classes/feedparser.php

   1 <?php
   2 class FeedParser {
   3         private $doc;
   4         private $error;
   5         private $items;
   6         private $link;
   7         private $title;
   8         private $type;
   9         private $xpath;
  10
  11         const FEED_RDF = 0;
  12         const FEED_RSS = 1;
  13         const FEED_ATOM = 2;
  14
  15         function __construct($data) {
  16                 libxml_use_internal_errors(true);
  17                 libxml_clear_errors();
  18                 $this->doc = new DOMDocument();
  19                 $this->doc->loadXML($data);
  20                 $this->error = $this->format_error(libxml_get_last_error());
  21                 libxml_clear_errors();
  22
  23                 $this->items = array();
  24         }
  25
  26         function init() {
  27                 $root = $this->doc->firstChild;
  28                 $xpath = new DOMXPath($this->doc);
  29                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
  30                 $xpath->registerNamespace('atom03', 'http://purl.org/atom/ns#');
  31                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
  32                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
  33                 $xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/');
  34                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
  35                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
  36
  37                 $this->xpath = $xpath;
  38
  39                 $root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)")->item(0);
  40
  41                 if ($root) {
  42                         switch (mb_strtolower($root->tagName)) {
  43                         case "rdf:rdf":
  44                                 $this->type = $this::FEED_RDF;
  45                                 break;
  46                         case "channel":
  47                                 $this->type = $this::FEED_RSS;
  48                                 break;
  49                         case "feed":
  50                                 $this->type = $this::FEED_ATOM;
  51                                 break;
  52                         default:
  53                                 if( !isset($this->error) ){
  54                                         $this->error = "Unknown/unsupported feed type";
  55                                 }
  56                                 return;
  57                         }
  58
  59                         switch ($this->type) {
  60                         case $this::FEED_ATOM:
  61
  62                                 $title = $xpath->query("//atom:feed/atom:title")->item(0);
  63
  64                                 if (!$title)
  65                                         $title = $xpath->query("//atom03:feed/atom03:title")->item(0);
  66
  67
  68                                 if ($title) {
  69                                         $this->title = $title->nodeValue;
  70                                 }
  71
  72                                 $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
  73
  74                                 if (!$link)
  75                                         $link = $xpath->query("//atom03:feed/atom03:link[not(@rel)]")->item(0);
  76
  77
  78                                 if ($link && $link->hasAttributes()) {
  79                                         $this->link = $link->getAttribute("href");
  80                                 }
  81
  82                                 $articles = $xpath->query("//atom:entry");
  83
  84                                 if (!$articles || $articles->length == 0)
  85                                         $articles = $xpath->query("//atom03:entry");
  86
  87                                 foreach ($articles as $article) {
  88                                         array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath));
  89                                 }
  90
  91                                 break;
  92                         case $this::FEED_RSS:
  93
  94                                 $title = $xpath->query("//channel/title")->item(0);
  95
  96                                 if ($title) {
  97                                         $this->title = $title->nodeValue;
  98                                 }
  99
 100                                 $link = $xpath->query("//channel/link")->item(0);
 101
 102                                 if ($link && $link->hasAttributes()) {
 103                                         $this->link = $link->getAttribute("href");
 104                                 }
 105
 106                                 $articles = $xpath->query("//channel/item");
 107
 108                                 foreach ($articles as $article) {
 109                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 110                                 }
 111
 112                                 break;
 113                         case $this::FEED_RDF:
 114                                 $xpath->registerNamespace('rssfake', 'http://purl.org/rss/1.0/');
 115
 116                                 $title = $xpath->query("//rssfake:channel/rssfake:title")->item(0);
 117
 118                                 if ($title) {
 119                                         $this->title = $title->nodeValue;
 120                                 }
 121
 122                                 $link = $xpath->query("//rssfake:channel/rssfake:link")->item(0);
 123
 124                                 if ($link) {
 125                                         $this->link = $link->nodeValue;
 126                                 }
 127
 128                                 $articles = $xpath->query("//rssfake:item");
 129
 130                                 foreach ($articles as $article) {
 131                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 132                                 }
 133
 134                                 break;
 135
 136                         }
 137                 } else {
 138                         if( !isset($this->error) ){
 139                                 $this->error = "Unknown/unsupported feed type";
 140                         }
 141                         return;
 142                 }
 143         }
 144
 145         function format_error($error) {
 146                 if ($error) {
 147                         return sprintf("LibXML error %s at line %d (column %d): %s",
 148                                 $error->code, $error->line, $error->column,
 149                                 $error->message);
 150                 } else {
 151                         return "";
 152                 }
 153         }
 154
 155         function error() {
 156                 return $this->error;
 157         }
 158
 159         function get_link() {
 160                 return $this->link;
 161         }
 162
 163         function get_title() {
 164                 return $this->title;
 165         }
 166
 167         function get_items() {
 168                 return $this->items;
 169         }
 170
 171         function get_links($rel) {
 172                 $rv = array();
 173
 174                 switch ($this->type) {
 175                 case $this::FEED_ATOM:
 176                         $links = $this->xpath->query("//atom:feed/atom:link");
 177
 178                         foreach ($links as $link) {
 179                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 180                                         array_push($rv, $link->getAttribute('href'));
 181                                 }
 182                         }
 183                         break;
 184                 case $this::FEED_RSS:
 185                         $links = $this->xpath->query("//channel/link");
 186                         foreach ($links as $link) {
 187                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 188                                         array_push($rv, $link->getAttribute('href'));
 189                                 }
 190                         }
 191                         break;
 192                 }
 193
 194                 return $rv;
 195         }
 196 } ?>