ast_grep_language/
html.rs1use super::pre_process_pattern;
2use ast_grep_core::matcher::{Pattern, PatternBuilder, PatternError};
3use ast_grep_core::tree_sitter::{LanguageExt, StrDoc, TSLanguage, TSRange};
4use ast_grep_core::Language;
5use ast_grep_core::{matcher::KindMatcher, Doc, Node};
6use std::collections::HashMap;
7
8#[derive(Clone, Copy, Debug)]
11pub struct Html;
12impl Language for Html {
13 fn expando_char(&self) -> char {
14 'z'
15 }
16 fn pre_process_pattern<'q>(&self, query: &'q str) -> std::borrow::Cow<'q, str> {
17 pre_process_pattern(self.expando_char(), query)
18 }
19 fn kind_to_id(&self, kind: &str) -> u16 {
20 crate::parsers::language_html().id_for_node_kind(kind, true)
21 }
22 fn field_to_id(&self, field: &str) -> Option<u16> {
23 crate::parsers::language_html()
24 .field_id_for_name(field)
25 .map(|f| f.get())
26 }
27 fn build_pattern(&self, builder: &PatternBuilder) -> Result<Pattern, PatternError> {
28 builder.build(|src| StrDoc::try_new(src, *self))
29 }
30}
31impl LanguageExt for Html {
32 fn get_ts_language(&self) -> TSLanguage {
33 crate::parsers::language_html()
34 }
35 fn injectable_languages(&self) -> Option<&'static [&'static str]> {
36 Some(&["css", "js", "ts", "tsx", "scss", "less", "stylus", "coffee"])
37 }
38 fn extract_injections<L: LanguageExt>(
39 &self,
40 root: Node<StrDoc<L>>,
41 ) -> HashMap<String, Vec<TSRange>> {
42 let lang = root.lang();
43 let mut map = HashMap::new();
44 let matcher = KindMatcher::new("script_element", lang.clone());
45 for script in root.find_all(matcher) {
46 let injected = find_lang(&script).unwrap_or_else(|| "js".into());
47 let content = script.children().find(|c| c.kind() == "raw_text");
48 if let Some(content) = content {
49 map
50 .entry(injected)
51 .or_insert_with(Vec::new)
52 .push(node_to_range(&content));
53 };
54 }
55 let matcher = KindMatcher::new("style_element", lang.clone());
56 for style in root.find_all(matcher) {
57 let injected = find_lang(&style).unwrap_or_else(|| "css".into());
58 let content = style.children().find(|c| c.kind() == "raw_text");
59 if let Some(content) = content {
60 map
61 .entry(injected)
62 .or_insert_with(Vec::new)
63 .push(node_to_range(&content));
64 };
65 }
66 map
67 }
68}
69
70fn find_lang<D: Doc>(node: &Node<D>) -> Option<String> {
71 let html = node.lang();
72 let attr_matcher = KindMatcher::new("attribute", html.clone());
73 let name_matcher = KindMatcher::new("attribute_name", html.clone());
74 let val_matcher = KindMatcher::new("attribute_value", html.clone());
75 node.find_all(attr_matcher).find_map(|attr| {
76 let name = attr.find(&name_matcher)?;
77 if name.text() != "lang" {
78 return None;
79 }
80 let val = attr.find(&val_matcher)?;
81 Some(val.text().to_string())
82 })
83}
84
85fn node_to_range<D: Doc>(node: &Node<D>) -> TSRange {
86 let r = node.range();
87 let start = node.start_pos();
88 let sp = start.byte_point();
89 let sp = tree_sitter::Point::new(sp.0, sp.1);
90 let end = node.end_pos();
91 let ep = end.byte_point();
92 let ep = tree_sitter::Point::new(ep.0, ep.1);
93 TSRange {
94 start_byte: r.start,
95 end_byte: r.end,
96 start_point: sp,
97 end_point: ep,
98 }
99}
100
101#[cfg(test)]
102mod test {
103 use super::*;
104
105 fn test_match(query: &str, source: &str) {
106 use crate::test::test_match_lang;
107 test_match_lang(query, source, Html);
108 }
109
110 fn test_non_match(query: &str, source: &str) {
111 use crate::test::test_non_match_lang;
112 test_non_match_lang(query, source, Html);
113 }
114
115 #[test]
116 fn test_html_match() {
117 test_match("<input>", "<input>");
118 test_match("<$TAG>", "<input>");
119 test_match("<$TAG class='foo'>$$$</$TAG>", "<div class='foo'></div>");
120 test_match("<div>$$$</div>", "<div>123</div>");
121 test_non_match("<$TAG class='foo'>$$$</$TAG>", "<div></div>");
122 test_non_match("<div>$$$</div>", "<div class='foo'>123</div>");
123 }
124
125 fn test_replace(src: &str, pattern: &str, replacer: &str) -> String {
126 use crate::test::test_replace_lang;
127 test_replace_lang(src, pattern, replacer, Html)
128 }
129
130 #[test]
131 fn test_html_replace() {
132 let ret = test_replace(
133 r#"<div class='foo'>bar</div>"#,
134 r#"<$TAG class='foo'>$$$B</$TAG>"#,
135 r#"<$TAG class='$$$B'>foo</$TAG>"#,
136 );
137 assert_eq!(ret, r#"<div class='bar'>foo</div>"#);
138 }
139
140 fn extract(src: &str) -> HashMap<String, Vec<TSRange>> {
141 let root = Html.ast_grep(src);
142 Html.extract_injections(root.root())
143 }
144
145 #[test]
146 fn test_html_extraction() {
147 let map = extract("<script>a</script><style>.a{}</style>");
148 assert!(map.contains_key("css"));
149 assert!(map.contains_key("js"));
150 assert_eq!(map["css"].len(), 1);
151 assert_eq!(map["js"].len(), 1);
152 }
153
154 #[test]
155 fn test_explicit_lang() {
156 let map = extract("<script lang='ts'>a</script><script lang=ts>.a{}</script><style lang=scss></style><style lang=\"scss\"></style>");
157 assert!(map.contains_key("ts"));
158 assert_eq!(map["ts"].len(), 2);
159 assert_eq!(map["scss"].len(), 2);
160 }
161}