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                 $this->error = $this->format_error(libxml_get_last_error());
  21                 libxml_clear_errors();
  22
  23                 $this->items = array();
  24         }
  25
  26         function init() {
  27                 $root = $this->doc->firstChild;
  28                 $xpath = new DOMXPath($this->doc);
  29                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
  30                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
  31                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
  32                 $xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/');
  33                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
  34
  35                 $this->xpath = $xpath;
  36
  37                 $root = $xpath->query("(//atom:feed|//channel|//rdf:rdf|//rdf:RDF)")->item(0);
  38
  39                 if ($root) {
  40                         switch (mb_strtolower($root->tagName)) {
  41                         case "rdf:rdf":
  42                                 $this->type = $this::FEED_RDF;
  43                                 break;
  44                         case "channel":
  45                                 $this->type = $this::FEED_RSS;
  46                                 break;
  47                         case "feed":
  48                                 $this->type = $this::FEED_ATOM;
  49                                 break;
  50                         default:
  51                                 $this->error = "Unknown/unsupported feed type";
  52                                 return;
  53                         }
  54
  55                         switch ($this->type) {
  56                         case $this::FEED_ATOM:
  57
  58                                 $title = $xpath->query("//atom:feed/atom:title")->item(0);
  59
  60                                 if ($title) {
  61                                         $this->title = $title->nodeValue;
  62                                 }
  63
  64                                 $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
  65
  66                                 if ($link && $link->hasAttributes()) {
  67                                         $this->link = $link->getAttribute("href");
  68                                 }
  69
  70                                 $articles = $xpath->query("//atom:entry");
  71
  72                                 foreach ($articles as $article) {
  73                                         array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath));
  74                                 }
  75
  76                                 break;
  77                         case $this::FEED_RSS:
  78
  79                                 $title = $xpath->query("//channel/title")->item(0);
  80
  81                                 if ($title) {
  82                                         $this->title = $title->nodeValue;
  83                                 }
  84
  85                                 $link = $xpath->query("//channel/link")->item(0);
  86
  87                                 if ($link && $link->hasAttributes()) {
  88                                         $this->link = $link->getAttribute("href");
  89                                 }
  90
  91                                 $articles = $xpath->query("//channel/item");
  92
  93                                 foreach ($articles as $article) {
  94                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
  95                                 }
  96
  97                                 break;
  98                         case $this::FEED_RDF:
  99                                 $xpath->registerNamespace('rssfake', 'http://purl.org/rss/1.0/');
 100
 101                                 $title = $xpath->query("//rssfake:channel/rssfake:title")->item(0);
 102
 103                                 if ($title) {
 104                                         $this->title = $title->nodeValue;
 105                                 }
 106
 107                                 $link = $xpath->query("//rssfake:channel/rssfake:link")->item(0);
 108
 109                                 if ($link) {
 110                                         $this->link = $link->nodeValue;
 111                                 }
 112
 113                                 $articles = $xpath->query("//rssfake:item");
 114
 115                                 foreach ($articles as $article) {
 116                                         array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
 117                                 }
 118
 119                                 break;
 120
 121                         }
 122                 } else {
 123                         $this->error = "Unknown/unsupported feed type";
 124                         return;
 125                 }
 126         }
 127
 128         function format_error($error) {
 129                 if ($error) {
 130                         return sprintf("LibXML error %s at line %d (column %d): %s",
 131                                 $error->code, $error->line, $error->column,
 132                                 $error->message);
 133                 } else {
 134                         return "";
 135                 }
 136         }
 137
 138         function error() {
 139                 return $this->error;
 140         }
 141
 142         function get_link() {
 143                 return $this->link;
 144         }
 145
 146         function get_title() {
 147                 return $this->title;
 148         }
 149
 150         function get_items() {
 151                 return $this->items;
 152         }
 153
 154         function get_links($rel) {
 155                 $rv = array();
 156
 157                 switch ($this->type) {
 158                 case $this::FEED_ATOM:
 159                         $links = $this->xpath->query("//atom:feed/atom:link");
 160
 161                         foreach ($links as $link) {
 162                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 163                                         array_push($rv, $link->getAttribute('href'));
 164                                 }
 165                         }
 166                         break;
 167                 case $this::FEED_RSS:
 168                         $links = $this->xpath->query("//channel/link");
 169                         foreach ($links as $link) {
 170                                 if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
 171                                         array_push($rv, $link->getAttribute('href'));
 172                                 }
 173                         }
 174                         break;
 175                 }
 176
 177                 return $rv;
 178         }
 179 } ?>