PHP Encoding::is_utf8方法代码示例

 /**
  * Read the next object from the resource.  In this case, the 'row' is the
  * row in the csv file, and 'line' returns the line number in the file that
  * the object STARTED on.  (CSV objects may span multiple lines, if they
  * contain newlines).
  *
  * @return array|false
  */
 protected function read_object()
 {
     $obj = fgetcsv($this->csv_fp, null, $this->csv_delim, $this->csv_encl);
     $num = $this->line_num;
     if (!$obj) {
         return false;
     } else {
         // optionally remove the last, probably blank, column
         if (count($obj) && $this->lastcol_skip) {
             array_pop($obj);
         }
         // Increment line number, including any newlines in the object.
         // Also check for UTF-8 encoding
         $this->line_num++;
         foreach ($obj as $i => $col) {
             $this->line_num += substr_count($col, "\n");
             if (!Encoding::is_utf8($col)) {
                 $obj[$i] = Encoding::convert_to_utf8($col);
             }
         }
         return array('row' => $this->position + ($this->csv_skip ? 2 : 1), 'data' => $obj, 'line' => $num);
     }
 }

/**
 *
 *
 * @param unknown $model
 */
function air2_model_prevalidate(&$model)
{
    $table = $model->getTable();
    $col_defs = $table->getColumns();
    $now = air2_date();
    foreach ($col_defs as $col => $def) {
        if (preg_match('/_uuid$/', $col) && isset($def['unique']) && $def['unique']) {
            if (!isset($model->{$col}) || !strlen($model->{$col})) {
                $model->{$col} = air2_generate_uuid(12);
            }
        } elseif (preg_match('/_cre_dtim$/', $col)) {
            if (!$model->exists() && !strlen($model->{$col})) {
                $model->{$col} = $now;
            }
        } elseif (preg_match('/_upd_dtim$/', $col)) {
            $model->{$col} = $now;
        } elseif (preg_match('/_cre_user$/', $col)) {
            if (!$model->exists() && !strlen($model->{$col})) {
                if (defined('AIR2_REMOTE_USER_ID')) {
                    $model->{$col} = AIR2_REMOTE_USER_ID;
                } else {
                    $model->{$col} = 1;
                    //sysuser -- TODO: check for shell users
                }
            }
        } elseif (preg_match('/_upd_user$/', $col)) {
            if (defined('AIR2_REMOTE_USER_ID')) {
                $model->{$col} = AIR2_REMOTE_USER_ID;
            } else {
                $model->{$col} = 1;
                //sysuser -- TODO: check for shell users
            }
        } elseif ($def['type'] === 'timestamp') {
            // validate the string for non cre/upd timestamps
            if (isset($model->{$col}) && !is_null($model->{$col})) {
                //NOTE: whenever you set a column in Doctrine, it checks that
                //the value has changed.  Since '2010-01-01 01:01:01' is equal
                //to '2010-01-01T01:01:01', we need to NULL it first
                if (strpos($model->{$col}, 'T') !== false) {
                    $airformat = air2_date(strtotime($model->{$col}));
                    $model->{$col} = null;
                    $model->{$col} = $airformat;
                }
            }
        } elseif ($def['type'] === 'date') {
            // validate the string for non cre/upd timestamps
            if (isset($model->{$col}) && !is_null($model->{$col})) {
                //NOTE: whenever you set a column in Doctrine, it checks that
                //the value has changed.  Since '2010-01-01 01:01:01' is equal
                //to '2010-01-01T01:01:01', we need to NULL it first
                if (strpos($model->{$col}, 'T') !== false) {
                    $airformat = air2_date(strtotime($model->{$col}));
                    $model->{$col} = null;
                    $model->{$col} = $airformat;
                }
            }
        }
        // default values
        if (strlen($table->getDefaultValueOf($col)) && !strlen($model->{$col})) {
            $model->{$col} = $table->getDefaultValueOf($col);
        }
        // lastly, check strings for valid UTF8 chars
        if ($def['type'] === 'string' && !Encoding::is_utf8($model->{$col})) {
            $modelname = get_class($model);
            throw new Doctrine_Exception("non-UTF8 string found in {$modelname}->{$col}: " . $model->{$col});
        }
    }
}

 /**
  * Convert non-UTF8 strings to UTF8.  Strings already in UTF8 will be
  * returned intact.
  *
  * @param string  $str
  * @return string
  */
 public static function convert_to_utf8($str)
 {
     if (Encoding::is_ascii($str)) {
         // valid! nothing to do
     } elseif (Encoding::is_utf8($str)) {
         // valid! nothing to do
     } elseif (Encoding::is_cp1252($str)) {
         $str = utf8_encode($str);
         $str = Encoding::fix_cp1252_codepoints_in_utf8($str);
     } elseif (Encoding::is_latin1($str)) {
         $str = utf8_encode($str);
     }
     return $str;
 }