mirror of https://github.com/CIRCL/AIL-framework
fix: [Crawler] faup
parent
3a347a79b7
commit
6daa750e3b
|
@ -289,7 +289,9 @@ if __name__ == '__main__':
|
||||||
|
|
||||||
print(f'SPLASH Name: {splash_name}')
|
print(f'SPLASH Name: {splash_name}')
|
||||||
print(f'Proxy Name: {proxy_name}')
|
print(f'Proxy Name: {proxy_name}')
|
||||||
|
print(f'Crawler Type: {get_splash_crawler_type(splash_name)})
|
||||||
|
|
||||||
|
#time.sleep(10)
|
||||||
#sys.exit(0)
|
#sys.exit(0)
|
||||||
|
|
||||||
#rotation_mode = deque(['onion', 'regular'])
|
#rotation_mode = deque(['onion', 'regular'])
|
||||||
|
@ -330,7 +332,7 @@ if __name__ == '__main__':
|
||||||
db=p.config.getint("ARDB_Onion", "db"),
|
db=p.config.getint("ARDB_Onion", "db"),
|
||||||
decode_responses=True)
|
decode_responses=True)
|
||||||
|
|
||||||
faup = Faup()
|
faup = crawlers.get_faup()
|
||||||
|
|
||||||
# get HAR files
|
# get HAR files
|
||||||
default_crawler_har = p.config.getboolean("Crawler", "default_crawler_har")
|
default_crawler_har = p.config.getboolean("Crawler", "default_crawler_har")
|
||||||
|
|
|
@ -67,6 +67,10 @@ def is_valid_onion_domain(domain):
|
||||||
return True
|
return True
|
||||||
return False
|
return False
|
||||||
|
|
||||||
|
# TEMP FIX
|
||||||
|
def get_faup():
|
||||||
|
return faup
|
||||||
|
|
||||||
################################################################################
|
################################################################################
|
||||||
|
|
||||||
# # TODO: handle prefix cookies
|
# # TODO: handle prefix cookies
|
||||||
|
@ -894,7 +898,7 @@ def get_splash_manager_session_uuid():
|
||||||
return res['session_uuid']
|
return res['session_uuid']
|
||||||
else:
|
else:
|
||||||
print(req.json())
|
print(req.json())
|
||||||
except requests.exceptions.ConnectionError:
|
except requests.exceptions.ConnectionError, requests.exceptions.MissingSchema:
|
||||||
# splash manager unreachable
|
# splash manager unreachable
|
||||||
update_splash_manager_connection_status(False)
|
update_splash_manager_connection_status(False)
|
||||||
|
|
||||||
|
|
Loading…
Reference in New Issue