CLSS is a DOM traversal engine based on CSS selectors. It makes use of the Plump-DOM and is used by lQuery.
Load CLSS through Quicklisp or ASDF:
(ql:quickload :clss)
Using a standard CSS selector you can retrieve a vector of nodes from the DOM:
(clss:select "img" (plump:parse "<div><p>A beautiful image: <img src="//example.com/image.png" alt="image" /></p></div>"))
CLSS implements Level 3 selectors and offers most of the features from the spec. Some things were left out as they make no sense outside a CSS context.
As Plump supports XML as well as HTML, it also includes special handling for a few nodes that are not elements and are thus not reachable by standard CSS selectors. In order to solve this problem, CLSS adds an extra operator, the ^ caret. The caret is followed by a Plump-DOM class-name and will then match any elements that conform to a typep test against it.
(clss:select "^CDATA" (plump:parse "<foo><![CDATA[bar]]></foo>"))
CSS selectors in themselves also don't support XML namespaces due to the ambiguity arising with pseudo-selectors. CLSS solves this by interpreting a double colon as a name. Thus, a tag with the name of foo:bar is selected by foo::bar.
CLSS attempts to be a fast engine and various parts of it have been tuned for this purpose, which limits the extensibility of CLSS a bit. However, it is still possible to f.e. programmatically construct a selector.
Using define-pseudo-selector you can add your own extensions to CLSS:
(clss:define-pseudo-selector outside-link (node)
(let ((href (plump:attribute node "href")))
(and href (cl-ppcre:scan "^(http|https)://" href))))
(clss:select "a:outside-link" (plump:parse "<foo><a href=\"/baloney\"/><a href=\"http://example.com\"/></foo>"))
