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