test_runner: parse non-ascii character correctly

PR-URL: #45736 Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Moshe Atlow <moshe@atlow.co.il>
nodejs · Feb 20, 2023 · ddf5002 · ddf5002
1 parent fa600fe
commit ddf5002
Show file tree

Hide file tree

Showing 3 changed files with 74 additions and 24 deletions.
diff --git a/lib/internal/test_runner/tap_lexer.js b/lib/internal/test_runner/tap_lexer.js
@@ -5,13 +5,15 @@ const {
   ArrayPrototypePush,
   MathMax,
   SafeSet,
-  StringPrototypeIncludes,
+  StringPrototypeCodePointAt,
   StringPrototypeTrim,
 } = primordials;
 const {
   codes: { ERR_TAP_LEXER_ERROR },
 } = require('internal/errors');
 
+const { isZeroWidthCodePoint } = require('internal/util/inspect');
+
 const kEOL = '';
 const kEOF = '';
 
@@ -474,18 +476,28 @@ class TapLexer {
   }
 
   #isLiteralSymbol(char) {
-    return (
-      (char >= 'a' && char <= 'z') ||
-      (char >= 'A' && char <= 'Z') ||
-      this.#isSpecialCharacterSymbol(char)
-    );
-  }
-
-  #isSpecialCharacterSymbol(char) {
-    // We deliberately do not include "# \ + -"" in this list
-    // these are used for comments/reasons explanations, pragma and escape characters
-    // whitespace is not included because it is handled separately
-    return StringPrototypeIncludes('!"$%&\'()*,./:;<=>?@[]^_`{|}~', char);
+    if (typeof char !== 'string') return false;
+    const charCode = StringPrototypeCodePointAt(char);
+
+    if (isZeroWidthCodePoint(charCode)) return false;
+    if (this.#isWhitespaceSymbol(char)) return false;
+    const MAX_ASCII_CHAR_CODE = 0b111_1111; // ASCII is 7-bit long
+    // Allow all non-latin characters.
+    if (charCode > MAX_ASCII_CHAR_CODE) return true;
+    const ZERO = 48; // 0
+    const NINE = 58; // 9
+    // Disallow numeric values
+    if (charCode >= ZERO && char <= NINE) return false;
+
+    // Disallow characters with special meaning in TAP
+    const HASH = 35; // #
+    const BACKSLASH = 92; // \
+    const PLUS = 43; // +
+    const DASH = 45; // -
+
+    // Disallow characters with special meaning in TAP
+    return charCode !== HASH && charCode !== BACKSLASH &&
+           charCode !== PLUS && charCode !== DASH;
   }
 
   #isWhitespaceSymbol(char) {

diff --git a/lib/internal/util/inspect.js b/lib/internal/util/inspect.js
@@ -2295,6 +2295,18 @@ function formatWithOptionsInternal(inspectOptions, args) {
   return str;
 }
 
+function isZeroWidthCodePoint(code) {
+  return code <= 0x1F || // C0 control codes
+    (code >= 0x7F && code <= 0x9F) || // C1 control codes
+    (code >= 0x300 && code <= 0x36F) || // Combining Diacritical Marks
+    (code >= 0x200B && code <= 0x200F) || // Modifying Invisible Characters
+    // Combining Diacritical Marks for Symbols
+    (code >= 0x20D0 && code <= 0x20FF) ||
+    (code >= 0xFE00 && code <= 0xFE0F) || // Variation Selectors
+    (code >= 0xFE20 && code <= 0xFE2F) || // Combining Half Marks
+    (code >= 0xE0100 && code <= 0xE01EF); // Variation Selectors
+}
+
 if (internalBinding('config').hasIntl) {
   const icu = internalBinding('icu');
   // icu.getStringWidth(string, ambiguousAsFullWidth, expandEmojiSequence)
@@ -2384,17 +2396,6 @@ if (internalBinding('config').hasIntl) {
     );
   };
 
-  const isZeroWidthCodePoint = (code) => {
-    return code <= 0x1F || // C0 control codes
-      (code >= 0x7F && code <= 0x9F) || // C1 control codes
-      (code >= 0x300 && code <= 0x36F) || // Combining Diacritical Marks
-      (code >= 0x200B && code <= 0x200F) || // Modifying Invisible Characters
-      // Combining Diacritical Marks for Symbols
-      (code >= 0x20D0 && code <= 0x20FF) ||
-      (code >= 0xFE00 && code <= 0xFE0F) || // Variation Selectors
-      (code >= 0xFE20 && code <= 0xFE2F) || // Combining Half Marks
-      (code >= 0xE0100 && code <= 0xE01EF); // Variation Selectors
-  };
 }
 
 /**
@@ -2414,4 +2415,5 @@ module.exports = {
   formatWithOptions,
   getStringWidth,
   stripVTControlCharacters,
+  isZeroWidthCodePoint,
 };
diff --git a/test/parallel/test-runner-tap-lexer.js b/test/parallel/test-runner-tap-lexer.js
@@ -444,3 +444,39 @@ ok 1
     assert.strictEqual(tokens[index].value, token.value);
   });
 }
+
+// Test isLiteralSymbol method
+{
+  const tokens = TAPLexer('ok 1 - description أتث讲演講👍🔥');
+
+  [
+    { kind: TokenKind.TAP_TEST_OK, value: 'ok' },
+    { kind: TokenKind.WHITESPACE, value: ' ' },
+    { kind: TokenKind.NUMERIC, value: '1' },
+    { kind: TokenKind.WHITESPACE, value: ' ' },
+    { kind: TokenKind.DASH, value: '-' },
+    { kind: TokenKind.WHITESPACE, value: ' ' },
+    { kind: TokenKind.LITERAL, value: 'description' },
+    { kind: TokenKind.WHITESPACE, value: ' ' },
+    { kind: TokenKind.LITERAL, value: 'أتث讲演講👍🔥' },
+    { kind: TokenKind.EOL, value: '' },
+  ].forEach((token, index) => {
+    assert.strictEqual(tokens[index].kind, token.kind);
+    assert.strictEqual(tokens[index].value, token.value);
+  });
+}
+
+{
+  const tokens = TAPLexer('# comment أتث讲演講👍🔥');
+  [
+    { kind: TokenKind.COMMENT, value: '#' },
+    { kind: TokenKind.WHITESPACE, value: ' ' },
+    { kind: TokenKind.LITERAL, value: 'comment' },
+    { kind: TokenKind.WHITESPACE, value: ' ' },
+    { kind: TokenKind.LITERAL, value: 'أتث讲演講👍🔥' },
+    { kind: TokenKind.EOL, value: '' },
+  ].forEach((token, index) => {
+    assert.strictEqual(tokens[index].kind, token.kind);
+    assert.strictEqual(tokens[index].value, token.value);
+  });
+}