Html::lint is a pure-perl html parser and checker for syntactic legitmacy.

the "weblint" script - provided in the separate package "weblint-perl" - is a wrapper around the html::lint.

you can also look into apache::lint which is a mod_perl wrapper around html::lint.

finally, for those of you doing automated testing with test::more and its brethren, test::html::lint lets you automate html checking.