Skip to main content

sdivi_lang_python/
lib.rs

1//! Python language adapter for sdivi-rust.
2//!
3//! Implements [`sdivi_parsing::adapter::LanguageAdapter`] for `.py` source files
4//! using the `tree-sitter-python` grammar.
5//!
6//! # Thread safety
7//!
8//! `tree_sitter::Parser` is not `Send`. Parsers are stored in `thread_local!`
9//! storage so that `PythonAdapter` itself can be `Send + Sync` and participate
10//! in rayon parallel parsing.
11
12mod extract;
13
14use std::cell::RefCell;
15use std::path::Path;
16
17use sdivi_parsing::adapter::LanguageAdapter;
18use sdivi_parsing::feature_record::FeatureRecord;
19
20use extract::{collect_hints, extract_exports, extract_imports, extract_signatures};
21
22thread_local! {
23    static PARSER: RefCell<tree_sitter::Parser> = RefCell::new({
24        let mut p = tree_sitter::Parser::new();
25        p.set_language(&tree_sitter_python::language())
26            .expect("tree-sitter-python grammar failed to load");
27        p
28    });
29}
30
31/// Language adapter for Python source files.
32///
33/// Parses `.py` files with the `tree-sitter-python` grammar and extracts:
34/// - `imports` from `import` and `from … import` statements
35/// - `exports` from top-level non-underscore definitions
36/// - `signatures` from function and class definitions
37/// - `pattern_hints` for the patterns stage
38///
39/// # Examples
40///
41/// ```rust
42/// use sdivi_lang_python::PythonAdapter;
43/// use sdivi_parsing::adapter::LanguageAdapter;
44///
45/// let adapter = PythonAdapter;
46/// assert_eq!(adapter.language_name(), "python");
47/// assert!(adapter.file_extensions().contains(&".py"));
48/// ```
49pub struct PythonAdapter;
50
51impl LanguageAdapter for PythonAdapter {
52    fn language_name(&self) -> &'static str {
53        "python"
54    }
55
56    fn file_extensions(&self) -> &[&'static str] {
57        &[".py"]
58    }
59
60    /// Parses `content` and returns a [`FeatureRecord`].
61    ///
62    /// The tree-sitter CST is created, traversed, and **dropped** before this
63    /// method returns. No tree-sitter type escapes into the returned record.
64    fn parse_file(&self, path: &Path, content: String) -> FeatureRecord {
65        let source = content.as_bytes();
66
67        let (imports, exports, signatures, pattern_hints) = PARSER.with(|cell| {
68            let mut parser = cell.borrow_mut();
69            let tree = parser
70                .parse(source, None)
71                .expect("tree-sitter-python failed to parse");
72
73            let root = tree.root_node();
74            let imports = extract_imports(root, source);
75            let exports = extract_exports(root, source);
76            let signatures = extract_signatures(root, source);
77            let hints = collect_hints(root, source);
78            // tree is dropped here — CST does not escape this closure.
79            (imports, exports, signatures, hints)
80        });
81
82        FeatureRecord {
83            path: path.to_path_buf(),
84            language: "python".to_string(),
85            imports,
86            exports,
87            signatures,
88            pattern_hints,
89        }
90    }
91}