clover's git

markodown

1	//! Regexps to match html elements
2	//!
3	#![allow(non_upper_case_globals)]
4	use const_format::formatcp;
5	use once_cell::sync::Lazy;
6	use regex::Regex;
7
8	const attr_name : &str = r#"[a-zA-Z_:][a-zA-Z0-9:._-]*"#;
9
10	const unquoted : &str = r#"[^"'=<>`\x00-\x20]+"#;
11	const single_quoted : &str = r#"'[^']*'"#;
12	const double_quoted : &str = r#""[^"]*""#;
13
14	const attr_value : &str = formatcp!("(?:{unquoted}|{single_quoted}|{double_quoted})");
15
16	const attribute : &str = formatcp!("(?:\\s+{attr_name}(?:\\s*=\\s*{attr_value})?)");
17
18	const open_tag : &str = formatcp!("<[A-Za-z][A-Za-z0-9\\-]*{attribute}*\\s*/?>");
19
20	const close_tag : &str = r#"</[A-Za-z][A-Za-z0-9\-]*\s*>"#;
21	const comment : &str = r#"<!---->|<!--(?:-?[^>-])(?:-?[^-])*-->"#;
22	const processing : &str = r#"<[?][\s\S]*?[?]>"#;
23	const declaration : &str = r#"<![A-Z]+\s+[^>]*>"#;
24	const cdata : &str = r#"<!\[CDATA\[[\s\S]*?\]\]>"#;
25
26	#[allow(clippy::double_parens)]
27	pub static HTML_TAG_RE : Lazy<Regex> = Lazy::new(|| {
28	Regex::new(
29	formatcp!("^(?:{open_tag}|{close_tag}|{comment}|{processing}|{declaration}|{cdata})")
30	).unwrap()
31	});
32
33	#[allow(clippy::double_parens)]
34	pub static HTML_OPEN_CLOSE_TAG_RE : Lazy<Regex> = Lazy::new(|| {
35	Regex::new(
36	formatcp!("^(?:{open_tag}|{close_tag})")
37	).unwrap()
38	});
39
40	pub static HTML_LINK_OPEN : Lazy<Regex> = Lazy::new(|| {
41	Regex::new(r#"^<a[>\s]"#).unwrap()
42	});
43
44	pub static HTML_LINK_CLOSE : Lazy<Regex> = Lazy::new(|| {
45	Regex::new(r#"^</a\s*>"#).unwrap()
46	});