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
  21                 $error = libxml_get_last_error();
  22
  23                 if ($error && $error->code == 9) {
  24                         libxml_clear_errors();
  25
  26                         // we might want to try guessing input encoding here too
  27                         $data = iconv("UTF-8", "UTF-8//IGNORE", $data);
  28
  29                         $this->doc = new DOMDocument();
  30                         $this->doc->loadXML($data);
  31
  32                         $error = libxml_get_last_error();
  33                 }
  34
  35                 $this->error = $this->format_error($error);
  36                 libxml_clear_errors();
  37
  38                 $this->items = array();
  39         }
  40
  41         function init() {
  42                 $root = $this->doc->firstChild;
  43                 $xpath = new DOMXPath($this->doc);
  44                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
  45                 $xpath->registerNamespace('atom03', 'http://purl.org/atom/ns#');
  46                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
  47                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
  48                 $xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/');
  49                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
  50                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
  51
  52                 $this->xpath = $xpath;
  53
  54                 $root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)");
  55
  56                 if ($root) {
  57                         $root = $root->item(0);
  58
  59                         if ($root) {
  60                                 switch (mb_strtolower($root->tagName)) {
  61                                 case "rdf:rdf":
  62                                         $this->type = $this::FEED_RDF;
  63                                         break;
  64                                 case "channel":
  65                                         $this->type = $this::FEED_RSS;
  66                                         break;
  67                                 case "feed":
  68                                         $this->type = $this::FEED_ATOM;
  69                                         break;
  70                                 default:
  71                                         if( !isset($this->error) ){
  72                                                 $this->error = "Unknown/unsupported feed type";
  73                                         }
  74                                         return;
  75                                 }
  76                         }
  77
  78                         switch ($this->type) {
  79                         case $this::FEED_ATOM:
  80
  81                                 $title = $xpath->query("//atom:feed/atom:title")->item(0);
  82
  83                                 if (!$title)
  84                                         $title = $xpath->query("//atom03:feed/atom03:title")->item(0);
  85
  86
  87                                 if ($title) {
  88                                         $this->title = $title->nodeValue;
  89                                 }
  90
  91                                 $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
  92
  93                                 if (!$link)
  94                                         $link = $xpath->query("//atom03:feed/atom03:link[not(@rel)]")->item(0);
  95
  96
  97                                 if ($link && $link->hasAttributes()) {
  98                                         $this->link = $link->getAttribute("href");
  99                                 }
 100
 101                                 $articles = $xpath->query("//atom:entry");
 102
 103                                 if (!$articles || $articles->length == 0)
 104                                         $articles = $xpath->query("//atom03:entry");
 105
 106                                 foreach ($articles as $article) {
 107                                         array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath));
 108                                 }
 109
 110                                 break;
 111                         case $this::FEED_RSS:
 112                                 $title = $xpath->query("//channel/title")->item(0);
 113
 114                                 if ($title) {
 115                                         $this->title = $title->nodeValue;
 116                                 }
 117
 118                                 $link = $xpath->query("//channel/link")->item(0);
 119
 120                                 if ($link) {
 121                                         if ($link->getAttribute("href"))
 122                                                 $this->link = $link->getAttribute("href");
 123                                         else if ($link->nodeValue)
 124                                                 $this->link = $link->nodeValue;
 125                                 }
 126
 127                                 $articles = $xpath->query("//channel/item");
 128
 129                                 foreach ($articles as $article) {
 130                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 131                                 }
 132
 133                                 break;
 134                         case $this::FEED_RDF:
 135                                 $xpath->registerNamespace('rssfake', 'http://purl.org/rss/1.0/');
 136
 137                                 $title = $xpath->query("//rssfake:channel/rssfake:title")->item(0);
 138
 139                                 if ($title) {
 140                                         $this->title = $title->nodeValue;
 141                                 }
 142
 143                                 $link = $xpath->query("//rssfake:channel/rssfake:link")->item(0);
 144
 145                                 if ($link) {
 146                                         $this->link = $link->nodeValue;
 147                                 }
 148
 149                                 $articles = $xpath->query("//rssfake:item");
 150
 151                                 foreach ($articles as $article) {
 152                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 153                                 }
 154
 155                                 break;
 156
 157                         }
 158                 } else {
 159                         if( !isset($this->error) ){
 160                                 $this->error = "Unknown/unsupported feed type";
 161                         }
 162                         return;
 163                 }
 164         }
 165
 166         function format_error($error) {
 167                 if ($error) {
 168                         return sprintf("LibXML error %s at line %d (column %d): %s",
 169                                 $error->code, $error->line, $error->column,
 170                                 $error->message);
 171                 } else {
 172                         return "";
 173                 }
 174         }
 175
 176         function error() {
 177                 return $this->error;
 178         }
 179
 180         function get_link() {
 181                 return $this->link;
 182         }
 183
 184         function get_title() {
 185                 return $this->title;
 186         }
 187
 188         function get_items() {
 189                 return $this->items;
 190         }
 191
 192         function get_links($rel) {
 193                 $rv = array();
 194
 195                 switch ($this->type) {
 196                 case $this::FEED_ATOM:
 197                         $links = $this->xpath->query("//atom:feed/atom:link");
 198
 199                         foreach ($links as $link) {
 200                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 201                                         array_push($rv, $link->getAttribute('href'));
 202                                 }
 203                         }
 204                         break;
 205                 case $this::FEED_RSS:
 206                         $links = $this->xpath->query("//atom:link");
 207
 208                         foreach ($links as $link) {
 209                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 210                                         array_push($rv, $link->getAttribute('href'));
 211                                 }
 212                         }
 213                         break;
 214                 }
 215
 216                 return $rv;
 217         }
 218 } ?>