updated with v10.0.0 of unicode.org confusables.txt file

Author: codebox

generator/main.py

@@ -3,6 +3,7 @@
 from output_char_codes import OutputCharCodes
 from output_chars import OutputChars
 from output_js import OutputJS
+from output_js_tests import OutputJSTests
 
 TEMPLATES_DIR = 'generator/templates'
 DATA_DIR      = 'generator/source_data'
@@ -18,3 +19,4 @@
     OutputChars('raw_data', TEMPLATES_DIR).create(cm)
     OutputJS('homoglyph.js', 'javascript/src', TEMPLATES_DIR).create(cm, CHARS)
     OutputJS('index.js','node', TEMPLATES_DIR).create(cm, CHARS)
+    OutputJSTests('DataTests.js','javascript/tests/js/tests', TEMPLATES_DIR).create(cm, CHARS)

generator/output_js_tests.py

@@ -0,0 +1,20 @@
+from output_builder import OutputBuilder
+
+class OutputJSTests(OutputBuilder):
+    def __init__(self, file_name, output_dir, template_dir):
+        self.file_name = file_name
+        OutputBuilder.__init__(self, output_dir, template_dir)
+
+    def create(self, char_manager, chars):
+        check_statements = []
+        for char in chars:
+            char_homoglyphs = char_manager.get_set_for_char(char)
+            char_homoglyphs_as_unicode = []
+            for char_homoglyph in char_homoglyphs:
+                char_homoglyphs_as_unicode.append('"\\u{' + '{:0>4}'.format(self._hex_code_for_char(char_homoglyph)) + '}"')
+
+            check_statements.append('    check("{}", [{}]);'.format(char, ' ,'.join(char_homoglyphs_as_unicode)))
+
+        text = self._get_template_text().replace('[[check_statements]]', '\n'.join(check_statements))
+
+        self._write_output(text)

generator/source_data/confusables.txt

@@ -0,0 +1,16 @@
+describe("Homoglyph Search - Data Tests", function () {
+    function check(targetChar, homoglyphs){
+        describe("Homoglyphs of '" + targetChar + "'", function(){
+            homoglyphs.forEach(function (c) {
+                it("Checking '" + c + "'", function () {
+                    var textContainingHomoglyph = 'xx' + c + 'xx',
+                        targetWord = 'xx' + targetChar + 'xx';
+                    expect(search(textContainingHomoglyph, [targetWord])).toEqual([{match: textContainingHomoglyph, word: targetWord, index: 0}]);
+                })
+            });
+        })
+    }
+
+[[check_statements]]
+
+});