info.yml

# Settings

# load & output
# The path of the data file to be loaded. Create a new search if `load_path` is empty.
load_path:
output_dir: res/apple_cn_search3000
# The destination folder to be output, the default value is `res/YYYY-MM-dd HH:mm:ss/`.
# Number of searched required.
search_num: 3000
timeout: 10

# The image can't be output only if the value of `is_output_image` is 'false' or empty.
is_output_image: false
output_image_size: 50

# start with
# start searching at this URL.
url: https://apple.com.cn/
user_agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/88.0.4324.150 Safari/537.36

# URL filter
# If you do not want to search all the links contained in the site, but filter some of the links you want.
# For example `netloc_regex: zh.wikipedia.org` means that only sites with the domain name 'zh.wikipedia.org' are allowed to search
url_filter:
  url_regex:
  netloc_regex:
  path_regex:

# site settings
site_settings:
  title_regex: # 匹配 ' -' 前的内容