def exposed_head(url, ref):
'''
Do a phantomjs HEAD request for url `url`, passing the referrer `ref`
'''
rpc_interface = common.get_rpyc.RemoteJobInterface("Test_Interface!")
print('wat?')
print(rpc_interface)
raw_job = buildjob(
module = 'NUWebRequest',
call = 'getHeadPhantomJS',
dispatchKey = "fetcher",
jobid = -1,
args = [url, ref],
kwargs = {},
additionalData = {'mode' : 'fetch'},
postDelay = 0,
unique_id = url
)
rpc_interface.put_job(raw_job)
while True:
try:
resp = rpc_interface.get_job()
print_response(resp)
if not resp:
time.sleep(1)
except queue.Empty:
print("No response yet?")
def exposed_head(url, ref):
'''
Do a phantomjs HEAD request for url `url`, passing the referrer `ref`
'''
rpc_interface = common.get_rpyc.RemoteJobInterface("Test_Interface!")
print('wat?')
print(rpc_interface)
raw_job = buildjob(
module = 'NUWebRequest',
call = 'getHeadPhantomJS',
dispatchKey = "fetcher",
jobid = -1,
args = [url, ref],
kwargs = {},
additionalData = {'mode' : 'fetch'},
postDelay = 0,
unique_id = url
)
rpc_interface.put_job(raw_job)
while True:
try:
resp = rpc_interface.get_job()
print_response(resp)
if not resp:
time.sleep(1)
except queue.Empty:
print("No response yet?")
def main():
logSetup.initLogging()
remote = rpyc.connect("localhost",
12345,
config=rpyc.core.protocol.DEFAULT_CONFIG)
raw_job = buildjob(module='WebRequest',
call='getItem',
dispatchKey="fetcher",
jobid=-1,
args=['http://www.google.com'],
kwargs={},
additionalData={'mode': 'fetch'},
postDelay=0)
print(remote)
print(remote.root.putJob('wat', raw_job))
while 1:
try:
j = remote.root.getJob("wat")
print("Got job!")
except queue.Empty:
time.sleep(1)
print("No message")
except rpyc.core.vinegar.GenericException as e:
# this is horrible
if 'queue.Empty' in rpyc.core.vinegar._generic_exceptions_cache:
if isinstance(
e, rpyc.core.vinegar.
_generic_exceptions_cache['queue.Empty']):
print("Empty exception")
continue
print("type", type(e))
print("instance", issubclass(type(e), queue.Empty))
import inspect
print(inspect.getmro(type(e)))
# extp = rpyc.core.vinegar._get_exception_class(queue.Empty)
# print(extp)
# print("instance", isinstance(e, extp))
# print("instance", isinstance(type(e), extp))
# print("type", type(extp()))
fakemodule = {
"__module__": "%s/%s" % ("rpyc.core.vinegar", "queue")
}
extp = type("queue.Empty", (rpyc.core.vinegar.GenericException, ),
fakemodule)
print(extp)
print(isinstance(e, extp))
print(isinstance(e, rpyc.core.vinegar.GenericException))
print(rpyc.core.vinegar._generic_exceptions_cache)
raise e
remote.close()
def main():
logSetup.initLogging()
remote = rpyc.connect("localhost", 12345, config = rpyc.core.protocol.DEFAULT_CONFIG)
raw_job = buildjob(
module = 'WebRequest',
call = 'getItem',
dispatchKey = "fetcher",
jobid = -1,
args = ['http://www.google.com'],
kwargs = {},
additionalData = {'mode' : 'fetch'},
postDelay = 0
)
print(remote)
print(remote.root.putJob('wat', raw_job))
while 1:
try:
j = remote.root.getJob("wat")
print("Got job!")
except queue.Empty:
time.sleep(1)
print("No message")
except rpyc.core.vinegar.GenericException as e:
# this is horrible
if 'queue.Empty' in rpyc.core.vinegar._generic_exceptions_cache:
if isinstance(e, rpyc.core.vinegar._generic_exceptions_cache['queue.Empty']):
print("Empty exception")
continue
print("type", type(e))
print("instance", issubclass(type(e), queue.Empty))
import inspect
print(inspect.getmro(type(e)))
# extp = rpyc.core.vinegar._get_exception_class(queue.Empty)
# print(extp)
# print("instance", isinstance(e, extp))
# print("instance", isinstance(type(e), extp))
# print("type", type(extp()))
fakemodule = {"__module__" : "%s/%s" % ("rpyc.core.vinegar", "queue")}
extp = type("queue.Empty", (rpyc.core.vinegar.GenericException,), fakemodule)
print(extp)
print(isinstance(e, extp))
print(isinstance(e, rpyc.core.vinegar.GenericException))
print(rpyc.core.vinegar._generic_exceptions_cache )
raise e
remote.close()
def main():
import logSetup
from WebMirror.NewJobQueue import buildjob
logSetup.initLogging()
raw_job = buildjob(
module = 'WebRequest',
call = 'getItem',
dispatchKey = "fetcher",
jobid = -1,
args = ['http://www.google.com'],
kwargs = {},
additionalData = {'mode' : 'fetch'},
postDelay = 0
)
rint = RemoteJobInterface("wat")
print(rint.put_job(raw_job))
print(rint)
while 1:
try:
j = rint.get_job()
if j:
print("Got job!", j)
except queue.Empty:
time.sleep(1)
print("No message")
except Exception as e:
# except pyjsonrpc.JsonRpcError as err:
print("type", type(e))
print("instance", issubclass(type(e), queue.Empty))
import inspect
print(inspect.getmro(type(e)))
raise e
remote.close()
def main():
import logSetup
from WebMirror.NewJobQueue import buildjob
logSetup.initLogging()
raw_job = buildjob(module='WebRequest',
call='getItem',
dispatchKey="fetcher",
jobid=-1,
args=['http://www.google.com'],
kwargs={},
additionalData={'mode': 'fetch'},
postDelay=0)
rint = RemoteJobInterface("wat")
print(rint.put_job(raw_job))
print(rint)
while 1:
try:
j = rint.get_job()
if j:
print("Got job!", j)
except queue.Empty:
time.sleep(1)
print("No message")
except Exception as e:
# except pyjsonrpc.JsonRpcError as err:
print("type", type(e))
print("instance", issubclass(type(e), queue.Empty))
import inspect
print(inspect.getmro(type(e)))
raise e
remote.close()
def put_job(self, put=3):
self.log.info("Loading a row to fetch...")
# haveq = self.db_sess.query(db.NuReleaseItem) \
# .outerjoin(db.NuResolvedOutbound) \
# .filter(db.NuReleaseItem.validated == False) \
# .having(func.count(db.NuResolvedOutbound.parent) < 3) \
# .order_by(func.random()) \
# .group_by(db.NuReleaseItem.id) \
# .limit(max(100, put*3))
# # .order_by(desc(db.NuReleaseItem.first_seen)) \
# moar = haveq.all()
self.log.info("Loading a row to fetch...")
haveq = self.db_sess.query(db.NuReleaseItem) \
.outerjoin(db.NuResolvedOutbound) \
.filter(db.NuReleaseItem.validated == False) \
.having(func.count(db.NuResolvedOutbound.parent) < 3) \
.order_by(desc(db.NuReleaseItem.first_seen)) \
.group_by(db.NuReleaseItem.id) \
.limit(max(100, put*3))
haveset = haveq.all()
# haveset += moar
if not haveset:
self.log.info("No jobs to remote HEAD.")
return
# We pick a large number of items, and randomly choose one of them.
# This lets us weight the fetch preferentially to the recent items, but still
# have some variability.
haveset = random.sample(haveset, min(put, len(haveset)))
for have in haveset:
if len(list(have.resolved)) >= 3:
raise RuntimeError("Overresolved item that's not valid.")
if (have.referrer == "http://www.novelupdates.com" or
have.referrer == "https://www.novelupdates.com" or
have.referrer == "https://www.novelupdates.com/" or
have.referrer == "http://www.novelupdates.com/"):
self.log.error("Wat?")
self.log.error("Bad Referrer URL got into the input queue!")
self.log.error("Id: %s", have.id)
continue
self.log.info("Putting job for url '%s'", have.outbound_wrapper)
self.log.info("Referring page '%s'", have.referrer)
raw_job = buildjob(
module = 'NUWebRequest',
call = 'getHeadTitlePhantomJS',
dispatchKey = "fetcher",
jobid = -1,
args = [have.outbound_wrapper, have.referrer],
kwargs = {},
additionalData = {
'mode' : 'fetch',
'wrapper_url' : have.outbound_wrapper,
'referrer' : have.referrer
},
postDelay = 0,
unique_id = have.outbound_wrapper
)
self.rpc.put_job(raw_job)
def put_job(self, put=3):
self.log.info("Loading rows to fetch...")
recent_d = datetime.datetime.now() - datetime.timedelta(hours=72)
recentq = self.db_sess.query(db.NuReleaseItem) \
.outerjoin(db.NuResolvedOutbound) \
.filter(db.NuReleaseItem.validated == False) \
.filter(db.NuReleaseItem.first_seen >= recent_d) \
.having(func.count(db.NuResolvedOutbound.parent) < 3) \
.order_by(desc(db.NuReleaseItem.first_seen)) \
.group_by(db.NuReleaseItem.id) \
.limit(max(100, put*3))
bulkq = self.db_sess.query(db.NuReleaseItem) \
.outerjoin(db.NuResolvedOutbound) \
.filter(db.NuReleaseItem.validated == False) \
.having(func.count(db.NuResolvedOutbound.parent) < 3) \
.order_by(desc(db.NuReleaseItem.first_seen)) \
.group_by(db.NuReleaseItem.id) \
.limit(max(100, put))
bulkset = bulkq.all()
recentset = recentq.all()
self.log.info("Have %s recent items, %s long-term items to fetch",
len(recentset), len(bulkset))
haveset = bulkset + recentset
# haveset = recentset
# haveset += moar
if not haveset:
self.log.info("No jobs to remote HEAD.")
return
# We pick a large number of items, and randomly choose one of them.
# This lets us weight the fetch preferentially to the recent items, but still
# have some variability.
haveset = random.sample(haveset, min(put, len(haveset)))
for have in haveset:
if len(list(have.resolved)) >= 3:
raise RuntimeError("Overresolved item that's not valid.")
if (have.referrer == "http://www.novelupdates.com"
or have.referrer == "https://www.novelupdates.com"
or have.referrer == "https://www.novelupdates.com/"
or have.referrer == "http://www.novelupdates.com/"):
self.log.error("Wat?")
self.log.error("Bad Referrer URL got into the input queue!")
self.log.error("Id: %s", have.id)
continue
self.log.info("Putting job for url '%s'", have.outbound_wrapper)
self.log.info("Referring page '%s'", have.referrer)
raw_job = buildjob(
module='WebRequest',
call='getHeadTitleChromium',
dispatchKey="fetcher",
jobid=-1,
args=[have.outbound_wrapper, have.referrer],
kwargs={},
additionalData={
'mode': 'fetch',
'wrapper_url': have.outbound_wrapper,
'referrer': have.referrer
},
postDelay=0,
unique_id=have.outbound_wrapper,
serialize=True,
)
# rval = random.random()
# if rval >= 0.5:
# raw_job = buildjob(
# module = 'NUWebRequest',
# call = 'getHeadTitlePhantomJS',
# dispatchKey = "fetcher",
# jobid = -1,
# args = [have.outbound_wrapper, have.referrer],
# kwargs = {},
# additionalData = {
# 'mode' : 'fetch',
# 'wrapper_url' : have.outbound_wrapper,
# 'referrer' : have.referrer
# },
# postDelay = 0,
# unique_id = have.outbound_wrapper
# )
# else:
# raw_job = buildjob(
# module = 'WebRequest',
# call = 'getHeadTitleChromium',
# dispatchKey = "fetcher",
# jobid = -1,
# args = [have.outbound_wrapper, have.referrer],
# kwargs = {},
# additionalData = {
# 'mode' : 'fetch',
# 'wrapper_url' : have.outbound_wrapper,
# 'referrer' : have.referrer
# },
# postDelay = 0,
# unique_id = have.outbound_wrapper
# )
self.rpc.put_job(raw_job)