follows()
fetcher.n_connections = 0
print 'crawl fans'
crawler = CnWeiboCrawler(fetcher, store_path, uid=uid)
crawler.crawl_fans()
cost_time = int(time.time() - start)
print 'finished: # connections: %s, cost time: %s' %(fetcher.n_connections, cost_time)
def on_login(self, event):
website = SITE_CHOICES[self.website.GetCurrentSelection()]
account = self.account.GetValue().strip().encode('UTF-8')
password = self.password.GetValue().strip().encode('UTF-8')
if (account is None or len(account) == 0 or password is None
or len(password) == 0):
wx.MessageBox(
message='Account/Password cannot be blank. Please retry!',
caption='Warning',
style=wx.OK | wx.ICON_INFORMATION)
else:
if website in [settings.COMWEIBO, settings.CNWEIBO]:
if website == settings.COMWEIBO:
fetcher = ComWeiboFetcher(username=account,
password=password,
window=self)
elif website == settings.CNWEIBO:
fetcher = CnWeiboFetcher(username=account,
password=password,
window=self)
else:
return
if fetcher.check_cookie():
fetcher.window = None
self.Destroy()
wx.GetApp().ExitMainLoop()
import win_local_crawler as wlc
wlc.main(account_display=account,
website_display=website,
fetcher=fetcher)
elif website == settings.TWITTER:
wx.MessageBox(
message='For Twitter: not implemented. Please retry!',
caption='Error',
style=wx.OK | wx.ICON_INFORMATION)
elif website == settings.FACEBOOK:
wx.MessageBox(
message='For Facebook: not implemented. Please retry!',
caption='Error',
style=wx.OK | wx.ICON_INFORMATION)
def TestWeibo__init__(user, pwd, weibo_com):
if weibo_com:
fetcher = ComWeiboFetcher(username=user, password=pwd)
else:
fetcher = CnWeiboFetcher(username=user, password=pwd)
login_ok = fetcher.check_cookie()
if not login_ok:
print 'login failed.'
sys.exit()
uids = [1000000253, 10057, 10029]
msg_urls = [
'http://weibo.com/1000000253/ezC36cq3i6G',
'http://weibo.com/1713926427/A2V5CENGU'
]
start = time.time()
print 'crawl weibos'
sina_weibo.main(fetcher,
fetch_data='weibos',
store_path='./file/',
uids=uids,
weibo_com=weibo_com)
print 'crawl follows'
sina_weibo.main(fetcher,
fetch_data='follows',
store_path='./file/',
uids=uids,
weibo_com=weibo_com)
print 'crawl fans'
sina_weibo.main(fetcher,
fetch_data='fans',
store_path='./file/',
uids=uids,
weibo_com=weibo_com)
print 'crawl infos'
sina_weibo.main(fetcher,
fetch_data='infos',
store_path='./file/',
uids=uids,
weibo_com=weibo_com)
print 'crawl reposts'
sina_weibo.main(fetcher,
store_path='./file/',
msg_urls=msg_urls,
fetch_data='repost',
weibo_com=weibo_com)
print 'crawl comments'
sina_weibo.main(fetcher,
store_path='./file/',
msg_urls=msg_urls,
fetch_data='comment',
weibo_com=weibo_com)
cost_time = int(time.time() - start)
print 'finished: # connections: %s, cost time: %s' % (
fetcher.n_connections, cost_time)
