crawler.properties

利用广度优先遍历搜索一定范围内的所有网页,可用于建立搜索引擎和查找网络错误.

### database_server_name: IP address of the machine on which you installed PostgreSQL database.
database_server_name = localhost
### database_server_port: PostgreSQL database TCP port. 5432 is the default value.
database_server_port = 5432
### database_name: do not change this if you use dbsetup.sql unmodified.
database_name = crawler
### database_user_name: it depends on your PostgreSQL configuration.
database_user_name = postgres
### log_file: if you want to keep a log file in a different file name, specify it here.
log_file = crawler.txt
#start_url = http://www.cs.usu.edu/~lim/index.html
### start_url: staring page to crawl. Point to a specific HTML file name.
#start_url = http://www.cs.usu.edu/index.html
### start_url_context: The context of the staring web page.
start_url_context = http://www.cs.usu.edu/