mirror of
https://github.com/ail-project/ail-framework.git
synced 2024-11-26 15:57:16 +00:00
fix: [Crawler splash ResponseNeverReceived] add retry
This commit is contained in:
parent
00573c9401
commit
672bb02bbf
1 changed files with 20 additions and 1 deletions
|
@ -172,7 +172,26 @@ class TorSplashCrawler():
|
|||
# LUA ERROR # # TODO: print/display errors
|
||||
elif 'error' in response.data:
|
||||
if(response.data['error'] == 'network99'):
|
||||
print('Connection to proxy refused')
|
||||
## splash restart ##
|
||||
error_retry = request.meta.get('error_retry', 0)
|
||||
if error_retry < 3:
|
||||
error_retry += 1
|
||||
url= request.meta['current_url']
|
||||
father = request.meta['father']
|
||||
|
||||
self.logger.error('Splash, ResponseNeverReceived for %s, retry in 10s ...', url)
|
||||
time.sleep(10)
|
||||
yield SplashRequest(
|
||||
url,
|
||||
self.parse,
|
||||
errback=self.errback_catcher,
|
||||
endpoint='execute',
|
||||
cache_args=['lua_source'],
|
||||
meta={'father': father, 'current_url': url, 'error_retry' = error_retry},
|
||||
args=self.build_request_arg(response.cookiejar)
|
||||
)
|
||||
else:
|
||||
print('Connection to proxy refused')
|
||||
else:
|
||||
print(response.data['error'])
|
||||
|
||||
|
|
Loading…
Reference in a new issue