requireCapitalizedComments: correct letter recognition

Fixes #842
Closes gh-847

Author: AlexanderZeilmann

lib/rules/disallow-capitalized-comments.js

@@ -1,4 +1,13 @@
 var assert = require('assert');
+var regenerate = require('regenerate');
+var Lu = require('unicode-6.3.0/categories/Lu/code-points');
+var Ll = require('unicode-6.3.0/categories/Ll/code-points');
+var Lt = require('unicode-6.3.0/categories/Lt/code-points');
+var Lm = require('unicode-6.3.0/categories/Lm/code-points');
+var Lo = require('unicode-6.3.0/categories/Lo/code-points');
+
+var letterPattern = regenerate(Lu, Ll, Lt, Lm, Lo).toRegExp();
+var lowerCasePattern = regenerate(Ll).toRegExp();
 
 module.exports = function() {};
 
@@ -15,14 +24,11 @@ module.exports.prototype = {
     },
 
     check: function(file, errors) {
-        var lowerCasePattern = /[a-z]/;
-        var alphabeticalPattern = /[a-z|A-Z]/;
-
         file.getComments().forEach(function(comment) {
             var stripped = comment.value.replace(/[\n\s\*]/g, '');
             var firstChar = stripped[0];
 
-            if (alphabeticalPattern.test(firstChar) && !lowerCasePattern.test(firstChar)) {
+            if (letterPattern.test(firstChar) && !lowerCasePattern.test(firstChar)) {
                 errors.add(
                     'Comments must start with a lowercase letter',
                     comment.loc.start

lib/rules/require-capitalized-comments.js

@@ -1,7 +1,13 @@
 var assert = require('assert');
+var regenerate = require('regenerate');
+var Lu = require('unicode-6.3.0/categories/Lu/code-points');
+var Ll = require('unicode-6.3.0/categories/Ll/code-points');
+var Lt = require('unicode-6.3.0/categories/Lt/code-points');
+var Lm = require('unicode-6.3.0/categories/Lm/code-points');
+var Lo = require('unicode-6.3.0/categories/Lo/code-points');
 
-var upperCasePattern = /[A-Z]/;
-var alphabeticalPattern = /[a-z|A-Z]/;
+var letterPattern = regenerate(Lu, Ll, Lt, Lm, Lo).toRegExp();
+var upperCasePattern = regenerate(Lu).toRegExp();
 
 module.exports = function() {};
 
@@ -23,7 +29,7 @@ module.exports.prototype = {
         file.getComments().forEach(function(comment) {
             var stripped = comment.value.replace(/[\n\s\*]/g, '');
             var firstChar = stripped[0];
-            var isLetter = firstChar && alphabeticalPattern.test(firstChar);
+            var isLetter = firstChar && letterPattern.test(firstChar);
 
             if (!isLetter) {
                 inTextBlock = false;

package.json

@@ -39,7 +39,9 @@
     "vow": "~0.4.3",
     "vow-fs": "~0.3.1",
     "xmlbuilder": "~2.4.0",
-    "supports-color": "~1.2.0"
+    "supports-color": "~1.2.0",
+    "unicode-6.3.0": "~0.1.5",
+    "regenerate": "~1.0.1"
   },
   "devDependencies": {
     "browserify": "~6.2.0",

test/rules/disallow-capitalized-comments.js

@@ -17,12 +17,16 @@ describe('rules/disallow-capitalized-comments', function() {
         assert(checker.checkString('/**\n * Invalid\n */').getErrorCount() === 1);
         assert(checker.checkString('/* Invalid */').getErrorCount() === 1);
         assert(checker.checkString('/*\n Invalid\n */').getErrorCount() === 1);
+        assert(checker.checkString('//Über').getErrorCount() === 1);
+        assert(checker.checkString('//Π').getErrorCount() === 1);
     });
 
     it('should not report on a lowercase start of a comment', function() {
         assert(checker.checkString('//valid').isEmpty());
         assert(checker.checkString('// valid').isEmpty());
         assert(checker.checkString('/** valid */').isEmpty());
+        assert(checker.checkString('//über').isEmpty());
+        assert(checker.checkString('//π').isEmpty());
     });
 
     it('should not report on comments that start with a non-alphabetical character', function() {

test/rules/require-capitalized-comments.js

@@ -22,6 +22,8 @@ describe('rules/require-capitalized-comments', function() {
         assert(checker.checkString('/**\n * invalid\n*/').getErrorCount() === 1);
         assert(checker.checkString('//\n// invalid\n//').getErrorCount() === 1);
         assert(checker.checkString('/*\ninvalid\n*/').getErrorCount() === 1);
+        assert(checker.checkString('//über').getErrorCount() === 1);
+        assert(checker.checkString('//π').getErrorCount() === 1);
     });
 
     it('should not report an uppercase start of a comment', function() {
@@ -30,6 +32,8 @@ describe('rules/require-capitalized-comments', function() {
         assertEmpty('/** Valid */');
         assertEmpty('//\n// Valid\n//');
         assertEmpty('/*\nValid\n*/');
+        assertEmpty('//Über');
+        assertEmpty('//∏');
     });
 
     it('should not report on comments that start with a non-alphabetical character', function() {
@@ -89,5 +93,10 @@ describe('rules/require-capitalized-comments', function() {
             '// and has one or more non-capitalized lines',
             '// afterwards'
         ].join('\n'));
+
+        assert(checker.checkString([
+            '// Über',
+            '// diese Funktion'
+        ].join('\n')).isEmpty());
     });
 });