Automated link-checker for system testing
I often have to work with fragile legacy websites that break in unexpected ways when logic or configuration are updated.
I don't have the time or knowledge of the system needed to create a Selenium script. Besides, I don't want to check a specific use case - I want to verify every link and page on the site.
I would like to create an automated system test that will spider through a site and check for broken links and crashes. Ideally, there would be a tool that I could use to achieve this. It should have as many as possible of the following features, in descending order of priority:
- Triggered via script
- Does not require human interaction
- Follows all links including anchor tags and links to CSS and js files
- Produces a log of all found 404s, 500s etc.
- Can be deployed locally to check sites on intranets
- Supports cookie/form-based authentication
- Free/Open source
There are many partial solutions out there, like FitNesse, Firefox's LinkChecker and the W3C link checker, but none of them do everything I need.
I would like to use this test with projects using a range of technologies and platforms, so the more portable the solution the better.
I realise this is no substitute for proper system testing, but it would be very useful if I had a convenient and automatable way of verifying that no part of the site was obviously broken.
Answer
We use and really like Linkchecker:
http://wummel.github.io/linkchecker/
It's open-source, Python, command-line, internally deployable, and outputs to a variety of formats. The developer has been very helpful when we've contacted him with issues.
We have a Ruby script that queries our database of internal websites, kicks off LinkChecker with appropriate parameters for each site, and parses the XML that LinkChecker gives us to create a custom error report for each site in our CMS.