SimplePie_Content_Type_Sniffer

Content-type sniffing.

Defined (1)

The class is defined in the following location(s).

/wp-includes/SimplePie/Content/Type/Sniffer.php  
  1. class SimplePie_Content_Type_Sniffer 
  2. { 
  3.     /** 
  4.      * File object 
  5.      * 
  6.      * @var SimplePie_File 
  7.      */ 
  8.     var $file; 
  9.  
  10.     /** 
  11.      * Create an instance of the class with the input file 
  12.      * 
  13.      * @param SimplePie_Content_Type_Sniffer $file Input file 
  14.      */ 
  15.     public function __construct($file) 
  16.     { 
  17.         $this->file = $file; 
  18.     } 
  19.  
  20.     /** 
  21.      * Get the Content-Type of the specified file 
  22.      * 
  23.      * @return string Actual Content-Type 
  24.      */ 
  25.     public function get_type() 
  26.     { 
  27.         if (isset($this->file->headers['content-type'])) 
  28.         { 
  29.             if (!isset($this->file->headers['content-encoding']) 
  30.                 && ($this->file->headers['content-type'] === 'text/plain' 
  31.                     || $this->file->headers['content-type'] === 'text/plain; charset=ISO-8859-1' 
  32.                     || $this->file->headers['content-type'] === 'text/plain; charset=iso-8859-1' 
  33.                     || $this->file->headers['content-type'] === 'text/plain; charset=UTF-8')) 
  34.             { 
  35.                 return $this->text_or_binary(); 
  36.             } 
  37.  
  38.             if (($pos = strpos($this->file->headers['content-type'], ';')) !== false) 
  39.             { 
  40.                 $official = substr($this->file->headers['content-type'], 0, $pos); 
  41.             } 
  42.             else 
  43.             { 
  44.                 $official = $this->file->headers['content-type']; 
  45.             } 
  46.             $official = trim(strtolower($official)); 
  47.  
  48.             if ($official === 'unknown/unknown' 
  49.                 || $official === 'application/unknown') 
  50.             { 
  51.                 return $this->unknown(); 
  52.             } 
  53.             elseif (substr($official, -4) === '+xml' 
  54.                 || $official === 'text/xml' 
  55.                 || $official === 'application/xml') 
  56.             { 
  57.                 return $official; 
  58.             } 
  59.             elseif (substr($official, 0, 6) === 'image/') 
  60.             { 
  61.                 if ($return = $this->image()) 
  62.                 { 
  63.                     return $return; 
  64.                 } 
  65.                 else 
  66.                 { 
  67.                     return $official; 
  68.                 } 
  69.             } 
  70.             elseif ($official === 'text/html') 
  71.             { 
  72.                 return $this->feed_or_html(); 
  73.             } 
  74.             else 
  75.             { 
  76.                 return $official; 
  77.             } 
  78.         } 
  79.         else 
  80.         { 
  81.             return $this->unknown(); 
  82.         } 
  83.     } 
  84.  
  85.     /** 
  86.      * Sniff text or binary 
  87.      * 
  88.      * @return string Actual Content-Type 
  89.      */ 
  90.     public function text_or_binary() 
  91.     { 
  92.         if (substr($this->file->body, 0, 2) === "\xFE\xFF" 
  93.             || substr($this->file->body, 0, 2) === "\xFF\xFE" 
  94.             || substr($this->file->body, 0, 4) === "\x00\x00\xFE\xFF" 
  95.             || substr($this->file->body, 0, 3) === "\xEF\xBB\xBF") 
  96.         { 
  97.             return 'text/plain'; 
  98.         } 
  99.         elseif (preg_match('/[\x00-\x08\x0E-\x1A\x1C-\x1F]/', $this->file->body)) 
  100.         { 
  101.             return 'application/octect-stream'; 
  102.         } 
  103.         else 
  104.         { 
  105.             return 'text/plain'; 
  106.         } 
  107.     } 
  108.  
  109.     /** 
  110.      * Sniff unknown 
  111.      * 
  112.      * @return string Actual Content-Type 
  113.      */ 
  114.     public function unknown() 
  115.     { 
  116.         $ws = strspn($this->file->body, "\x09\x0A\x0B\x0C\x0D\x20"); 
  117.         if (strtolower(substr($this->file->body, $ws, 14)) === '<!doctype html' 
  118.             || strtolower(substr($this->file->body, $ws, 5)) === '<html' 
  119.             || strtolower(substr($this->file->body, $ws, 7)) === '<script') 
  120.         { 
  121.             return 'text/html'; 
  122.         } 
  123.         elseif (substr($this->file->body, 0, 5) === '%PDF-') 
  124.         { 
  125.             return 'application/pdf'; 
  126.         } 
  127.         elseif (substr($this->file->body, 0, 11) === '%!PS-Adobe-') 
  128.         { 
  129.             return 'application/postscript'; 
  130.         } 
  131.         elseif (substr($this->file->body, 0, 6) === 'GIF87a' 
  132.             || substr($this->file->body, 0, 6) === 'GIF89a') 
  133.         { 
  134.             return 'image/gif'; 
  135.         } 
  136.         elseif (substr($this->file->body, 0, 8) === "\x89\x50\x4E\x47\x0D\x0A\x1A\x0A") 
  137.         { 
  138.             return 'image/png'; 
  139.         } 
  140.         elseif (substr($this->file->body, 0, 3) === "\xFF\xD8\xFF") 
  141.         { 
  142.             return 'image/jpeg'; 
  143.         } 
  144.         elseif (substr($this->file->body, 0, 2) === "\x42\x4D") 
  145.         { 
  146.             return 'image/bmp'; 
  147.         } 
  148.         elseif (substr($this->file->body, 0, 4) === "\x00\x00\x01\x00") 
  149.         { 
  150.             return 'image/vnd.microsoft.icon'; 
  151.         } 
  152.         else 
  153.         { 
  154.             return $this->text_or_binary(); 
  155.         } 
  156.     } 
  157.  
  158.     /** 
  159.      * Sniff images 
  160.      * 
  161.      * @return string Actual Content-Type 
  162.      */ 
  163.     public function image() 
  164.     { 
  165.         if (substr($this->file->body, 0, 6) === 'GIF87a' 
  166.             || substr($this->file->body, 0, 6) === 'GIF89a') 
  167.         { 
  168.             return 'image/gif'; 
  169.         } 
  170.         elseif (substr($this->file->body, 0, 8) === "\x89\x50\x4E\x47\x0D\x0A\x1A\x0A") 
  171.         { 
  172.             return 'image/png'; 
  173.         } 
  174.         elseif (substr($this->file->body, 0, 3) === "\xFF\xD8\xFF") 
  175.         { 
  176.             return 'image/jpeg'; 
  177.         } 
  178.         elseif (substr($this->file->body, 0, 2) === "\x42\x4D") 
  179.         { 
  180.             return 'image/bmp'; 
  181.         } 
  182.         elseif (substr($this->file->body, 0, 4) === "\x00\x00\x01\x00") 
  183.         { 
  184.             return 'image/vnd.microsoft.icon'; 
  185.         } 
  186.         else 
  187.         { 
  188.             return false; 
  189.         } 
  190.     } 
  191.  
  192.     /** 
  193.      * Sniff HTML 
  194.      * 
  195.      * @return string Actual Content-Type 
  196.      */ 
  197.     public function feed_or_html() 
  198.     { 
  199.         $len = strlen($this->file->body); 
  200.         $pos = strspn($this->file->body, "\x09\x0A\x0D\x20"); 
  201.  
  202.         while ($pos < $len) 
  203.         { 
  204.             switch ($this->file->body[$pos]) 
  205.             { 
  206.                 case "\x09": 
  207.                 case "\x0A": 
  208.                 case "\x0D": 
  209.                 case "\x20": 
  210.                     $pos += strspn($this->file->body, "\x09\x0A\x0D\x20", $pos); 
  211.                     continue 2; 
  212.  
  213.                 case '<': 
  214.                     $pos++; 
  215.                     break; 
  216.  
  217.                 default: 
  218.                     return 'text/html'; 
  219.             } 
  220.  
  221.             if (substr($this->file->body, $pos, 3) === '!--') 
  222.             { 
  223.                 $pos += 3; 
  224.                 if ($pos < $len && ($pos = strpos($this->file->body, '-->', $pos)) !== false) 
  225.                 { 
  226.                     $pos += 3; 
  227.                 } 
  228.                 else 
  229.                 { 
  230.                     return 'text/html'; 
  231.                 } 
  232.             } 
  233.             elseif (substr($this->file->body, $pos, 1) === '!') 
  234.             { 
  235.                 if ($pos < $len && ($pos = strpos($this->file->body, '>', $pos)) !== false) 
  236.                 { 
  237.                     $pos++; 
  238.                 } 
  239.                 else 
  240.                 { 
  241.                     return 'text/html'; 
  242.                 } 
  243.             } 
  244.             elseif (substr($this->file->body, $pos, 1) === '?') 
  245.             { 
  246.                 if ($pos < $len && ($pos = strpos($this->file->body, '?>', $pos)) !== false) 
  247.                 { 
  248.                     $pos += 2; 
  249.                 } 
  250.                 else 
  251.                 { 
  252.                     return 'text/html'; 
  253.                 } 
  254.             } 
  255.             elseif (substr($this->file->body, $pos, 3) === 'rss' 
  256.                 || substr($this->file->body, $pos, 7) === 'rdf:RDF') 
  257.             { 
  258.                 return 'application/rss+xml'; 
  259.             } 
  260.             elseif (substr($this->file->body, $pos, 4) === 'feed') 
  261.             { 
  262.                 return 'application/atom+xml'; 
  263.             } 
  264.             else 
  265.             { 
  266.                 return 'text/html'; 
  267.             } 
  268.         } 
  269.  
  270.         return 'text/html'; 
  271.     } 
  272. }