use crate::language::{DefaultVisibility, LanguageSpec};
use crate::model::Visibility;
use std::path::{Path, PathBuf};
use std::sync::LazyLock;
fn resolve_js_import(raw: &str, source_dir: &Path, _project_root: &Path) -> Option<PathBuf> {
use crate::language::import_resolver::{JS_EXTS, resolve_js_family_import};
resolve_js_family_import(raw, source_dir, JS_EXTS, &|p| p.is_file())
}
static JS_QUERY: LazyLock<tree_sitter::Query> = LazyLock::new(|| {
crate::language::common::compile_query(
&tree_sitter_javascript::LANGUAGE.into(),
r#"
(function_declaration
"async"? @async
name: (identifier) @name
parameters: (formal_parameters) @signature
) @kind.function
(generator_function_declaration
"async"? @async
name: (identifier) @name
parameters: (formal_parameters) @signature
) @kind.function
(class_declaration
name: (identifier) @name
) @kind.class
(method_definition
"async"? @async
name: [
(property_identifier)
(identifier)
] @name
parameters: (formal_parameters) @signature
) @kind.method
(export_statement
[
(function_declaration
"async"? @async
name: (identifier) @name
parameters: (formal_parameters) @signature
) @kind.function
(class_declaration
name: (identifier) @name
) @kind.class
]
)
"#,
"JavaScript",
)
});
fn js_query() -> &'static tree_sitter::Query {
&JS_QUERY
}
const JS_IMPORT_QUERY_STR: &str = r#"
(import_statement
source: (string) @import.path)
(import_statement
(import_clause
(named_imports
(import_specifier
name: (identifier) @import.symbol
alias: (identifier)? @import.alias))))
(import_statement
(import_clause
(identifier) @import.symbol))
(import_statement
(import_clause
(namespace_import
(identifier) @import.symbol)))
(call_expression
function: (identifier) @call.name
arguments: (arguments . (string) @import.path .)
(#eq? @call.name "require"))
"#;
const JS_REFERENCE_QUERY_STR: &str = crate::language::typescript::TS_FAMILY_REFERENCE_QUERY;
static JS_IMPORT_REF_QUERY: LazyLock<tree_sitter::Query> = LazyLock::new(|| {
crate::language::common::compile_query(
&tree_sitter_javascript::LANGUAGE.into(),
&format!("{}\n{}", JS_IMPORT_QUERY_STR, JS_REFERENCE_QUERY_STR),
"JavaScript combined import+ref",
)
});
fn js_import_ref_query() -> &'static tree_sitter::Query {
&JS_IMPORT_REF_QUERY
}
pub(crate) const JS_SPEC: LanguageSpec = LanguageSpec {
extensions: &["js", "mjs", "cjs"],
grammar_fn: || tree_sitter_javascript::LANGUAGE.into(),
query_fn: js_query,
import_path_resolver: resolve_js_import,
import_ref_query_fn: js_import_ref_query,
class_like_parents: &["class_declaration", "class"],
ancestor_visibility_rules: &[("export_statement", Visibility::Public)],
visibility_from_name: None,
import_statement_kinds: &["import_statement"],
default_visibility: DefaultVisibility::PrivateByDefault,
doc_comment_config: Some(crate::language::C_LIKE_DOC_COMMENT),
};
#[cfg(feature = "dataflow")]
pub(crate) const JS_DATAFLOW_QUERY_STR: &str = r#"
; Variable declarator: name position in a `let/const/var` binding.
(variable_declarator
name: (identifier) @def.var)
; Function parameters (JS grammar: identifier directly in formal_parameters,
; possibly wrapped by assignment_pattern for default values).
(formal_parameters
(identifier) @def.param)
(formal_parameters
(assignment_pattern
left: (identifier) @def.param))
; Identifier references in expression position.
(call_expression
function: (identifier) @use.var)
(call_expression
arguments: (arguments
(identifier) @use.var))
(binary_expression
left: (identifier) @use.var)
(binary_expression
right: (identifier) @use.var)
(member_expression
object: (identifier) @use.var)
(return_statement
(identifier) @use.var)
(assignment_expression
right: (identifier) @use.var)
"#;
#[cfg(feature = "dataflow")]
pub(crate) const TS_FAMILY_DATAFLOW_QUERY: &str = r#"
; Variable declarator: name position in a `let/const/var` binding.
(variable_declarator
name: (identifier) @def.var)
; Function parameters (TS grammar: required_parameter / optional_parameter).
(required_parameter
(identifier) @def.param)
(optional_parameter
(identifier) @def.param)
; Identifier references in expression position.
(call_expression
function: (identifier) @use.var)
(call_expression
arguments: (arguments
(identifier) @use.var))
(binary_expression
left: (identifier) @use.var)
(binary_expression
right: (identifier) @use.var)
(member_expression
object: (identifier) @use.var)
(return_statement
(identifier) @use.var)
(assignment_expression
right: (identifier) @use.var)
"#;
#[cfg(feature = "dataflow")]
pub(crate) fn extract_js_family_dataflow_with_query(
tree: &tree_sitter::Tree,
source: &[u8],
query: &tree_sitter::Query,
function_kinds: &[&str],
id_gen: &crate::model::IdGenerator<crate::model::DataNodeId>,
) -> (Vec<crate::model::DataNode>, Vec<crate::model::FlowEdge>) {
crate::language::common::extract_def_use_dataflow(tree, source, query, function_kinds, id_gen)
}
#[cfg(feature = "dataflow")]
static JS_DATAFLOW_QUERY: LazyLock<tree_sitter::Query> = LazyLock::new(|| {
crate::language::common::compile_query(
&tree_sitter_javascript::LANGUAGE.into(),
JS_DATAFLOW_QUERY_STR,
"JavaScript dataflow",
)
});
#[cfg(feature = "dataflow")]
pub(crate) const JS_FUNCTION_KINDS: &[&str] = crate::language::common::JS_FAMILY_FUNCTION_KINDS;
#[cfg(feature = "dataflow")]
pub fn extract_javascript_dataflow(
tree: &tree_sitter::Tree,
source: &[u8],
id_gen: &crate::model::IdGenerator<crate::model::DataNodeId>,
) -> (Vec<crate::model::DataNode>, Vec<crate::model::FlowEdge>) {
extract_js_family_dataflow_with_query(
tree,
source,
&JS_DATAFLOW_QUERY,
JS_FUNCTION_KINDS,
id_gen,
)
}
#[cfg(test)]
mod tests {
use crate::language::{LangId, extract_symbols_for, grammar_for};
use crate::model::{SymbolKind, Visibility};
fn parse(source: &[u8]) -> tree_sitter::Tree {
let mut parser = tree_sitter::Parser::new();
parser
.set_language(&grammar_for(LangId::JavaScript))
.unwrap();
parser.parse(source, None).unwrap()
}
#[test]
fn extract_function_declaration() {
let src = b"function hello() {}";
let tree = parse(src);
let symbols = extract_symbols_for(LangId::JavaScript, &tree, src);
assert_eq!(symbols.len(), 1);
assert_eq!(symbols[0].name, "hello");
assert!(matches!(symbols[0].kind, SymbolKind::Function));
}
#[test]
fn extract_async_function() {
let src = b"async function fetch() {}";
let tree = parse(src);
let symbols = extract_symbols_for(LangId::JavaScript, &tree, src);
assert_eq!(symbols.len(), 1);
assert!(symbols[0].is_async);
}
#[test]
fn extract_class_and_methods() {
let src = b"class Foo {\n constructor() {}\n bar() {}\n}";
let tree = parse(src);
let symbols = extract_symbols_for(LangId::JavaScript, &tree, src);
let class = symbols.iter().find(|s| s.name == "Foo").unwrap();
assert!(matches!(class.kind, SymbolKind::Class));
let methods: Vec<_> = symbols
.iter()
.filter(|s| matches!(s.kind, SymbolKind::Method))
.collect();
assert_eq!(methods.len(), 2);
}
#[test]
fn extract_exported_class() {
let src = b"export class Foo { bar() {} }";
let tree = parse(src);
let symbols = extract_symbols_for(LangId::JavaScript, &tree, src);
let class = symbols.iter().find(|s| s.name == "Foo").unwrap();
assert_eq!(class.visibility, Some(Visibility::Public));
}
#[test]
fn extract_named_imports() {
use crate::language::extract_imports_and_references_for;
let src = b"import { foo, bar } from 'utils';";
let tree = parse(src);
let (imports, _) = extract_imports_and_references_for(
LangId::JavaScript,
&tree,
src,
&std::path::PathBuf::from("test.js"),
);
let named: Vec<_> = imports.iter().filter(|i| i.symbol.is_some()).collect();
assert_eq!(
named.len(),
2,
"expected 2 named import records for foo and bar"
);
for imp in &named {
assert_eq!(imp.import_specifier, "'utils'");
}
assert_eq!(named[0].symbol.as_deref(), Some("foo"));
assert_eq!(named[1].symbol.as_deref(), Some("bar"));
}
#[test]
fn extract_default_import() {
use crate::language::extract_imports_and_references_for;
let src = b"import React from 'react';";
let tree = parse(src);
let (imports, _) = extract_imports_and_references_for(
LangId::JavaScript,
&tree,
src,
&std::path::PathBuf::from("test.js"),
);
let named: Vec<_> = imports.iter().filter(|i| i.symbol.is_some()).collect();
assert_eq!(named.len(), 1);
assert_eq!(named[0].import_specifier, "'react'");
assert_eq!(named[0].symbol.as_deref(), Some("React"));
}
#[test]
fn extract_side_effect_import() {
use crate::language::extract_imports_and_references_for;
let src = b"import 'styles.css';";
let tree = parse(src);
let (imports, _) = extract_imports_and_references_for(
LangId::JavaScript,
&tree,
src,
&std::path::PathBuf::from("test.js"),
);
assert_eq!(imports.len(), 1);
assert_eq!(imports[0].import_specifier, "'styles.css'");
assert!(imports[0].symbol.is_none());
}
#[test]
fn js_docstring_extraction() {
let src = b"/** JSDoc comment. */\nfunction documented() {}";
let tree = parse(src);
let symbols = extract_symbols_for(LangId::JavaScript, &tree, src);
let func = symbols.iter().find(|s| s.name == "documented").unwrap();
assert!(func.docstring.is_some(), "documented should have docstring");
assert!(func.docstring.as_ref().unwrap().contains("JSDoc comment"));
}
#[test]
fn js_insta_snapshot() {
let src = std::fs::read_to_string(
std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR"))
.join("tests/fixtures/javascript/functions.js"),
)
.unwrap();
let tree = parse(src.as_bytes());
let symbols = extract_symbols_for(LangId::JavaScript, &tree, src.as_bytes());
insta::assert_json_snapshot!(symbols);
}
#[cfg(feature = "dataflow")]
mod dataflow_tests {
use super::*;
use crate::language::javascript::extract_javascript_dataflow;
use crate::model::{DataScope, FlowKind};
fn extract(source: &[u8]) -> (Vec<crate::model::DataNode>, Vec<crate::model::FlowEdge>) {
let id_gen = crate::model::IdGenerator::new();
extract_javascript_dataflow(&parse(source), source, &id_gen)
}
#[test]
fn const_declaration_captured_as_local() {
let src = b"function f() { const x = 42; return x; }";
let (nodes, _) = extract(src);
let x = nodes
.iter()
.find(|n| n.name.as_deref() == Some("x") && n.scope == DataScope::Local)
.expect("const x should be captured as Local");
assert_eq!(x.scope, DataScope::Local);
}
#[test]
fn function_parameter_captured_as_parameter() {
let src = b"function add(a, b) { return a + b; }";
let (nodes, edges) = extract(src);
let params: Vec<_> = nodes
.iter()
.filter(|n| n.scope == DataScope::Parameter)
.collect();
assert_eq!(params.len(), 2, "expected 2 parameters");
let names: Vec<_> = params.iter().map(|n| n.name.as_deref()).collect();
assert!(names.contains(&Some("a")));
assert!(names.contains(&Some("b")));
assert!(
!edges.is_empty(),
"parameter usages should produce def-use edges"
);
for edge in &edges {
assert_eq!(edge.kind, FlowKind::DefUse);
assert!(
(edge.confidence - 0.9).abs() < f32::EPSILON,
"confidence should be 0.9, got {}",
edge.confidence
);
}
}
#[test]
fn def_use_edge_anchored_in_graph() {
let src = b"function f() { let x = 1; let y = x; }";
let (nodes, edges) = extract(src);
let ids: std::collections::HashSet<_> = nodes.iter().map(|n| n.id).collect();
for edge in &edges {
assert!(
ids.contains(&edge.source),
"edge source {:?} not in nodes",
edge.source
);
assert!(
ids.contains(&edge.target),
"edge target {:?} not in nodes (dangling edge)",
edge.target
);
}
}
#[test]
fn no_cross_function_def_use_leak() {
let src = b"function outer() { let x = 1; function inner() { let x = 2; return x; } return x; }";
let (nodes, edges) = extract(src);
let defs_for_x: Vec<_> = nodes
.iter()
.filter(|n| n.name.as_deref() == Some("x") && n.scope == DataScope::Local)
.collect();
assert_eq!(defs_for_x.len(), 4, "2 def + 2 use nodes for `x` expected");
let defs: Vec<_> = nodes
.iter()
.filter(|n| {
n.name.as_deref() == Some("x")
&& n.scope == DataScope::Local
&& !edges.iter().any(|e| e.target == n.id)
})
.collect();
assert_eq!(defs.len(), 2, "two distinct `x` defs expected");
let use_nodes: Vec<_> = nodes
.iter()
.filter(|n| {
n.name.as_deref() == Some("x")
&& n.scope == DataScope::Local
&& edges.iter().any(|e| e.target == n.id)
})
.collect();
assert_eq!(use_nodes.len(), 2);
let outer_def = defs
.iter()
.min_by_key(|n| n.source_range.byte_start)
.unwrap();
let inner_def = defs
.iter()
.max_by_key(|n| n.source_range.byte_start)
.unwrap();
let use_for_inner_x = use_nodes
.iter()
.min_by_key(|n| n.source_range.byte_start)
.unwrap();
let use_for_outer_x = use_nodes
.iter()
.max_by_key(|n| n.source_range.byte_start)
.unwrap();
let edge_for_inner = edges
.iter()
.find(|e| e.target == use_for_inner_x.id)
.expect("inner x-use must have an edge");
let edge_for_outer = edges
.iter()
.find(|e| e.target == use_for_outer_x.id)
.expect("outer x-use must have an edge");
assert_eq!(
edge_for_inner.source, inner_def.id,
"inner use must bind to inner def"
);
assert_eq!(
edge_for_outer.source, outer_def.id,
"outer use must bind to outer def"
);
assert_ne!(edge_for_inner.source, edge_for_outer.source);
}
#[test]
fn arrow_function_creates_new_scope() {
let src = b"function outer() { let x = 1; const f = () => { let x = 2; return x; }; return x; }";
let (_nodes, edges) = extract(src);
assert!(!edges.is_empty());
}
#[test]
fn no_edges_for_undefined_names() {
let src = b"function f() { return undefinedSymbol; }";
let (_nodes, edges) = extract(src);
assert!(edges.is_empty(), "unresolved identifiers produce no edges");
}
#[test]
fn empty_function_yields_no_nodes() {
let src = b"function f() {}";
let (nodes, edges) = extract(src);
assert!(nodes.is_empty());
assert!(edges.is_empty());
}
#[test]
fn dataflow_against_fixture_file() {
let src = std::fs::read_to_string(
std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR"))
.join("tests/fixtures/javascript/functions.js"),
)
.unwrap();
let (nodes, edges) = extract(src.as_bytes());
assert!(!nodes.is_empty(), "fixture must yield data nodes");
assert!(!edges.is_empty(), "fixture must yield flow edges");
let ids: std::collections::HashSet<_> = nodes.iter().map(|n| n.id).collect();
for edge in &edges {
assert!(ids.contains(&edge.source));
assert!(ids.contains(&edge.target));
}
}
}
}