package
0.0.0-20241121195025-46bed5c1815d
Repository: https://github.com/niklasfasching/x.git
Documentation: pkg.go.dev
# README
- css selectors for go Yet another package implementing a css selector compiler for go - see e.g. [[https://github.com/ericchiang/css/][github.com/ericchiang/css]] and [[https://github.com/andybalholm/cascadia][github.com/andybalholm/cascadia]] for prior art.
Differentiating features:
- Extensible (allows for user defined pseudo-classes/functions, matchers and combinators)
- Conversion from compiled Selector back to Selector string
- Small (~800 LOC), simple/modular (separation into lexing, parsing & selecting) and fast (benchmarks put it head to head with cascadia)
#+begin_src go import ( "log" "strings"
"github.com/niklasfasching/x/css"
"golang.org/x/net/html"
)
func main() {
doc, _ := html.Parse(strings.NewReader( <p> <span class="a">apple</span> <span class="b">banana</span> <span class="b">berry</span> <span class="c">pear</span> </p> ))
selector := css.MustCompile("p > span.b")
nodes := css.All(selector, doc)
for _, n := range nodes {
var s strings.Builder
html.Render(&s, n)
log.Println(s.String())
}
// <span class="b">banana</span>
// <span class="b">berry</span>
log.Printf("Converted back to string: %s", selector) // Converted back to string: p > span.b
// easy to add your own custom pseudo classes, pseudo functions, matchers & combinators
css.PseudoClasses["my-pseudo-p"] = func(n *html.Node) bool { return n.Data == "p" },
selector = css.MustCompile(":my-pseudo-p")
var s strings.Builder
html.Render(&s, css.First(selector, doc))
log.Println(s.String()) // <p>...</p>
} #+end_src
** but why? for fun
- It seemed easy enough to do
- I've been really into writing parsers lately and this felt a belt more complicated than a lisp
- I enjoy web scraping and wanted to learn more about underlying concepts
- I wanted to learn more about profiling go - and this seemed like a good playground
- The existing css libraries cannot be extended / customized
- resources