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                 $error = libxml_get_last_error();
  22
  23                 // libxml compiled without iconv?
  24                 if ($error && $error->code == 32) {
  25                         if (preg_match('/^(<\?xml[\t\n\r ].*?encoding=["\'])(.+?)(["\'].*?\?>)/s', $data, $matches) === 1) {
  26                                 libxml_clear_errors();
  27
  28                                 $enc = $matches[2];
  29
  30                                 $data = iconv($enc, 'UTF-8//IGNORE', $data);
  31                                 $data = preg_replace('/^<\?xml[\t\n\r ].*?\?>/s', $matches[1] . "UTF-8" . $matches[3] , $data);
  32
  33                                 $this->doc = new DOMDocument();
  34                                 $this->doc->loadXML($data);
  35
  36                                 $error = libxml_get_last_error();
  37                    }
  38                 }
  39
  40                 // some terrible invalid unicode entity?
  41                 if ($error && $error->code == 9) {
  42                         libxml_clear_errors();
  43
  44                         // we might want to try guessing input encoding here too
  45                         $data = iconv("UTF-8", "UTF-8//IGNORE", $data);
  46
  47                         $this->doc = new DOMDocument();
  48                         $this->doc->loadXML($data);
  49
  50                         $error = libxml_get_last_error();
  51                 }
  52
  53                 $this->error = $this->format_error($error);
  54                 libxml_clear_errors();
  55
  56                 $this->items = array();
  57         }
  58
  59         function init() {
  60                 $root = $this->doc->firstChild;
  61                 $xpath = new DOMXPath($this->doc);
  62                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
  63                 $xpath->registerNamespace('atom03', 'http://purl.org/atom/ns#');
  64                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
  65                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
  66                 $xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/');
  67                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
  68                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
  69
  70                 $this->xpath = $xpath;
  71
  72                 $root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)");
  73
  74                 if ($root) {
  75                         $root = $root->item(0);
  76
  77                         if ($root) {
  78                                 switch (mb_strtolower($root->tagName)) {
  79                                 case "rdf:rdf":
  80                                         $this->type = $this::FEED_RDF;
  81                                         break;
  82                                 case "channel":
  83                                         $this->type = $this::FEED_RSS;
  84                                         break;
  85                                 case "feed":
  86                                         $this->type = $this::FEED_ATOM;
  87                                         break;
  88                                 default:
  89                                         if( !isset($this->error) ){
  90                                                 $this->error = "Unknown/unsupported feed type";
  91                                         }
  92                                         return;
  93                                 }
  94                         }
  95
  96                         switch ($this->type) {
  97                         case $this::FEED_ATOM:
  98
  99                                 $title = $xpath->query("//atom:feed/atom:title")->item(0);
 100
 101                                 if (!$title)
 102                                         $title = $xpath->query("//atom03:feed/atom03:title")->item(0);
 103
 104
 105                                 if ($title) {
 106                                         $this->title = $title->nodeValue;
 107                                 }
 108
 109                                 $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
 110
 111                                 if (!$link)
 112                                         $link = $xpath->query("//atom03:feed/atom03:link[not(@rel)]")->item(0);
 113
 114
 115                                 if ($link && $link->hasAttributes()) {
 116                                         $this->link = $link->getAttribute("href");
 117                                 }
 118
 119                                 $articles = $xpath->query("//atom:entry");
 120
 121                                 if (!$articles || $articles->length == 0)
 122                                         $articles = $xpath->query("//atom03:entry");
 123
 124                                 foreach ($articles as $article) {
 125                                         array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath));
 126                                 }
 127
 128                                 break;
 129                         case $this::FEED_RSS:
 130                                 $title = $xpath->query("//channel/title")->item(0);
 131
 132                                 if ($title) {
 133                                         $this->title = $title->nodeValue;
 134                                 }
 135
 136                                 $link = $xpath->query("//channel/link")->item(0);
 137
 138                                 if ($link) {
 139                                         if ($link->getAttribute("href"))
 140                                                 $this->link = $link->getAttribute("href");
 141                                         else if ($link->nodeValue)
 142                                                 $this->link = $link->nodeValue;
 143                                 }
 144
 145                                 $articles = $xpath->query("//channel/item");
 146
 147                                 foreach ($articles as $article) {
 148                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 149                                 }
 150
 151                                 break;
 152                         case $this::FEED_RDF:
 153                                 $xpath->registerNamespace('rssfake', 'http://purl.org/rss/1.0/');
 154
 155                                 $title = $xpath->query("//rssfake:channel/rssfake:title")->item(0);
 156
 157                                 if ($title) {
 158                                         $this->title = $title->nodeValue;
 159                                 }
 160
 161                                 $link = $xpath->query("//rssfake:channel/rssfake:link")->item(0);
 162
 163                                 if ($link) {
 164                                         $this->link = $link->nodeValue;
 165                                 }
 166
 167                                 $articles = $xpath->query("//rssfake:item");
 168
 169                                 foreach ($articles as $article) {
 170                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 171                                 }
 172
 173                                 break;
 174
 175                         }
 176                 } else {
 177                         if( !isset($this->error) ){
 178                                 $this->error = "Unknown/unsupported feed type";
 179                         }
 180                         return;
 181                 }
 182         }
 183
 184         function format_error($error) {
 185                 if ($error) {
 186                         return sprintf("LibXML error %s at line %d (column %d): %s",
 187                                 $error->code, $error->line, $error->column,
 188                                 $error->message);
 189                 } else {
 190                         return "";
 191                 }
 192         }
 193
 194         function error() {
 195                 return $this->error;
 196         }
 197
 198         function get_link() {
 199                 return $this->link;
 200         }
 201
 202         function get_title() {
 203                 return $this->title;
 204         }
 205
 206         function get_items() {
 207                 return $this->items;
 208         }
 209
 210         function get_links($rel) {
 211                 $rv = array();
 212
 213                 switch ($this->type) {
 214                 case $this::FEED_ATOM:
 215                         $links = $this->xpath->query("//atom:feed/atom:link");
 216
 217                         foreach ($links as $link) {
 218                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 219                                         array_push($rv, $link->getAttribute('href'));
 220                                 }
 221                         }
 222                         break;
 223                 case $this::FEED_RSS:
 224                         $links = $this->xpath->query("//atom:link");
 225
 226                         foreach ($links as $link) {
 227                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 228                                         array_push($rv, $link->getAttribute('href'));
 229                                 }
 230                         }
 231                         break;
 232                 }
 233
 234                 return $rv;
 235         }
 236 } ?>