| 1 | import { parse } from "@babel/parser"; |
| 2 | import _traverse from "@babel/traverse"; |
| 3 | const traverse = _traverse.default || _traverse; |
| 4 | import fs from "fs"; |
| 5 | import path from "path"; |
| 6 | import fg from "fast-glob"; |
| 7 | const { globSync } = fg; |
| 8 | |
| 9 | const FIXTURE_DIR = process.argv[2]; // source dir with JS/TS files |
| 10 | const OUTPUT_DIR = process.argv[3]; // output dir for JSON files |
| 11 | |
| 12 | if (!FIXTURE_DIR || !OUTPUT_DIR) { |
| 13 | console.error( |
| 14 | "Usage: node babel-ast-to-json.mjs <fixtures-dir> <output-dir>" |
| 15 | ); |
| 16 | process.exit(1); |
| 17 | } |
| 18 | |
| 19 | // Find all fixture source files |
| 20 | const fixtures = globSync("**/*.{js,ts,tsx,jsx}", { cwd: FIXTURE_DIR }); |
| 21 | |
| 22 | function getScopeKind(babelScope) { |
| 23 | const blockType = babelScope.block.type; |
| 24 | switch (blockType) { |
| 25 | case "Program": |
| 26 | return "program"; |
| 27 | case "FunctionDeclaration": |
| 28 | case "FunctionExpression": |
| 29 | case "ArrowFunctionExpression": |
| 30 | case "ObjectMethod": |
| 31 | case "ClassMethod": |
| 32 | case "ClassPrivateMethod": |
| 33 | return "function"; |
| 34 | case "BlockStatement": |
| 35 | return "block"; |
| 36 | case "ForStatement": |
| 37 | case "ForInStatement": |
| 38 | case "ForOfStatement": |
| 39 | return "for"; |
| 40 | case "ClassDeclaration": |
| 41 | case "ClassExpression": |
| 42 | return "class"; |
| 43 | case "SwitchStatement": |
| 44 | return "switch"; |
| 45 | case "CatchClause": |
| 46 | return "catch"; |
| 47 | default: |
| 48 | return "block"; |
| 49 | } |
| 50 | } |
| 51 | |
| 52 | function getBindingKind(babelKind) { |
| 53 | switch (babelKind) { |
| 54 | case "var": |
| 55 | return "var"; |
| 56 | case "let": |
| 57 | return "let"; |
| 58 | case "const": |
| 59 | return "const"; |
| 60 | case "param": |
| 61 | return "param"; |
| 62 | case "module": |
| 63 | return "module"; |
| 64 | case "hoisted": |
| 65 | return "hoisted"; |
| 66 | case "local": |
| 67 | return "local"; |
| 68 | default: |
| 69 | return "unknown"; |
| 70 | } |
| 71 | } |
| 72 | |
| 73 | function getImportData(binding) { |
| 74 | if (binding.path.isImportSpecifier()) { |
| 75 | const imported = binding.path.node.imported; |
| 76 | return { |
| 77 | source: binding.path.parent.source.value, |
| 78 | kind: "named", |
| 79 | imported: imported.type === "StringLiteral" ? imported.value : imported.name, |
| 80 | }; |
| 81 | } else if (binding.path.isImportDefaultSpecifier()) { |
| 82 | return { |
| 83 | source: binding.path.parent.source.value, |
| 84 | kind: "default", |
| 85 | }; |
| 86 | } else if (binding.path.isImportNamespaceSpecifier()) { |
| 87 | return { |
| 88 | source: binding.path.parent.source.value, |
| 89 | kind: "namespace", |
| 90 | }; |
| 91 | } |
| 92 | return null; |
| 93 | } |
| 94 | |
| 95 | function collectScopeInfo(ast) { |
| 96 | const scopeMap = new Map(); // Babel scope -> ScopeId |
| 97 | const bindingMap = new Map(); // Babel binding -> BindingId |
| 98 | const scopes = []; |
| 99 | const bindings = []; |
| 100 | const nodeToScope = {}; |
| 101 | const nodeToScopeEnd = {}; |
| 102 | const referenceToBinding = {}; |
| 103 | const refNodeIdToBinding = {}; |
| 104 | const nodeIdToScope = {}; |
| 105 | let nextScopeId = 0; |
| 106 | let nextBindingId = 0; |
| 107 | let nextNodeId = 1; |
| 108 | |
| 109 | function getOrAssignNodeId(node) { |
| 110 | if (node._nodeId == null) { |
| 111 | node._nodeId = nextNodeId++; |
| 112 | } |
| 113 | return node._nodeId; |
| 114 | } |
| 115 | |
| 116 | function mapRef(start, bindingId, node) { |
| 117 | referenceToBinding[String(start)] = bindingId; |
| 118 | const nodeId = getOrAssignNodeId(node); |
| 119 | refNodeIdToBinding[String(nodeId)] = bindingId; |
| 120 | } |
| 121 | |
| 122 | function ensureScope(babelScope) { |
| 123 | if (scopeMap.has(babelScope)) return scopeMap.get(babelScope); |
| 124 | |
| 125 | // Ensure parent is registered first (preorder: parent gets lower ID) |
| 126 | if (babelScope.parent) { |
| 127 | ensureScope(babelScope.parent); |
| 128 | } |
| 129 | |
| 130 | const id = nextScopeId++; |
| 131 | scopeMap.set(babelScope, id); |
| 132 | |
| 133 | const parentId = babelScope.parent ? scopeMap.get(babelScope.parent) : null; |
| 134 | const kind = getScopeKind(babelScope); |
| 135 | const bindingsMap = {}; |
| 136 | |
| 137 | // Register all bindings in this scope |
| 138 | for (const [name, binding] of Object.entries(babelScope.bindings)) { |
| 139 | if (!bindingMap.has(binding)) { |
| 140 | const bid = nextBindingId++; |
| 141 | bindingMap.set(binding, bid); |
| 142 | const declarationNodeId = getOrAssignNodeId(binding.identifier); |
| 143 | const bindingData = { |
| 144 | id: bid, |
| 145 | name, |
| 146 | kind: getBindingKind(binding.kind), |
| 147 | scope: id, |
| 148 | declarationType: binding.path.node.type, |
| 149 | declarationStart: binding.identifier.start, |
| 150 | declarationNodeId, |
| 151 | }; |
| 152 | |
| 153 | // Import bindings |
| 154 | if (binding.kind === "module") { |
| 155 | bindingData.import = getImportData(binding); |
| 156 | } |
| 157 | |
| 158 | bindings.push(bindingData); |
| 159 | } |
| 160 | bindingsMap[name] = bindingMap.get(binding); |
| 161 | } |
| 162 | |
| 163 | scopes.push({ |
| 164 | id, |
| 165 | parent: parentId, |
| 166 | kind, |
| 167 | bindings: bindingsMap, |
| 168 | }); |
| 169 | |
| 170 | // Record node_to_scope and node_to_scope_end |
| 171 | const blockNode = babelScope.block; |
| 172 | if (blockNode.start != null) { |
| 173 | nodeToScope[String(blockNode.start)] = id; |
| 174 | if (blockNode.end != null) { |
| 175 | nodeToScopeEnd[String(blockNode.start)] = blockNode.end; |
| 176 | } |
| 177 | const scopeNodeId = getOrAssignNodeId(blockNode); |
| 178 | nodeIdToScope[String(scopeNodeId)] = id; |
| 179 | } |
| 180 | |
| 181 | return id; |
| 182 | } |
| 183 | |
| 184 | traverse(ast, { |
| 185 | enter(path) { |
| 186 | ensureScope(path.scope); |
| 187 | }, |
| 188 | Identifier(path) { |
| 189 | getOrAssignNodeId(path.node); |
| 190 | if (!path.isReferencedIdentifier()) return; |
| 191 | const binding = path.scope.getBinding(path.node.name); |
| 192 | if (binding && bindingMap.has(binding) && path.node.start != null) { |
| 193 | mapRef(path.node.start, bindingMap.get(binding), path.node); |
| 194 | } |
| 195 | }, |
| 196 | JSXIdentifier(path) { |
| 197 | getOrAssignNodeId(path.node); |
| 198 | }, |
| 199 | AssignmentExpression(path) { |
| 200 | const left = path.get("left"); |
| 201 | if (left.isLVal()) { |
| 202 | mapLValToBindings(left, bindingMap); |
| 203 | } |
| 204 | }, |
| 205 | UpdateExpression(path) { |
| 206 | const argument = path.get("argument"); |
| 207 | if (argument.isLVal()) { |
| 208 | mapLValToBindings(argument, bindingMap); |
| 209 | } |
| 210 | }, |
| 211 | }); |
| 212 | |
| 213 | // Map identifiers in assignment targets (LVal positions) to their bindings. |
| 214 | function mapLValToBindings(lvalPath, bindingMap) { |
| 215 | const node = lvalPath.node; |
| 216 | if (!node) return; |
| 217 | switch (node.type) { |
| 218 | case "Identifier": { |
| 219 | const binding = lvalPath.scope.getBinding(node.name); |
| 220 | if (binding && bindingMap.has(binding) && node.start != null) { |
| 221 | mapRef(node.start, bindingMap.get(binding), node); |
| 222 | } |
| 223 | break; |
| 224 | } |
| 225 | case "ArrayPattern": { |
| 226 | for (const element of lvalPath.get("elements")) { |
| 227 | if (element.node) mapLValToBindings(element, bindingMap); |
| 228 | } |
| 229 | break; |
| 230 | } |
| 231 | case "ObjectPattern": { |
| 232 | for (const property of lvalPath.get("properties")) { |
| 233 | if (property.isObjectProperty()) { |
| 234 | mapLValToBindings(property.get("value"), bindingMap); |
| 235 | } else if (property.isRestElement()) { |
| 236 | mapLValToBindings(property, bindingMap); |
| 237 | } |
| 238 | } |
| 239 | break; |
| 240 | } |
| 241 | case "AssignmentPattern": { |
| 242 | mapLValToBindings(lvalPath.get("left"), bindingMap); |
| 243 | break; |
| 244 | } |
| 245 | case "RestElement": { |
| 246 | mapLValToBindings(lvalPath.get("argument"), bindingMap); |
| 247 | break; |
| 248 | } |
| 249 | default: |
| 250 | break; |
| 251 | } |
| 252 | } |
| 253 | |
| 254 | // Record declaration identifiers in reference_to_binding and refNodeIdToBinding |
| 255 | for (const [binding, bid] of bindingMap) { |
| 256 | if (binding.identifier && binding.identifier.start != null) { |
| 257 | mapRef(binding.identifier.start, bid, binding.identifier); |
| 258 | } |
| 259 | } |
| 260 | |
| 261 | const result = { |
| 262 | scopes, |
| 263 | bindings, |
| 264 | nodeToScope, |
| 265 | referenceToBinding, |
| 266 | programScope: 0, |
| 267 | }; |
| 268 | // Only include new fields when non-empty, matching Rust skip_serializing_if |
| 269 | if (Object.keys(nodeToScopeEnd).length > 0) { |
| 270 | result.nodeToScopeEnd = nodeToScopeEnd; |
| 271 | } |
| 272 | if (Object.keys(refNodeIdToBinding).length > 0) { |
| 273 | result.refNodeIdToBinding = refNodeIdToBinding; |
| 274 | } |
| 275 | if (Object.keys(nodeIdToScope).length > 0) { |
| 276 | result.nodeIdToScope = nodeIdToScope; |
| 277 | } |
| 278 | return result; |
| 279 | } |
| 280 | |
| 281 | function renameIdentifiers(ast, scopeInfo) { |
| 282 | traverse(ast, { |
| 283 | Identifier(path) { |
| 284 | const nodeId = path.node._nodeId; |
| 285 | if (nodeId != null && String(nodeId) in scopeInfo.refNodeIdToBinding) { |
| 286 | const bindingId = scopeInfo.refNodeIdToBinding[String(nodeId)]; |
| 287 | const binding = scopeInfo.bindings[bindingId]; |
| 288 | path.node.name = `${path.node.name}_${binding.scope}_${bindingId}`; |
| 289 | } |
| 290 | }, |
| 291 | }); |
| 292 | } |
| 293 | |
| 294 | let parsed = 0; |
| 295 | let errors = 0; |
| 296 | |
| 297 | for (const fixture of fixtures) { |
| 298 | const input = fs.readFileSync(path.join(FIXTURE_DIR, fixture), "utf8"); |
| 299 | const isFlow = input.includes("@flow"); |
| 300 | |
| 301 | const plugins = isFlow ? ["flow", "jsx"] : ["typescript", "jsx"]; |
| 302 | // Default to module unless there's an indicator it should be script |
| 303 | const sourceType = "module"; |
| 304 | |
| 305 | try { |
| 306 | const ast = parse(input, { |
| 307 | sourceFilename: fixture, |
| 308 | plugins, |
| 309 | sourceType, |
| 310 | allowReturnOutsideFunction: true, |
| 311 | errorRecovery: true, |
| 312 | }); |
| 313 | |
| 314 | // Collect scope info first — this assigns _nodeId to Identifier nodes |
| 315 | const scopeInfo = collectScopeInfo(ast); |
| 316 | |
| 317 | // Serialize AST after scope collection so _nodeId fields are included |
| 318 | const outPath = path.join(OUTPUT_DIR, fixture + ".json"); |
| 319 | fs.mkdirSync(path.dirname(outPath), { recursive: true }); |
| 320 | fs.writeFileSync(outPath, JSON.stringify(ast, null, 2)); |
| 321 | |
| 322 | // Write scope info |
| 323 | const scopeOutPath = path.join(OUTPUT_DIR, fixture + ".scope.json"); |
| 324 | fs.writeFileSync(scopeOutPath, JSON.stringify(scopeInfo, null, 2)); |
| 325 | |
| 326 | // Create renamed AST for scope resolution verification |
| 327 | renameIdentifiers(ast, scopeInfo); |
| 328 | const renamedOutPath = path.join(OUTPUT_DIR, fixture + ".renamed.json"); |
| 329 | fs.writeFileSync(renamedOutPath, JSON.stringify(ast, null, 2)); |
| 330 | |
| 331 | parsed++; |
| 332 | } catch (e) { |
| 333 | // Parse errors are expected for some fixtures |
| 334 | const outPath = path.join(OUTPUT_DIR, fixture + ".parse-error"); |
| 335 | fs.mkdirSync(path.dirname(outPath), { recursive: true }); |
| 336 | fs.writeFileSync(outPath, e.message); |
| 337 | errors++; |
| 338 | } |
| 339 | } |
| 340 | |
| 341 | console.log( |
| 342 | `Parsed ${parsed} fixtures, ${errors} parse errors, ${fixtures.length} total` |
| 343 | ); |