Invalid Identifier by Unicode escape

mozilla · Aug 27, 2023 · 90ac409 · 90ac409
1 parent 86a01e1
commit 90ac409
Show file tree

Hide file tree

Showing 2 changed files with 50 additions and 20 deletions.
diff --git a/src/org/mozilla/javascript/TokenStream.java b/src/org/mozilla/javascript/TokenStream.java
@@ -576,6 +576,25 @@ private static int stringToKeywordForES(String name, boolean isStrict) {
         return id & 0xff;
     }
 
+    private static boolean isValidIdentifierName(String str) {
+        int i = 0;
+        for (int c : str.codePoints().toArray()) {
+            if (i++ == 0) {
+                if (c != '$' && c != '_' && !Character.isUnicodeIdentifierStart(c)) {
+                    return false;
+                }
+            } else {
+                if (c != '$'
+                        && c != '\u200c'
+                        && c != '\u200d'
+                        && !Character.isUnicodeIdentifierPart(c)) {
+                    return false;
+                }
+            }
+        }
+        return true;
+    }
+
     final String getSourceString() {
         return sourceString;
     }
@@ -779,6 +798,14 @@ final int getToken() throws IOException {
                     // we convert the last character back to unicode
                     str = convertLastCharToHex(str);
                 }
+
+                if (containsEscape
+                        && parser.compilerEnv.getLanguageVersion() >= Context.VERSION_ES6
+                        && !isValidIdentifierName(str)) {
+                    parser.reportError("msg.invalid.escape");
+                    return Token.ERROR;
+                }
+
                 this.string = (String) allStrings.intern(str);
                 return Token.NAME;
             }

diff --git a/testsrc/test262.properties b/testsrc/test262.properties
@@ -5067,31 +5067,48 @@ language/global-code 29/41 (70.73%)
 
 language/identifier-resolution 0/13 (0.0%)
 
-language/identifiers 28/188 (14.89%)
+language/identifiers 45/188 (23.94%)
     other_id_continue.js
+    other_id_continue-escaped.js
     other_id_start.js
+    other_id_start-escaped.js
     part-unicode-10.0.0.js
+    part-unicode-10.0.0-escaped.js
     part-unicode-11.0.0.js
+    part-unicode-11.0.0-escaped.js
     part-unicode-12.0.0.js
+    part-unicode-12.0.0-escaped.js
     part-unicode-13.0.0.js
+    part-unicode-13.0.0-escaped.js
     part-unicode-5.2.0.js
+    part-unicode-5.2.0-escaped.js
     part-unicode-6.0.0.js
     part-unicode-6.1.0.js
     part-unicode-7.0.0.js
+    part-unicode-7.0.0-escaped.js
     part-unicode-8.0.0.js
+    part-unicode-8.0.0-escaped.js
     part-unicode-9.0.0.js
+    part-unicode-9.0.0-escaped.js
     start-unicode-10.0.0.js
+    start-unicode-10.0.0-escaped.js
     start-unicode-11.0.0.js
+    start-unicode-11.0.0-escaped.js
     start-unicode-12.0.0.js
+    start-unicode-12.0.0-escaped.js
     start-unicode-13.0.0.js
+    start-unicode-13.0.0-escaped.js
     start-unicode-5.2.0.js
+    start-unicode-5.2.0-escaped.js
     start-unicode-6.0.0.js
     start-unicode-6.1.0.js
+    start-unicode-6.1.0-escaped.js
     start-unicode-7.0.0.js
+    start-unicode-7.0.0-escaped.js
     start-unicode-8.0.0.js
+    start-unicode-8.0.0-escaped.js
     start-unicode-9.0.0.js
-    start-zwj-escaped.js
-    start-zwnj-escaped.js
+    start-unicode-9.0.0-escaped.js
     vertical-tilde-continue.js
     vertical-tilde-continue-escaped.js
     vertical-tilde-start.js
@@ -5101,11 +5118,7 @@ language/identifiers 28/188 (14.89%)
 
 language/keywords 0/25 (0.0%)
 
-language/line-terminators 4/41 (9.76%)
-    S7.3_A6_T1.js
-    S7.3_A6_T2.js
-    S7.3_A6_T3.js
-    S7.3_A6_T4.js
+language/line-terminators 0/41 (0.0%)
 
 language/literals 96/434 (22.12%)
     bigint/numeric-separators/numeric-separator-literal-nonoctal-08-err.js non-strict
@@ -5152,12 +5165,7 @@ language/literals 96/434 (22.12%)
 
 ~language/module-code
 
-language/punctuators 5/11 (45.45%)
-    S7.7_A2_T1.js
-    S7.7_A2_T2.js
-    S7.7_A2_T3.js
-    S7.7_A2_T4.js
-    S7.7_A2_T5.js
+language/punctuators 0/11 (0.0%)
 
 language/reserved-words 2/27 (7.41%)
     await-module.js {unsupported: [module]}
@@ -6301,11 +6309,6 @@ language/types 9/113 (7.96%)
     undefined/S8.1_A3_T1.js
     undefined/S8.1_A3_T2.js non-strict
 
-language/white-space 7/42 (16.67%)
+language/white-space 2/42 (4.76%)
     mongolian-vowel-separator.js {unsupported: [u180e]}
     mongolian-vowel-separator-eval.js {unsupported: [u180e]}
-    S7.2_A5_T1.js
-    S7.2_A5_T2.js
-    S7.2_A5_T3.js
-    S7.2_A5_T4.js
-    S7.2_A5_T5.js