35

is there a chance to stop crawling when specific if condition is true (like scrap_item_id == predefine_value ). My problem is similar to Scrapy - how to identify already scraped urls but I want to 'force' my scrapy spider to stop crawling after discover the last scraped item.

Community
  • 1
  • 1
no1
  • 835
  • 2
  • 9
  • 9
  • 2
    Ok got it ... I'm pretty sure there is a better solution but from scrapy.project import crawler crawler.engine.close_spider(spider, 'closespider_blee') works – no1 Dec 16 '10 at 15:59
  • That solution seems fine. It's used in the scrapy source too (e.g. contrib/closespider.py) – Shane Evans Aug 31 '11 at 01:21

4 Answers4

40

In the latest version of Scrapy, available on GitHub, you can raise a CloseSpider exception to manually close a spider.

In the 0.14 release note doc is mentioned: "Added CloseSpider exception to manually close spiders (r2691)"

Example as per the docs:

def parse_page(self, response):
  if 'Bandwidth exceeded' in response.body:
    raise CloseSpider('bandwidth_exceeded')

See also: http://readthedocs.org/docs/scrapy/en/latest/topics/exceptions.html?highlight=closeSpider

Sjaak Trekhaak
  • 4,596
  • 27
  • 36
11

This question was asked 8 months ago but I was wondering the same thing and have found another (not great) solution. Hopefully this can help the future readers.

I'm connecting to a database in my Pipeline file, if the database connection is unsuccessful, I wanted the Spider to stop crawling (no point in collecting data if there's nowhere to send it). What I ended up doing was using:

from scrapy.project import crawler
crawler._signal_shutdown(9,0) #Run this if the cnxn fails.

This causes the Spider to do the following:

[scrapy] INFO: Received SIGKILL, shutting down gracefully. Send again to force unclean shutdown.

I just kind of pieced this together after reading your comment and looking through the "/usr/local/lib/python2.7/dist-packages/Scrapy-0.12.0.2543-py2.7.egg/scrapy/crawler.py" file. I'm not totally sure what it's doing, the first number delivered to the function is the signame (for example, using 3,0 instead of 9,0 returns error [scrapy] INFO: Received SIGKILL...

Seems to work well enough though. Happy scraping.

EDIT: I also suppose that you could just force your program to shut down with something like:

import sys
sys.exit("SHUT DOWN EVERYTHING!")
Nicolò Gasparini
  • 1,762
  • 2
  • 18
  • 38
alukach
  • 3,780
  • 3
  • 34
  • 33
  • 2
    Thanks for mentioning extension - right now it is really way to go. Here are docs: http://readthedocs.org/docs/scrapy/en/0.12/topics/extensions.html#module-scrapy.contrib.closespider – Victor Farazdagi Sep 13 '11 at 10:15
  • The thing that I dislike about the Close Spider extension is that it can only be initiated by four conditions (timeout, itempassed, pagecount, errorcount; as far as I know). What would be nice is if you could make your own conditions to close the spider, so that it's closed when there is a specific occurance (ex. a certain word is scraped). – alukach Sep 13 '11 at 19:32
  • 3
    the link to the extension is down –  Dec 08 '15 at 10:37
  • The new link to the extension is: https://github.com/scrapy/scrapy/blob/master/scrapy/extensions/closespider.py – matiskay Dec 14 '17 at 20:43
3

From a pipeline, I prefer the following solution.

class MongoDBPipeline(object):

def process_item(self, item, spider):
    spider.crawler.engine.close_spider(self, reason='duplicate')

Source: Force spider to stop in scrapy

Macbric
  • 442
  • 4
  • 10
0

Tried lots of options nothing works. This dirty hack do the trick for Linux:

os.kill(os.getpid(), signal.SIGINT)
os.kill(os.getpid(), signal.SIGINT)

This sends SIGINT signal two times to scrapy. Second signal forces shutdown

Alex
  • 869
  • 6
  • 17