1
0
mirror of https://github.com/scrapy/scrapy.git synced 2025-02-25 00:24:31 +00:00

make SpiderState extension always available, regardless of whether there is a job dir, and make sure it always set spider.state attribute, for consistency between in-memory and on-disk runs

This commit is contained in:
Pablo Hoffman 2011-09-23 12:56:44 -03:00
parent 2559f211a8
commit cfddc314ce
2 changed files with 20 additions and 9 deletions

View File

@ -9,26 +9,28 @@ from scrapy.xlib.pydispatch import dispatcher
class SpiderState(object):
"""Store and load spider state during a scraping job"""
def __init__(self, jobdir):
self.statefn = os.path.join(jobdir, 'spider.state')
def __init__(self, jobdir=None):
self.jobdir = jobdir
@classmethod
def from_crawler(cls, crawler):
jobdir = crawler.settings.get('JOBDIR')
if not jobdir:
raise NotConfigured
obj = cls(jobdir)
obj = cls(crawler.settings.get('JOBDIR'))
dispatcher.connect(obj.spider_closed, signal=signals.spider_closed)
dispatcher.connect(obj.spider_opened, signal=signals.spider_opened)
return obj
def spider_closed(self, spider):
with open(self.statefn, 'wb') as f:
pickle.dump(spider.state, f, protocol=2)
if self.jobdir:
with open(self.statefn, 'wb') as f:
pickle.dump(spider.state, f, protocol=2)
def spider_opened(self, spider):
if os.path.exists(self.statefn):
if self.jobdir and os.path.exists(self.statefn):
with open(self.statefn) as f:
spider.state = pickle.load(f)
else:
spider.state = {}
@property
def statefn(self):
return os.path.join(self.jobdir, 'spider.state')

View File

@ -25,3 +25,12 @@ class SpiderStateTest(unittest.TestCase):
ss2.spider_opened(spider2)
self.assertEqual(spider.state, {'one': 1, 'dt': dt})
ss2.spider_closed(spider2)
def test_state_attribute(self):
# state attribute must be present if jobdir is not set, to provide a
# consistent interface
spider = BaseSpider(name='default')
ss = SpiderState()
ss.spider_opened(spider)
self.assertEqual(spider.state, {})
ss.spider_closed(spider)