Added possibility to parse yaml for value when encoding is not utf-8

Author: jeromemacias

lib/addon/sfData.class.php

@@ -53,7 +53,7 @@ public function getDeleteCurrentData()
   protected function doLoadDataFromFile($file)
   {
     // import new datas
-    $data = sfYaml::load($file);
+    $data = sfYaml::load($file, sfConfig::get('sf_charset', 'UTF-8'));
 
     $this->loadDataFromArray($data);
   }

lib/config/sfYamlConfigHandler.class.php

@@ -69,7 +69,7 @@ static public function parseYaml($configFile)
     }
 
     // parse our config
-    $config = sfYaml::load($configFile);
+    $config = sfYaml::load($configFile, sfConfig::get('sf_charset', 'UTF-8'));
 
     if ($config === false)
     {

lib/i18n/extract/sfI18nYamlGeneratorExtractor.class.php

@@ -29,7 +29,7 @@ public function extract($content)
   {
     $this->strings = array();
 
-    $config = sfYaml::load($content);
+    $config = sfYaml::load($content, sfConfig::get('sf_charset', 'UTF-8'));
 
     if (!isset($config['generator']['param']['config']))
     {
@@ -82,7 +82,7 @@ public function extract($content)
         }
       }
     }
-  
+
     return $this->strings;
   }
 

lib/i18n/extract/sfI18nYamlValidateExtractor.class.php

@@ -27,7 +27,7 @@ public function extract($content)
   {
     $strings = array();
 
-    $config = sfYaml::load($content);
+    $config = sfYaml::load($content, sfConfig::get('sf_charset', 'UTF-8'));
 
     // New validate.yml format
 

lib/yaml/sfYaml.php

@@ -64,7 +64,7 @@ static public function getSpecVersion()
    *
    * @throws InvalidArgumentException If the YAML is not valid
    */
-  public static function load($input)
+  public static function load($input, $encoding = 'UTF-8')
   {
     $file = '';
 
@@ -87,6 +87,14 @@ public static function load($input)
       return $input;
     }
 
+    $mbConvertEncoding = false;
+    $encoding = strtoupper($encoding);
+    if ('UTF-8' != $encoding && function_exists('mb_convert_encoding'))
+    {
+      $input = mb_convert_encoding($input, 'UTF-8', $encoding);
+      $mbConvertEncoding = true;
+    }
+
     require_once dirname(__FILE__).'/sfYamlParser.php';
 
     $yaml = new sfYamlParser();
@@ -100,6 +108,11 @@ public static function load($input)
       throw new InvalidArgumentException(sprintf('Unable to parse %s: %s', $file ? sprintf('file "%s"', $file) : 'string', $e->getMessage()));
     }
 
+    if ($ret && $mbConvertEncoding)
+    {
+      $ret = self::arrayConvertEncoding($ret, $encoding);
+    }
+
     return $ret;
   }
 
@@ -122,6 +135,25 @@ public static function dump($array, $inline = 2)
 
     return $yaml->dump($array, $inline);
   }
+
+  /**
+   * Converts all kayes and values from UTF-8 to given encoding
+   *
+   * @param  array  $result   Original result
+   * @param  string $encoding The expected encoding
+   * @param  array $convertedResult Converted result
+   * @return array
+   */
+  protected static function arrayConvertEncoding(array $result, $encoding, &$convertedResult = array())
+  {
+    foreach ($result as $key => $value)
+    {
+      $key = mb_convert_encoding($key, 'UTF-8', $encoding);
+      $convertedResult[$key] = is_array($value) ? self::arrayConvertEncoding($value, $encoding, $convertedArray) : mb_convert_encoding($value, $encoding, 'UTF-8');
+    }
+
+    return $convertedResult;
+  }
 }
 
 /**

lib/yaml/sfYamlParser.php

@@ -57,6 +57,11 @@ public function parse($value)
     $this->currentLine = '';
     $this->lines = explode("\n", $this->cleanup($value));
 
+    if (function_exists('mb_detect_encoding') && false === mb_detect_encoding($value, 'UTF-8', true))
+    {
+      throw new InvalidArgumentException('The YAML value does not appear to be valid UTF-8.');
+    }
+
     if (function_exists('mb_internal_encoding') && ((int) ini_get('mbstring.func_overload')) & 2)
     {
       $mbEncoding = mb_internal_encoding();