Update removeNonWord pattern to cover more chars. closes mout#156. cl…

…oses mout#163

_build/pattern-removeNonWord.js

@@ -0,0 +1,19 @@
+// Run this script with Node to generate the pattern that is included in
+// src/string/removeNonWord.js.
+//
+//     node _build/pattern-removeNonWord
+//
+
+var regenerate = require('regenerate');
+
+var pattern = regenerate()
+    .addRange('a', 'z')
+    .addRange('A', 'Z')
+    .addRange('0', '9')
+    .addRange(0xC0, 0xFF)
+    .add(' ', '-', '_')
+    .remove(0xF7, 0xD7)
+    .toString();
+
+console.log('// Pattern for string/removeNonWord:');
+console.log('/' + pattern + '/g');

package.json

@@ -51,7 +51,8 @@
     "commander": "~1.0.5",
     "rocambole": "~0.2.3",
     "jshint": "2.x",
-    "rimraf": "~2.2.2"
+    "rimraf": "~2.2.2",
+    "regenerate": "~0.5.4"
   },
   "testling": {
     "preprocess": "node build testling",
@@ -95,4 +96,4 @@
       "tests/runner.js"
     ]
   }
-}
+}

src/string/removeNonWord.js

@@ -1,10 +1,13 @@
 define(['../lang/toString'], function(toString){
+    // This pattern is generated by the _build/pattern-removeNonWord.js script
+    var PATTERN = /[^\x20\x2D0-9A-Z\x5Fa-z\xC0-\xD6\xD8-\xF6\xF8-\xFF]/g;
+
     /**
      * Remove non-word chars.
      */
     function removeNonWord(str){
         str = toString(str);
-        return str.replace(/[^0-9a-zA-Z\xC0-\xFF \-_]/g, '');
+        return str.replace(PATTERN, '');
     }
 
     return removeNonWord;

tests/spec/string/spec-removeNonWord.js

@@ -3,7 +3,7 @@ define(['mout/string/removeNonWord'], function (removeNonWord) {
     describe('string/removeNonWord()', function(){
 
         it('should remove non word chars', function(){
-            var str = 'lorem ~!@#$%^&*()_+`-={}[]|\\:";\'/?><., ipsum';
+            var str = 'lorem ~!@#$%^&*()_+`-={}[]|\\:";\'/?><., ipsum\xD7';
             expect( removeNonWord(str) ).toEqual('lorem _- ipsum');
         });