include/feed.php

   1 <?php
   2 use Friendica\Database\DBM;
   3 require_once("include/html2bbcode.php");
   4 require_once("include/items.php");
   5
   6 /**
   7  * @brief Read a RSS/RDF/Atom feed and create an item entry for it
   8  *
   9  * @param string $xml The feed data
  10  * @param array $importer The user record of the importer
  11  * @param array $contact The contact record of the feed
  12  * @param string $hub Unused dummy value for compatibility reasons
  13  * @param bool $simulate If enabled, no data is imported
  14  *
  15  * @return array In simulation mode it returns the header and the first item
  16  */
  17 function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
  18
  19         $a = get_app();
  20
  21         if (!$simulate) {
  22                 logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
  23         } else {
  24                 logger("Test Atom/RSS feed", LOGGER_DEBUG);
  25         }
  26         if ($xml == "") {
  27                 logger('XML is empty.', LOGGER_DEBUG);
  28                 return;
  29         }
  30
  31         if (!empty($contact['poll'])) {
  32                 $basepath = $contact['poll'];
  33         } elseif (!empty($contact['url'])) {
  34                 $basepath = $contact['url'];
  35         } else {
  36                 $basepath = '';
  37         }
  38
  39         $doc = new DOMDocument();
  40         @$doc->loadXML(trim($xml));
  41         $xpath = new DomXPath($doc);
  42         $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  43         $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  44         $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  45         $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  46         $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  47         $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  48         $xpath->registerNamespace('poco', NAMESPACE_POCO);
  49
  50         $author = array();
  51
  52         // Is it RDF?
  53         if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  54                 $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  55                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  56
  57                 if ($author["author-name"] == "") {
  58                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  59                 }
  60                 $entries = $xpath->query('/rdf:RDF/rss:item');
  61         }
  62
  63         // Is it Atom?
  64         if ($xpath->query('/atom:feed')->length > 0) {
  65                 $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  66                 if (is_object($alternate)) {
  67                         foreach ($alternate AS $attributes) {
  68                                 if ($attributes->name == "href") {
  69                                         $author["author-link"] = $attributes->textContent;
  70                                 }
  71                         }
  72                 }
  73
  74                 if ($author["author-link"] == "") {
  75                         $author["author-link"] = $author["author-id"];
  76                 }
  77                 if ($author["author-link"] == "") {
  78                         $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  79                         if (is_object($self)) {
  80                                 foreach ($self AS $attributes) {
  81                                         if ($attributes->name == "href") {
  82                                                 $author["author-link"] = $attributes->textContent;
  83                                         }
  84                                 }
  85                         }
  86                 }
  87
  88                 if ($author["author-link"] == "") {
  89                         $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
  90                 }
  91                 $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
  92
  93                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
  94
  95                 if ($author["author-name"] == "") {
  96                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
  97                 }
  98                 if ($author["author-name"] == "") {
  99                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
 100                 }
 101                 $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
 102                 if ($value != "") {
 103                         $author["author-name"] = $value;
 104                 }
 105                 if ($simulate) {
 106                         $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
 107
 108                         $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 109                         if ($value != "") {
 110                                 $author["author-nick"] = $value;
 111                         }
 112                         $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
 113                         if ($value != "") {
 114                                 $author["author-location"] = $value;
 115                         }
 116                         $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 117                         if ($value != "") {
 118                                 $author["author-about"] = $value;
 119                         }
 120                         $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
 121                         if (is_object($avatar)) {
 122                                 foreach ($avatar AS $attributes) {
 123                                         if ($attributes->name == "href") {
 124                                                 $author["author-avatar"] = $attributes->textContent;
 125                                         }
 126                                 }
 127                         }
 128                 }
 129
 130                 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 131
 132                 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 133
 134                 $entries = $xpath->query('/atom:feed/atom:entry');
 135         }
 136
 137         // Is it RSS?
 138         if ($xpath->query('/rss/channel')->length > 0) {
 139                 $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 140
 141                 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 142                 $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 143
 144                 if ($author["author-name"] == "") {
 145                         $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 146                 }
 147                 if ($author["author-name"] == "") {
 148                         $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 149                 }
 150                 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 151
 152                 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 153
 154                 $entries = $xpath->query('/rss/channel/item');
 155         }
 156
 157         if (!$simulate) {
 158                 $author["author-link"] = $contact["url"];
 159
 160                 if ($author["author-name"] == "") {
 161                         $author["author-name"] = $contact["name"];
 162                 }
 163                 $author["author-avatar"] = $contact["thumb"];
 164
 165                 $author["owner-link"] = $contact["url"];
 166                 $author["owner-name"] = $contact["name"];
 167                 $author["owner-avatar"] = $contact["thumb"];
 168         }
 169
 170         $header = array();
 171         $header["uid"] = $importer["uid"];
 172         $header["network"] = NETWORK_FEED;
 173         $header["type"] = "remote";
 174         $header["wall"] = 0;
 175         $header["origin"] = 0;
 176         $header["gravity"] = GRAVITY_PARENT;
 177         $header["private"] = 2;
 178         $header["verb"] = ACTIVITY_POST;
 179         $header["object-type"] = ACTIVITY_OBJ_NOTE;
 180
 181         $header["contact-id"] = $contact["id"];
 182
 183         if (!strlen($contact["notify"])) {
 184                 // one way feed - no remote comment ability
 185                 $header["last-child"] = 0;
 186         }
 187
 188         if (!is_object($entries)) {
 189                 logger("There are no entries in this feed.", LOGGER_DEBUG);
 190                 return;
 191         }
 192
 193         $items = array();
 194
 195         $entrylist = array();
 196
 197         foreach ($entries AS $entry) {
 198                 $entrylist[] = $entry;
 199         }
 200         foreach (array_reverse($entrylist) AS $entry) {
 201                 $item = array_merge($header, $author);
 202
 203                 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 204                 if (!is_object($alternate)) {
 205                         $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 206                 }
 207                 if (is_object($alternate)) {
 208                         foreach ($alternate AS $attributes) {
 209                                 if ($attributes->name == "href") {
 210                                         $item["plink"] = $attributes->textContent;
 211                                 }
 212                         }
 213                 }
 214                 if ($item["plink"] == "") {
 215                         $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 216                 }
 217                 if ($item["plink"] == "") {
 218                         $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 219                 }
 220
 221                 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 222
 223                 if ($item["uri"] == "") {
 224                         $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 225                 }
 226                 if ($item["uri"] == "") {
 227                         $item["uri"] = $item["plink"];
 228                 }
 229
 230                 $orig_plink = $item["plink"];
 231
 232                 $item["plink"] = original_url($item["plink"]);
 233
 234                 $item["parent-uri"] = $item["uri"];
 235
 236                 if (!$simulate) {
 237                         $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
 238                                 intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
 239                         if (DBM::is_result($r)) {
 240                                 logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
 241                                 continue;
 242                         }
 243                 }
 244
 245                 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 246
 247                 if ($item["title"] == "") {
 248                         $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 249                 }
 250                 if ($item["title"] == "") {
 251                         $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 252                 }
 253                 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 254
 255                 if ($published == "") {
 256                         $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 257                 }
 258                 if ($published == "") {
 259                         $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 260                 }
 261                 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 262
 263                 if ($updated == "") {
 264                         $updated = $published;
 265                 }
 266                 if ($published != "") {
 267                         $item["created"] = $published;
 268                 }
 269                 if ($updated != "") {
 270                         $item["edited"] = $updated;
 271                 }
 272                 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 273
 274                 if ($creator == "") {
 275                         $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 276                 }
 277                 if ($creator == "") {
 278                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 279                 }
 280                 if ($creator != "") {
 281                         $item["author-name"] = $creator;
 282                 }
 283                 if ($pubDate != "") {
 284                         $item["edited"] = $item["created"] = $pubDate;
 285                 }
 286                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 287
 288                 if ($creator != "") {
 289                         $item["author-name"] = $creator;
 290                 }
 291
 292                 /// @TODO ?
 293                 // <category>Ausland</category>
 294                 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 295
 296                 $attachments = array();
 297
 298                 $enclosures = $xpath->query("enclosure", $entry);
 299                 foreach ($enclosures AS $enclosure) {
 300                         $href = "";
 301                         $length = "";
 302                         $type = "";
 303                         $title = "";
 304
 305                         foreach ($enclosure->attributes AS $attributes) {
 306                                 if ($attributes->name == "url") {
 307                                         $href = $attributes->textContent;
 308                                 } elseif ($attributes->name == "length") {
 309                                         $length = $attributes->textContent;
 310                                 } elseif ($attributes->name == "type") {
 311                                         $type = $attributes->textContent;
 312                                 }
 313                         }
 314                         if (strlen($item["attach"])) {
 315                                 $item["attach"] .= ',';
 316                         }
 317
 318                         $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 319
 320                         $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 321                 }
 322
 323                 $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 324
 325                 if ($body == "") {
 326                         $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 327                 }
 328                 if ($body == "") {
 329                         $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 330                 }
 331                 if ($body == "") {
 332                         $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 333                 }
 334
 335                 // remove the content of the title if it is identically to the body
 336                 // This helps with auto generated titles e.g. from tumblr
 337                 if (title_is_body($item["title"], $body)) {
 338                         $item["title"] = "";
 339                 }
 340                 $item["body"] = html2bbcode($body, $basepath);
 341
 342                 if (($item["body"] == '') && ($item["title"] != '')) {
 343                         $item["body"] = $item["title"];
 344                         $item["title"] = '';
 345                 }
 346
 347                 if ($contact["fetch_further_information"]) {
 348                         $preview = "";
 349
 350                         // Handle enclosures and treat them as preview picture
 351                         foreach ($attachments AS $attachment) {
 352                                 if ($attachment["type"] == "image/jpeg") {
 353                                         $preview = $attachment["link"];
 354                                 }
 355                         }
 356
 357                         // Remove a possible link to the item itself
 358                         $item["body"] = str_replace($item["plink"], '', $item["body"]);
 359                         $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
 360
 361                         // Replace the content when the title is longer than the body
 362                         $replace = (strlen($item["title"]) > strlen($item["body"]));
 363
 364                         // Replace it, when there is an image in the body
 365                         if (strstr($item["body"], '[/img]')) {
 366                                 $replace = true;
 367                         }
 368
 369                         // Replace it, when there is a link in the body
 370                         if (strstr($item["body"], '[/url]')) {
 371                                 $replace = true;
 372                         }
 373
 374                         if ($replace) {
 375                                 $item["body"] = $item["title"];
 376                         }
 377                         // We always strip the title since it will be added in the page information
 378                         $item["title"] = "";
 379                         $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 380                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 381                         $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 382                         unset($item["attach"]);
 383                 } else {
 384                         if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
 385                                 $item["body"] .= "[hr][url]".$item['plink']."[/url]";
 386                         }
 387                 }
 388
 389                 if (!$simulate) {
 390                         logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 391
 392                         $notify = item_is_remote_self($contact, $item);
 393
 394                         // Distributed items should have a well formatted URI.
 395                         // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 396                         if ($notify) {
 397                                 $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
 398                                 unset($item['uri']);
 399                                 unset($item['parent-uri']);
 400                         }
 401
 402                         $id = item_store($item, false, $notify);
 403
 404                         logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 405                 } else {
 406                         $items[] = $item;
 407                 }
 408                 if ($simulate) {
 409                         break;
 410                 }
 411         }
 412
 413         if ($simulate) {
 414                 return array("header" => $author, "items" => $items);
 415         }
 416 }