2023-07-15 09:11:53 +00:00
|
|
|
# Desc: base config
|
2023-06-27 15:38:30 +00:00
|
|
|
PLATFORM = "xhs"
|
2023-08-16 11:49:41 +00:00
|
|
|
KEYWORDS = "python,golang"
|
2023-07-30 13:30:26 +00:00
|
|
|
LOGIN_TYPE = "qrcode" # qrcode or phone or cookie
|
|
|
|
|
COOKIES = "" # login by cookie, if login_type is cookie, you must set this value
|
2023-11-18 05:38:11 +00:00
|
|
|
CRAWLER_TYPE = "search"
|
2023-06-27 15:38:30 +00:00
|
|
|
|
|
|
|
|
# enable ip proxy
|
|
|
|
|
ENABLE_IP_PROXY = False
|
|
|
|
|
|
|
|
|
|
# retry_interval
|
|
|
|
|
RETRY_INTERVAL = 60 * 30 # 30 minutes
|
|
|
|
|
|
|
|
|
|
# playwright headless
|
|
|
|
|
HEADLESS = True
|
2023-07-15 09:11:53 +00:00
|
|
|
|
|
|
|
|
# save login state
|
2023-07-24 12:59:43 +00:00
|
|
|
SAVE_LOGIN_STATE = True
|
2023-07-15 09:11:53 +00:00
|
|
|
|
|
|
|
|
# save user data dir
|
|
|
|
|
USER_DATA_DIR = "%s_user_data_dir" # %s will be replaced by platform name
|
|
|
|
|
|
2023-07-29 07:35:40 +00:00
|
|
|
# crawler max notes count
|
|
|
|
|
CRAWLER_MAX_NOTES_COUNT = 20
|
2023-07-15 14:25:56 +00:00
|
|
|
|
|
|
|
|
# max concurrency num
|
|
|
|
|
MAX_CONCURRENCY_NUM = 10
|
2023-11-18 05:38:11 +00:00
|
|
|
|
|
|
|
|
|
2023-11-18 07:53:10 +00:00
|
|
|
# xhs specified note id list
|
|
|
|
|
XHS_SPECIFIED_ID_LIST = [
|
2023-11-18 05:38:11 +00:00
|
|
|
"6422c2750000000027000d88",
|
|
|
|
|
"64ca1b73000000000b028dd2",
|
|
|
|
|
"630d5b85000000001203ab41",
|
|
|
|
|
]
|
