This code extracts HTML that matches CSS3-ish selectors.

• selectorp.g, the grammar file for Amit J. Patel's Yapps 2.
• htmlp.py, the SGMLParser that uses the generated selectorp.py (I can't put the selectorp.py up because you need the yappsrt.py module anyway).

I won't say I outright stole Mark Pilgrim's HTML parser code, but how many ways are there to use sgmllib? I certainly used his code as a model though.

It doesn't support combinators (foo + bar, foo > bar, foo ~ bar), though it does descendence (foo bar) just fine. Actually it'll parse the combinators, it just won't denote them in any way; the HTML parser will ignore them.

It doesn't do pseudos (e.g., :first-child) either. That was too much engineering up front to accomplish in one thunk, and, well... this is already a step up from the application it's replacing.

The attribute comparators it supports are:

=

Attribute must exactly equal the value.

~=

Attribute must contain the value in a space-separated list (e.g., class attributes).

|=

Attribute must contain the value in a hyphen-separated list (i.e., lang attributes).

^=

Attribute must begin with the value.

$=

Attribute must end with the value.

*=

Attribute must contain, somewhere, the value.

For example, selector img[src*="foo"] will match all img tags with the text "foo" in their src URLs.

Yes, this is for Stapler.py. Thanks for asking.