classes/feedparser.php

   1 <?php
   2 class FeedParser {
   3         private $doc;
   4         private $error;
   5         private $libxml_errors = array();
   6         private $items;
   7         private $link;
   8         private $title;
   9         private $type;
  10         private $xpath;
  11
  12         const FEED_RDF = 0;
  13         const FEED_RSS = 1;
  14         const FEED_ATOM = 2;
  15
  16         function normalize_encoding($data) {
  17                 if (preg_match('/^(<\?xml[\t\n\r ].*?encoding[\t\n\r ]*=[\t\n\r ]*["\'])(.+?)(["\'].*?\?>)/s', $data, $matches) === 1) {
  18                         $data = mb_convert_encoding($data, 'UTF-8', $matches[2]);
  19
  20                         $data = preg_replace('/^<\?xml[\t\n\r ].*?\?>/s', $matches[1] . "UTF-8" . $matches[3] , $data);
  21                 }
  22
  23                 return $data;
  24         }
  25
  26         function __construct($data) {
  27                 libxml_use_internal_errors(true);
  28                 libxml_clear_errors();
  29                 $this->doc = new DOMDocument();
  30                 $this->doc->preserveWhiteSpace = false;
  31                 $this->doc->loadXML($data);
  32
  33                 mb_substitute_character("none");
  34
  35                 $error = libxml_get_last_error();
  36
  37                 // libxml compiled without iconv?
  38                 if ($error && $error->code == 32) {
  39                         $data = $this->normalize_encoding($data);
  40
  41                         if ($data) {
  42                                 libxml_clear_errors();
  43
  44                                 $this->doc = new DOMDocument();
  45                                 $this->doc->preserveWhiteSpace = false;
  46                                 $this->doc->loadXML($data);
  47
  48                                 $error = libxml_get_last_error();
  49                         }
  50                 }
  51
  52                 // some terrible invalid unicode entity?
  53                 if ($error) {
  54                         foreach (libxml_get_errors() as $err) {
  55                                 if ($err->code == 9) {
  56                                         // if the source feed is not in utf8, next conversion will fail
  57                                         $data = $this->normalize_encoding($data);
  58
  59                                         // remove dangling bytes
  60                                         $data = mb_convert_encoding($data, 'UTF-8', 'UTF-8');
  61
  62                                         // apparently not all UTF-8 characters are valid for XML
  63                                         $data = preg_replace('/[^\x{0009}\x{000a}\x{000d}\x{0020}-\x{D7FF}\x{E000}-\x{FFFD}]+/u', ' ', $data);
  64
  65                                         if ($data) {
  66                                                 libxml_clear_errors();
  67
  68                                                 $this->doc = new DOMDocument();
  69                                                 $this->doc->preserveWhiteSpace = false;
  70                                                 $this->doc->loadXML($data);
  71
  72                                                 $error = libxml_get_last_error();
  73                                         }
  74                                         break;
  75                                 }
  76                         }
  77                 }
  78
  79                 if ($error) {
  80                         foreach (libxml_get_errors() as $error) {
  81                                 if ($error->level == LIBXML_ERR_FATAL) {
  82                                         if(!isset($this->error)) //currently only the first error is reported
  83                                                 $this->error = $this->format_error($error);
  84                                         $this->libxml_errors [] = $this->format_error($error);
  85                                 }
  86                         }
  87                 }
  88                 libxml_clear_errors();
  89
  90                 $this->items = array();
  91         }
  92
  93         function init() {
  94                 $root = $this->doc->firstChild;
  95                 $xpath = new DOMXPath($this->doc);
  96                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
  97                 $xpath->registerNamespace('atom03', 'http://purl.org/atom/ns#');
  98                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
  99                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
 100                 $xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/');
 101                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
 102                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
 103                 $xpath->registerNamespace('thread', 'http://purl.org/syndication/thread/1.0');
 104
 105                 $this->xpath = $xpath;
 106
 107                 $root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)");
 108
 109                 if ($root && $root->length > 0) {
 110                         $root = $root->item(0);
 111
 112                         if ($root) {
 113                                 switch (mb_strtolower($root->tagName)) {
 114                                 case "rdf:rdf":
 115                                         $this->type = $this::FEED_RDF;
 116                                         break;
 117                                 case "channel":
 118                                         $this->type = $this::FEED_RSS;
 119                                         break;
 120                                 case "feed":
 121                                         $this->type = $this::FEED_ATOM;
 122                                         break;
 123                                 default:
 124                                         if( !isset($this->error) ){
 125                                                 $this->error = "Unknown/unsupported feed type";
 126                                         }
 127                                         return;
 128                                 }
 129                         }
 130
 131                         switch ($this->type) {
 132                         case $this::FEED_ATOM:
 133
 134                                 $title = $xpath->query("//atom:feed/atom:title")->item(0);
 135
 136                                 if (!$title)
 137                                         $title = $xpath->query("//atom03:feed/atom03:title")->item(0);
 138
 139
 140                                 if ($title) {
 141                                         $this->title = $title->nodeValue;
 142                                 }
 143
 144                                 $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
 145
 146                                 if (!$link)
 147                                         $link = $xpath->query("//atom03:feed/atom03:link[not(@rel)]")->item(0);
 148
 149
 150                                 if ($link && $link->hasAttributes()) {
 151                                         $this->link = $link->getAttribute("href");
 152                                 }
 153
 154                                 $articles = $xpath->query("//atom:entry");
 155
 156                                 if (!$articles || $articles->length == 0)
 157                                         $articles = $xpath->query("//atom03:entry");
 158
 159                                 foreach ($articles as $article) {
 160                                         array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath));
 161                                 }
 162
 163                                 break;
 164                         case $this::FEED_RSS:
 165                                 $title = $xpath->query("//channel/title")->item(0);
 166
 167                                 if ($title) {
 168                                         $this->title = $title->nodeValue;
 169                                 }
 170
 171                                 $link = $xpath->query("//channel/link")->item(0);
 172
 173                                 if ($link) {
 174                                         if ($link->getAttribute("href"))
 175                                                 $this->link = $link->getAttribute("href");
 176                                         else if ($link->nodeValue)
 177                                                 $this->link = $link->nodeValue;
 178                                 }
 179
 180                                 $articles = $xpath->query("//channel/item");
 181
 182                                 foreach ($articles as $article) {
 183                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 184                                 }
 185
 186                                 break;
 187                         case $this::FEED_RDF:
 188                                 $xpath->registerNamespace('rssfake', 'http://purl.org/rss/1.0/');
 189
 190                                 $title = $xpath->query("//rssfake:channel/rssfake:title")->item(0);
 191
 192                                 if ($title) {
 193                                         $this->title = $title->nodeValue;
 194                                 }
 195
 196                                 $link = $xpath->query("//rssfake:channel/rssfake:link")->item(0);
 197
 198                                 if ($link) {
 199                                         $this->link = $link->nodeValue;
 200                                 }
 201
 202                                 $articles = $xpath->query("//rssfake:item");
 203
 204                                 foreach ($articles as $article) {
 205                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 206                                 }
 207
 208                                 break;
 209
 210                         }
 211                 } else {
 212                         if( !isset($this->error) ){
 213                                 $this->error = "Unknown/unsupported feed type";
 214                         }
 215                         return;
 216                 }
 217         }
 218
 219         function format_error($error) {
 220                 if ($error) {
 221                         return sprintf("LibXML error %s at line %d (column %d): %s",
 222                                 $error->code, $error->line, $error->column,
 223                                 $error->message);
 224                 } else {
 225                         return "";
 226                 }
 227         }
 228
 229         function error() {
 230                 return $this->error;
 231         }
 232
 233         function errors() {
 234                 return $this->libxml_errors;
 235         }
 236
 237         function get_link() {
 238                 return $this->link;
 239         }
 240
 241         function get_title() {
 242                 return $this->title;
 243         }
 244
 245         function get_items() {
 246                 return $this->items;
 247         }
 248
 249         function get_links($rel) {
 250                 $rv = array();
 251
 252                 switch ($this->type) {
 253                 case $this::FEED_ATOM:
 254                         $links = $this->xpath->query("//atom:feed/atom:link");
 255
 256                         foreach ($links as $link) {
 257                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 258                                         array_push($rv, $link->getAttribute('href'));
 259                                 }
 260                         }
 261                         break;
 262                 case $this::FEED_RSS:
 263                         $links = $this->xpath->query("//atom:link");
 264
 265                         foreach ($links as $link) {
 266                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 267                                         array_push($rv, $link->getAttribute('href'));
 268                                 }
 269                         }
 270                         break;
 271                 }
 272
 273                 return $rv;
 274         }
 275 } ?>