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                 mb_substitute_character("none");
  22
  23                 $error = libxml_get_last_error();
  24
  25                 // libxml compiled without iconv?
  26                 if ($error && ($error->code == 32 || $error->code == 9)) {
  27                         if (preg_match('/^(<\?xml[\t\n\r ].*?encoding=["\'])(.+?)(["\'].*?\?>)/s', $data, $matches) === 1) {
  28                                 $enc = $matches[2];
  29
  30                                 $data = mb_convert_encoding($data, 'UTF-8', $enc);
  31
  32                                 $data = preg_replace('/^<\?xml[\t\n\r ].*?\?>/s', $matches[1] . "UTF-8" . $matches[3] , $data);
  33
  34
  35                                 // apparently not all UTF-8 characters are valid for XML
  36                                 $data = preg_replace('/[^\x{0009}\x{000a}\x{000d}\x{0020}-\x{D7FF}\x{E000}-\x{FFFD}]+/u', ' ', $data);
  37
  38                                 if ($data) {
  39                                         libxml_clear_errors();
  40
  41                                         $this->doc = new DOMDocument();
  42                                         $this->doc->loadXML($data);
  43
  44                                         $error = libxml_get_last_error();
  45                                 }
  46                    }
  47                 }
  48
  49                 // some terrible invalid unicode entity?
  50                 if ($error && $error->code == 9) {
  51                         $data = mb_convert_encoding($data, 'UTF-8', 'UTF-8');
  52
  53                         if ($data) {
  54                                 libxml_clear_errors();
  55
  56                                 $this->doc = new DOMDocument();
  57                                 $this->doc->loadXML($data);
  58
  59                                 $error = libxml_get_last_error();
  60                         }
  61                 }
  62
  63                 $this->error = $this->format_error($error);
  64                 libxml_clear_errors();
  65
  66                 $this->items = array();
  67         }
  68
  69         function init() {
  70                 $root = $this->doc->firstChild;
  71                 $xpath = new DOMXPath($this->doc);
  72                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
  73                 $xpath->registerNamespace('atom03', 'http://purl.org/atom/ns#');
  74                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
  75                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
  76                 $xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/');
  77                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
  78                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
  79
  80                 $this->xpath = $xpath;
  81
  82                 $root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)");
  83
  84                 if ($root) {
  85                         $root = $root->item(0);
  86
  87                         if ($root) {
  88                                 switch (mb_strtolower($root->tagName)) {
  89                                 case "rdf:rdf":
  90                                         $this->type = $this::FEED_RDF;
  91                                         break;
  92                                 case "channel":
  93                                         $this->type = $this::FEED_RSS;
  94                                         break;
  95                                 case "feed":
  96                                         $this->type = $this::FEED_ATOM;
  97                                         break;
  98                                 default:
  99                                         if( !isset($this->error) ){
 100                                                 $this->error = "Unknown/unsupported feed type";
 101                                         }
 102                                         return;
 103                                 }
 104                         }
 105
 106                         switch ($this->type) {
 107                         case $this::FEED_ATOM:
 108
 109                                 $title = $xpath->query("//atom:feed/atom:title")->item(0);
 110
 111                                 if (!$title)
 112                                         $title = $xpath->query("//atom03:feed/atom03:title")->item(0);
 113
 114
 115                                 if ($title) {
 116                                         $this->title = $title->nodeValue;
 117                                 }
 118
 119                                 $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
 120
 121                                 if (!$link)
 122                                         $link = $xpath->query("//atom03:feed/atom03:link[not(@rel)]")->item(0);
 123
 124
 125                                 if ($link && $link->hasAttributes()) {
 126                                         $this->link = $link->getAttribute("href");
 127                                 }
 128
 129                                 $articles = $xpath->query("//atom:entry");
 130
 131                                 if (!$articles || $articles->length == 0)
 132                                         $articles = $xpath->query("//atom03:entry");
 133
 134                                 foreach ($articles as $article) {
 135                                         array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath));
 136                                 }
 137
 138                                 break;
 139                         case $this::FEED_RSS:
 140                                 $title = $xpath->query("//channel/title")->item(0);
 141
 142                                 if ($title) {
 143                                         $this->title = $title->nodeValue;
 144                                 }
 145
 146                                 $link = $xpath->query("//channel/link")->item(0);
 147
 148                                 if ($link) {
 149                                         if ($link->getAttribute("href"))
 150                                                 $this->link = $link->getAttribute("href");
 151                                         else if ($link->nodeValue)
 152                                                 $this->link = $link->nodeValue;
 153                                 }
 154
 155                                 $articles = $xpath->query("//channel/item");
 156
 157                                 foreach ($articles as $article) {
 158                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 159                                 }
 160
 161                                 break;
 162                         case $this::FEED_RDF:
 163                                 $xpath->registerNamespace('rssfake', 'http://purl.org/rss/1.0/');
 164
 165                                 $title = $xpath->query("//rssfake:channel/rssfake:title")->item(0);
 166
 167                                 if ($title) {
 168                                         $this->title = $title->nodeValue;
 169                                 }
 170
 171                                 $link = $xpath->query("//rssfake:channel/rssfake:link")->item(0);
 172
 173                                 if ($link) {
 174                                         $this->link = $link->nodeValue;
 175                                 }
 176
 177                                 $articles = $xpath->query("//rssfake:item");
 178
 179                                 foreach ($articles as $article) {
 180                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 181                                 }
 182
 183                                 break;
 184
 185                         }
 186                 } else {
 187                         if( !isset($this->error) ){
 188                                 $this->error = "Unknown/unsupported feed type";
 189                         }
 190                         return;
 191                 }
 192         }
 193
 194         function format_error($error) {
 195                 if ($error) {
 196                         return sprintf("LibXML error %s at line %d (column %d): %s",
 197                                 $error->code, $error->line, $error->column,
 198                                 $error->message);
 199                 } else {
 200                         return "";
 201                 }
 202         }
 203
 204         function error() {
 205                 return $this->error;
 206         }
 207
 208         function get_link() {
 209                 return $this->link;
 210         }
 211
 212         function get_title() {
 213                 return $this->title;
 214         }
 215
 216         function get_items() {
 217                 return $this->items;
 218         }
 219
 220         function get_links($rel) {
 221                 $rv = array();
 222
 223                 switch ($this->type) {
 224                 case $this::FEED_ATOM:
 225                         $links = $this->xpath->query("//atom:feed/atom:link");
 226
 227                         foreach ($links as $link) {
 228                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 229                                         array_push($rv, $link->getAttribute('href'));
 230                                 }
 231                         }
 232                         break;
 233                 case $this::FEED_RSS:
 234                         $links = $this->xpath->query("//atom:link");
 235
 236                         foreach ($links as $link) {
 237                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 238                                         array_push($rv, $link->getAttribute('href'));
 239                                 }
 240                         }
 241                         break;
 242                 }
 243
 244                 return $rv;
 245         }
 246 } ?>