include/feed.php

   1 <?php
   2 require_once("include/html2bbcode.php");
   3 require_once("include/items.php");
   4
   5 /**
   6  * @brief Read a RSS/RDF/Atom feed and create an item entry for it
   7  *
   8  * @param string $xml The feed data
   9  * @param array $importer The user record of the importer
  10  * @param array $contact The contact record of the feed
  11  * @param string $hub Unused dummy value for compatibility reasons
  12  * @param bool $simulate If enabled, no data is imported
  13  *
  14  * @return array In simulation mode it returns the header and the first item
  15  */
  16 function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
  17
  18         $a = get_app();
  19
  20         if (!$simulate) {
  21                 logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
  22         } else {
  23                 logger("Test Atom/RSS feed", LOGGER_DEBUG);
  24         }
  25         if ($xml == "") {
  26                 logger('XML is empty.', LOGGER_DEBUG);
  27                 return;
  28         }
  29
  30         $doc = new DOMDocument();
  31         @$doc->loadXML($xml);
  32         $xpath = new DomXPath($doc);
  33         $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  34         $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  35         $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  36         $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  37         $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  38         $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  39         $xpath->registerNamespace('poco', NAMESPACE_POCO);
  40
  41         $author = array();
  42
  43         // Is it RDF?
  44         if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  45                 $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  46                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  47
  48                 if ($author["author-name"] == "") {
  49                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  50                 }
  51                 $entries = $xpath->query('/rdf:RDF/rss:item');
  52         }
  53
  54         // Is it Atom?
  55         if ($xpath->query('/atom:feed')->length > 0) {
  56                 $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  57                 if (is_object($alternate)) {
  58                         foreach ($alternate AS $attributes) {
  59                                 if ($attributes->name == "href") {
  60                                         $author["author-link"] = $attributes->textContent;
  61                                 }
  62                         }
  63                 }
  64
  65                 if ($author["author-link"] == "") {
  66                         $author["author-link"] = $author["author-id"];
  67                 }
  68                 if ($author["author-link"] == "") {
  69                         $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  70                         if (is_object($self)) {
  71                                 foreach ($self AS $attributes) {
  72                                         if ($attributes->name == "href") {
  73                                                 $author["author-link"] = $attributes->textContent;
  74                                         }
  75                                 }
  76                         }
  77                 }
  78
  79                 if ($author["author-link"] == "") {
  80                         $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
  81                 }
  82                 $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
  83
  84                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
  85
  86                 if ($author["author-name"] == "") {
  87                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
  88                 }
  89                 if ($author["author-name"] == "") {
  90                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
  91                 }
  92                 $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
  93                 if ($value != "") {
  94                         $author["author-name"] = $value;
  95                 }
  96                 if ($simulate) {
  97                         $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
  98
  99                         $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 100                         if ($value != "") {
 101                                 $author["author-nick"] = $value;
 102                         }
 103                         $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
 104                         if ($value != "") {
 105                                 $author["author-location"] = $value;
 106                         }
 107                         $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 108                         if ($value != "") {
 109                                 $author["author-about"] = $value;
 110                         }
 111                         $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
 112                         if (is_object($avatar)) {
 113                                 foreach ($avatar AS $attributes) {
 114                                         if ($attributes->name == "href") {
 115                                                 $author["author-avatar"] = $attributes->textContent;
 116                                         }
 117                                 }
 118                         }
 119                 }
 120
 121                 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 122
 123                 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 124
 125                 $entries = $xpath->query('/atom:feed/atom:entry');
 126         }
 127
 128         // Is it RSS?
 129         if ($xpath->query('/rss/channel')->length > 0) {
 130                 $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 131
 132                 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 133                 $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 134
 135                 if ($author["author-name"] == "") {
 136                         $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 137                 }
 138                 if ($author["author-name"] == "") {
 139                         $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 140                 }
 141                 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 142
 143                 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 144
 145                 $entries = $xpath->query('/rss/channel/item');
 146         }
 147
 148         if (!$simulate) {
 149                 $author["author-link"] = $contact["url"];
 150
 151                 if ($author["author-name"] == "") {
 152                         $author["author-name"] = $contact["name"];
 153                 }
 154                 $author["author-avatar"] = $contact["thumb"];
 155
 156                 $author["owner-link"] = $contact["url"];
 157                 $author["owner-name"] = $contact["name"];
 158                 $author["owner-avatar"] = $contact["thumb"];
 159         }
 160
 161         $header = array();
 162         $header["uid"] = $importer["uid"];
 163         $header["network"] = NETWORK_FEED;
 164         $header["type"] = "remote";
 165         $header["wall"] = 0;
 166         $header["origin"] = 0;
 167         $header["gravity"] = GRAVITY_PARENT;
 168         $header["private"] = 2;
 169         $header["verb"] = ACTIVITY_POST;
 170         $header["object-type"] = ACTIVITY_OBJ_NOTE;
 171
 172         $header["contact-id"] = $contact["id"];
 173
 174         if (!strlen($contact["notify"])) {
 175                 // one way feed - no remote comment ability
 176                 $header["last-child"] = 0;
 177         }
 178
 179         if (!is_object($entries)) {
 180                 logger("There are no entries in this feed.", LOGGER_DEBUG);
 181                 return;
 182         }
 183
 184         $items = array();
 185
 186         $entrylist = array();
 187
 188         foreach ($entries AS $entry) {
 189                 $entrylist[] = $entry;
 190         }
 191         foreach (array_reverse($entrylist) AS $entry) {
 192                 $item = array_merge($header, $author);
 193
 194                 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 195                 if (!is_object($alternate)) {
 196                         $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 197                 }
 198                 if (is_object($alternate)) {
 199                         foreach ($alternate AS $attributes) {
 200                                 if ($attributes->name == "href") {
 201                                         $item["plink"] = $attributes->textContent;
 202                                 }
 203                         }
 204                 }
 205                 if ($item["plink"] == "") {
 206                         $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 207                 }
 208                 if ($item["plink"] == "") {
 209                         $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 210                 }
 211
 212                 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 213
 214                 if ($item["uri"] == "") {
 215                         $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 216                 }
 217                 if ($item["uri"] == "") {
 218                         $item["uri"] = $item["plink"];
 219                 }
 220
 221                 $orig_plink = $item["plink"];
 222
 223                 $item["plink"] = original_url($item["plink"]);
 224
 225                 $item["parent-uri"] = $item["uri"];
 226
 227                 if (!$simulate) {
 228                         $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
 229                                 intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
 230                         if (dbm::is_result($r)) {
 231                                 logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
 232                                 continue;
 233                         }
 234                 }
 235
 236                 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 237
 238                 if ($item["title"] == "") {
 239                         $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 240                 }
 241                 if ($item["title"] == "") {
 242                         $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 243                 }
 244                 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 245
 246                 if ($published == "") {
 247                         $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 248                 }
 249                 if ($published == "") {
 250                         $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 251                 }
 252                 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 253
 254                 if ($updated == "") {
 255                         $updated = $published;
 256                 }
 257                 if ($published != "") {
 258                         $item["created"] = $published;
 259                 }
 260                 if ($updated != "") {
 261                         $item["edited"] = $updated;
 262                 }
 263                 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 264
 265                 if ($creator == "") {
 266                         $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 267                 }
 268                 if ($creator == "") {
 269                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 270                 }
 271                 if ($creator != "") {
 272                         $item["author-name"] = $creator;
 273                 }
 274                 if ($pubDate != "") {
 275                         $item["edited"] = $item["created"] = $pubDate;
 276                 }
 277                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 278
 279                 if ($creator != "") {
 280                         $item["author-name"] = $creator;
 281                 }
 282
 283                 /// @TODO ?
 284                 // <category>Ausland</category>
 285                 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 286
 287                 $attachments = array();
 288
 289                 $enclosures = $xpath->query("enclosure", $entry);
 290                 foreach ($enclosures AS $enclosure) {
 291                         $href = "";
 292                         $length = "";
 293                         $type = "";
 294                         $title = "";
 295
 296                         foreach ($enclosure->attributes AS $attributes) {
 297                                 if ($attributes->name == "url") {
 298                                         $href = $attributes->textContent;
 299                                 } elseif ($attributes->name == "length") {
 300                                         $length = $attributes->textContent;
 301                                 } elseif ($attributes->name == "type") {
 302                                         $type = $attributes->textContent;
 303                                 }
 304                         }
 305                         if (strlen($item["attach"])) {
 306                                 $item["attach"] .= ',';
 307                         }
 308
 309                         $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 310
 311                         $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 312                 }
 313
 314                 if ($contact["fetch_further_information"]) {
 315                         $preview = "";
 316
 317                         // Handle enclosures and treat them as preview picture
 318                         foreach ($attachments AS $attachment) {
 319                                 if ($attachment["type"] == "image/jpeg") {
 320                                         $preview = $attachment["link"];
 321                                 }
 322                         }
 323
 324                         $item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 325                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 326                         $item["title"] = "";
 327                         $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 328                         unset($item["attach"]);
 329                 } else {
 330                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 331
 332                         if ($body == "") {
 333                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 334                         }
 335                         if ($body == "") {
 336                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 337                         }
 338                         if ($body == "") {
 339                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 340                         }
 341
 342                         // remove the content of the title if it is identically to the body
 343                         // This helps with auto generated titles e.g. from tumblr
 344                         if (title_is_body($item["title"], $body)) {
 345                                 $item["title"] = "";
 346                         }
 347                         $item["body"] = html2bbcode($body);
 348
 349                         if ($item["body"] == "") {
 350                                 if ($item["title"] != '') {
 351                                         $item["body"] = $item["title"];
 352                                         $item["title"] = '';
 353                                 }
 354                         }
 355
 356                         if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
 357                                 $item["body"] .= "\n[hr]\n[url]".$item['plink']."[/url]";
 358                         }
 359                 }
 360
 361                 if (!$simulate) {
 362                         logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 363
 364                         $notify = item_is_remote_self($contact, $item);
 365
 366                         // Distributed items should have a well formatted URI.
 367                         // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 368                         if ($notify) {
 369                                 $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
 370                                 unset($item['uri']);
 371                                 unset($item['parent-uri']);
 372                         }
 373
 374                         $id = item_store($item, false, $notify);
 375
 376                         logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 377                 } else {
 378                         $items[] = $item;
 379                 }
 380                 if ($simulate) {
 381                         break;
 382                 }
 383         }
 384
 385         if ($simulate) {
 386                 return array("header" => $author, "items" => $items);
 387         }
 388 }