def execute(cls, stage, state, data, next_allowed_exec_time=None):
"""Execute the operation, rate limiting allowing."""
try:
context = Context.from_state(state, stage)
now = datetime.utcnow()
if next_allowed_exec_time and now < next_allowed_exec_time:
# task not allowed to run yet; put it back in the queue
Queue.queue(stage, state, data, delay=next_allowed_exec_time)
elif context.crawler.disabled:
pass
elif context.stage.rate_limit:
try:
with rate_limiter(context):
context.execute(data)
except RateLimitException:
delay = max(1, 1.0 / context.stage.rate_limit)
delay = random.randint(1, int(delay))
context.log.info("Rate limit exceeded, delaying %d sec.",
delay)
Queue.queue(stage, state, data, delay=delay)
else:
context.execute(data)
except Exception:
log.exception("Task failed to execute:")
finally:
# Decrease the pending task count after excuting a task.
Queue.decr_pending(context.crawler)
# If we don't have anymore tasks to execute, time to clean up.
if not context.crawler.is_running:
context.crawler.aggregate(context)
def execute(cls, stage, state, data, next_allowed_exec_time=None):
"""Execute the operation, rate limiting allowing."""
now = datetime.utcnow()
if next_allowed_exec_time and now < next_allowed_exec_time:
# task not allowed to run yet; put it back in the queue
Queue.queue(stage, state, data, delay=next_allowed_exec_time)
return
context = Context.from_state(state, stage)
if context.crawler.disabled:
return
if context.stage.rate_limit:
try:
with rate_limiter(context):
context.execute(data)
return
except RateLimitException:
delay = max(1, 1.0 / context.stage.rate_limit)
delay = random.randint(1, int(delay))
context.log.info("Rate limit exceeded, delaying %d sec.",
delay)
Queue.queue(stage, state, data, delay=delay)
context.execute(data)
def after_task(self, task):
if task.job.is_done():
stage = CrawlerStage.detach_namespace(task.stage.stage)
state = task.context
context = Context.from_state(state, stage)
context.crawler.aggregate(context)
self.timeout_expiration_check()
def handle(self, task):
apply_task_context(task)
data = task.payload
stage = CrawlerStage.detach_namespace(task.stage.stage)
state = task.context
context = Context.from_state(state, stage)
context.execute(data)
def handle(self, task):
data = task.payload
stage = task.stage.stage
state = task.context
context = Context.from_state(state, stage)
if context.crawler.disabled:
return
context.execute(data)
def test_dump_load_state(self, context, crawler, stage):
dump = context.dump_state()
new_context = Context.from_state(dump, stage.name)
assert isinstance(new_context, Context)
assert new_context.run_id == context.run_id
assert new_context.crawler.name == crawler.name
assert new_context.stage.name == stage.name
assert all((k, v) in new_context.state.items()
for k, v in context.state.items())
def context():
ctx = Context(crawler(), stage(), {"foo": "bar"})
# Assign a fake operation id, so that the DB doesn't complain about
# NotNullContraint while saving events etc.
ctx.operation_id = randint(1, 99999)
return ctx
def get_context():
ctx = Context(get_crawler(), get_stage(), {"foo": "bar"})
ctx.run_id = str(uuid.uuid4())
return ctx
def after_task(self, task):
if task.job.is_done():
stage = task.stage.stage
state = task.context
context = Context.from_state(state, stage)
context.crawler.aggregate(context)
def handle(self, task):
data = task.payload
stage = task.stage.stage
state = task.context
context = Context.from_state(state, stage)
context.execute(data)