class Scheduler(object):
def __init__(self, dupefilter, jobdir=None, dqclass=None, mqclass=None, logunser=False, stats=None):
self.df = dupefilter
self.dqdir = self._dqdir(jobdir)
self.dqclass = dqclass
self.mqclass = mqclass
self.logunser = logunser
self.stats = stats
@classmethod
def from_crawler(cls, crawler):
settings = crawler.settings
dupefilter_cls = load_object(settings['DUPEFILTER_CLASS'])
dupefilter = dupefilter_cls.from_settings(settings)
dqclass = load_object(settings['SCHEDULER_DISK_QUEUE'])
mqclass = load_object(settings['SCHEDULER_MEMORY_QUEUE'])
logunser = settings.getbool('LOG_UNSERIALIZABLE_REQUESTS')
return cls(dupefilter, job_dir(settings), dqclass, mqclass, logunser, crawler.stats)
def has_pending_requests(self):
return len(self) > 0
def open(self, spider):
self.spider = spider
self.mqs = PriorityQueue(self._newmq)
self.dqs = self._dq() if self.dqdir else None
return self.df.open()
def close(self, reason):
if self.dqs:
prios = self.dqs.close()
with open(join(self.dqdir, 'active.json'), 'w') as f:
json.dump(prios, f)
return self.df.close(reason)
def enqueue_request(self, request):
if not request.dont_filter and self.df.request_seen(request):
return
if not self._dqpush(request):
self._mqpush(request)
def next_request(self):
return self.mqs.pop() or self._dqpop()
def __len__(self):
return len(self.dqs) + len(self.mqs) if self.dqs else len(self.mqs)
def _dqpush(self, request):
if self.dqs is None:
return
try:
reqd = request_to_dict(request, self.spider)
self.dqs.push(reqd, -request.priority)
except ValueError, e: # non serializable request
if self.logunser:
log.msg("Unable to serialize request: %s - reason: %s" % \
(request, str(e)), level=log.ERROR, spider=self.spider)
return
else:
class Scheduler(object):
def __init__(self, dupefilter, jobdir=None, dqclass=None, mqclass=None, logunser=False):
self.df = dupefilter
self.dqdir = self._dqdir(jobdir)
self.dqclass = dqclass
self.mqclass = mqclass
self.logunser = logunser
@classmethod
def from_settings(cls, settings):
dupefilter_cls = load_object(settings['DUPEFILTER_CLASS'])
dupefilter = dupefilter_cls.from_settings(settings)
dqclass = load_object(settings['SCHEDULER_DISK_QUEUE'])
mqclass = load_object(settings['SCHEDULER_MEMORY_QUEUE'])
logunser = settings.getbool('LOG_UNSERIALIZABLE_REQUESTS')
return cls(dupefilter, job_dir(settings), dqclass, mqclass, logunser)
def has_pending_requests(self):
return len(self) > 0
def open(self, spider):
self.spider = spider
self.mqs = PriorityQueue(self._newmq)
self.dqs = self._dq() if self.dqdir else None
return self.df.open()
def close(self, reason):
if self.dqs:
prios = self.dqs.close()
with open(join(self.dqdir, 'active.json'), 'w') as f:
json.dump(prios, f)
return self.df.close(reason)
def enqueue_request(self, request):
if not request.dont_filter and self.df.request_seen(request):
return
if not self._dqpush(request):
self._mqpush(request)
def next_request(self):
return self.mqs.pop() or self._dqpop()
def __len__(self):
return len(self.dqs) + len(self.mqs) if self.dqs else len(self.mqs)
def _dqpush(self, request):
if self.dqs is None:
return
try:
reqd = request_to_dict(request, self.spider)
self.dqs.push(reqd, -request.priority)
except ValueError, e: # non serializable request
if self.logunser:
log.msg("Unable to serialize request: %s - reason: %s" % \
(request, str(e)), level=log.ERROR, spider=self.spider)
return
else:
def _dq(self):
activef = join(self.dqdir, 'active.json')
if exists(activef):
with open(activef) as f:
prios = json.load(f)
else:
prios = ()
q = PriorityQueue(self._newdq, startprios=prios)
if q:
log.msg("Resuming crawl (%d requests scheduled)" % len(q), \
spider=self.spider)
return q
def open(self, spider):
self.spider = spider
self.mqs = PriorityQueue(self._newmq)
self.dqs = self._dq() if self.dqdir else None
return self.df.open()
def setUp(self):
self.q = PriorityQueue(self.qfactory)
class FifoMemoryPriorityQueueTest(unittest.TestCase):
def setUp(self):
self.q = PriorityQueue(self.qfactory)
def qfactory(self, prio):
return track_closed(FifoMemoryQueue)()
def test_push_pop_noprio(self):
self.q.push('a')
self.q.push('b')
self.q.push('c')
self.assertEqual(self.q.pop(), 'a')
self.assertEqual(self.q.pop(), 'b')
self.assertEqual(self.q.pop(), 'c')
self.assertEqual(self.q.pop(), None)
def test_push_pop_prio(self):
self.q.push('a', 3)
self.q.push('b', 1)
self.q.push('c', 2)
self.q.push('d', 1)
self.assertEqual(self.q.pop(), 'b')
self.assertEqual(self.q.pop(), 'd')
self.assertEqual(self.q.pop(), 'c')
self.assertEqual(self.q.pop(), 'a')
self.assertEqual(self.q.pop(), None)
def test_len_nonzero(self):
assert not self.q
self.assertEqual(len(self.q), 0)
self.q.push('a', 3)
assert self.q
self.q.push('b', 1)
self.q.push('c', 2)
self.q.push('d', 1)
self.assertEqual(len(self.q), 4)
self.q.pop()
self.q.pop()
self.q.pop()
self.q.pop()
assert not self.q
self.assertEqual(len(self.q), 0)
def test_close(self):
self.q.push('a', 3)
self.q.push('b', 1)
self.q.push('c', 2)
self.q.push('d', 1)
iqueues = self.q.queues.values()
self.assertEqual(sorted(self.q.close()), [1, 2, 3])
assert all(q.closed for q in iqueues)
def test_close_return_active(self):
self.q.push('b', 1)
self.q.push('c', 2)
self.q.push('a', 3)
self.q.pop()
self.assertEqual(sorted(self.q.close()), [2, 3])
def test_popped_internal_queues_closed(self):
self.q.push('a', 3)
self.q.push('b', 1)
self.q.push('c', 2)
p1queue = self.q.queues[1]
self.assertEqual(self.q.pop(), 'b')
self.q.close()
assert p1queue.closed
class FifoDiskPriorityQueueTest(FifoMemoryPriorityQueueTest):
def setUp(self):
self.q = PriorityQueue(self.qfactory)
def qfactory(self, prio):
return track_closed(FifoDiskQueue)(self.mktemp())
def test_nonserializable_object_one(self):
self.assertRaises(TypeError, self.q.push, lambda x: x, 0)
self.assertEqual(self.q.close(), [])
def test_nonserializable_object_many_close(self):
self.q.push('a', 3)
self.q.push('b', 1)
self.assertRaises(TypeError, self.q.push, lambda x: x, 0)
self.q.push('c', 2)
self.assertEqual(self.q.pop(), 'b')
self.assertEqual(sorted(self.q.close()), [2, 3])
def test_nonserializable_object_many_pop(self):
self.q.push('a', 3)
self.q.push('b', 1)
self.assertRaises(TypeError, self.q.push, lambda x: x, 0)
self.q.push('c', 2)
self.assertEqual(self.q.pop(), 'b')
self.assertEqual(self.q.pop(), 'c')
self.assertEqual(self.q.pop(), 'a')
self.assertEqual(self.q.pop(), None)
self.assertEqual(self.q.close(), [])
class Scheduler(object):
def __init__(self, dupefilter, jobdir=None, dqclass=None):
self.df = dupefilter
self.dqdir = join(jobdir, 'requests.queue') if jobdir else None
self.dqclass = dqclass
@classmethod
def from_settings(cls, settings):
dupefilter_cls = load_object(settings['DUPEFILTER_CLASS'])
dupefilter = dupefilter_cls.from_settings(settings)
dqclass = load_object(settings['SCHEDULER_DISK_QUEUE'])
return cls(dupefilter, job_dir(settings), dqclass)
def has_pending_requests(self):
return len(self) > 0
def open(self, spider):
self.spider = spider
self.mqs = PriorityQueue(self._newmq)
self.dqs = self._dq() if self.dqdir else None
return self.df.open()
def close(self, reason):
if self.dqs:
prios = self.dqs.close()
with open(join(self.dqdir, 'active.json'), 'w') as f:
json.dump(prios, f)
return self.df.close()
def enqueue_request(self, request):
if not request.dont_filter and self.df.request_seen(request):
return
if not self._dqpush(request):
self._mqpush(request)
def next_request(self):
return self.mqs.pop() or self._dqpop()
def __len__(self):
return len(self.dqs) + len(self.mqs) if self.dqs else len(self.mqs)
def _dqpush(self, request):
if self.dqs is None:
return
try:
reqd = request_to_dict(request, self.spider)
self.dqs.push(reqd, request.priority)
except ValueError: # non serializable request
return
else:
stats.inc_value('scheduler/disk_enqueued', spider=self.spider)
return True
def _mqpush(self, request):
stats.inc_value('scheduler/memory_enqueued', spider=self.spider)
self.mqs.push(request, request.priority)
def _dqpop(self):
if self.dqs:
d = self.dqs.pop()
if d:
return request_from_dict(d, self.spider)
def _newmq(self, priority):
return MemoryQueue()
def _newdq(self, priority):
return self.dqclass(join(self.dqdir, 'p%s' % priority))
def _dq(self):
activef = join(self.dqdir, 'active.json')
if exists(activef):
with open(activef) as f:
prios = json.load(f)
else:
prios = ()
q = PriorityQueue(self._newdq, startprios=prios)
if q:
log.msg("Resuming crawl (%d requests scheduled)" % len(q), \
spider=self.spider)
return q
def setUp(self):
qfactory = lambda x: TestMemoryQueue()
self.q = PriorityQueue(qfactory)
class PriorityQueueTest(unittest.TestCase):
def setUp(self):
qfactory = lambda x: TestMemoryQueue()
self.q = PriorityQueue(qfactory)
def test_push_pop_noprio(self):
self.q.push('a')
self.q.push('b')
self.q.push('c')
self.assertEqual(self.q.pop(), 'a')
self.assertEqual(self.q.pop(), 'b')
self.assertEqual(self.q.pop(), 'c')
self.assertEqual(self.q.pop(), None)
def test_push_pop_prio(self):
self.q.push('a', 3)
self.q.push('b', 1)
self.q.push('c', 2)
self.q.push('d', 1)
self.assertEqual(self.q.pop(), 'b')
self.assertEqual(self.q.pop(), 'd')
self.assertEqual(self.q.pop(), 'c')
self.assertEqual(self.q.pop(), 'a')
self.assertEqual(self.q.pop(), None)
def test_len_nonzero(self):
assert not self.q
self.assertEqual(len(self.q), 0)
self.q.push('a', 3)
assert self.q
self.q.push('b', 1)
self.q.push('c', 2)
self.q.push('d', 1)
self.assertEqual(len(self.q), 4)
self.q.pop()
self.q.pop()
self.q.pop()
self.q.pop()
assert not self.q
self.assertEqual(len(self.q), 0)
def test_close(self):
self.q.push('a', 3)
self.q.push('b', 1)
self.q.push('c', 2)
self.q.push('d', 1)
iqueues = self.q.queues.values()
self.assertEqual(sorted(self.q.close()), [1, 2, 3])
assert all(q.closed for q in iqueues)
def test_popped_internal_queues_closed(self):
self.q.push('a', 3)
self.q.push('b', 1)
self.q.push('c', 2)
p1queue = self.q.queues[1]
self.assertEqual(self.q.pop(), 'b')
self.q.close()
assert p1queue.closed
class Scheduler(SettingObject):
dupfilter_class = StringField(default="scrapy.dupefilter.RFPDupeFilter")
schedule_disk_queue = StringField(default="scrapy.squeue.PickleLifoDiskQueue")
schedule_memory_queue = StringField(default="scrapy.squeue.LifoMemoryQueue")
log_unserailizable_requests = BooleanField(default=False)
jobdir = StringField(default="")
def __init__(self, settings):
super(Scheduler, self).__init__(settings)
dupefilter_cls = load_object(self.dupfilter_class.to_value())
dupefilter = dupefilter_cls(self.metas)
dqclass = load_object(self.schedule_disk_queue.to_value())
mqclass = load_object(self.schedule_memory_queue.to_value())
logunser = self.log_unserailizable_requests.to_value()
self.df = dupefilter
self.jobpath = self.__job_dir(self.jobdir.to_value())
self.dqdir = self._dqdir(self.jobpath)
self.dqclass = dqclass
self.mqclass = mqclass
self.logunser = logunser
def __job_dir(self, path):
if path and not os.path.exists(path):
os.makedirs(path)
return path
def has_pending_requests(self):
return len(self) > 0
def open(self, spider):
self.spider = spider
self.mqs = PriorityQueue(self._newmq)
self.dqs = self._dq() if self.dqdir else None
return self.df.open()
def close(self, reason):
if self.dqs:
prios = self.dqs.close()
with open(join(self.dqdir, 'active.json'), 'w') as f:
json.dump(prios, f)
return self.df.close(reason)
def enqueue_request(self, request):
if not request.dont_filter and self.df.request_seen(request):
return
if not self._dqpush(request):
self._mqpush(request)
def next_request(self):
return self.mqs.pop() or self._dqpop()
def __len__(self):
return len(self.dqs) + len(self.mqs) if self.dqs else len(self.mqs)
def _dqpush(self, request):
if self.dqs is None:
return
try:
reqd = request_to_dict(request, self.spider)
self.dqs.push(reqd, -request.priority)
except ValueError, e: # non serializable request
if self.logunser:
log.msg("Unable to serialize request: %s - reason: %s" % \
(request, str(e)), level=log.ERROR, spider=self.spider)
return
else:
