def handle(self):
self.re_title = re.compile("<title1>([^\\x00]+?)</title1>\\s*<title2>([^\\x00]+?)</title2>", re.I)
self.googlechecker = checker();
# receive data
print "From:", self.client_address
datapieces = []
datapiece = self.request.recv(10240)
datapieces.append(datapiece)
print 'first get ', len(datapiece)
piece_last = ''
self.request.setblocking(0)
while(len(datapiece) > 0):
try:
#print 'second get ', len(datapiece)
datapiece = self.request.recv(10240)
datapieces.append(datapiece)
except Exception, e:
print e
time.sleep(0.5)
#break
last2pieces = ''.join((piece_last, datapiece))
if last2pieces.endswith('</root>'):
break
piece_last = datapiece
def handle(self):
self.re_title = re.compile(
"<title1>([^\\x00]+?)</title1>\\s*<title2>([^\\x00]+?)</title2>",
re.I)
self.googlechecker = checker()
# receive data
print "From:", self.client_address
datapieces = []
datapiece = self.request.recv(10240)
datapieces.append(datapiece)
print 'first get ', len(datapiece)
piece_last = ''
self.request.setblocking(0)
while (len(datapiece) > 0):
try:
#print 'second get ', len(datapiece)
datapiece = self.request.recv(10240)
datapieces.append(datapiece)
except Exception, e:
print e
time.sleep(0.5)
#break
last2pieces = ''.join((piece_last, datapiece))
if last2pieces.endswith('</root>'):
break
piece_last = datapiece
def start(self):
host = ''
port = 55555
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
s.bind((host, port))
s.listen(1)
googlechecker = checker()
re_title = re.compile("<title>([^\\x00]+?)</title>", re.I)
while 1:
try:
clientsock, clientaddr = s.accept()
print "Got connection from", clientsock.getpeername()
while 1:
try:
data = clientsock.recv(4096)
# if not len(data):
# break
matchs = re.findall(re_title, data)
if len(matchs) < 2:
print "ERROR NOTENOUGH TITLE:", data
continue
r = googlechecker.isInSamePage(matchs[0], matchs[1])
print 'Result is:', r
except Exception, e:
print e
finally:
clientsock.sendall(str(r))
clientsock.close()
clientsock.close()
def start(self):
host = ""
port = 55555
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
s.bind((host, port))
s.listen(1)
googlechecker = checker()
re_title = re.compile("<title>([^\\x00]+?)</title>", re.I)
while 1:
try:
clientsock, clientaddr = s.accept()
print "Got connection from", clientsock.getpeername()
while 1:
try:
data = clientsock.recv(4096)
# if not len(data):
# break
matchs = re.findall(re_title, data)
if len(matchs) < 2:
print "ERROR NOTENOUGH TITLE:", data
continue
r = googlechecker.isInSamePage(matchs[0], matchs[1])
print "Result is:", r
except Exception, e:
print e
finally:
clientsock.sendall(str(r))
clientsock.close()
clientsock.close()
def start(self):
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
s.bind((self.host, self.port))
s.listen(1)
googlechecker = checker()
re_title = re.compile(
"<title1>([^\\x00]+?)</title1>\\s*<title2>([^\\x00]+?)</title2>",
re.I)
while 1:
try:
clientsock, clientaddr = s.accept()
print "Got connection from", clientsock.getpeername()
while 1:
try:
data = clientsock.recv(409600000)
if not len(data):
break
matchs = re.findall(re_title, data)
print '- XML -----------------------------------------'
print data
print '- XML -----------------------------------------'
paircount = 0
print '- Receive title pair --------------------------'
for title1, title2 in matchs:
print "%s: %s\n\t%s" % (paircount, title1, title2)
paircount += 1
print '< Receive title pair --------------------------'
# if len(matchs) < 2:
# print "ERROR NOTENOUGH TITLE:", data
# continue
# [('t1', 't2'), ('b1', 'b2')]
r = googlechecker.isInSamePageMulti(matchs)
print 'Result is:', str(r)
except Exception, e:
print e
finally:
clientsock.send(str(r))
clientsock.send("\n")
clientsock.close()
clientsock.close()
def start(self):
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
s.bind((self.host, self.port))
s.listen(1)
googlechecker = checker();
re_title = re.compile("<title1>([^\\x00]+?)</title1>\\s*<title2>([^\\x00]+?)</title2>", re.I)
while 1:
try:
clientsock, clientaddr = s.accept()
print "Got connection from", clientsock.getpeername()
while 1:
try:
data = clientsock.recv(409600000)
if not len(data):
break
matchs = re.findall(re_title, data)
print '- XML -----------------------------------------'
print data
print '- XML -----------------------------------------'
paircount = 0;
print '- Receive title pair --------------------------'
for title1, title2 in matchs:
print "%s: %s\n\t%s" % (paircount, title1, title2)
paircount += 1
print '< Receive title pair --------------------------'
# if len(matchs) < 2:
# print "ERROR NOTENOUGH TITLE:", data
# continue
# [('t1', 't2'), ('b1', 'b2')]
r = googlechecker.isInSamePageMulti(matchs)
print 'Result is:', str(r)
except Exception, e:
print e
finally:
clientsock.send(str(r))
clientsock.send("\n")
clientsock.close()
clientsock.close()
def __init__(self):
self.re_title = re.compile("<title1>([^\\x00]+?)</title1>\\s*<title2>([^\\x00]+?)</title2>", re.I)
self.googlechecker = checker();
def __init__(self): self.max_threads = 2 self.google_checker = checker()
def __init__(self):
self.re_title = re.compile(
"<title1>([^\\x00]+?)</title1>\\s*<title2>([^\\x00]+?)</title2>",
re.I)
self.googlechecker = checker()
def __init__(self): self.settings = Settings.getInstance() self.parsegoogle = GoogleResultParser() self.htmlRetriever = HtmlRetriever(self.settings.use_proxy) self.checker = checker()
def __init__(self):
self.settings = Settings.getInstance()
self.parsegoogle = GoogleResultParser()
self.htmlRetriever = HtmlRetriever(self.settings.use_proxy)
self.checker = checker()