yuku-parser 0.5.19 → 0.5.21

package/README.md

@@ -25,10 +25,7 @@ For JavaScript and JSX, the AST is fully conformant with the [ESTree](https://gi
 
 For TypeScript, the AST conforms to the [TypeScript-ESTree](https://www.npmjs.com/package/@typescript-eslint/typescript-estree) format used by `@typescript-eslint`.
 
-Yuku also matches [Oxc](https://oxc.rs) for both JS and TS, with two intentional deviations from the base specs:
-
-- **Comments are attached to the AST nodes they belong to** rather than exposed as a separate offset-indexed array. See [Comments](#comments) for why.
-- **`Identifier` carries a `kind` field** (`"reference" | "binding" | "name" | "label" | "this"`) naming its role in the tree. ESTree collapses these distinct roles onto one `Identifier` type, so consumers normally have to walk the parent context to tell them apart. The parser already knows which is which, so it surfaces the answer as a free field.
+Yuku produces exactly the AST that [Oxc](https://oxc.rs) produces, for both JS and TS.
 
 On top of the base specs, the AST also carries:
 
@@ -74,6 +71,16 @@ const { line, column } = result.locOf(result.program.body[0].start);
 
 Lines are 1-based and columns are 0-based, matching ESTree's `loc` convention. The lookup is an O(log n) binary search: tens of nanoseconds per call, safe to invoke per node during a walk.
 
+When resolving many offsets in roughly source order, `locNear` is faster: it scans from a hint line instead of binary searching. Pass the previously returned `line` as the next `hintLine` so each lookup starts near the answer.
+
+```ts
+let hintLine = 1;
+for (const node of nodesInSourceOrder) {
+  const { line, column } = result.locNear(node.start, hintLine);
+  hintLine = line;
+}
+```
+
 ## Walking the AST
 
 [`yuku-ast`](https://www.npmjs.com/package/yuku-ast) is the companion toolkit built for this AST: a typed walker, node builders (`b`), type guards (`is`), and identifier validators.
@@ -134,6 +141,7 @@ interface ParseResult {
   diagnostics: Diagnostic[];
   lineStarts: number[];
   locOf(offset: number): { line: number; column: number };
+  locNear(offset: number, hintLine: number): { line: number; column: number };
 }
 ```
 

package/decode.js

@@ -1,6 +1,4 @@
 // generated by tools/gen_estree_decoder.zig, do not edit
-// `_u32` is an Int32Array so NodeIndex NULL reads as -1 and fits in
-// an SMI/int32, avoiding heap-number boxing on every field read in JSC.
 const NULL = -1;
 const _td = new TextDecoder("utf-8", { ignoreBOM: true });
 const BINARY_OPS = ["==", "!=", "===", "!==", "<", "<=", ">", ">=", "+", "-", "*", "/", "%", "**", "|", "^", "&", "<<", ">>", ">>>", "in", "instanceof"];
@@ -21,11 +19,6 @@ const TS_METHOD_SIGNATURE_KINDS = ["method", "get", "set"];
 const TS_MODULE_KINDS = ["namespace", "module"];
 const TS_MAPPED_OPTIONAL = [false, true, "+", "-"];
 const TS_MAPPED_READONLY = [null, true, "+", "-"];
-// builds a partial pos-map covering bytes [startByte, byteLen]. for any
-// byte position v >= startByte, m[v - startByte] is the corresponding
-// UTF-16 code-unit index into `src`. positions below startByte are an
-// identity map (byte position == UTF-16 position because the prefix is
-// pure ASCII), so the caller checks `v < startByte` and returns v directly.
 function buildPosMap(src, byteLen, startByte) {
   const m = new Uint32Array(byteLen - startByte + 1);
   const len = src.length;
@@ -80,8 +73,6 @@ function decode(buffer, source) {
   const eOff = _nodesOff + nodeCount * 48;
   const _extraBase = eOff >> 2;
   const _spOff = eOff + extraCount * 4;
-  // sections after the variable-length string pool may be at any
-  // byte alignment, so reads against them use `dv`, not `_u32`.
   const dv = new DataView(buffer);
   const _aoOff = _spOff + spLen;
   const _acOff = _attached ? _aoOff + (nodeCount + 1) * 4 : _aoOff;
@@ -115,22 +106,12 @@ function decode(buffer, source) {
     }
     return r;
   }
-  // `_p` and `str` translate UTF-8 byte offsets into JS UTF-16 indices.
-  // for any offset below `_firstNa` (the prefix of the source that is pure
-  // ASCII) the mapping is the identity, so we short-circuit and skip
-  // building a pos-map. when offsets cross into non-ASCII territory we
-  // lazily build a pos-map for `[_firstNa, _srcLen]` on first use.
-  let pm = null;
-  const _p = v => {
-    if (v < _firstNa) return v;
-    if (pm === null) pm = buildPosMap(_src, _srcLen, _firstNa);
-    return pm[v - _firstNa];
-  };
+  const pm = _firstNa < _srcLen ? buildPosMap(_src, _srcLen, _firstNa) : null;
+  const _p = v => v <= _firstNa ? v : pm[v - _firstNa];
   const str = (s, e) => {
     if (s === e) return "";
     if (s >= _srcLen) return _poolDecode(s, e);
     if (e <= _firstNa) return _src.slice(s, e);
-    if (pm === null) pm = buildPosMap(_src, _srcLen, _firstNa);
     const ss = s < _firstNa ? s : pm[s - _firstNa];
     return _src.slice(ss, pm[e - _firstNa]);
   };
@@ -175,8 +156,6 @@ function decode(buffer, source) {
   }
   function nodeWithComments(i) {
     const r = _decode(i);
-    // skip unwrap cases whose inner node already carries its own
-    // comments (e.g. formal_parameter returning its pattern)
     if (r && r.type !== undefined && r.comments === undefined) {
       const off = _aoOff + i * 4;
       const a = dv.getUint32(off, true), e = dv.getUint32(off + 4, true);
@@ -194,7 +173,9 @@ function decode(buffer, source) {
           f3 = _u32[b + 4], f4 = _u32[b + 5],
           f5 = _u32[b + 6], f6 = _u32[b + 7],
           f7 = _u32[b + 8], f8 = _u32[b + 9];
-    const start = _p(_u32[b + 10]), end = _p(_u32[b + 11]);
+    const _ss = _u32[b + 10], _se = _u32[b + 11];
+    const start = _ss <= _firstNa ? _ss : pm[_ss - _firstNa];
+    const end = _se <= _firstNa ? _se : pm[_se - _firstNa];
     switch (tag) {
     case 0: return { type: "SequenceExpression", start, end, expressions: nodeArr(f1, f0) };
     case 1: return { type: "ParenthesizedExpression", start, end, expression: f1 !== NULL ? node(f1) : null };
@@ -377,11 +358,11 @@ function decode(buffer, source) {
         tail: tl,
       };
     }
-    case 42: { const r = { type: "Identifier", start, end, name: str(f1, f2), kind: "reference" }; if (_isTs) { r.decorators = []; r.optional = false; r.typeAnnotation = null; } return r; }
+    case 42: { const r = { type: "Identifier", start, end, name: str(f1, f2) }; if (_isTs) { r.decorators = []; r.optional = false; r.typeAnnotation = null; } return r; }
     case 43: return { type: "PrivateIdentifier", start, end, name: str(f1, f2) };
-    case 44: { const r = { type: "Identifier", start, end, name: str(f1, f2), kind: "binding" }; if (_isTs) { r.decorators = nodeArr(f3, f0); r.typeAnnotation = f4 !== NULL ? node(f4) : null; r.optional = !!(flags & 1); } return r; }
-    case 45: { const r = { type: "Identifier", start, end, name: str(f1, f2), kind: "name" }; if (_isTs) { r.decorators = []; r.optional = false; r.typeAnnotation = null; } return r; }
-    case 46: { const r = { type: "Identifier", start, end, name: str(f1, f2), kind: "label" }; if (_isTs) { r.decorators = []; r.optional = false; r.typeAnnotation = null; } return r; }
+    case 44: { const r = { type: "Identifier", start, end, name: str(f1, f2) }; if (_isTs) { r.decorators = nodeArr(f3, f0); r.typeAnnotation = f4 !== NULL ? node(f4) : null; r.optional = !!(flags & 1); } return r; }
+    case 45: { const r = { type: "Identifier", start, end, name: str(f1, f2) }; if (_isTs) { r.decorators = []; r.optional = false; r.typeAnnotation = null; } return r; }
+    case 46: { const r = { type: "Identifier", start, end, name: str(f1, f2) }; if (_isTs) { r.decorators = []; r.optional = false; r.typeAnnotation = null; } return r; }
     case 47: { const r = { type: "ExpressionStatement", start, end, expression: f1 !== NULL ? node(f1) : null }; if (_isTs) { r.directive = null; } return r; }
     case 48: return { type: "IfStatement", start, end, test: f1 !== NULL ? node(f1) : null, consequent: f2 !== NULL ? node(f2) : null, alternate: f3 !== NULL ? node(f3) : null };
     case 49: return { type: "SwitchStatement", start, end, discriminant: f1 !== NULL ? node(f1) : null, cases: nodeArr(f2, f0) };
@@ -579,7 +560,7 @@ function decode(buffer, source) {
     case 146: return {
       type: "Identifier", start, end,
       decorators: [],
-      name: "this", kind: "this", optional: false,
+      name: "this", optional: false,
       typeAnnotation: f1 !== NULL ? node(f1) : null,
     };
     case 147: return { type: "TSAsExpression", start, end, expression: f1 !== NULL ? node(f1) : null, typeAnnotation: f2 !== NULL ? node(f2) : null };
@@ -640,7 +621,6 @@ function decode(buffer, source) {
       }
       return out;
     }
-    if (pm === null) pm = buildPosMap(_src, _srcLen, _firstNa);
     for (let j = 0; j < lineStartsCount; j++) {
       const v = dv.getUint32(lsOff + j * 4, true);
       out[j] = v < _firstNa ? v : (v >= _srcLen ? pm[pm.length - 1] : pm[v - _firstNa]);

package/index.d.ts

@@ -271,28 +271,9 @@ type PropertyDefinitionType = "PropertyDefinition" | "TSAbstractPropertyDefiniti
 
 type AccessorPropertyType = "AccessorProperty" | "TSAbstractAccessorProperty";
 
-/**
- * Role of an `Identifier` within the AST. Yuku exposes the four ESTree-collapsed
- * identifier flavors as a discriminator, plus `"this"` for a TS `this`-parameter
- * (which TS-ESTree also encodes as an `Identifier` with `name: "this"`).
- *
- * - `"reference"`: identifier read as a value (`x` in `x + 1`, `console`).
- * - `"binding"`: identifier being declared (parameters, `let x`, function/class names).
- * - `"name"`: identifier in a non-binding name slot (property/method names like `log` in `console.log`).
- * - `"label"`: loop label (`outer` in `outer: for (...) {}`).
- * - `"this"`: the TS `this`-parameter (`function f(this: T)`).
- */
-type IdentifierKind = "reference" | "binding" | "name" | "label" | "this";
-
 interface Identifier extends BaseNode {
   type: "Identifier";
   name: string;
-  /**
-   * ESTree collapses several distinct identifier
-   * roles onto one `Identifier` node, `kind` lets you tell them apart without
-   * walking the parent context. See {@link IdentifierKind}.
-   */
-  kind: IdentifierKind;
   decorators?: Decorator[];
   optional?: boolean;
   typeAnnotation?: TSTypeAnnotation | null;
@@ -1821,7 +1802,6 @@ export type {
   ModuleExportName,
   PropertyKey,
   Identifier,
-  IdentifierKind,
   IdentifierName,
   IdentifierReference,
   BindingIdentifier,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "yuku-parser",
-  "version": "0.5.19",
+  "version": "0.5.21",
   "description": "High-performance JavaScript/TypeScript parser",
   "license": "MIT",
   "repository": {
@@ -17,17 +17,17 @@
     "decode.js"
   ],
   "optionalDependencies": {
-    "@yuku-parser/binding-linux-x64-gnu": "0.5.19",
-    "@yuku-parser/binding-linux-arm64-gnu": "0.5.19",
-    "@yuku-parser/binding-linux-arm-gnu": "0.5.19",
-    "@yuku-parser/binding-linux-x64-musl": "0.5.19",
-    "@yuku-parser/binding-linux-arm64-musl": "0.5.19",
-    "@yuku-parser/binding-linux-arm-musl": "0.5.19",
-    "@yuku-parser/binding-darwin-x64": "0.5.19",
-    "@yuku-parser/binding-darwin-arm64": "0.5.19",
-    "@yuku-parser/binding-win32-x64": "0.5.19",
-    "@yuku-parser/binding-win32-arm64": "0.5.19",
-    "@yuku-parser/binding-freebsd-x64": "0.5.19"
+    "@yuku-parser/binding-linux-x64-gnu": "0.5.21",
+    "@yuku-parser/binding-linux-arm64-gnu": "0.5.21",
+    "@yuku-parser/binding-linux-arm-gnu": "0.5.21",
+    "@yuku-parser/binding-linux-x64-musl": "0.5.21",
+    "@yuku-parser/binding-linux-arm64-musl": "0.5.21",
+    "@yuku-parser/binding-linux-arm-musl": "0.5.21",
+    "@yuku-parser/binding-darwin-x64": "0.5.21",
+    "@yuku-parser/binding-darwin-arm64": "0.5.21",
+    "@yuku-parser/binding-win32-x64": "0.5.21",
+    "@yuku-parser/binding-win32-arm64": "0.5.21",
+    "@yuku-parser/binding-freebsd-x64": "0.5.21"
   },
   "keywords": [
     "acorn",