plugins/Oembed/lib/opengraphhelper.php

   1 <?php
   2
   3 if (!defined('GNUSOCIAL')) { exit(1); }
   4
   5
   6 /**
   7  * Utility class to get OpenGraph data from HTML DOMs etc.
   8  */
   9 class OpenGraphHelper
  10 {
  11     const KEY_REGEX = '/^og\:(\w+(?:\:\w+)?)/';
  12     protected static $property_map = [
  13                             'site_name'      => 'provider_name',
  14                             'title'          => 'title',
  15                             'description'    => 'html',
  16                             'type'           => 'type',
  17                             'url'            => 'url',
  18                             'image'          => 'thumbnail_url',
  19                             'image:height'   => 'thumbnail_height',
  20                             'image:width'    => 'thumbnail_width',
  21                             ];
  22
  23     // This regex map has:    /pattern(match)/ => matchindex | string
  24     protected static $type_regex_map = [
  25                             '/^(video)/'   => 1,
  26                             '/^image/'     => 'photo',
  27                             ];
  28
  29     static function ogFromHtml(DOMDocument $dom) {
  30         $obj = new stdClass();
  31         $obj->version = '1.0';  // fake it til u make it
  32
  33         $nodes = $dom->getElementsByTagName('meta');
  34         for ($i = 0; $i < $nodes->length; $i++) {
  35             $node = $nodes->item($i);
  36             if (!$node->hasAttributes()) {
  37                 continue;
  38             }
  39             $property = $node->attributes->getNamedItem('property');
  40             $matches = array();
  41             if ($property === null || !preg_match(self::KEY_REGEX, $property->value, $matches)) {
  42                 // not property="og:something"
  43                 continue;
  44             }
  45             if (!isset(self::$property_map[$matches[1]])) {
  46                 // unknown metadata property, nothing we would care about anyway
  47                 continue;
  48             }
  49
  50             $prop = self::$property_map[$matches[1]];
  51             $obj->{$prop} = $node->attributes->getNamedItem('content')->value;
  52             // I don't care right now if they're empty
  53         }
  54         if (isset($obj->type)) {
  55             // Loop through each known OpenGraph type where we have a match in oEmbed
  56             foreach (self::$type_regex_map as $pattern=>$replacement) {
  57                 $matches = array();
  58                 if (preg_match($pattern, $obj->type, $matches)) {
  59                     $obj->type = is_int($replacement)
  60                                     ? $matches[$replacement]
  61                                     : $replacement;
  62                     break;
  63                 }
  64             }
  65             // If it's not known to our type map, we just pass it through in hopes of it getting handled anyway
  66         } elseif (isset($obj->url)) {
  67             // If no type is set but we have a URL, let's set type=link
  68             $obj->type = 'link';
  69         }
  70         return $obj;
  71     }
  72 }