How to generate the start_urls dynamically in crawling?

I am crawling a site which may contain a lot of start_urls, like:

http://www.a.com/list_1_2_3.htm

I want to populate start_urls like [list_\d+_\d+_\d+\.htm], and extract items from URLs like [node_\d+\.htm] during crawling.

Can I use CrawlSpider to realize this function? And how can I generate the start_urls dynamically in crawling?