This document describes the current stable version of Celery (3.1). For development docs, go here.
celery.contrib.batches¶
Experimental task class that buffers messages and processes them as a list.
Warning
For this to work you have to set
CELERYD_PREFETCH_MULTIPLIER
to zero, or some value where
the final multiplied value is higher than flush_every
.
In the future we hope to add the ability to direct batching tasks to a channel with different QoS requirements than the task channel.
Simple Example
A click counter that flushes the buffer every 100 messages, and every 10 seconds. Does not do anything with the data, but can easily be modified to store it in a database.
# Flush after 100 messages, or 10 seconds.
@app.task(base=Batches, flush_every=100, flush_interval=10)
def count_click(requests):
from collections import Counter
count = Counter(request.kwargs['url'] for request in requests)
for url, count in count.items():
print('>>> Clicks: {0} -> {1}'.format(url, count))
Then you can ask for a click to be counted by doing:
>>> count_click.delay(url='http://example.com')
Example returning results
An interface to the Web of Trust API that flushes the buffer every 100 messages, and every 10 seconds.
import requests
from urlparse import urlparse
from celery.contrib.batches import Batches
wot_api_target = 'https://api.mywot.com/0.4/public_link_json'
@app.task(base=Batches, flush_every=100, flush_interval=10)
def wot_api(requests):
sig = lambda url: url
reponses = wot_api_real(
(sig(*request.args, **request.kwargs) for request in requests)
)
# use mark_as_done to manually return response data
for response, request in zip(reponses, requests):
app.backend.mark_as_done(request.id, response)
def wot_api_real(urls):
domains = [urlparse(url).netloc for url in urls]
response = requests.get(
wot_api_target,
params={'hosts': ('/').join(set(domains)) + '/'}
)
return [response.json()[domain] for domain in domains]
Using the API is done as follows:
>>> wot_api.delay('http://example.com')
Note
If you don’t have an app
instance then use the current app proxy
instead:
from celery import current_app
app.backend.mark_as_done(request.id, response)
API
-
class
celery.contrib.batches.
Batches
[source]¶ -
-
flush_every
= 10¶ Maximum number of message in buffer.
-
flush_interval
= 30¶ Timeout in seconds before buffer is flushed anyway.
-
-
class
celery.contrib.batches.
SimpleRequest
(id, name, args, kwargs, delivery_info, hostname)[source]¶ Pickleable request.
-
args
= ()¶ positional arguments
-
delivery_info
= None¶ message delivery information.
-
hostname
= None¶ worker node name
-
id
= None¶ task id
-
kwargs
= {}¶ keyword arguments
-
name
= None¶ task name
-