Merge pull request #149 from babel/fix-annex-b-quantifiable-assertion

fix: only ±lookbehind is quantifiable in annexB
jviereck · Oct 25, 2024 · 4dda218 · 4dda218
2 parents 77216e7 + 91d119f
commit 4dda218
Show file tree

Hide file tree

Showing 4 changed files with 102 additions and 34 deletions.
diff --git a/parser.js b/parser.js
@@ -567,51 +567,71 @@
     function parseTerm() {
       // Term ::
       //      Anchor
-      //      Anchor Quantifier (see https://github.com/jviereck/regjsparser/issues/130)
       //      Atom
       //      Atom Quantifier
 
+      // Term (Annex B)::
+      //      [~UnicodeMode] QuantifiableAssertion Quantifier (see https://github.com/jviereck/regjsparser/issues/130)
+      //      [~UnicodeMode] ExtendedAtom Quantifier
+
+      // QuantifiableAssertion::
+      //      (?= Disjunction[~UnicodeMode, ~UnicodeSetsMode, ?NamedCaptureGroups] )
+      //      (?! Disjunction[~UnicodeMode, ~UnicodeSetsMode, ?NamedCaptureGroups] ) 
+
       if (pos >= str.length || current('|') || current(')')) {
         return null; /* Means: The term is empty */
       }
 
-      var anchorOrAtom = parseAnchor();
+      var anchor = parseAnchor();
 
-      // If there is no Anchor, try to parse an atom.
-      if (!anchorOrAtom) {
-        var atom = parseAtomAndExtendedAtom();
-        var quantifier;
-        if (!atom) {
-          // Check if a quantifier is following. A quantifier without an atom
-          // is an error.
-          var pos_backup = pos
-          quantifier = parseQuantifier() || false;
-          if (quantifier) {
-            pos = pos_backup
-            bail('Expected atom');
+      if (anchor) {
+        var pos_backup = pos;
+        quantifier = parseQuantifier() || false;
+        if (quantifier) {
+          // Annex B
+          if (!isUnicodeMode && anchor.type === "group") {
+            quantifier.body = flattenBody(anchor);
+            // The quantifier contains the anchor. Therefore, the beginning of the
+            // quantifier range is given by the beginning of the anchor.
+            updateRawStart(quantifier, anchor.range[0]);
+            return quantifier;
           }
+          pos = pos_backup;
+          bail("Expected atom");
+        }
+        return anchor;
+      }
 
-          // If no unicode flag, then try to parse ExtendedAtom -> ExtendedPatternCharacter.
-          //      ExtendedPatternCharacter
-          var res;
-          if (!isUnicodeMode && (res = matchReg(/^\{/))) {
-            atom = createCharacter(res);
-          } else {
-            bail('Expected atom');
-          }
+      // If there is no Anchor, try to parse an atom.
+      var atom = parseAtomAndExtendedAtom();
+      var quantifier;
+      if (!atom) {
+        // Check if a quantifier is following. A quantifier without an atom
+        // is an error.
+        pos_backup = pos;
+        quantifier = parseQuantifier() || false;
+        if (quantifier) {
+          pos = pos_backup;
+          bail("Expected atom");
+        }
+
+        // If no unicode flag, then try to parse ExtendedAtom -> ExtendedPatternCharacter.
+        //      ExtendedPatternCharacter
+        var res;
+        if (!isUnicodeMode && (res = matchReg(/^\{/))) {
+          atom = createCharacter(res);
+        } else {
+          bail("Expected atom");
         }
-        anchorOrAtom = atom;
       }
 
       quantifier = parseQuantifier() || false;
       if (quantifier) {
-        var type = anchorOrAtom.type, behavior = anchorOrAtom.behavior;
+        var type = atom.type, behavior = atom.behavior;
         if (
           type === "group" &&
           (behavior === "negativeLookbehind" ||
-            behavior === "lookbehind" ||
-            (isUnicodeMode &&
-              (behavior === "negativeLookahead" || behavior === "lookahead")))
+            behavior === "lookbehind")
         ) {
           bail(
             "Invalid quantifier",
@@ -620,13 +640,13 @@
             quantifier.range[1]
           );
         }
-        quantifier.body = flattenBody(anchorOrAtom);
+        quantifier.body = flattenBody(atom);
         // The quantifier contains the atom. Therefore, the beginning of the
         // quantifier range is given by the beginning of the atom.
-        updateRawStart(quantifier, anchorOrAtom.range[0]);
+        updateRawStart(quantifier, atom.range[0]);
         return quantifier;
       }
-      return anchorOrAtom;
+      return atom;
     }
 
     function parseGroup(matchA, typeA, matchB, typeB) {

diff --git a/test/test-data-unicode-set.json b/test/test-data-unicode-set.json
@@ -1499,13 +1499,13 @@
   ".(?=.){2,3}": {
     "type": "error",
     "name": "SyntaxError",
-    "message": "Invalid quantifier at position 6\n    .(?=.){2,3}\n          ^",
+    "message": "Expected atom at position 6\n    .(?=.){2,3}\n          ^",
     "input": ".(?=.){2,3}"
   },
   ".(?!.){2,3}": {
     "type": "error",
     "name": "SyntaxError",
-    "message": "Invalid quantifier at position 6\n    .(?!.){2,3}\n          ^",
+    "message": "Expected atom at position 6\n    .(?!.){2,3}\n          ^",
     "input": ".(?!.){2,3}"
   },
   "[\\__]": {

diff --git a/test/test-data-unicode.json b/test/test-data-unicode.json
@@ -1232,13 +1232,13 @@
   ".(?=.){2,3}": {
     "type": "error",
     "name": "SyntaxError",
-    "message": "Invalid quantifier at position 6\n    .(?=.){2,3}\n          ^",
+    "message": "Expected atom at position 6\n    .(?=.){2,3}\n          ^",
     "input": ".(?=.){2,3}"
   },
   ".(?!.){2,3}": {
     "type": "error",
     "name": "SyntaxError",
-    "message": "Invalid quantifier at position 6\n    .(?!.){2,3}\n          ^",
+    "message": "Expected atom at position 6\n    .(?!.){2,3}\n          ^",
     "input": ".(?!.){2,3}"
   },
   "[&&]": {
@@ -1848,5 +1848,29 @@
       4
     ],
     "raw": "[~~]"
+  },
+  "^*": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Expected atom at position 1\n    ^*\n     ^",
+    "input": "^*"
+  },
+  "$+": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Expected atom at position 1\n    $+\n     ^",
+    "input": "$+"
+  },
+  "\\b?": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Expected atom at position 2\n    \\b?\n      ^",
+    "input": "\\b?"
+  },
+  "\\B{1}": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Expected atom at position 2\n    \\B{1}\n      ^",
+    "input": "\\B{1}"
   }
 }
diff --git a/test/test-data.json b/test/test-data.json
@@ -38626,5 +38626,29 @@
       11
     ],
     "raw": ".(?!.){2,3}"
+  },
+  "^*": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Expected atom at position 1\n    ^*\n     ^",
+    "input": "^*"
+  },
+  "$+": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Expected atom at position 1\n    $+\n     ^",
+    "input": "$+"
+  },
+  "\\b?": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Expected atom at position 2\n    \\b?\n      ^",
+    "input": "\\b?"
+  },
+  "\\B{1}": {
+    "type": "error",
+    "name": "SyntaxError",
+    "message": "Expected atom at position 2\n    \\B{1}\n      ^",
+    "input": "\\B{1}"
   }
 }