Deferred requests in scrapy - scrapy

Question

I would like to repeatedly scrape the same URLs with different delays. After researching the issue it seemed that the appropriate solution was to use something like
nextreq = scrapy.Request(url, dont_filter=True)
d = defer.Deferred()
delay = 1
reactor.callLater(delay, d.callback, nextreq)
yield d
in parse.
However, I have been unable to make this work. I am getting the error message
ERROR: Spider must return Request, BaseItem, dict or None, got 'Deferred'
I am not familiar with twisted so I hope I am just missing something obvious
Is there a better way of achieving my goal that doesn't fight the framework so much?

Answer 1

I finally found an answer in an old PR
def parse():
req = scrapy.Request(...)
delay = 0
reactor.callLater(delay, self.crawler.engine.schedule, request=req, spider=self)
However, the spider can exit due to being idle too early. Based on the outdated middleware https://github.com/ArturGaspar/scrapy-delayed-requests, this can be remedied with
from scrapy import signals
from scrapy.exceptions import DontCloseSpider
class ImmortalSpiderMiddleware(object):
#classmethod
def from_crawler(cls, crawler):
s = cls()
crawler.signals.connect(s.spider_idle, signal=signals.spider_idle)
return s
#classmethod
def spider_idle(cls, spider):
raise DontCloseSpider()
The final option, updating the middleware by ArturGaspar, led to:
from weakref import WeakKeyDictionary
from scrapy import signals
from scrapy.exceptions import DontCloseSpider
from twisted.internet import reactor
class DelayedRequestsMiddleware(object):
requests = WeakKeyDictionary()
#classmethod
def from_crawler(cls, crawler):
ext = cls()
crawler.signals.connect(ext.spider_idle, signal=signals.spider_idle)
return ext
#classmethod
def spider_idle(cls, spider):
if cls.requests.get(spider):
spider.log("delayed requests pending, not closing spider")
raise DontCloseSpider()
def process_request(self, request, spider):
delay = request.meta.pop('delay_request', None)
if delay:
self.requests.setdefault(spider, 0)
self.requests[spider] += 1
reactor.callLater(delay, self.schedule_request, request.copy(),
spider)
raise IgnoreRequest()
def schedule_request(self, request, spider):
spider.crawler.engine.schedule(request, spider)
self.requests[spider] -= 1
And can be used in parse like:
yield Request(..., meta={'delay_request': 5})