Add Modified Hepburn romanji system.

README.md

@@ -8,6 +8,7 @@
   - Find dictionary entries (lemmas) for conjugated words
   - Get readings and pronunciations for words
   - Build fuirgana for words
+  - Convert Japanese to romanji (English lettering)
 
 ## Contents  
   - [Quick Guide](#quick-guide)
@@ -512,15 +513,24 @@ echo $results->plugin('Furigana'); // Output: <ruby>東京<rt>とうきょう</r
 
 ### Romanji
 
-The Romanji plugin converts words from Japanese to romanji (English letters).  Currently, only [traditional hepburn](https://en.wikipedia.org/wiki/Hepburn_romanization) romanization is available, but other options are coming soon.  
-
+The Romanji plugin converts words from Japanese to romanji (English letters).  A few different romanji systems are available and can be set in the config.php file in the 'style' key. 
+```php
+'Romanji' => [
+  'style' => 'hepburn_modified'
+]
+```
+
+The plugin supports the following romanji systems:  
+  - [Traditional Hepburn](https://en.wikipedia.org/wiki/Hepburn_romanization): traditional_hepburn
+  - [Modified Hepburn](https://en.wikipedia.org/wiki/Hepburn_romanization): modified_hepburn
+
 To get romanji for a string, parse it and access it on the LimelightResults object.
 ```php
 $limelight = new Limelight();
 
 $results = $limelight->parse('東京に行きます');
 
-echo $results->plugin('Romanji'); // Output: Toukyou ni ikimasu
+echo $results->plugin('Romanji'); // Output: Tōkyō ni ikimasu
 ```
 Strings on the LimelightResults object are space seperated.  
 
@@ -536,7 +546,7 @@ foreach ($results->getNext() as $word) {
 
 // Output
 //
-// Toukyouniikimasu
+// Tōkyōniikimasu
 ```
 
 Proper nouns are capitalized.  

composer.json

@@ -2,7 +2,7 @@
     "name": "nihongodera/limelight",
     "type": "project",
     "description": "A php Japanese language text analyzer and parser.",
-    "keywords": ["japanese", "language", "mecab", "furigana", "kanji"],
+    "keywords": ["japanese", "language", "mecab", "furigana", "kanji", "romanji", "parse"],
     "homepage": "https://github.com/zachleigh/limelight",
     "license": "MIT",
     "authors": [

limelight_console

@@ -10,17 +10,4 @@ $limelight = new Limelight();
 
 $results = $limelight->parse('東京に行きます');
 
-// $word = $results->getByIndex(0);
-
-// $romanji = $word->reading()->toRomanji()->get();
-
-// echo $romanji;
-
-// var_dump($results);
-
-foreach ($results->getNext() as $word) {
-    echo $word->romanji;
-}
-
-// echo $word->romanji;
-// 
+var_dump($results);

src/Plugins/Library/Romanji/Lib/HepburnModified.php → src/Plugins/Library/Romanji/Lib/Hepburn.php

@@ -55,5 +55,6 @@
         'うぁ'=>'wha', 'いぇ'=>'ye', 'うぉ'=>'who',
         'ぁ'=>'xa', 'ぃ'=>'xi', 'ぅ'=>'xu', 'ぇ'=>'xe', 'ぉ'=>'xo',
         'ゕ'=>'xka', 'ゖ'=>'xke', 'ゎ'=>'xwa',
-        '。'=>'.'
+        '。'=>'.', '、'=>',', '？'=>'?', '！'=>'!',
+        '「' =>'"', '」'=>'"'
 ];

src/Plugins/Library/Romanji/Lib/HepburnTraditional.php

@@ -1,4 +1,5 @@
 <?php
+
 return [
         'あ'=>'a', 'い'=>'i', 'う'=>'u', 'え'=>'e', 'お'=>'o',
         'か'=>'ka', 'き'=>'ki', 'く'=>'ku', 'け'=>'ke', 'こ'=>'ko',
@@ -43,5 +44,6 @@
         'うぁ'=>'wha', 'いぇ'=>'ye', 'うぉ'=>'who',
         'ぁ'=>'xa', 'ぃ'=>'xi', 'ぅ'=>'xu', 'ぇ'=>'xe', 'ぉ'=>'xo',
         'ゕ'=>'xka', 'ゖ'=>'xke', 'ゎ'=>'xwa',
-        '。'=>'.'
+        '。'=>'.', '、'=>',', '？'=>'?', '！'=>'!',
+        '「' =>'"', '」'=>'"'
 ];

src/Plugins/Library/Romanji/Romanji.php

@@ -4,7 +4,6 @@
 
 use Limelight\Limelight;
 use Limelight\Plugins\Plugin;
-use Limelight\Plugins\Library\Romanji\Styles\Hepburn;
 
 class Romanji extends Plugin
 {
@@ -15,18 +14,14 @@ class Romanji extends Plugin
      */
     public function handle()
     {
-        $options = $this->config->get('Romanji');
-
-        $styleClass = 'Limelight\\Plugins\\Library\\Romanji\\Styles\\' . ucfirst($this->underscoreToCamelCase($options['style']));
-
-        $style = new $styleClass();
+        $decorator = $this->makeDecoratorClass();
 
         $romanjiString = '';
 
         foreach ($this->words as $word) {
             $hiraganaWord = mb_convert_kana($word->reading, 'c');
 
-            $romanjiWord = $style->convert($hiraganaWord, $word);
+            $romanjiWord = $decorator->convert($hiraganaWord, $word);
 
             $word->setPluginData('Romanji', $romanjiWord);
 
@@ -37,17 +32,41 @@ public function handle()
             $romanjiString .= $romanjiWord;
         }
 
-        return ucfirst(trim($romanjiString));
+        $romanjiString = trim($romanjiString);
+
+        return $this->uppercaseFirst($romanjiString);
+    }
+
+    /**
+     * Make decorator class from config value.
+     *
+     * @return Limelight\Plugins\Library\Romanji\StyleDecorator
+     */
+    private function makeDecoratorClass()
+    {
+        $options = $this->config->get('Romanji');
+
+        $style = $this->underscoreToCamelCase($options['style']);
+
+        $decoratorClass = 'Limelight\\Plugins\\Library\\Romanji\\Styles\\'.ucfirst($style);
+
+        if (class_exists($decoratorClass)) {
+            $converter = new RomanjiConverter();
+
+            return new $decoratorClass($converter);
+        }
+
+        throw new LimelightPluginErrorException("Style {$style} does not exist.  Check config.php file.");
     }
 
     /**
      * Make an underscored word camel-case.
      *
-     * @param   string  $string
+     * @param string $string
      *
-     * @return  string
+     * @return string
      */
-    function underscoreToCamelCase($string)
+    public function underscoreToCamelCase($string)
     {
         $string = strtolower($string);
 
@@ -67,4 +86,20 @@ function underscoreToCamelCase($string)
 
         return $string;
     }
+
+    /**
+     * Multibyte safe ucfirst.
+     *
+     * @param string $string
+     *
+     * @return string
+     */
+    public function uppercaseFirst($string)
+    {
+        $firstChar = mb_substr($string, 0, 1);
+
+        $rest = mb_substr($string, 1);
+
+        return mb_convert_case($firstChar, MB_CASE_UPPER, 'UTF-8').$rest;
+    }
 }