include/feed.php

   1 <?php
   2 require_once("include/html2bbcode.php");
   3 require_once("include/items.php");
   4
   5 function feed_import($xml,$importer,&$contact, &$hub) {
   6
   7         $a = get_app();
   8
   9         logger("Import Atom/RSS feed", LOGGER_DEBUG);
  10
  11         if ($xml == "")
  12                 return;
  13
  14         $doc = new DOMDocument();
  15         @$doc->loadXML($xml);
  16         $xpath = new DomXPath($doc);
  17         $xpath->registerNamespace('atom', "http://www.w3.org/2005/Atom");
  18         $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  19         $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  20         $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  21         $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  22         $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  23
  24         $author = array();
  25
  26         // Is it RDF?
  27         if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  28                 //$author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  29                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  30
  31                 if ($author["author-name"] == "")
  32                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  33
  34                 $entries = $xpath->query('/rdf:RDF/rss:item');
  35         }
  36
  37         // Is it Atom?
  38         if ($xpath->query('/atom:feed/atom:entry')->length > 0) {
  39                 //$self = $xpath->query("/atom:feed/atom:link[@rel='self']")->item(0)->attributes;
  40                 //if (is_object($self))
  41                 //      foreach($self AS $attributes)
  42                 //              if ($attributes->name == "href")
  43                 //                      $author["author-link"] = $attributes->textContent;
  44
  45                 //if ($author["author-link"] == "") {
  46                 //      $alternate = $xpath->query("/atom:feed/atom:link[@rel='alternate']")->item(0)->attributes;
  47                 //      if (is_object($alternate))
  48                 //              foreach($alternate AS $attributes)
  49                 //                      if ($attributes->name == "href")
  50                 //                              $author["author-link"] = $attributes->textContent;
  51                 //}
  52
  53                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
  54
  55                 if ($author["author-name"] == "")
  56                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
  57
  58                 if ($author["author-name"] == "")
  59                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
  60
  61                 //$author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
  62
  63                 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
  64
  65                 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
  66
  67                 $entries = $xpath->query('/atom:feed/atom:entry');
  68         }
  69
  70         // Is it RSS?
  71         if ($xpath->query('/rss/channel')->length > 0) {
  72                 //$author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
  73                 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
  74                 //$author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
  75
  76                 if ($author["author-name"] == "")
  77                         $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
  78
  79                 if ($author["author-name"] == "")
  80                         $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
  81
  82                 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
  83
  84                 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
  85
  86                 $entries = $xpath->query('/rss/channel/item');
  87         }
  88
  89         //if ($author["author-link"] == "")
  90                 $author["author-link"] = $contact["url"];
  91
  92         if ($author["author-name"] == "")
  93                 $author["author-name"] = $contact["name"];
  94
  95         //if ($author["author-avatar"] == "")
  96                 $author["author-avatar"] = $contact["thumb"];
  97
  98         $author["owner-link"] = $contact["url"];
  99         $author["owner-name"] = $contact["name"];
 100         $author["owner-avatar"] = $contact["thumb"];
 101
 102         $header = array();
 103         $header["uid"] = $importer["uid"];
 104         $header["network"] = NETWORK_FEED;
 105         $header["type"] = "remote";
 106         $header["wall"] = 0;
 107         $header["origin"] = 0;
 108         $header["gravity"] = GRAVITY_PARENT;
 109
 110         $header["contact-id"] = $contact["id"];
 111
 112         if (!is_object($entries))
 113                 return;
 114
 115         foreach ($entries AS $entry)
 116                 $entrylist[] = $entry;
 117
 118         foreach (array_reverse($entrylist) AS $entry) {
 119                 $item = array_merge($header, $author);
 120
 121                 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 122
 123                 if ($item["title"] == "")
 124                         $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 125
 126                 if ($item["title"] == "")
 127                         $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 128
 129                 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 130                 if (!is_object($alternate))
 131                         $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 132
 133                 if (is_object($alternate))
 134                         foreach($alternate AS $attributes)
 135                                 if ($attributes->name == "href")
 136                                         $item["plink"] = $attributes->textContent;
 137
 138                 if ($item["plink"] == "")
 139                         $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 140
 141                 if ($item["plink"] == "")
 142                         $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 143
 144                 $item["plink"] = original_url($item["plink"]);
 145
 146                 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 147
 148                 if ($item["uri"] == "")
 149                         $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 150
 151                 if ($item["uri"] == "")
 152                         $item["uri"] = $item["plink"];
 153
 154                 $item["parent-uri"] = $item["uri"];
 155
 156                 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 157
 158                 if ($published == "")
 159                         $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 160
 161                 if ($published == "")
 162                         $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 163
 164                 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 165
 166                 if ($updated == "")
 167                         $updated = $published;
 168
 169                 if ($published != "")
 170                         $item["created"] = $published;
 171
 172                 if ($updated != "")
 173                         $item["edited"] = $updated;
 174
 175                 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 176
 177                 if ($creator == "")
 178                         $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 179
 180                 if ($creator == "")
 181                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 182
 183                 if ($creator != "")
 184                         $item["author-name"] = $creator;
 185
 186                 if ($pubDate != "")
 187                         $item["edited"] = $item["created"] = $pubDate;
 188
 189                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 190
 191                 if ($creator != "")
 192                         $item["author-name"] = $creator;
 193
 194                 //$item["object"] = $xml;
 195
 196                 $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s'",
 197                         intval($importer["uid"]), dbesc($item["uri"]));
 198                 if ($r) {
 199                         logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
 200                         continue;
 201                 }
 202
 203                 // To-Do?
 204                 // <category>Ausland</category>
 205                 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 206
 207                 $attachments = array();
 208
 209                 $enclosures = $xpath->query("enclosure", $entry);
 210                 foreach ($enclosures AS $enclosure) {
 211                         $href = "";
 212                         $length = "";
 213                         $type = "";
 214                         $title = "";
 215
 216                         foreach($enclosure->attributes AS $attributes) {
 217                                 if ($attributes->name == "url")
 218                                         $href = $attributes->textContent;
 219                                 elseif ($attributes->name == "length")
 220                                         $length = $attributes->textContent;
 221                                 elseif ($attributes->name == "type")
 222                                         $type = $attributes->textContent;
 223                         }
 224                         if(strlen($item["attach"]))
 225                                 $item["attach"] .= ',';
 226
 227                         $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 228
 229                         $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 230                 }
 231
 232                 if ($contact["fetch_further_information"]) {
 233                         $preview = "";
 234
 235                         // Handle enclosures and treat them as preview picture
 236                         foreach ($attachments AS $attachment)
 237                                 if ($attachment["type"] == "image/jpeg")
 238                                         $preview = $attachment["link"];
 239
 240                         $item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 241                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 242                         $item["title"] = "";
 243                         $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 244                         unset($item["attach"]);
 245                 } else {
 246                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 247
 248                         if ($body == "")
 249                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 250
 251                         if ($body == "")
 252                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 253
 254                         if ($body == "")
 255                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 256
 257                         // remove the content of the title if it is identically to the body
 258                         // This helps with auto generated titles e.g. from tumblr
 259                         if (title_is_body($item["title"], $body))
 260                                 $item["title"] = "";
 261
 262                         $item["body"] = html2bbcode($body);
 263                 }
 264
 265                 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 266
 267                 $notify = item_is_remote_self($contact, $item);
 268                 $id = item_store($item, false, $notify);
 269
 270                 //print_r($item);
 271
 272                 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 273         }
 274 }
 275 ?>