Merge pull request #122 from nicolo-ribaudo/fix-unicode-refs

Fix parsing of group references in unicode mode
jviereck · Dec 14, 2021 · f09b912 · f09b912
2 parents 450cb3f + 15cf7a5
commit f09b912
Show file tree

Hide file tree

Showing 3 changed files with 122 additions and 17 deletions.
diff --git a/parser.js b/parser.js
@@ -851,16 +851,6 @@
 
       var res, match, from = pos;
 
-      if (hasUnicodeFlag) {
-        if (res = matchReg(/^\d/)) {
-          if (res[0] !== "0" || (res = matchReg(/^\d/)) ) {
-            bail("Invalid decimal escape in unicode mode", null, from, pos);
-          }
-          return createEscaped('null', 0x0000, '0', 1);
-        }
-        return false;
-      }
-
       if (res = matchReg(/^(?!0)\d+/)) {
         match = res[0];
         var refIdx = parseInt(res[0], 10);
@@ -878,6 +868,16 @@
           // a second time with the total normal-groups.
           backrefDenied.push(refIdx);
 
+          // \1 octal escapes are disallowed in unicode mode, but they might
+          // be references to groups which haven't been parsed yet.
+          // We must parse a second time to determine if \1 is a reference
+          // or an octal scape, and then we can report the error.
+          if (firstIteration) {
+            shouldReparse = true;
+          } else {
+            bailOctalEscapeIfUnicode(from, pos);
+          }
+
           // Reset the position again, as maybe only parts of the previous
           // matched numbers are actual octal numbers. E.g. in '019' only
           // the '01' should be matched.
@@ -904,6 +904,9 @@
       //   /\091/.exec('\091')[0].length === 3
       else if (res = matchReg(/^[0-7]{1,3}/)) {
         match = res[0];
+        if (match !== '0') {
+          bailOctalEscapeIfUnicode(from, pos);
+        }
         if (/^0{1,3}$/.test(match)) {
           // If they are all zeros, then only take the first one.
           return createEscaped('null', 0x0000, '0', match.length);
@@ -914,6 +917,12 @@
       return false;
     }
 
+    function bailOctalEscapeIfUnicode(from, pos) {
+      if (hasUnicodeFlag || hasUnicodeSetFlag) {
+        bail("Invalid decimal escape in unicode mode", null, from, pos);
+      }
+    }
+
     function parseCharacterClassEscape() {
       // CharacterClassEscape :: one of d D s S w W
       var res;
@@ -1500,6 +1509,7 @@
     var backrefDenied = [];
     var closedCaptureCounter = 0;
     var firstIteration = true;
+    var shouldReparse = false;
     var hasUnicodeFlag = (flags || "").indexOf("u") !== -1;
     var hasUnicodeSetFlag = (flags || "").indexOf("v") !== -1;
     var pos = 0;
@@ -1532,13 +1542,14 @@
     // the total number of capture groups set.
     //
     // SEE: https://github.com/jviereck/regjsparser/issues/70
-    for (var i = 0; i < backrefDenied.length; i++) {
-      if (backrefDenied[i] <= closedCaptureCounter) {
-        // Parse the input a second time.
-        pos = 0;
-        firstIteration = false;
-        return parseDisjunction();
-      }
+    shouldReparse = shouldReparse || backrefDenied.some(function (ref) {
+      return ref <= closedCaptureCounter;
+    });
+    if (shouldReparse) {
+      // Parse the input a second time.
+      pos = 0;
+      firstIteration = false;
+      return parseDisjunction();
     }
 
     return result;

diff --git a/test/test-data-unicode-set.json b/test/test-data-unicode-set.json
@@ -1493,5 +1493,11 @@
       9
     ],
     "raw": "[^\\q{AB}]"
+  },
+  "\\1": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Invalid decimal escape in unicode mode at position 1\n    \\1\n     ^",
+    "input": "\\1"
   }
 }
diff --git a/test/test-data-unicode.json b/test/test-data-unicode.json
@@ -1140,5 +1140,93 @@
       2
     ],
     "raw": "\\0"
+  },
+  "(.)\\1": {
+    "type": "alternative",
+    "body": [
+      {
+        "type": "group",
+        "behavior": "normal",
+        "body": [
+          {
+            "type": "dot",
+            "range": [
+              1,
+              2
+            ],
+            "raw": "."
+          }
+        ],
+        "range": [
+          0,
+          3
+        ],
+        "raw": "(.)"
+      },
+      {
+        "type": "reference",
+        "matchIndex": 1,
+        "range": [
+          3,
+          5
+        ],
+        "raw": "\\1"
+      }
+    ],
+    "range": [
+      0,
+      5
+    ],
+    "raw": "(.)\\1"
+  },
+  "\\1(.)": {
+    "type": "alternative",
+    "body": [
+      {
+        "type": "reference",
+        "matchIndex": 1,
+        "range": [
+          0,
+          2
+        ],
+        "raw": "\\1"
+      },
+      {
+        "type": "group",
+        "behavior": "normal",
+        "body": [
+          {
+            "type": "dot",
+            "range": [
+              3,
+              4
+            ],
+            "raw": "."
+          }
+        ],
+        "range": [
+          2,
+          5
+        ],
+        "raw": "(.)"
+      }
+    ],
+    "range": [
+      0,
+      5
+    ],
+    "raw": "\\1(.)"
+  },
+  "(.)\\2": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Invalid decimal escape in unicode mode at position 4\n    (.)\\2\n        ^",
+    "input": "(.)\\2"
+  },
+  "\\2(.)": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Invalid decimal escape in unicode mode at position 1\n    \\2(.)\n     ^",
+    "input": "\\2(.)"
   }
 }