Merge pull request #102 from jviereck/jviereck/fix101

Fix regression with using closing brackets inside of class range in non-unicode mode
jviereck · Mar 11, 2020 · 1325cad · 1325cad
2 parents 0f22de3 + e929267
commit 1325cad
Show file tree

Hide file tree

Showing 4 changed files with 110 additions and 15 deletions.
diff --git a/package.json b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "regjsparser",
-  "version": "0.6.2",
+  "version": "0.6.3",
   "author": "'Julian Viereck' <[email protected]>",
   "license": "BSD-2-Clause",
   "main": "./parser",

diff --git a/parser.js b/parser.js
@@ -233,12 +233,6 @@
       var _char = matches[0];
       var first = _char.charCodeAt(0);
       if (hasUnicodeFlag) {
-        if (_char === '}') {
-          bail("unescaped or unmatched closing brace");
-        }
-        if (_char === ']') {
-          bail("unescaped or unmatched closing bracket");
-        }
         var second;
         if (_char.length === 1 && first >= 0xD800 && first <= 0xDBFF) {
           second = lookahead().charCodeAt(0);
@@ -487,7 +481,7 @@
         return anchor;
       }
 
-      var atom = parseAtom();
+      var atom = parseAtomAndExtendedAtom();
       if (!atom) {
         bail('Expected atom');
       }
@@ -611,24 +605,36 @@
       return quantifier;
     }
 
-    function parseAtom() {
+    function parseAtomAndExtendedAtom() {
+      // Parsing Atom and ExtendedAtom together due to redundancy.
+      // ExtendedAtom is defined in Apendix B of the ECMA-262 standard.
+      //
+      // SEE: https://www.ecma-international.org/ecma-262/10.0/index.html#prod-annexB-ExtendedPatternCharacter
+      //
       // Atom ::
       //      PatternCharacter
       //      .
       //      \ AtomEscape
       //      CharacterClass
       //      ( GroupSpecifier Disjunction )
       //      ( ? : Disjunction )
+      // ExtendedAtom ::
+      //      ExtendedPatternCharacter
+      // ExtendedPatternCharacter ::
+      //      SourceCharacter but not one of ^$\.*+?()[|
 
       var res;
 
       // jviereck: allow ']', '}' here as well to be compatible with browser's
       //   implementations: ']'.match(/]/);
-      // if (res = matchReg(/^[^^$\\.*+?()[\]{}|]/)) {
-      if (res = matchReg(/^[^^$\\.*+?(){[|]/)) {
+      if (res = matchReg(/^[^^$\\.*+?()[\]{}|]/)) {
         //      PatternCharacter
         return createCharacter(res);
       }
+      else if (!hasUnicodeFlag && (res = matchReg(/^(?:]|})/))) {
+        //      ExtendedPatternCharacter
+        return createCharacter(res);
+      }
       else if (match('.')) {
         //      .
         return createDot();

diff --git a/test/test-data-unicode.json b/test/test-data-unicode.json
@@ -914,13 +914,13 @@
   "}": {
     "type": "error",
     "name": "SyntaxError",
-    "message": "unescaped or unmatched closing brace at position 1\n    }\n     ^",
+    "message": "Expected atom at position 0\n    }\n    ^",
     "input": "}"
   },
   "]": {
     "type": "error",
     "name": "SyntaxError",
-    "message": "unescaped or unmatched closing bracket at position 1\n    ]\n     ^",
+    "message": "Expected atom at position 0\n    ]\n    ^",
     "input": "]"
   },
   "[\\-]": {
@@ -930,12 +930,60 @@
         "type": "value",
         "kind": "singleEscape",
         "codePoint": 45,
-        "range": [1, 3],
+        "range": [
+          1,
+          3
+        ],
         "raw": "\\-"
       }
     ],
     "negative": false,
-    "range": [0, 4],
+    "range": [
+      0,
+      4
+    ],
     "raw": "[\\-]"
+  },
+  "[}]": {
+    "type": "characterClass",
+    "body": [
+      {
+        "type": "value",
+        "kind": "symbol",
+        "codePoint": 125,
+        "range": [
+          1,
+          2
+        ],
+        "raw": "}"
+      }
+    ],
+    "negative": false,
+    "range": [
+      0,
+      3
+    ],
+    "raw": "[}]"
+  },
+  "[^}]": {
+    "type": "characterClass",
+    "body": [
+      {
+        "type": "value",
+        "kind": "symbol",
+        "codePoint": 125,
+        "range": [
+          2,
+          3
+        ],
+        "raw": "}"
+      }
+    ],
+    "negative": true,
+    "range": [
+      0,
+      4
+    ],
+    "raw": "[^}]"
   }
 }
diff --git a/test/test-data.json b/test/test-data.json
@@ -37472,5 +37472,46 @@
       2
     ],
     "raw": "a."
+  },
+  "}": {
+    "type": "value",
+    "kind": "symbol",
+    "codePoint": 125,
+    "range": [
+      0,
+      1
+    ],
+    "raw": "}"
+  },
+  "]": {
+    "type": "value",
+    "kind": "symbol",
+    "codePoint": 93,
+    "range": [
+      0,
+      1
+    ],
+    "raw": "]"
+  },
+  "[}]": {
+    "type": "characterClass",
+    "body": [
+      {
+        "type": "value",
+        "kind": "symbol",
+        "codePoint": 125,
+        "range": [
+          1,
+          2
+        ],
+        "raw": "}"
+      }
+    ],
+    "negative": false,
+    "range": [
+      0,
+      3
+    ],
+    "raw": "[}]"
   }
 }