improve performance

.changeset/slow-chefs-study.md

@@ -0,0 +1,5 @@
+---
+"toml-eslint-parser": patch
+---
+
+improve performance

.eslintrc.js

@@ -22,6 +22,7 @@ module.exports = {
     "no-warning-comments": "warn",
     "no-lonely-if": "off",
     "no-param-reassign": "off",
+    "no-shadow": "off",
   },
   overrides: [
     {
@@ -32,6 +33,7 @@ module.exports = {
         project: "./tsconfig.json",
       },
       rules: {
+        "@typescript-eslint/no-shadow": "error",
         "@typescript-eslint/naming-convention": [
           "error",
           {

.github/workflows/NewOldBenchmark.yml

@@ -0,0 +1,22 @@
+name: NewOldBenchmark
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+
+jobs:
+  benchmark:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v4
+    - uses: actions/setup-node@v4
+    - name: Setup
+      run: npm run setup
+    - name: Install Packages
+      run: npm install -f
+    - name: Build
+      run: npm run build
+    - name: Benchmark
+      run: npm run benchmark

package.json

@@ -80,6 +80,7 @@
     "nyc": "^15.1.0",
     "prettier": "^3.0.0",
     "semver": "^7.3.4",
+    "toml-eslint-parser": "^0.8.0",
     "ts-node": "^10.0.0",
     "typescript": "~5.0.0",
     "vue-eslint-parser": "^9.0.0"

src/tokenizer/code-point-iterator.ts

@@ -1,4 +1,4 @@
-import { NULL, EOF, LINE_FEED, CARRIAGE_RETURN } from "./code-point";
+import { CodePoint } from "./code-point";
 
 type Position = {
   offset: number;
@@ -9,7 +9,7 @@ type Position = {
 export class CodePointIterator {
   public readonly text: string;
 
-  private lastCodePoint: number = NULL;
+  private lastCodePoint: number = CodePoint.NULL;
 
   public start: Position = {
     offset: -1,
@@ -31,31 +31,31 @@ export class CodePointIterator {
   }
 
   public next(): number {
-    if (this.lastCodePoint === EOF) {
-      return EOF;
+    if (this.lastCodePoint === CodePoint.EOF) {
+      return CodePoint.EOF;
     }
 
     this.start.offset = this.end.offset;
     this.start.line = this.end.line;
     this.start.column = this.end.column;
 
-    const cp = this.text.codePointAt(this.start.offset) ?? EOF;
-    if (cp === EOF) {
+    const cp = this.text.codePointAt(this.start.offset) ?? CodePoint.EOF;
+    if (cp === CodePoint.EOF) {
       this.end = this.start;
       return (this.lastCodePoint = cp);
     }
     const shift = cp >= 0x10000 ? 2 : 1;
     this.end.offset = this.start.offset + shift;
-    if (cp === LINE_FEED) {
+    if (cp === CodePoint.LINE_FEED) {
       this.end.line = this.start.line + 1;
       this.end.column = 0;
-    } else if (cp === CARRIAGE_RETURN) {
-      if (this.text.codePointAt(this.end.offset) === LINE_FEED) {
+    } else if (cp === CodePoint.CARRIAGE_RETURN) {
+      if (this.text.codePointAt(this.end.offset) === CodePoint.LINE_FEED) {
         this.end.offset++;
         this.end.line = this.start.line + 1;
         this.end.column = 0;
       }
-      return (this.lastCodePoint = LINE_FEED);
+      return (this.lastCodePoint = CodePoint.LINE_FEED);
     } else {
       this.end.column = this.start.column + shift;
     }
@@ -66,15 +66,15 @@ export class CodePointIterator {
   public *iterateSubCodePoints(): IterableIterator<number> {
     let index = this.end.offset;
     while (true) {
-      let cp = this.text.codePointAt(index) ?? EOF;
-      if (cp === CARRIAGE_RETURN) {
-        if (this.text.codePointAt(index) === LINE_FEED) {
-          cp = this.text.codePointAt(++index) ?? EOF;
+      let cp = this.text.codePointAt(index) ?? CodePoint.EOF;
+      if (cp === CodePoint.CARRIAGE_RETURN) {
+        if (this.text.codePointAt(index) === CodePoint.LINE_FEED) {
+          cp = this.text.codePointAt(++index) ?? CodePoint.EOF;
         } else {
-          cp = LINE_FEED;
+          cp = CodePoint.LINE_FEED;
         }
       }
-      if (cp === EOF) {
+      if (cp === CodePoint.EOF) {
         return;
       }
       yield cp;
@@ -92,12 +92,12 @@ export class CodePointIterator {
     return {
       next() {
         if (end) {
-          return EOF;
+          return CodePoint.EOF;
         }
         const r = sub.next();
         if (r.done) {
           end = true;
-          return EOF;
+          return CodePoint.EOF;
         }
         count++;
         return r.value;

src/tokenizer/code-point.ts

@@ -1,125 +1,127 @@
-export const EOF = -1;
-export const NULL = 0x00;
-export const SOH = 0x01;
-export const BACKSPACE = 0x08;
-export const TABULATION = 0x09;
-export const LINE_FEED = 0x0a;
-export const FORM_FEED = 0x0c;
-export const CARRIAGE_RETURN = 0x0d;
-export const ESCAPE = 0x1b;
-export const SO = 0x0e;
-export const US = 0x1f;
-export const SPACE = 0x20;
-export const QUOTATION_MARK = 0x22;
-export const HASH = 0x23;
-export const SINGLE_QUOTE = 0x27;
-export const PLUS_SIGN = 0x2b;
-export const COMMA = 0x2c;
-export const DASH = 0x2d;
-export const DOT = 0x2e;
-export const DIGIT_0 = 0x30;
-export const DIGIT_1 = 0x31;
-export const DIGIT_2 = 0x32;
-export const DIGIT_3 = 0x33;
-export const DIGIT_7 = 0x37;
-export const DIGIT_9 = 0x39;
-export const COLON = 0x3a;
-export const EQUALS_SIGN = 0x3d;
-export const LATIN_CAPITAL_A = 0x41;
-export const LATIN_CAPITAL_E = 0x45;
-export const LATIN_CAPITAL_F = 0x46;
-export const LATIN_CAPITAL_T = 0x54;
-export const LATIN_CAPITAL_U = 0x55;
-export const LATIN_CAPITAL_Z = 0x5a;
-export const LEFT_BRACKET = 0x5b; // [
-export const BACKSLASH = 0x5c;
-export const RIGHT_BRACKET = 0x5d; // ]
-export const UNDERSCORE = 0x5f;
-export const LATIN_SMALL_A = 0x61;
-export const LATIN_SMALL_B = 0x62;
-export const LATIN_SMALL_E = 0x65;
-export const LATIN_SMALL_F = 0x66;
-export const LATIN_SMALL_I = 0x69;
-export const LATIN_SMALL_L = 0x6c;
-export const LATIN_SMALL_N = 0x6e;
-export const LATIN_SMALL_O = 0x6f;
-export const LATIN_SMALL_R = 0x72;
-export const LATIN_SMALL_S = 0x73;
-export const LATIN_SMALL_T = 0x74;
-export const LATIN_SMALL_U = 0x75;
-export const LATIN_SMALL_X = 0x78;
-export const LATIN_SMALL_Z = 0x7a;
-export const LEFT_BRACE = 0x7b; // {
-export const RIGHT_BRACE = 0x7d; // }
-export const DELETE = 0x7f;
-export const PAD = 0x80;
-export const SUPERSCRIPT_TWO = 0xb2;
-export const SUPERSCRIPT_THREE = 0xb3;
-export const SUPERSCRIPT_ONE = 0xb9;
-export const VULGAR_FRACTION_ONE_QUARTER = 0xbc;
-export const VULGAR_FRACTION_THREE_QUARTERS = 0xbe;
-export const LATIN_CAPITAL_LETTER_A_WITH_GRAVE = 0xc0;
-export const LATIN_CAPITAL_LETTER_O_WITH_DIAERESIS = 0xd6;
-export const LATIN_CAPITAL_LETTER_O_WITH_STROKE = 0xd8;
-export const LATIN_SMALL_LETTER_O_WITH_DIAERESIS = 0xf6;
-export const LATIN_SMALL_LETTER_O_WITH_STROKE = 0xf8;
-export const GREEK_SMALL_REVERSED_DOTTED_LUNATE_SIGMA_SYMBOL = 0x37b;
-export const GREEK_CAPITAL_LETTER_YOT = 0x37f;
-export const CP_1FFF = 0x1fff;
-export const ZERO_WIDTH_NON_JOINER = 0x200c;
-export const ZERO_WIDTH_JOINER = 0x200d;
-export const UNDERTIE = 0x203f;
-export const CHARACTER_TIE = 0x2040;
-export const SUPERSCRIPT_ZERO = 0x2070;
-export const CP_218F = 0x218f;
-export const CIRCLED_DIGIT_ONE = 0x2460;
-export const NEGATIVE_CIRCLED_DIGIT_ZERO = 0x24ff;
-export const GLAGOLITIC_CAPITAL_LETTER_AZU = 0x2c00;
-export const CP_2FEF = 0x2fef;
-export const IDEOGRAPHIC_COMMA = 0x3001;
-export const CP_D7FF = 0xd7ff;
-export const CP_E000 = 0xe000;
-export const CJK_COMPATIBILITY_IDEOGRAPH_F900 = 0xf900;
-export const ARABIC_LIGATURE_SALAAMUHU_ALAYNAA = 0xfdcf;
-export const ARABIC_LIGATURE_SALLA_USED_AS_KORANIC_STOP_SIGN_ISOLATED_FORM = 0xfdf0;
-export const REPLACEMENT_CHARACTER = 0xfffd;
-export const LINEAR_B_SYLLABLE_B008_A = 0x10000;
-export const CP_EFFFF = 0xeffff;
-export const CP_10FFFF = 0x10ffff;
+export const enum CodePoint {
+  EOF = -1,
+  NULL = 0x00,
+  SOH = 0x01,
+  BACKSPACE = 0x08,
+  TABULATION = 0x09,
+  LINE_FEED = 0x0a,
+  FORM_FEED = 0x0c,
+  CARRIAGE_RETURN = 0x0d,
+  ESCAPE = 0x1b,
+  SO = 0x0e,
+  US = 0x1f,
+  SPACE = 0x20,
+  QUOTATION_MARK = 0x22,
+  HASH = 0x23,
+  SINGLE_QUOTE = 0x27,
+  PLUS_SIGN = 0x2b,
+  COMMA = 0x2c,
+  DASH = 0x2d,
+  DOT = 0x2e,
+  DIGIT_0 = 0x30,
+  DIGIT_1 = 0x31,
+  DIGIT_2 = 0x32,
+  DIGIT_3 = 0x33,
+  DIGIT_7 = 0x37,
+  DIGIT_9 = 0x39,
+  COLON = 0x3a,
+  EQUALS_SIGN = 0x3d,
+  LATIN_CAPITAL_A = 0x41,
+  LATIN_CAPITAL_E = 0x45,
+  LATIN_CAPITAL_F = 0x46,
+  LATIN_CAPITAL_T = 0x54,
+  LATIN_CAPITAL_U = 0x55,
+  LATIN_CAPITAL_Z = 0x5a,
+  LEFT_BRACKET = 0x5b, // [
+  BACKSLASH = 0x5c,
+  RIGHT_BRACKET = 0x5d, // ]
+  UNDERSCORE = 0x5f,
+  LATIN_SMALL_A = 0x61,
+  LATIN_SMALL_B = 0x62,
+  LATIN_SMALL_E = 0x65,
+  LATIN_SMALL_F = 0x66,
+  LATIN_SMALL_I = 0x69,
+  LATIN_SMALL_L = 0x6c,
+  LATIN_SMALL_N = 0x6e,
+  LATIN_SMALL_O = 0x6f,
+  LATIN_SMALL_R = 0x72,
+  LATIN_SMALL_S = 0x73,
+  LATIN_SMALL_T = 0x74,
+  LATIN_SMALL_U = 0x75,
+  LATIN_SMALL_X = 0x78,
+  LATIN_SMALL_Z = 0x7a,
+  LEFT_BRACE = 0x7b, // {
+  RIGHT_BRACE = 0x7d, // }
+  DELETE = 0x7f,
+  PAD = 0x80,
+  SUPERSCRIPT_TWO = 0xb2,
+  SUPERSCRIPT_THREE = 0xb3,
+  SUPERSCRIPT_ONE = 0xb9,
+  VULGAR_FRACTION_ONE_QUARTER = 0xbc,
+  VULGAR_FRACTION_THREE_QUARTERS = 0xbe,
+  LATIN_CAPITAL_LETTER_A_WITH_GRAVE = 0xc0,
+  LATIN_CAPITAL_LETTER_O_WITH_DIAERESIS = 0xd6,
+  LATIN_CAPITAL_LETTER_O_WITH_STROKE = 0xd8,
+  LATIN_SMALL_LETTER_O_WITH_DIAERESIS = 0xf6,
+  LATIN_SMALL_LETTER_O_WITH_STROKE = 0xf8,
+  GREEK_SMALL_REVERSED_DOTTED_LUNATE_SIGMA_SYMBOL = 0x37b,
+  GREEK_CAPITAL_LETTER_YOT = 0x37f,
+  CP_1FFF = 0x1fff,
+  ZERO_WIDTH_NON_JOINER = 0x200c,
+  ZERO_WIDTH_JOINER = 0x200d,
+  UNDERTIE = 0x203f,
+  CHARACTER_TIE = 0x2040,
+  SUPERSCRIPT_ZERO = 0x2070,
+  CP_218F = 0x218f,
+  CIRCLED_DIGIT_ONE = 0x2460,
+  NEGATIVE_CIRCLED_DIGIT_ZERO = 0x24ff,
+  GLAGOLITIC_CAPITAL_LETTER_AZU = 0x2c00,
+  CP_2FEF = 0x2fef,
+  IDEOGRAPHIC_COMMA = 0x3001,
+  CP_D7FF = 0xd7ff,
+  CP_E000 = 0xe000,
+  CJK_COMPATIBILITY_IDEOGRAPH_F900 = 0xf900,
+  ARABIC_LIGATURE_SALAAMUHU_ALAYNAA = 0xfdcf,
+  ARABIC_LIGATURE_SALLA_USED_AS_KORANIC_STOP_SIGN_ISOLATED_FORM = 0xfdf0,
+  REPLACEMENT_CHARACTER = 0xfffd,
+  LINEAR_B_SYLLABLE_B008_A = 0x10000,
+  CP_EFFFF = 0xeffff,
+  CP_10FFFF = 0x10ffff,
+}
 
 /**
  * Check whether the code point is a control character.
  */
 export function isControl(cp: number): boolean {
-  return cp >= NULL && cp <= US;
+  return cp >= CodePoint.NULL && cp <= CodePoint.US;
 }
 
 /**
  * Check whether the code point is a whitespace.
  */
 export function isWhitespace(cp: number): boolean {
-  return cp === TABULATION || cp === SPACE;
+  return cp === CodePoint.TABULATION || cp === CodePoint.SPACE;
 }
 
 /**
  * Check whether the code point is a end of line.
  */
 export function isEOL(cp: number): boolean {
-  return cp === LINE_FEED || cp === CARRIAGE_RETURN;
+  return cp === CodePoint.LINE_FEED || cp === CodePoint.CARRIAGE_RETURN;
 }
 
 /**
  * Check whether the code point is an uppercase letter character.
  */
 function isUpperLetter(cp: number): boolean {
-  return cp >= LATIN_CAPITAL_A && cp <= LATIN_CAPITAL_Z;
+  return cp >= CodePoint.LATIN_CAPITAL_A && cp <= CodePoint.LATIN_CAPITAL_Z;
 }
 
 /**
  * Check whether the code point is a lowercase letter character.
  */
 function isLowerLetter(cp: number): boolean {
-  return cp >= LATIN_SMALL_A && cp <= LATIN_SMALL_Z;
+  return cp >= CodePoint.LATIN_SMALL_A && cp <= CodePoint.LATIN_SMALL_Z;
 }
 
 /**
@@ -133,7 +135,7 @@ export function isLetter(cp: number): boolean {
  * Check whether the code point is a digit character.
  */
 export function isDigit(cp: number): boolean {
-  return cp >= DIGIT_0 && cp <= DIGIT_9;
+  return cp >= CodePoint.DIGIT_0 && cp <= CodePoint.DIGIT_9;
 }
 
 /**
@@ -142,15 +144,15 @@ export function isDigit(cp: number): boolean {
 export function isHexDig(cp: number): boolean {
   return (
     isDigit(cp) ||
-    (cp >= LATIN_SMALL_A && cp <= LATIN_SMALL_F) ||
-    (cp >= LATIN_CAPITAL_A && cp <= LATIN_CAPITAL_F)
+    (cp >= CodePoint.LATIN_SMALL_A && cp <= CodePoint.LATIN_SMALL_F) ||
+    (cp >= CodePoint.LATIN_CAPITAL_A && cp <= CodePoint.LATIN_CAPITAL_F)
   );
 }
 /**
  * Check whether the code point is a octal digit character.
  */
 export function isOctalDig(cp: number): boolean {
-  return cp >= DIGIT_0 && cp <= DIGIT_7;
+  return cp >= CodePoint.DIGIT_0 && cp <= CodePoint.DIGIT_7;
 }
 
 /**