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         $header["private"] = 2;
 110         $header["verb"] = ACTIVITY_POST;
 111         $header["object-type"] = ACTIVITY_OBJ_NOTE;
 112
 113         $header["contact-id"] = $contact["id"];
 114
 115         if(!strlen($contact["notify"])) {
 116                 // one way feed - no remote comment ability
 117                 $header["last-child"] = 0;
 118         }
 119
 120         if (!is_object($entries))
 121                 return;
 122
 123         $entrylist = array();
 124
 125         foreach ($entries AS $entry)
 126                 $entrylist[] = $entry;
 127
 128         foreach (array_reverse($entrylist) AS $entry) {
 129                 $item = array_merge($header, $author);
 130
 131                 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 132
 133                 if ($item["title"] == "")
 134                         $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 135
 136                 if ($item["title"] == "")
 137                         $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 138
 139                 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 140                 if (!is_object($alternate))
 141                         $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 142
 143                 if (is_object($alternate))
 144                         foreach($alternate AS $attributes)
 145                                 if ($attributes->name == "href")
 146                                         $item["plink"] = $attributes->textContent;
 147
 148                 if ($item["plink"] == "")
 149                         $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 150
 151                 if ($item["plink"] == "")
 152                         $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 153
 154                 $item["plink"] = original_url($item["plink"]);
 155
 156                 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 157
 158                 if ($item["uri"] == "")
 159                         $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 160
 161                 if ($item["uri"] == "")
 162                         $item["uri"] = $item["plink"];
 163
 164                 $item["parent-uri"] = $item["uri"];
 165
 166                 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 167
 168                 if ($published == "")
 169                         $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 170
 171                 if ($published == "")
 172                         $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 173
 174                 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 175
 176                 if ($updated == "")
 177                         $updated = $published;
 178
 179                 if ($published != "")
 180                         $item["created"] = $published;
 181
 182                 if ($updated != "")
 183                         $item["edited"] = $updated;
 184
 185                 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 186
 187                 if ($creator == "")
 188                         $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 189
 190                 if ($creator == "")
 191                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 192
 193                 if ($creator != "")
 194                         $item["author-name"] = $creator;
 195
 196                 if ($pubDate != "")
 197                         $item["edited"] = $item["created"] = $pubDate;
 198
 199                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 200
 201                 if ($creator != "")
 202                         $item["author-name"] = $creator;
 203
 204                 //$item["object"] = $xml;
 205
 206                 $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
 207                         intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
 208                 if ($r) {
 209                         logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
 210                         continue;
 211                 }
 212
 213                 /// @TODO ?
 214                 // <category>Ausland</category>
 215                 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 216
 217                 $attachments = array();
 218
 219                 $enclosures = $xpath->query("enclosure", $entry);
 220                 foreach ($enclosures AS $enclosure) {
 221                         $href = "";
 222                         $length = "";
 223                         $type = "";
 224                         $title = "";
 225
 226                         foreach($enclosure->attributes AS $attributes) {
 227                                 if ($attributes->name == "url")
 228                                         $href = $attributes->textContent;
 229                                 elseif ($attributes->name == "length")
 230                                         $length = $attributes->textContent;
 231                                 elseif ($attributes->name == "type")
 232                                         $type = $attributes->textContent;
 233                         }
 234                         if(strlen($item["attach"]))
 235                                 $item["attach"] .= ',';
 236
 237                         $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 238
 239                         $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 240                 }
 241
 242                 if ($contact["fetch_further_information"]) {
 243                         $preview = "";
 244
 245                         // Handle enclosures and treat them as preview picture
 246                         foreach ($attachments AS $attachment)
 247                                 if ($attachment["type"] == "image/jpeg")
 248                                         $preview = $attachment["link"];
 249
 250                         $item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 251                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 252                         $item["title"] = "";
 253                         $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 254                         unset($item["attach"]);
 255                 } else {
 256                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 257
 258                         if ($body == "")
 259                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 260
 261                         if ($body == "")
 262                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 263
 264                         if ($body == "")
 265                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 266
 267                         // remove the content of the title if it is identically to the body
 268                         // This helps with auto generated titles e.g. from tumblr
 269                         if (title_is_body($item["title"], $body))
 270                                 $item["title"] = "";
 271
 272                         $item["body"] = html2bbcode($body);
 273                 }
 274
 275                 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 276
 277                 $notify = item_is_remote_self($contact, $item);
 278                 $id = item_store($item, false, $notify);
 279
 280                 //print_r($item);
 281
 282                 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 283         }
 284 }
 285 ?>