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, $simulate = false) {
   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 = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  40                 if (is_object($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 = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  47                         if (is_object($alternate))
  48                                 foreach($alternate AS $attributes)
  49                                         if ($attributes->name == "href")
  50                                                 $author["author-link"] = $attributes->textContent;
  51                 }
  52
  53                 if ($author["author-link"] == "")
  54                         $author["author-link"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
  55                 if ($author["author-link"] == "")
  56                         $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
  57
  58                 $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
  59
  60                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
  61
  62                 if ($author["author-name"] == "")
  63                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
  64
  65                 if ($author["author-name"] == "")
  66                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
  67
  68                 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
  69
  70                 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
  71
  72                 $entries = $xpath->query('/atom:feed/atom:entry');
  73         }
  74
  75         // Is it RSS?
  76         if ($xpath->query('/rss/channel')->length > 0) {
  77                 $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
  78
  79                 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
  80                 $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
  81
  82                 if ($author["author-name"] == "")
  83                         $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
  84
  85                 if ($author["author-name"] == "")
  86                         $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
  87
  88                 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
  89
  90                 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
  91
  92                 $entries = $xpath->query('/rss/channel/item');
  93         }
  94
  95         if (is_array($contact)) {
  96                 $author["author-link"] = $contact["url"];
  97
  98                 if ($author["author-name"] == "")
  99                         $author["author-name"] = $contact["name"];
 100
 101                 $author["author-avatar"] = $contact["thumb"];
 102
 103                 $author["owner-link"] = $contact["url"];
 104                 $author["owner-name"] = $contact["name"];
 105                 $author["owner-avatar"] = $contact["thumb"];
 106         }
 107
 108         $header = array();
 109         $header["uid"] = $importer["uid"];
 110         $header["network"] = NETWORK_FEED;
 111         $header["type"] = "remote";
 112         $header["wall"] = 0;
 113         $header["origin"] = 0;
 114         $header["gravity"] = GRAVITY_PARENT;
 115         $header["private"] = 2;
 116         $header["verb"] = ACTIVITY_POST;
 117         $header["object-type"] = ACTIVITY_OBJ_NOTE;
 118
 119         $header["contact-id"] = $contact["id"];
 120
 121         if(!strlen($contact["notify"])) {
 122                 // one way feed - no remote comment ability
 123                 $header["last-child"] = 0;
 124         }
 125
 126         if (!is_object($entries))
 127                 return;
 128
 129         $items = array();
 130
 131         $entrylist = array();
 132
 133         foreach ($entries AS $entry)
 134                 $entrylist[] = $entry;
 135
 136         foreach (array_reverse($entrylist) AS $entry) {
 137                 $item = array_merge($header, $author);
 138
 139                 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 140
 141                 if ($item["title"] == "")
 142                         $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 143
 144                 if ($item["title"] == "")
 145                         $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 146
 147                 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 148                 if (!is_object($alternate))
 149                         $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 150
 151                 if (is_object($alternate))
 152                         foreach($alternate AS $attributes)
 153                                 if ($attributes->name == "href")
 154                                         $item["plink"] = $attributes->textContent;
 155
 156                 if ($item["plink"] == "")
 157                         $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 158
 159                 if ($item["plink"] == "")
 160                         $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 161
 162                 $item["plink"] = original_url($item["plink"]);
 163
 164                 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 165
 166                 if ($item["uri"] == "")
 167                         $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 168
 169                 if ($item["uri"] == "")
 170                         $item["uri"] = $item["plink"];
 171
 172                 $item["parent-uri"] = $item["uri"];
 173
 174                 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 175
 176                 if ($published == "")
 177                         $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 178
 179                 if ($published == "")
 180                         $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 181
 182                 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 183
 184                 if ($updated == "")
 185                         $updated = $published;
 186
 187                 if ($published != "")
 188                         $item["created"] = $published;
 189
 190                 if ($updated != "")
 191                         $item["edited"] = $updated;
 192
 193                 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 194
 195                 if ($creator == "")
 196                         $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 197
 198                 if ($creator == "")
 199                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 200
 201                 if ($creator != "")
 202                         $item["author-name"] = $creator;
 203
 204                 if ($pubDate != "")
 205                         $item["edited"] = $item["created"] = $pubDate;
 206
 207                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 208
 209                 if ($creator != "")
 210                         $item["author-name"] = $creator;
 211
 212                 if (!$simulate) {
 213                         $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
 214                                 intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
 215                         if ($r) {
 216                                 logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
 217                                 continue;
 218                         }
 219                 }
 220
 221                 /// @TODO ?
 222                 // <category>Ausland</category>
 223                 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 224
 225                 $attachments = array();
 226
 227                 $enclosures = $xpath->query("enclosure", $entry);
 228                 foreach ($enclosures AS $enclosure) {
 229                         $href = "";
 230                         $length = "";
 231                         $type = "";
 232                         $title = "";
 233
 234                         foreach($enclosure->attributes AS $attributes) {
 235                                 if ($attributes->name == "url")
 236                                         $href = $attributes->textContent;
 237                                 elseif ($attributes->name == "length")
 238                                         $length = $attributes->textContent;
 239                                 elseif ($attributes->name == "type")
 240                                         $type = $attributes->textContent;
 241                         }
 242                         if(strlen($item["attach"]))
 243                                 $item["attach"] .= ',';
 244
 245                         $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 246
 247                         $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 248                 }
 249
 250                 if ($contact["fetch_further_information"]) {
 251                         $preview = "";
 252
 253                         // Handle enclosures and treat them as preview picture
 254                         foreach ($attachments AS $attachment)
 255                                 if ($attachment["type"] == "image/jpeg")
 256                                         $preview = $attachment["link"];
 257
 258                         $item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 259                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 260                         $item["title"] = "";
 261                         $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 262                         unset($item["attach"]);
 263                 } else {
 264                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 265
 266                         if ($body == "")
 267                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 268
 269                         if ($body == "")
 270                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 271
 272                         if ($body == "")
 273                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 274
 275                         // remove the content of the title if it is identically to the body
 276                         // This helps with auto generated titles e.g. from tumblr
 277                         if (title_is_body($item["title"], $body))
 278                                 $item["title"] = "";
 279
 280                         $item["body"] = html2bbcode($body);
 281                 }
 282
 283                 if (!$simulate) {
 284                         logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 285
 286                         $notify = item_is_remote_self($contact, $item);
 287                         $id = item_store($item, false, $notify);
 288
 289                         logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 290                 } else
 291                         $items[] = $item;
 292
 293                 if ($simulate)
 294                         break;
 295         }
 296
 297         if ($simulate)
 298                 return array("header" => $author, "items" => $items);
 299 }
 300 ?>