lib.rs

// XML parsing adventure

#![license = "MIT"]
#![feature(phase)]


// extern crate simhash;

#[phase(plugin)]
extern crate peg_syntax_ext;

use html::{name, start_tag, end_tag, element, text};

enum Node {
    Text(String),
    Element(String, Vec<Node>)
}

peg! html(r#"
#[pub]
name -> String
  = [a-zA-Z0-9_]+ { match_str.to_string() }

#[pub]
start_tag -> String
  = "<" n:name ">" { n }

#[pub]
end_tag -> String
  = "</" n:name ">" { n }

#[pub]
element -> super::Node
  = s:start_tag c:content e:end_tag { super::Element(s, c) }

#[pub]
content -> Vec<super::Node>
  = (element / text)*

#[pub]
text -> super::Node
  = (!start_tag !end_tag .)+ { super::Text(match_str.to_string()) }
"#)

#[test]
fn test_parser() {
    /*
  assert!(name("div").is_ok());
  assert!(name("di'v").is_err());

  assert_eq!(start_tag("<div>"), Ok("div"));
  assert!(start_tag("</div>").is_err());

  assert_eq!(end_tag("</div>"), Ok("div"));
  assert!(end_tag("<div>").is_err());

  assert_eq!(
    element("<div>A</div>"),
    Ok(     "<div>A</div>"));
  assert!(element("<divA</").is_err());
  assert_eq!(
    element("<a>A<b>B</b></a>"),
    Ok(     "<a>A<b>B</b></a>"));

  assert!(text("lorem ipsum").is_ok());
  assert!(text("lorem ipsum<div>").is_err());
  */
}