26 lines
701 B
Python
26 lines
701 B
Python
"""
|
|
Download timeout middleware
|
|
|
|
See documentation in docs/topics/downloader-middleware.rst
|
|
"""
|
|
|
|
from scrapy import signals
|
|
|
|
|
|
class DownloadTimeoutMiddleware:
|
|
|
|
def __init__(self, timeout=180):
|
|
self._timeout = timeout
|
|
|
|
@classmethod
|
|
def from_crawler(cls, crawler):
|
|
o = cls(crawler.settings.getfloat('DOWNLOAD_TIMEOUT'))
|
|
crawler.signals.connect(o.spider_opened, signal=signals.spider_opened)
|
|
return o
|
|
|
|
def spider_opened(self, spider):
|
|
self._timeout = getattr(spider, 'download_timeout', self._timeout)
|
|
|
|
def process_request(self, request, spider):
|
|
if self._timeout:
|
|
request.meta.setdefault('download_timeout', self._timeout)
|