simplepie/create.php

   1 <?php
   2
   3 require_once 'simplepie.inc';
   4
   5 function normalize_character_set($charset)
   6 {
   7         return strtolower(preg_replace('/(?:[^a-zA-Z0-9]+|([^0-9])0+)/', '\1', $charset));
   8 }
   9
  10 function build_character_set_list()
  11 {
  12         $file = new SimplePie_File('http://www.iana.org/assignments/character-sets');
  13         if (!$file->success && !($file->method & SIMPLEPIE_FILE_SOURCE_REMOTE === 0 || ($file->status_code === 200 || $file->status_code > 206 && $file->status_code < 300)))
  14         {
  15                 return false;
  16         }
  17         else
  18         {
  19                 $data = explode("\n", $file->body);
  20                 unset($file);
  21
  22                 foreach ($data as $line)
  23                 {
  24                         // New character set
  25                         if (substr($line, 0, 5) === 'Name:')
  26                         {
  27                                 // If we already have one, push it on to the array
  28                                 if (isset($aliases))
  29                                 {
  30                                         for ($i = 0, $count = count($aliases); $i < $count; $i++)
  31                                         {
  32                                                 $aliases[$i] = normalize_character_set($aliases[$i]);
  33                                         }
  34                                         $charsets[$preferred] = array_unique($aliases);
  35                                         natsort($charsets[$preferred]);
  36                                 }
  37
  38                                 $start = 5 + strspn($line, "\x09\x0A\x0B\xC\x0D\x20", 5);
  39                                 $chars = strcspn($line, "\x09\x0A\x0B\xC\x0D\x20", $start);
  40                                 $aliases = array(substr($line, $start, $chars));
  41                                 $preferred = end($aliases);
  42                         }
  43                         // Another alias
  44                         elseif(substr($line, 0, 6) === 'Alias:')
  45                         {
  46                                 $start = 7 + strspn($line, "\x09\x0A\x0B\xC\x0D\x20", 7);
  47                                 $chars = strcspn($line, "\x09\x0A\x0B\xC\x0D\x20", $start);
  48                                 $aliases[] = substr($line, $start, $chars);
  49
  50                                 if (end($aliases) === 'None')
  51                                 {
  52                                         array_pop($aliases);
  53                                 }
  54                                 elseif (substr($line, 7 + $chars + 1, 21) === '(preferred MIME name)')
  55                                 {
  56                                         $preferred = end($aliases);
  57                                 }
  58                         }
  59                 }
  60
  61                 // Compatibility replacements
  62                 $compat = array(
  63                         'EUC-KR' => 'windows-949',
  64                         'GB2312' => 'GBK',
  65                         'GB_2312-80' => 'GBK',
  66                         'ISO-8859-1' => 'windows-1252',
  67                         'ISO-8859-9' => 'windows-1254',
  68                         'ISO-8859-11' => 'windows-874',
  69                         'KS_C_5601-1987' => 'windows-949',
  70                         'TIS-620' => 'windows-874',
  71                         //'US-ASCII' => 'windows-1252',
  72                         'x-x-big5' => 'Big5',
  73                 );
  74
  75                 foreach ($compat as $real => $replace)
  76                 {
  77                         if (isset($charsets[$real]) && isset($charsets[$replace]))
  78                         {
  79                                 $charsets[$replace] = array_merge($charsets[$replace], $charsets[$real]);
  80                                 unset($charsets[$real]);
  81                         }
  82                         elseif (isset($charsets[$real]))
  83                         {
  84                                 $charsets[$replace] = $charsets[$real];
  85                                 $charsets[$replace][] = normalize_character_set($replace);
  86                                 unset($charsets[$real]);
  87                         }
  88                         else
  89                         {
  90                                 $charsets[$replace][] = normalize_character_set($real);
  91                         }
  92                         $charsets[$replace] = array_unique($charsets[$replace]);
  93                         natsort($charsets[$replace]);
  94                 }
  95
  96                 // Sort it
  97                 uksort($charsets, 'strnatcasecmp');
  98
  99                 // Check that nothing matches more than one
 100                 $all = call_user_func_array('array_merge', $charsets);
 101                 $all_count = array_count_values($all);
 102                 if (max($all_count) > 1)
 103                 {
 104                         echo "Duplicated charsets:\n";
 105                         foreach ($all_count as $charset => $count)
 106                         {
 107                                 if ($count > 1)
 108                                 {
 109                                         echo "$charset\n";
 110                                 }
 111                         }
 112                 }
 113
 114                 // And we're done!
 115                 return $charsets;
 116         }
 117 }
 118
 119 function charset($charset)
 120 {
 121         $normalized_charset = normalize_character_set($charset);
 122         if ($charsets = build_character_set_list())
 123         {
 124                 foreach ($charsets as $preferred => $aliases)
 125                 {
 126                         if (in_array($normalized_charset, $aliases))
 127                         {
 128                                 return $preferred;
 129                         }
 130                 }
 131                 return $charset;
 132         }
 133         else
 134         {
 135                 return false;
 136         }
 137 }
 138
 139 function build_function()
 140 {
 141         if ($charsets = build_character_set_list())
 142         {
 143                 $return = <<<EOF
 144 function charset(\$charset)
 145 {
 146         // Normalization from UTS #22
 147         switch (strtolower(preg_replace('/(?:[^a-zA-Z0-9]+|([^0-9])0+)/', '\\1', \$charset)))
 148         {
 149
 150 EOF;
 151                 foreach ($charsets as $preferred => $aliases)
 152                 {
 153                         foreach ($aliases as $alias)
 154                         {
 155                                 $return .= "\t\tcase " . var_export($alias, true) . ":\n";
 156                         }
 157                         $return .= "\t\t\treturn " . var_export($preferred, true) . ";\n\n";
 158                 }
 159                 $return .= <<<EOF
 160                 default:
 161                         return \$charset;
 162         }
 163 }
 164 EOF;
 165                 return $return;
 166         }
 167         else
 168         {
 169                 return false;
 170         }
 171 }
 172
 173 if (php_sapi_name() === 'cli' && realpath($_SERVER['argv'][0]) === __FILE__)
 174 {
 175         echo build_function();
 176 }
 177
 178 ?>