Extremely experimental. Drag the following bookmarklet to your bookmark bar, and click it whenever you want to scrape a page. It'll let you click any element in the page, and you'll get its unique selector.
ScraperScript will block input to the page while it's active. To get control over your input again, simply close the ScraperScript bar using the X in the top right corner.
Want to use the resulting selectors with BeautifulSoup? You'll need to use my patched version, which has support for the :nth-of-type() pseudoselector and proper support for direct descendants (this is not part of the stock BeautifulSoup module).
If things break, send an email to scraperscript@cryto.net with a description of what's going wrong, and a link to the page you are trying to work with.
There's also a repository for this stuff on GitHub.