| 1 | //! Rust equivalent of the TypeScript `FindContextIdentifiers` pass. |
| 2 | //! |
| 3 | //! Determines which bindings need StoreContext/LoadContext semantics by |
| 4 | //! walking the AST with scope tracking to find variables that cross |
| 5 | //! function boundaries. |
| 6 | |
| 7 | use rustc_hash::{FxHashMap, FxHashSet}; |
| 8 | |
| 9 | use react_compiler_ast::expressions::*; |
| 10 | use react_compiler_ast::patterns::*; |
| 11 | use react_compiler_ast::scope::*; |
| 12 | use react_compiler_ast::statements::FunctionDeclaration; |
| 13 | use react_compiler_ast::visitor::AstWalker; |
| 14 | use react_compiler_ast::visitor::Visitor; |
| 15 | use react_compiler_diagnostics::CompilerError; |
| 16 | use react_compiler_diagnostics::CompilerErrorDetail; |
| 17 | use react_compiler_diagnostics::ErrorCategory; |
| 18 | use react_compiler_diagnostics::Position; |
| 19 | use react_compiler_diagnostics::SourceLocation; |
| 20 | use react_compiler_hir::environment::Environment; |
| 21 | |
| 22 | use crate::FunctionNode; |
| 23 | |
| 24 | #[derive(Default)] |
| 25 | struct BindingInfo { |
| 26 | reassigned: bool, |
| 27 | reassigned_by_inner_fn: bool, |
| 28 | referenced_by_inner_fn: bool, |
| 29 | } |
| 30 | |
| 31 | struct ContextIdentifierVisitor<'a> { |
| 32 | scope_info: &'a ScopeInfo, |
| 33 | env: &'a mut Environment, |
| 34 | /// Stack of inner function scopes encountered during traversal. |
| 35 | /// Empty when at the top level of the function being compiled. |
| 36 | function_stack: Vec<ScopeId>, |
| 37 | binding_info: FxHashMap<BindingId, BindingInfo>, |
| 38 | error: Option<CompilerError>, |
| 39 | } |
| 40 | |
| 41 | impl<'a> ContextIdentifierVisitor<'a> { |
| 42 | fn push_function_scope(&mut self, _start: Option<u32>, node_id: Option<u32>) { |
| 43 | let scope = self.scope_info.resolve_scope_for_node(node_id); |
| 44 | if let Some(scope) = scope { |
| 45 | self.function_stack.push(scope); |
| 46 | } |
| 47 | } |
| 48 | |
| 49 | fn pop_function_scope(&mut self, _start: Option<u32>, node_id: Option<u32>) { |
| 50 | let has_scope = self.scope_info.resolve_scope_for_node(node_id); |
| 51 | if has_scope.is_some() { |
| 52 | self.function_stack.pop(); |
| 53 | } |
| 54 | } |
| 55 | |
| 56 | fn check_captured_reference(&mut self, _start: Option<u32>, node_id: Option<u32>) { |
| 57 | let binding_id = match self.scope_info.resolve_reference_id_for_node(node_id) { |
| 58 | Some(id) => id, |
| 59 | None => return, |
| 60 | }; |
| 61 | let &fn_scope = match self.function_stack.last() { |
| 62 | Some(s) => s, |
| 63 | None => return, |
| 64 | }; |
| 65 | let binding = &self.scope_info.bindings[binding_id.0 as usize]; |
| 66 | if is_captured_by_function(self.scope_info, binding.scope, fn_scope) { |
| 67 | let info = self.binding_info.entry(binding_id).or_default(); |
| 68 | info.referenced_by_inner_fn = true; |
| 69 | } |
| 70 | } |
| 71 | |
| 72 | fn handle_reassignment_identifier(&mut self, name: &str, current_scope: ScopeId) { |
| 73 | if let Some(binding_id) = self.scope_info.get_binding(current_scope, name) { |
| 74 | let info = self.binding_info.entry(binding_id).or_default(); |
| 75 | info.reassigned = true; |
| 76 | if let Some(&fn_scope) = self.function_stack.last() { |
| 77 | let binding = &self.scope_info.bindings[binding_id.0 as usize]; |
| 78 | if is_captured_by_function(self.scope_info, binding.scope, fn_scope) { |
| 79 | info.reassigned_by_inner_fn = true; |
| 80 | } |
| 81 | } |
| 82 | } |
| 83 | } |
| 84 | } |
| 85 | |
| 86 | impl<'ast> Visitor<'ast> for ContextIdentifierVisitor<'_> { |
| 87 | fn enter_function_declaration(&mut self, node: &'ast FunctionDeclaration, _: &[ScopeId]) { |
| 88 | self.push_function_scope(node.base.start, node.base.node_id); |
| 89 | } |
| 90 | fn leave_function_declaration(&mut self, node: &'ast FunctionDeclaration, _: &[ScopeId]) { |
| 91 | self.pop_function_scope(node.base.start, node.base.node_id); |
| 92 | } |
| 93 | fn enter_function_expression(&mut self, node: &'ast FunctionExpression, _: &[ScopeId]) { |
| 94 | self.push_function_scope(node.base.start, node.base.node_id); |
| 95 | } |
| 96 | fn leave_function_expression(&mut self, node: &'ast FunctionExpression, _: &[ScopeId]) { |
| 97 | self.pop_function_scope(node.base.start, node.base.node_id); |
| 98 | } |
| 99 | fn enter_arrow_function_expression( |
| 100 | &mut self, |
| 101 | node: &'ast ArrowFunctionExpression, |
| 102 | _: &[ScopeId], |
| 103 | ) { |
| 104 | self.push_function_scope(node.base.start, node.base.node_id); |
| 105 | } |
| 106 | fn leave_arrow_function_expression( |
| 107 | &mut self, |
| 108 | node: &'ast ArrowFunctionExpression, |
| 109 | _: &[ScopeId], |
| 110 | ) { |
| 111 | self.pop_function_scope(node.base.start, node.base.node_id); |
| 112 | } |
| 113 | fn enter_object_method(&mut self, node: &'ast ObjectMethod, _: &[ScopeId]) { |
| 114 | self.push_function_scope(node.base.start, node.base.node_id); |
| 115 | } |
| 116 | fn leave_object_method(&mut self, node: &'ast ObjectMethod, _: &[ScopeId]) { |
| 117 | self.pop_function_scope(node.base.start, node.base.node_id); |
| 118 | } |
| 119 | |
| 120 | fn enter_identifier(&mut self, node: &'ast Identifier, _scope_stack: &[ScopeId]) { |
| 121 | self.check_captured_reference(node.base.start, node.base.node_id); |
| 122 | } |
| 123 | |
| 124 | fn enter_jsx_identifier( |
| 125 | &mut self, |
| 126 | node: &'ast react_compiler_ast::jsx::JSXIdentifier, |
| 127 | _scope_stack: &[ScopeId], |
| 128 | ) { |
| 129 | self.check_captured_reference(node.base.start, node.base.node_id); |
| 130 | } |
| 131 | |
| 132 | fn enter_assignment_expression( |
| 133 | &mut self, |
| 134 | node: &'ast AssignmentExpression, |
| 135 | scope_stack: &[ScopeId], |
| 136 | ) { |
| 137 | let current_scope = scope_stack |
| 138 | .last() |
| 139 | .copied() |
| 140 | .unwrap_or(self.scope_info.program_scope); |
| 141 | if self.error.is_none() { |
| 142 | if let Err(error) = walk_lval_for_reassignment(self, &node.left, current_scope) { |
| 143 | self.error = Some(error); |
| 144 | } |
| 145 | } |
| 146 | } |
| 147 | |
| 148 | fn enter_update_expression(&mut self, node: &'ast UpdateExpression, scope_stack: &[ScopeId]) { |
| 149 | if let Expression::Identifier(ident) = node.argument.as_ref() { |
| 150 | let current_scope = scope_stack |
| 151 | .last() |
| 152 | .copied() |
| 153 | .unwrap_or(self.scope_info.program_scope); |
| 154 | self.handle_reassignment_identifier(&ident.name, current_scope); |
| 155 | } |
| 156 | } |
| 157 | } |
| 158 | |
| 159 | /// Recursively walk an LVal pattern to find all reassignment target identifiers. |
| 160 | fn walk_lval_for_reassignment( |
| 161 | visitor: &mut ContextIdentifierVisitor<'_>, |
| 162 | pattern: &PatternLike, |
| 163 | current_scope: ScopeId, |
| 164 | ) -> Result<(), CompilerError> { |
| 165 | match pattern { |
| 166 | PatternLike::Identifier(ident) => { |
| 167 | visitor.handle_reassignment_identifier(&ident.name, current_scope); |
| 168 | } |
| 169 | PatternLike::ArrayPattern(pat) => { |
| 170 | for element in &pat.elements { |
| 171 | if let Some(el) = element { |
| 172 | walk_lval_for_reassignment(visitor, el, current_scope)?; |
| 173 | } |
| 174 | } |
| 175 | } |
| 176 | PatternLike::ObjectPattern(pat) => { |
| 177 | for prop in &pat.properties { |
| 178 | match prop { |
| 179 | ObjectPatternProperty::ObjectProperty(p) => { |
| 180 | walk_lval_for_reassignment(visitor, &p.value, current_scope)?; |
| 181 | } |
| 182 | ObjectPatternProperty::RestElement(p) => { |
| 183 | walk_lval_for_reassignment(visitor, &p.argument, current_scope)?; |
| 184 | } |
| 185 | } |
| 186 | } |
| 187 | } |
| 188 | PatternLike::AssignmentPattern(pat) => { |
| 189 | walk_lval_for_reassignment(visitor, &pat.left, current_scope)?; |
| 190 | } |
| 191 | PatternLike::RestElement(pat) => { |
| 192 | walk_lval_for_reassignment(visitor, &pat.argument, current_scope)?; |
| 193 | } |
| 194 | PatternLike::MemberExpression(_) => { |
| 195 | // Interior mutability - not a variable reassignment |
| 196 | } |
| 197 | PatternLike::TSAsExpression(node) => { |
| 198 | record_unsupported_lval( |
| 199 | visitor.env, |
| 200 | "TSAsExpression", |
| 201 | convert_opt_loc(&node.base.loc), |
| 202 | )?; |
| 203 | } |
| 204 | PatternLike::TSSatisfiesExpression(node) => { |
| 205 | record_unsupported_lval( |
| 206 | visitor.env, |
| 207 | "TSSatisfiesExpression", |
| 208 | convert_opt_loc(&node.base.loc), |
| 209 | )?; |
| 210 | } |
| 211 | PatternLike::TSNonNullExpression(node) => { |
| 212 | record_unsupported_lval( |
| 213 | visitor.env, |
| 214 | "TSNonNullExpression", |
| 215 | convert_opt_loc(&node.base.loc), |
| 216 | )?; |
| 217 | } |
| 218 | PatternLike::TSTypeAssertion(node) => { |
| 219 | record_unsupported_lval( |
| 220 | visitor.env, |
| 221 | "TSTypeAssertion", |
| 222 | convert_opt_loc(&node.base.loc), |
| 223 | )?; |
| 224 | } |
| 225 | PatternLike::TypeCastExpression(node) => { |
| 226 | record_unsupported_lval( |
| 227 | visitor.env, |
| 228 | "TypeCastExpression", |
| 229 | convert_opt_loc(&node.base.loc), |
| 230 | )?; |
| 231 | } |
| 232 | } |
| 233 | Ok(()) |
| 234 | } |
| 235 | |
| 236 | fn convert_loc(loc: &react_compiler_ast::common::SourceLocation) -> SourceLocation { |
| 237 | SourceLocation { |
| 238 | start: Position { |
| 239 | line: loc.start.line, |
| 240 | column: loc.start.column, |
| 241 | index: loc.start.index, |
| 242 | }, |
| 243 | end: Position { |
| 244 | line: loc.end.line, |
| 245 | column: loc.end.column, |
| 246 | index: loc.end.index, |
| 247 | }, |
| 248 | } |
| 249 | } |
| 250 | |
| 251 | fn convert_opt_loc( |
| 252 | loc: &Option<react_compiler_ast::common::SourceLocation>, |
| 253 | ) -> Option<SourceLocation> { |
| 254 | loc.as_ref().map(convert_loc) |
| 255 | } |
| 256 | |
| 257 | /// Record the TS-faithful Todo for an unsupported assignment-target wrapper |
| 258 | /// node, mirroring the TypeScript `FindContextIdentifiers` pass. TS throws |
| 259 | /// immediately (CompilerError.throwTodo in handleAssignment's default case), |
| 260 | /// aborting before BuildHIR ever runs or logs, so this must return Err rather |
| 261 | /// than record-and-continue: otherwise Rust emits HIR debug entries for a |
| 262 | /// function TS never lowered. |
| 263 | fn record_unsupported_lval( |
| 264 | env: &mut Environment, |
| 265 | type_name: &str, |
| 266 | loc: Option<SourceLocation>, |
| 267 | ) -> Result<(), CompilerError> { |
| 268 | let _ = env; |
| 269 | let mut err = CompilerError::new(); |
| 270 | err.push_error_detail(CompilerErrorDetail { |
| 271 | category: ErrorCategory::Todo, |
| 272 | reason: format!( |
| 273 | "[FindContextIdentifiers] Cannot handle Object destructuring assignment target {type_name}" |
| 274 | ), |
| 275 | description: None, |
| 276 | loc, |
| 277 | suggestions: None, |
| 278 | }); |
| 279 | Err(err) |
| 280 | } |
| 281 | |
| 282 | /// Check if a binding declared at `binding_scope` is captured by a function at `function_scope`. |
| 283 | /// Returns true if the binding is declared above the function (in the parent scope or higher). |
| 284 | fn is_captured_by_function( |
| 285 | scope_info: &ScopeInfo, |
| 286 | binding_scope: ScopeId, |
| 287 | function_scope: ScopeId, |
| 288 | ) -> bool { |
| 289 | let fn_parent = match scope_info.scopes[function_scope.0 as usize].parent { |
| 290 | Some(p) => p, |
| 291 | None => return false, |
| 292 | }; |
| 293 | if binding_scope == fn_parent { |
| 294 | return true; |
| 295 | } |
| 296 | // Walk up from fn_parent to see if binding_scope is an ancestor |
| 297 | let mut current = scope_info.scopes[fn_parent.0 as usize].parent; |
| 298 | while let Some(scope_id) = current { |
| 299 | if scope_id == binding_scope { |
| 300 | return true; |
| 301 | } |
| 302 | current = scope_info.scopes[scope_id.0 as usize].parent; |
| 303 | } |
| 304 | false |
| 305 | } |
| 306 | |
| 307 | /// Build a set of `(BindingId, position)` pairs that are declaration sites |
| 308 | /// in `reference_to_binding`, not true references. Uses node-ID comparison |
| 309 | /// when available (from `ref_node_id_to_binding` + `declaration_node_id`), |
| 310 | /// falling back to position comparison otherwise. |
| 311 | /// Build a set of (BindingId, node_id) pairs for declaration sites in |
| 312 | /// ref_node_id_to_binding. These are entries where the reference's node_id |
| 313 | /// matches the binding's declaration_node_id — i.e., the "reference" is |
| 314 | /// actually the declaration itself. |
| 315 | fn build_declaration_node_ids(scope_info: &ScopeInfo) -> FxHashSet<(BindingId, u32)> { |
| 316 | let mut result = FxHashSet::default(); |
| 317 | for (&ref_nid, &binding_id) in &scope_info.ref_node_id_to_binding { |
| 318 | let binding = &scope_info.bindings[binding_id.0 as usize]; |
| 319 | if binding.declaration_node_id == Some(ref_nid) { |
| 320 | result.insert((binding_id, ref_nid)); |
| 321 | } |
| 322 | } |
| 323 | result |
| 324 | } |
| 325 | |
| 326 | /// Find context identifiers for a function: variables that are captured across |
| 327 | /// function boundaries and need StoreContext/LoadContext semantics. |
| 328 | /// |
| 329 | /// A binding is a context identifier if: |
| 330 | /// - It is reassigned from inside a nested function (`reassignedByInnerFn`), OR |
| 331 | /// - It is reassigned AND referenced from inside a nested function |
| 332 | /// (`reassigned && referencedByInnerFn`) |
| 333 | /// |
| 334 | /// This is the Rust equivalent of the TypeScript `FindContextIdentifiers` pass. |
| 335 | pub fn find_context_identifiers( |
| 336 | func: &FunctionNode<'_>, |
| 337 | scope_info: &ScopeInfo, |
| 338 | env: &mut Environment, |
| 339 | identifier_locs: &crate::identifier_loc_index::IdentifierLocIndex, |
| 340 | ) -> Result<FxHashSet<BindingId>, CompilerError> { |
| 341 | let func_scope = scope_info |
| 342 | .resolve_scope_for_node(func.node_id()) |
| 343 | .unwrap_or(scope_info.program_scope); |
| 344 | |
| 345 | let mut visitor = ContextIdentifierVisitor { |
| 346 | scope_info, |
| 347 | env, |
| 348 | function_stack: Vec::new(), |
| 349 | binding_info: FxHashMap::default(), |
| 350 | error: None, |
| 351 | }; |
| 352 | let mut walker = AstWalker::with_initial_scope(scope_info, func_scope); |
| 353 | |
| 354 | // Walk params and body (like Babel's func.traverse()) |
| 355 | match func { |
| 356 | FunctionNode::FunctionDeclaration(d) => { |
| 357 | for param in &d.params { |
| 358 | walker.walk_pattern(&mut visitor, param); |
| 359 | } |
| 360 | walker.walk_block_statement(&mut visitor, &d.body); |
| 361 | } |
| 362 | FunctionNode::FunctionExpression(e) => { |
| 363 | for param in &e.params { |
| 364 | walker.walk_pattern(&mut visitor, param); |
| 365 | } |
| 366 | walker.walk_block_statement(&mut visitor, &e.body); |
| 367 | } |
| 368 | FunctionNode::ArrowFunctionExpression(a) => { |
| 369 | for param in &a.params { |
| 370 | walker.walk_pattern(&mut visitor, param); |
| 371 | } |
| 372 | match a.body.as_ref() { |
| 373 | ArrowFunctionBody::BlockStatement(block) => { |
| 374 | walker.walk_block_statement(&mut visitor, block); |
| 375 | } |
| 376 | ArrowFunctionBody::Expression(expr) => { |
| 377 | walker.walk_expression(&mut visitor, expr); |
| 378 | } |
| 379 | } |
| 380 | } |
| 381 | } |
| 382 | |
| 383 | if let Some(error) = visitor.error { |
| 384 | return Err(error); |
| 385 | } |
| 386 | |
| 387 | // Supplement the walker-based analysis with referenceToBinding data. |
| 388 | // The AST walker doesn't visit identifiers inside type annotations, |
| 389 | // but Babel's traverse (used by TS findContextIdentifiers) does. |
| 390 | // After scope extraction includes type annotation references, |
| 391 | // we check if any reassigned binding has references in nested function scopes |
| 392 | // via referenceToBinding — matching the TS behavior. |
| 393 | // |
| 394 | // We must skip declaration sites (e.g., the `x` in `function x() {}`), |
| 395 | // which are included in reference_to_binding but are not true references. |
| 396 | // Prefer node-ID comparison (immune to position-0 collisions from synthetic |
| 397 | // nodes), falling back to position when node-IDs are unavailable. |
| 398 | let declaration_node_ids = build_declaration_node_ids(scope_info); |
| 399 | for (&ref_nid, &binding_id) in &scope_info.ref_node_id_to_binding { |
| 400 | let info = match visitor.binding_info.get(&binding_id) { |
| 401 | Some(info) if info.reassigned && !info.referenced_by_inner_fn => info, |
| 402 | _ => continue, |
| 403 | }; |
| 404 | let _ = info; |
| 405 | if declaration_node_ids.contains(&(binding_id, ref_nid)) { |
| 406 | continue; |
| 407 | } |
| 408 | // Get the reference's position from identifier_locs for range checks |
| 409 | let ref_pos = match identifier_locs.get(&ref_nid) { |
| 410 | Some(entry) => entry.start, |
| 411 | None => continue, |
| 412 | }; |
| 413 | let binding = &scope_info.bindings[binding_id.0 as usize]; |
| 414 | // Check if ref_pos is inside a nested function scope |
| 415 | for (&scope_start, &scope_id) in &scope_info.node_to_scope { |
| 416 | if scope_start <= ref_pos { |
| 417 | if let Some(&scope_end) = scope_info.node_to_scope_end.get(&scope_start) { |
| 418 | if ref_pos < scope_end |
| 419 | && matches!( |
| 420 | scope_info.scopes[scope_id.0 as usize].kind, |
| 421 | ScopeKind::Function |
| 422 | ) |
| 423 | && is_captured_by_function(scope_info, binding.scope, scope_id) |
| 424 | { |
| 425 | visitor |
| 426 | .binding_info |
| 427 | .get_mut(&binding_id) |
| 428 | .unwrap() |
| 429 | .referenced_by_inner_fn = true; |
| 430 | break; |
| 431 | } |
| 432 | } |
| 433 | } |
| 434 | } |
| 435 | } |
| 436 | |
| 437 | // Collect results |
| 438 | Ok(visitor |
| 439 | .binding_info |
| 440 | .into_iter() |
| 441 | .filter(|(_, info)| { |
| 442 | info.reassigned_by_inner_fn || (info.reassigned && info.referenced_by_inner_fn) |
| 443 | }) |
| 444 | .map(|(id, _)| id) |
| 445 | .collect()) |
| 446 | } |