diffler_core/syntax/
scope.rs1use tree_sitter::{QueryCursor, StreamingIterator};
6
7use crate::syntax::registry::LanguageRegistry;
8use crate::syntax::{MAX_PARSE_BYTES, parse};
9
10#[derive(Debug, Clone, Default)]
13pub struct ScopeIndex {
14 defs: Vec<Def>,
15}
16
17#[derive(Debug, Clone)]
18struct Def {
19 start_row: usize,
20 end_row: usize,
21 name: String,
22}
23
24impl ScopeIndex {
25 pub fn is_empty(&self) -> bool {
26 self.defs.is_empty()
27 }
28
29 pub fn def_starts(&self) -> Vec<usize> {
32 let mut rows: Vec<usize> = self.defs.iter().map(|d| d.start_row).collect();
33 rows.sort_unstable();
34 rows.dedup();
35 rows
36 }
37
38 pub fn crumbs(&self, line: usize) -> Vec<String> {
41 let mut hits: Vec<&Def> = self
42 .defs
43 .iter()
44 .filter(|d| d.start_row <= line && line <= d.end_row)
45 .collect();
46 hits.sort_by(|a, b| {
47 a.start_row
48 .cmp(&b.start_row)
49 .then(b.end_row.cmp(&a.end_row))
50 });
51 hits.into_iter().map(|d| d.name.clone()).collect()
52 }
53}
54
55impl LanguageRegistry {
56 pub fn scope_index(&self, path: &str, content: &str) -> ScopeIndex {
60 if content.len() > MAX_PARSE_BYTES {
61 return ScopeIndex::default();
62 }
63 let Some(entry) = self.for_path(path) else {
64 return ScopeIndex::default();
65 };
66 let Some(query) = entry.tags.as_ref() else {
67 return ScopeIndex::default();
68 };
69 let Some(tree) = parse(entry, content) else {
70 return ScopeIndex::default();
71 };
72
73 let names = query.capture_names();
74 let bytes = content.as_bytes();
75 let mut cursor = QueryCursor::new();
76 let mut matches = cursor.matches(query, tree.root_node(), bytes);
77 let mut defs = Vec::new();
78 while let Some(m) = matches.next() {
79 let mut span: Option<(usize, usize)> = None;
80 let mut name: Option<String> = None;
81 for cap in m.captures {
82 let cname = names.get(cap.index as usize).copied().unwrap_or("");
83 if cname.starts_with("definition.") {
84 span = Some((cap.node.start_position().row, cap.node.end_position().row));
85 } else if cname == "name" {
86 name = cap.node.utf8_text(bytes).ok().map(str::to_owned);
87 }
88 }
89 if let (Some((start_row, end_row)), Some(name)) = (span, name) {
90 defs.push(Def {
91 start_row,
92 end_row,
93 name,
94 });
95 }
96 }
97 ScopeIndex { defs }
98 }
99}
100
101#[cfg(test)]
102mod tests {
103 use super::*;
104
105 #[test]
106 fn nested_python_scope_reads_class_then_method() {
107 let reg = LanguageRegistry::build();
108 let src = "class A:\n def method(self):\n x = 1\n return x\n";
109 let crumbs = reg.scope_index("a.py", src).crumbs(2);
110 let names: Vec<&str> = crumbs.iter().map(String::as_str).collect();
111 assert_eq!(names, ["A", "method"]);
112 }
113
114 #[test]
115 fn rust_function_scope() {
116 let reg = LanguageRegistry::build();
117 let src = "fn outer() {\n let y = 2;\n}\n";
118 let crumbs = reg.scope_index("a.rs", src).crumbs(1);
119 let names: Vec<&str> = crumbs.iter().map(String::as_str).collect();
120 assert_eq!(names, ["outer"]);
121 }
122
123 #[test]
124 fn scope_index_empty_for_unsupported_language() {
125 let reg = LanguageRegistry::build();
126 assert!(reg.scope_index("a.zzz-unknown", "whatever\n").is_empty());
127 }
128
129 #[test]
130 fn line_outside_any_definition_has_no_crumbs() {
131 let reg = LanguageRegistry::build();
132 let src = "import os\n\ndef f():\n pass\n";
133 assert!(reg.scope_index("a.py", src).crumbs(0).is_empty());
134 }
135}