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
  19                         $encoding = strtolower($matches[2]);
  20
  21                         if (in_array($encoding, array_map('strtolower', mb_list_encodings())))
  22                                 $data = mb_convert_encoding($data, 'UTF-8', $encoding);
  23
  24                         $data = preg_replace('/^<\?xml[\t\n\r ].*?\?>/s', $matches[1] . "UTF-8" . $matches[3] , $data);
  25                 }
  26
  27                 return $data;
  28         }
  29
  30         function __construct($data) {
  31                 libxml_use_internal_errors(true);
  32                 libxml_clear_errors();
  33                 $this->doc = new DOMDocument();
  34                 $this->doc->loadXML($data);
  35
  36                 mb_substitute_character("none");
  37
  38                 $error = libxml_get_last_error();
  39
  40                 // libxml compiled without iconv?
  41                 if ($error && $error->code == 32) {
  42                         $data = $this->normalize_encoding($data);
  43
  44                         if ($data) {
  45                                 libxml_clear_errors();
  46
  47                                 $this->doc = new DOMDocument();
  48                                 $this->doc->loadXML($data);
  49
  50                                 $error = libxml_get_last_error();
  51                         }
  52                 }
  53
  54                 // some terrible invalid unicode entity?
  55                 if ($error) {
  56                         foreach (libxml_get_errors() as $err) {
  57                                 if ($err->code == 9) {
  58                                         // if the source feed is not in utf8, next conversion will fail
  59                                         $data = $this->normalize_encoding($data);
  60
  61                                         // remove dangling bytes
  62                                         $data = mb_convert_encoding($data, 'UTF-8', 'UTF-8');
  63
  64                                         // apparently not all UTF-8 characters are valid for XML
  65                                         $data = preg_replace('/[^\x{0009}\x{000a}\x{000d}\x{0020}-\x{D7FF}\x{E000}-\x{FFFD}]+/u', ' ', $data);
  66
  67                                         if ($data) {
  68                                                 libxml_clear_errors();
  69
  70                                                 $this->doc = new DOMDocument();
  71                                                 $this->doc->loadXML($data);
  72
  73                                                 $error = libxml_get_last_error();
  74                                         }
  75                                         break;
  76                                 }
  77                         }
  78                 }
  79
  80                 if ($error) {
  81                         foreach (libxml_get_errors() as $error) {
  82                                 if ($error->level == LIBXML_ERR_FATAL) {
  83                                         if(!isset($this->error)) //currently only the first error is reported
  84                                                 $this->error = $this->format_error($error);
  85                                         $this->libxml_errors [] = $this->format_error($error);
  86                                 }
  87                         }
  88                 }
  89                 libxml_clear_errors();
  90
  91                 $this->items = array();
  92         }
  93
  94         function init() {
  95                 $root = $this->doc->firstChild;
  96                 $xpath = new DOMXPath($this->doc);
  97                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
  98                 $xpath->registerNamespace('atom03', 'http://purl.org/atom/ns#');
  99                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
 100                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
 101                 $xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/');
 102                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
 103                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
 104                 $xpath->registerNamespace('thread', 'http://purl.org/syndication/thread/1.0');
 105
 106                 $this->xpath = $xpath;
 107
 108                 $root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)");
 109
 110                 if ($root && $root->length > 0) {
 111                         $root = $root->item(0);
 112
 113                         if ($root) {
 114                                 switch (mb_strtolower($root->tagName)) {
 115                                 case "rdf:rdf":
 116                                         $this->type = $this::FEED_RDF;
 117                                         break;
 118                                 case "channel":
 119                                         $this->type = $this::FEED_RSS;
 120                                         break;
 121                                 case "feed":
 122                                 case "atom:feed":
 123                                         $this->type = $this::FEED_ATOM;
 124                                         break;
 125                                 default:
 126                                         if( !isset($this->error) ){
 127                                                 $this->error = "Unknown/unsupported feed type";
 128                                         }
 129                                         return;
 130                                 }
 131                         }
 132
 133                         switch ($this->type) {
 134                         case $this::FEED_ATOM:
 135
 136                                 $title = $xpath->query("//atom:feed/atom:title")->item(0);
 137
 138                                 if (!$title)
 139                                         $title = $xpath->query("//atom03:feed/atom03:title")->item(0);
 140
 141
 142                                 if ($title) {
 143                                         $this->title = $title->nodeValue;
 144                                 }
 145
 146                                 $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
 147
 148                                 if (!$link)
 149                                         $link = $xpath->query("//atom:feed/atom:link[@rel='alternate']")->item(0);
 150
 151                                 if (!$link)
 152                                         $link = $xpath->query("//atom03:feed/atom03:link[not(@rel)]")->item(0);
 153
 154                                 if (!$link)
 155                                         $link = $xpath->query("//atom03:feed/atom03:link[@rel='alternate']")->item(0);
 156
 157                                 if ($link && $link->hasAttributes()) {
 158                                         $this->link = $link->getAttribute("href");
 159                                 }
 160
 161                                 $articles = $xpath->query("//atom:entry");
 162
 163                                 if (!$articles || $articles->length == 0)
 164                                         $articles = $xpath->query("//atom03:entry");
 165
 166                                 foreach ($articles as $article) {
 167                                         array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath));
 168                                 }
 169
 170                                 break;
 171                         case $this::FEED_RSS:
 172                                 $title = $xpath->query("//channel/title")->item(0);
 173
 174                                 if ($title) {
 175                                         $this->title = $title->nodeValue;
 176                                 }
 177
 178                                 $link = $xpath->query("//channel/link")->item(0);
 179
 180                                 if ($link) {
 181                                         if ($link->getAttribute("href"))
 182                                                 $this->link = $link->getAttribute("href");
 183                                         else if ($link->nodeValue)
 184                                                 $this->link = $link->nodeValue;
 185                                 }
 186
 187                                 $articles = $xpath->query("//channel/item");
 188
 189                                 foreach ($articles as $article) {
 190                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 191                                 }
 192
 193                                 break;
 194                         case $this::FEED_RDF:
 195                                 $xpath->registerNamespace('rssfake', 'http://purl.org/rss/1.0/');
 196
 197                                 $title = $xpath->query("//rssfake:channel/rssfake:title")->item(0);
 198
 199                                 if ($title) {
 200                                         $this->title = $title->nodeValue;
 201                                 }
 202
 203                                 $link = $xpath->query("//rssfake:channel/rssfake:link")->item(0);
 204
 205                                 if ($link) {
 206                                         $this->link = $link->nodeValue;
 207                                 }
 208
 209                                 $articles = $xpath->query("//rssfake:item");
 210
 211                                 foreach ($articles as $article) {
 212                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 213                                 }
 214
 215                                 break;
 216
 217                         }
 218
 219                         if ($this->title) $this->title = trim($this->title);
 220                         if ($this->link) $this->link = trim($this->link);
 221
 222                 } else {
 223                         if( !isset($this->error) ){
 224                                 $this->error = "Unknown/unsupported feed type";
 225                         }
 226                         return;
 227                 }
 228         }
 229
 230         function format_error($error) {
 231                 if ($error) {
 232                         return sprintf("LibXML error %s at line %d (column %d): %s",
 233                                 $error->code, $error->line, $error->column,
 234                                 $error->message);
 235                 } else {
 236                         return "";
 237                 }
 238         }
 239
 240         function error() {
 241                 return $this->error;
 242         }
 243
 244         function errors() {
 245                 return $this->libxml_errors;
 246         }
 247
 248         function get_link() {
 249                 return $this->link;
 250         }
 251
 252         function get_title() {
 253                 return $this->title;
 254         }
 255
 256         function get_items() {
 257                 return $this->items;
 258         }
 259
 260         function get_links($rel) {
 261                 $rv = array();
 262
 263                 switch ($this->type) {
 264                 case $this::FEED_ATOM:
 265                         $links = $this->xpath->query("//atom:feed/atom:link");
 266
 267                         foreach ($links as $link) {
 268                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 269                                         array_push($rv, trim($link->getAttribute('href')));
 270                                 }
 271                         }
 272                         break;
 273                 case $this::FEED_RSS:
 274                         $links = $this->xpath->query("//atom:link");
 275
 276                         foreach ($links as $link) {
 277                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 278                                         array_push($rv, trim($link->getAttribute('href')));
 279                                 }
 280                         }
 281                         break;
 282                 }
 283
 284                 return $rv;
 285         }
 286 }