[[FIX]] Restrict IdentifierNames in ES5 code

jshint · May 27, 2018 · 5995a9f · 5995a9f
1 parent f2ce8fe
commit 5995a9f
Show file tree

Hide file tree

Showing 4 changed files with 61 additions and 4 deletions.
diff --git a/data/es5-identifier-names.js b/data/es5-identifier-names.js
diff --git a/scripts/generate-identifier-data.js b/scripts/generate-identifier-data.js
@@ -19,6 +19,18 @@ var ID_Start = get('Binary_Property/ID_Start');
 var ID_Continue = get('Binary_Property/ID_Continue');
 var Other_ID_Start = get('Binary_Property/Other_ID_Start');
 
+// Get the Unicode categories needed to construct the ES5 regex.
+var Lu = get('General_Category/Uppercase_Letter');
+var Ll = get('General_Category/Lowercase_Letter');
+var Lt = get('General_Category/Titlecase_Letter');
+var Lm = get('General_Category/Modifier_Letter');
+var Lo = get('General_Category/Other_Letter');
+var Nl = get('General_Category/Letter_Number');
+var Mn = get('General_Category/Nonspacing_Mark');
+var Mc = get('General_Category/Spacing_Mark');
+var Nd = get('General_Category/Decimal_Number');
+var Pc = get('General_Category/Connector_Punctuation');
+
 var generateData = function() { // ES 5.1
   // http://mathiasbynens.be/notes/javascript-identifiers#valid-identifier-names
   var identifierStart = regenerate(ID_Start)
@@ -40,6 +52,20 @@ var generateData = function() { // ES 5.1
   };
 };
 
+// Adapted from https://gist.github.com/mathiasbynens/6334847
+var generateES5Regex = function() { // ES 5.1
+  // https://mathiasbynens.be/notes/javascript-identifiers#valid-identifier-names
+  var identifierStart = regenerate('$', '_')
+    .add(Lu, Ll, Lt, Lm, Lo, Nl)
+    .removeRange(0x010000, 0x10FFFF); // Remove astral symbols.
+  var identifierPart = identifierStart.clone()
+    .add('\u200C', '\u200D', Mn, Mc, Nd, Pc)
+    .removeRange(0x010000, 0x10FFFF); // Remove astral symbols.
+
+  return '/^(?:' + identifierStart.toString() + ')' +
+    '(?:' + identifierPart.toString() + ')*$/';
+};
+
 var fs = require('fs');
 var writeFile = function(fileName, data) {
   fs.writeFileSync(
@@ -57,3 +83,7 @@ var writeFile = function(fileName, data) {
 var result = generateData();
 writeFile('./data/non-ascii-identifier-start.js', result.nonAsciiIdentifierStart);
 writeFile('./data/non-ascii-identifier-part-only.js', result.nonAsciiIdentifierPart);
+fs.writeFileSync(
+  './data/es5-identifier-names.js',
+  'module.exports = ' + generateES5Regex() + ';'
+);
diff --git a/src/lex.js b/src/lex.js
@@ -14,6 +14,8 @@ var asciiIdentifierStartTable = unicodeData.asciiIdentifierStartTable;
 var asciiIdentifierPartTable = unicodeData.asciiIdentifierPartTable;
 var nonAsciiIdentifierStartTable = require("../data/non-ascii-identifier-start.js");
 var nonAsciiIdentifierPartTable = require("../data/non-ascii-identifier-part-only.js");
+// Loading of this module is deferred as an optimization for ES2015 input
+var es5IdentifierNames;
 
 // Some of these token types are from JavaScript Parser API
 // while others are specific to JSHint parser.
@@ -582,10 +584,10 @@ Lexer.prototype = {
    * Extract a JavaScript identifier out of the next sequence of
    * characters or return 'null' if its not possible.
    */
-  scanIdentifier: function() {
+  scanIdentifier: function(checks) {
     var id = "";
     var index = 0;
-    var char;
+    var char, value;
 
     function isNonAsciiIdentifierStart(code) {
       return nonAsciiIdentifierStartTable.indexOf(code) > -1;
@@ -695,9 +697,29 @@ Lexer.prototype = {
       id += char;
     }
 
+    value = removeEscapeSequences(id)
+
+    if (!state.inES6(true)) {
+      es5IdentifierNames = require("../data/es5-identifier-names.js");
+
+      if (!es5IdentifierNames.test(value)) {
+        this.triggerAsync(
+          "warning",
+          {
+            code: "W119",
+            line: this.line,
+            character: this.char,
+            data: ["unicode 8", "6"]
+          },
+          checks,
+          function() { return true; }
+        );
+      }
+    }
+
     return {
       type: Token.Identifier,
-      value: removeEscapeSequences(id),
+      value: value,
       text: id,
       tokenLength: id.length
     };
@@ -1531,7 +1553,7 @@ Lexer.prototype = {
       this.scanRegExp(checks) ||
       this.scanPunctuator() ||
       this.scanKeyword() ||
-      this.scanIdentifier() ||
+      this.scanIdentifier(checks) ||
       this.scanNumericLiteral(checks);
 
     if (match) {

diff --git a/tests/unit/parser.js b/tests/unit/parser.js
@@ -8366,6 +8366,10 @@ exports["regression test for GH-3230"] = function (test) {
 };
 
 exports.unicode8 = function (test) {
+  TestRun(test)
+    .addError(1, 5, "'unicode 8' is only available in ES6 (use 'esversion: 6').")
+    .test("var Ϳ;", {esversion: 5});
+
   TestRun(test)
     .test("var Ϳ;", {esversion: 6});