1- from datetime import date , timedelta
21from pathlib import Path
32
43import requests
@@ -241,16 +240,18 @@ def get_spider_job_data(self, pid, sid, jid, datatype, last_chunk=None):
241240 self .check_status (response , 200 )
242241 return response .json ()
243242
244- def create_spider_job (self , pid , sid , args = [], env_vars = [], tags = [], day = None ):
243+ def create_spider_job (self , pid , sid , args = [], env_vars = [], tags = [], tier = None , day = None ):
245244 endpoint = "projects/{}/spiders/{}/jobs" .format (pid , sid )
246245 data = {
247246 "args" : args ,
248247 "env_vars" : env_vars ,
249248 "tags" : tags ,
250249 "data_status" : "PENDING" if day else "PERSISTENT" ,
251250 }
251+ if tier :
252+ data ["resource_tier" ] = tier
252253 if day :
253- data ["data_expiry_days" ] = f" { date . today () + timedelta ( days = day ) } "
254+ data ["data_expiry_days" ] = day
254255
255256 response = self .post (endpoint , data = data )
256257 self .check_status (response , 201 )
@@ -263,7 +264,7 @@ def stop_spider_job(self, pid, sid, jid):
263264 return response .json ()
264265
265266 def create_spider_cronjob (
266- self , pid , sid , schedule = "" , args = [], env_vars = [], tags = [], day = None
267+ self , pid , sid , schedule = "" , args = [], env_vars = [], tags = [], tier = None , day = None
267268 ):
268269 endpoint = "projects/{}/spiders/{}/cronjobs" .format (pid , sid )
269270 data = {
@@ -272,9 +273,10 @@ def create_spider_cronjob(
272273 "cenv_vars" : env_vars ,
273274 "ctags" : tags ,
274275 "data_status" : "PENDING" if day else "PERSISTENT" ,
276+ "data_expiry_days" : day if day else 1 ,
275277 }
276- if day :
277- data ["data_expiry_days " ] = f"0/ { day } "
278+ if tier :
279+ data ["resource_tier " ] = tier
278280
279281 response = self .post (endpoint , data = data )
280282 self .check_status (response , 201 )
0 commit comments