def run(self):
global work_mutex
global work_numbers
err_local = 0
try:
socket = TSocket(self.server_ip, int(self.server_port))
transport = TFramedTransport(socket)
transport.open()
protocol = TBinaryProtocol.TBinaryProtocol(transport)
client = ThriftNeloEventServer.Client(protocol)
stop_flag = True
while stop_flag:
#read thrift from file
f = file(file_name, 'r')
fd_transport = TFileObjectTransport(f)
buffered_transport = TBufferedTransport(fd_transport)
binary_protocol = TBinaryProtocol.TBinaryProtocol(
buffered_transport)
fd_transport.open()
stop_flag = False
try:
evt = ThriftNeloEvent()
while True:
evt.read(binary_protocol)
#send the log to each project name
for prjName, logCnt in prj_dict.items():
try:
if logCnt_dict.has_key(prjName):
if int(logCnt_dict[prjName]) < int(logCnt):
evt.projectName = prjName
evt.sendTime = int(time.time() * 1000)
err = client.ackedAppend(evt)
tps_remember(err)
err_local += err
logCnt_dict[
prjName] = logCnt_dict[prjName] + 1
stop_flag = True
else:
evt.projectName = prjName
err = client.ackedAppend(evt)
tps_remember(err)
err_local += err
logCnt_dict[prjName] = 1
stop_flag = True
except TException, msg:
print msg, prjName
except EOFError, msg:
buffered_transport.close() #close the transport
stop_flag = True
work_mutex.acquire()
work_numbers -= 1
work_mutex.release()
socket.close()
def run(self):
global work_mutex
global work_numbers
err_local = 0
try:
socket = TSocket(self.server_ip, int(self.server_port))
transport = TFramedTransport(socket)
transport.open()
protocol = TBinaryProtocol.TBinaryProtocol(transport)
client = ThriftNeloEventServer.Client(protocol)
stop_flag = True
while stop_flag:
#read thrift from file
f = file(file_name, 'r')
fd_transport = TFileObjectTransport(f)
buffered_transport = TBufferedTransport(fd_transport)
binary_protocol = TBinaryProtocol.TBinaryProtocol(buffered_transport)
fd_transport.open()
stop_flag = False
try:
evt = ThriftNeloEvent()
while True:
evt.read(binary_protocol)
#send the log to each project name
for prjName, logCnt in prj_dict.items():
try:
if logCnt_dict.has_key(prjName):
if int(logCnt_dict[prjName]) < int(logCnt):
evt.projectName = prjName
evt.sendTime = int(time.time() * 1000)
err = client.ackedAppend(evt)
tps_remember(err)
err_local += err
logCnt_dict[prjName] = logCnt_dict[prjName] + 1
stop_flag = True
else:
evt.projectName = prjName
err = client.ackedAppend(evt)
tps_remember(err)
err_local += err
logCnt_dict[prjName] = 1
stop_flag = True
except TException, msg:
print msg, prjName
except EOFError,msg:
buffered_transport.close() #close the transport
stop_flag = True
work_mutex.acquire()
work_numbers -= 1
work_mutex.release()
socket.close()
def get_mention_from_wikilink_thrift_file(fn):
f = open(args.thrift_data_dir + '/%03d' % fn)
out_val = defaultdict(list)
p = TBinaryProtocol.TBinaryProtocolAccelerated(TFileObjectTransport(f))
pp = WikiLinkItem()
while True:
try:
pp.read(p)
except EOFError:
break
for m in pp.mentions:
c = m.context
if c is not None:
url = simplify_wiki_url(m.wiki_url)
# Follow url redirect.
try:
url = redirect[hash(url)]
except KeyError:
pass
if url in POOL:
# if c.left.startswith('the Musical August 30th, 2009 | Author: operator Shrek the Musical is a musical with music by Jeanine Tesori and a book and lyrics'):
# print 1, fn
out_val[url].append([c.left, c.middle, c.right])
print fn, len(out_val), sum(len(e) for e in out_val.itervalues())
out_val.default_factory = None # FINALIZE out_val
return out_val
def _process_socket(self, client, address):
"""A greenlet for handling a single client."""
client = TFileObjectTransport(client.makefile())
itrans = self.inputTransportFactory.getTransport(client)
otrans = self.outputTransportFactory.getTransport(client)
iprot = self.inputProtocolFactory.getProtocol(itrans)
oprot = self.outputProtocolFactory.getProtocol(otrans)
try:
while True:
self.processor.process(iprot, oprot)
except EOFError:
pass
except Exception:
self.log.exception("caught exception while processing thrift request")
itrans.close()
otrans.close()
def _process_socket(self, client, address):
"""A greenlet for handling a single client."""
client = TFileObjectTransport(client.makefile())
itrans = self.inputTransportFactory.getTransport(client)
otrans = self.outputTransportFactory.getTransport(client)
iprot = self.inputProtocolFactory.getProtocol(itrans)
oprot = self.outputProtocolFactory.getProtocol(otrans)
try:
while True:
self.processor.process(iprot, oprot)
except EOFError:
pass
except Exception:
self.log.exception(
"caught exception while processing thrift request")
itrans.close()
otrans.close()
def _handle_request(self, listener_name, sock, addr):
client = TFileObjectTransport(sock.makefile())
itrans = self.tfactory.getTransport(client)
otrans = self.tfactory.getTransport(client)
iprot = self.pfactory.getProtocol(itrans)
oprot = self.pfactory.getProtocol(otrans)
try:
while True:
(name, type, seqid) = iprot.readMessageBegin()
request_start = time.time()
try:
timeout_con = Timeout(self.cfg.timeout, Timeout)
timeout_con.start()
if name not in self.wsgi._processMap:
iprot.skip(TType.STRUCT)
iprot.readMessageEnd()
x = TApplicationException(
TApplicationException.UNKNOWN_METHOD, "Unknown function %s" % (name))
oprot.writeMessageBegin(
name, TMessageType.EXCEPTION, seqid)
x.write(oprot)
oprot.writeMessageEnd()
oprot.trans.flush()
raise ThriftFuncNotFound
else:
self.wsgi._processMap[name](self.wsgi, seqid, iprot, oprot)
except ThriftFuncNotFound, ex:
self.log.error("Unknown function %s" % (name))
self.log.access(
addr, name, "FUNC_NOT_FOUND", time.time() - request_start)
break
except Timeout, ex:
self.log.error("A greenlet process timeout.")
self.log.access(
addr, name, "TIMEOUT", time.time() - request_start)
break
def read_thrift(file_obj, ttype):
"""Read a thrift structure from the given fo."""
from thrift.transport.TTransport import TFileObjectTransport, TBufferedTransport
starting_pos = file_obj.tell()
# set up the protocol chain
ft = TFileObjectTransport(file_obj)
bufsize = 2 ** 16
# for accelerated reading ensure that we wrap this so that the CReadable transport can be used.
bt = TBufferedTransport(ft, bufsize)
pin = TCompactProtocol(bt)
# read out type
obj = ttype()
obj.read(pin)
# The read will actually overshoot due to the buffering that thrift does. Seek backwards to the correct spot,.
buffer_pos = bt.cstringio_buf.tell()
ending_pos = file_obj.tell()
blocks = ((ending_pos - starting_pos) // bufsize) - 1
if blocks < 0:
blocks = 0
file_obj.seek(starting_pos + blocks * bufsize + buffer_pos)
return obj
arg_parser.add_argument('--out_fn',
default='data/wiki_link_url_counts.pkl',
type=str)
args = arg_parser.parse_args()
import sys
sys.path.append(args.thrift_class_dir)
from edu.umass.cs.iesl.wikilink.expanded.data.constants import WikiLinkItem
pp = WikiLinkItem()
from collections import defaultdict
out_val = defaultdict(int)
for fn in xrange(1, 110):
with open(args.thrift_data_dir + '/%03d' % fn) as f:
p = TBinaryProtocol.TBinaryProtocolAccelerated(TFileObjectTransport(f))
print fn, len(out_val)
while True:
try:
pp.read(p)
except EOFError:
break
for m in pp.mentions:
c = m.context
if c is not None:
url = m.wiki_url
out_val[url] += 1
import cPickle as pickle
with open(args.out_fn, 'wb') as out_f:
pickle.dump(dict(out_val), out_f, -1)