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)
