src/Protocol/Feed.php

   1 <?php
   2 /**
   3  * @file src/Protocol/Feed.php
   4  * @brief Imports RSS/RDF/Atom feeds
   5  *
   6  */
   7 namespace Friendica\Protocol;
   8
   9 use DOMDocument;
  10 use DOMXPath;
  11 use Friendica\Content\Text\HTML;
  12 use Friendica\Core\System;
  13 use Friendica\Database\DBA;
  14 use Friendica\Model\Item;
  15 use Friendica\Util\Network;
  16 use Friendica\Util\XML;
  17
  18 require_once 'include/dba.php';
  19 require_once 'include/items.php';
  20
  21 /**
  22  * @brief This class contain functions to import feeds
  23  *
  24  */
  25 class Feed {
  26         /**
  27          * @brief Read a RSS/RDF/Atom feed and create an item entry for it
  28          *
  29          * @param string $xml The feed data
  30          * @param array $importer The user record of the importer
  31          * @param array $contact The contact record of the feed
  32          * @param string $hub Unused dummy value for compatibility reasons
  33          * @param bool $simulate If enabled, no data is imported
  34          *
  35          * @return array In simulation mode it returns the header and the first item
  36          */
  37         public static function import($xml, $importer, &$contact, &$hub, $simulate = false) {
  38
  39                 $a = get_app();
  40
  41                 if (!$simulate) {
  42                         logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
  43                 } else {
  44                         logger("Test Atom/RSS feed", LOGGER_DEBUG);
  45                 }
  46                 if (empty($xml)) {
  47                         logger('XML is empty.', LOGGER_DEBUG);
  48                         return;
  49                 }
  50
  51                 if (!empty($contact['poll'])) {
  52                         $basepath = $contact['poll'];
  53                 } elseif (!empty($contact['url'])) {
  54                         $basepath = $contact['url'];
  55                 } else {
  56                         $basepath = '';
  57                 }
  58
  59                 $doc = new DOMDocument();
  60                 @$doc->loadXML(trim($xml));
  61                 $xpath = new DOMXPath($doc);
  62                 $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  63                 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  64                 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  65                 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  66                 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  67                 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  68                 $xpath->registerNamespace('poco', NAMESPACE_POCO);
  69
  70                 $author = [];
  71                 $entries = null;
  72
  73                 // Is it RDF?
  74                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  75                         $author["author-link"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
  76                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
  77
  78                         if (empty($author["author-name"])) {
  79                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
  80                         }
  81                         $entries = $xpath->query('/rdf:RDF/rss:item');
  82                 }
  83
  84                 // Is it Atom?
  85                 if ($xpath->query('/atom:feed')->length > 0) {
  86                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']");
  87                         if (is_object($alternate)) {
  88                                 foreach ($alternate AS $attribute) {
  89                                         if ($attribute->name == "href") {
  90                                                 $author["author-link"] = $attribute->textContent;
  91                                         }
  92                                 }
  93                         }
  94
  95                         if (empty($author["author-link"])) {
  96                                 $self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']");
  97                                 if (is_object($self)) {
  98                                         foreach ($self AS $attribute) {
  99                                                 if ($attribute->name == "href") {
 100                                                         $author["author-link"] = $attribute->textContent;
 101                                                 }
 102                                         }
 103                                 }
 104                         }
 105
 106                         if (empty($author["author-link"])) {
 107                                 $author["author-link"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()');
 108                         }
 109                         $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()');
 110
 111                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()');
 112
 113                         if (empty($author["author-name"])) {
 114                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
 115                         }
 116                         if (empty($author["author-name"])) {
 117                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
 118                         }
 119                         $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
 120                         if ($value != "") {
 121                                 $author["author-name"] = $value;
 122                         }
 123                         if ($simulate) {
 124                                 $author["author-id"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:uri/text()');
 125
 126                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()');
 127                                 if ($value != "") {
 128                                         $author["author-nick"] = $value;
 129                                 }
 130                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()');
 131                                 if ($value != "") {
 132                                         $author["author-location"] = $value;
 133                                 }
 134                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()');
 135                                 if ($value != "") {
 136                                         $author["author-about"] = $value;
 137                                 }
 138                                 $avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']");
 139                                 if (is_object($avatar)) {
 140                                         foreach ($avatar AS $attribute) {
 141                                                 if ($attribute->name == "href") {
 142                                                         $author["author-avatar"] = $attribute->textContent;
 143                                                 }
 144                                         }
 145                                 }
 146                         }
 147
 148                         $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()');
 149
 150                         $author["app"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()');
 151
 152                         $entries = $xpath->query('/atom:feed/atom:entry');
 153                 }
 154
 155                 // Is it RSS?
 156                 if ($xpath->query('/rss/channel')->length > 0) {
 157                         $author["author-link"] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
 158
 159                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
 160                         $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
 161
 162                         if (empty($author["author-name"])) {
 163                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
 164                         }
 165                         if (empty($author["author-name"])) {
 166                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
 167                         }
 168                         $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
 169
 170                         $author["app"] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
 171
 172                         $entries = $xpath->query('/rss/channel/item');
 173                 }
 174
 175                 if (!$simulate) {
 176                         $author["author-link"] = $contact["url"];
 177
 178                         if (empty($author["author-name"])) {
 179                                 $author["author-name"] = $contact["name"];
 180                         }
 181                         $author["author-avatar"] = $contact["thumb"];
 182
 183                         $author["owner-link"] = $contact["url"];
 184                         $author["owner-name"] = $contact["name"];
 185                         $author["owner-avatar"] = $contact["thumb"];
 186                 }
 187
 188                 $header = [];
 189                 $header["uid"] = $importer["uid"];
 190                 $header["network"] = NETWORK_FEED;
 191                 $header["wall"] = 0;
 192                 $header["origin"] = 0;
 193                 $header["gravity"] = GRAVITY_PARENT;
 194                 $header["private"] = 2;
 195                 $header["verb"] = ACTIVITY_POST;
 196                 $header["object-type"] = ACTIVITY_OBJ_NOTE;
 197
 198                 $header["contact-id"] = $contact["id"];
 199
 200                 if (!is_object($entries)) {
 201                         logger("There are no entries in this feed.", LOGGER_DEBUG);
 202                         return;
 203                 }
 204
 205                 $items = [];
 206                 // Importing older entries first
 207                 for($i = $entries->length - 1; $i >= 0;--$i) {
 208                         $entry = $entries->item($i);
 209
 210                         $item = array_merge($header, $author);
 211
 212                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry);
 213                         if (!is_object($alternate)) {
 214                                 $alternate = XML::getFirstAttributes($xpath, "atom:link", $entry);
 215                         }
 216                         if (is_object($alternate)) {
 217                                 foreach ($alternate AS $attribute) {
 218                                         if ($attribute->name == "href") {
 219                                                 $item["plink"] = $attribute->textContent;
 220                                         }
 221                                 }
 222                         }
 223                         if (empty($item["plink"])) {
 224                                 $item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
 225                         }
 226                         if (empty($item["plink"])) {
 227                                 $item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
 228                         }
 229
 230                         $item["uri"] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry);
 231
 232                         if (empty($item["uri"])) {
 233                                 $item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
 234                         }
 235                         if (empty($item["uri"])) {
 236                                 $item["uri"] = $item["plink"];
 237                         }
 238
 239                         $orig_plink = $item["plink"];
 240
 241                         $item["plink"] = Network::finalUrl($item["plink"]);
 242
 243                         $item["parent-uri"] = $item["uri"];
 244
 245                         if (!$simulate) {
 246                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 247                                         $importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
 248                                 $previous = Item::selectFirst(['id'], $condition);
 249                                 if (DBA::isResult($previous)) {
 250                                         logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG);
 251                                         continue;
 252                                 }
 253                         }
 254
 255                         $item["title"] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
 256
 257                         if (empty($item["title"])) {
 258                                 $item["title"] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
 259                         }
 260                         if (empty($item["title"])) {
 261                                 $item["title"] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
 262                         }
 263                         $published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry);
 264
 265                         if (empty($published)) {
 266                                 $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
 267                         }
 268                         if (empty($published)) {
 269                                 $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
 270                         }
 271                         $updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry);
 272
 273                         if (empty($updated)) {
 274                                 $updated = $published;
 275                         }
 276                         if ($published != "") {
 277                                 $item["created"] = $published;
 278                         }
 279                         if ($updated != "") {
 280                                 $item["edited"] = $updated;
 281                         }
 282                         $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
 283
 284                         if (empty($creator)) {
 285                                 $creator = XML::getFirstNodeValue($xpath, 'atom:author/atom:name/text()', $entry);
 286                         }
 287                         if (empty($creator)) {
 288                                 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 289                         }
 290                         if ($creator != "") {
 291                                 $item["author-name"] = $creator;
 292                         }
 293                         $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 294
 295                         if ($creator != "") {
 296                                 $item["author-name"] = $creator;
 297                         }
 298
 299                         /// @TODO ?
 300                         // <category>Ausland</category>
 301                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 302
 303                         $attachments = [];
 304
 305                         $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
 306                         foreach ($enclosures AS $enclosure) {
 307                                 $href = "";
 308                                 $length = "";
 309                                 $type = "";
 310                                 $title = "";
 311
 312                                 foreach ($enclosure->attributes AS $attribute) {
 313                                         if (in_array($attribute->name, ["url", "href"])) {
 314                                                 $href = $attribute->textContent;
 315                                         } elseif ($attribute->name == "length") {
 316                                                 $length = $attribute->textContent;
 317                                         } elseif ($attribute->name == "type") {
 318                                                 $type = $attribute->textContent;
 319                                         }
 320                                 }
 321                                 if (!empty($item["attach"])) {
 322                                         $item["attach"] .= ',';
 323                                 } else {
 324                                         $item["attach"] = '';
 325                                 }
 326
 327                                 $attachments[] = ["link" => $href, "type" => $type, "length" => $length];
 328
 329                                 $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 330                         }
 331
 332                         $tags = '';
 333                         $categories = $xpath->query("category", $entry);
 334                         foreach ($categories AS $category) {
 335                                 $hashtag = $category->nodeValue;
 336                                 if ($tags != '') {
 337                                         $tags .= ', ';
 338                                 }
 339
 340                                 $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
 341                                 $tags .= $taglink;
 342                         }
 343
 344                         $body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry));
 345
 346                         if (empty($body)) {
 347                                 $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
 348                         }
 349                         if (empty($body)) {
 350                                 $body = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
 351                         }
 352                         if (empty($body)) {
 353                                 $body = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry));
 354                         }
 355
 356                         // remove the content of the title if it is identically to the body
 357                         // This helps with auto generated titles e.g. from tumblr
 358                         if (self::titleIsBody($item["title"], $body)) {
 359                                 $item["title"] = "";
 360                         }
 361                         $item["body"] = HTML::toBBCode($body, $basepath);
 362
 363                         if (($item["body"] == '') && ($item["title"] != '')) {
 364                                 $item["body"] = $item["title"];
 365                                 $item["title"] = '';
 366                         }
 367
 368                         $preview = '';
 369                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 370                                 // Handle enclosures and treat them as preview picture
 371                                 foreach ($attachments AS $attachment) {
 372                                         if ($attachment["type"] == "image/jpeg") {
 373                                                 $preview = $attachment["link"];
 374                                         }
 375                                 }
 376
 377                                 // Remove a possible link to the item itself
 378                                 $item["body"] = str_replace($item["plink"], '', $item["body"]);
 379                                 $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
 380
 381                                 // Replace the content when the title is longer than the body
 382                                 $replace = (strlen($item["title"]) > strlen($item["body"]));
 383
 384                                 // Replace it, when there is an image in the body
 385                                 if (strstr($item["body"], '[/img]')) {
 386                                         $replace = true;
 387                                 }
 388
 389                                 // Replace it, when there is a link in the body
 390                                 if (strstr($item["body"], '[/url]')) {
 391                                         $replace = true;
 392                                 }
 393
 394                                 if ($replace) {
 395                                         $item["body"] = $item["title"];
 396                                 }
 397                                 // We always strip the title since it will be added in the page information
 398                                 $item["title"] = "";
 399                                 $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 400                                 $item["tag"] = add_page_keywords($item["plink"], $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 401                                 $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 402                                 unset($item["attach"]);
 403                         } else {
 404                                 if ($contact["fetch_further_information"] == 3) {
 405                                         if (!empty($tags)) {
 406                                                 $item["tag"] = $tags;
 407                                         } else {
 408                                                 // @todo $preview is never set in this case, is it intended? - @MrPetovan 2018-02-13
 409                                                 $item["tag"] = add_page_keywords($item["plink"], $preview, true, $contact["ffi_keyword_blacklist"]);
 410                                         }
 411                                         $item["body"] .= "\n".$item['tag'];
 412                                 }
 413                                 // Add the link to the original feed entry if not present in feed
 414                                 if (($item['plink'] != '') && !strstr($item["body"], $item['plink'])) {
 415                                         $item["body"] .= "[hr][url]".$item['plink']."[/url]";
 416                                 }
 417                         }
 418
 419                         if (!$simulate) {
 420                                 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 421
 422                                 $notify = Item::isRemoteSelf($contact, $item);
 423
 424                                 // Distributed items should have a well formatted URI.
 425                                 // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 426                                 if ($notify) {
 427                                         $item['guid'] = Item::guidFromUri($orig_plink, $a->get_hostname());
 428                                         unset($item['uri']);
 429                                         unset($item['parent-uri']);
 430
 431                                         // Set the delivery priority for "remote self" to "medium"
 432                                         $notify = PRIORITY_MEDIUM;
 433                                 }
 434
 435                                 $id = Item::insert($item, false, $notify);
 436
 437                                 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 438                         } else {
 439                                 $items[] = $item;
 440                         }
 441                         if ($simulate) {
 442                                 break;
 443                         }
 444                 }
 445
 446                 if ($simulate) {
 447                         return ["header" => $author, "items" => $items];
 448                 }
 449         }
 450
 451         private static function titleIsBody($title, $body)
 452         {
 453                 $title = strip_tags($title);
 454                 $title = trim($title);
 455                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 456                 $title = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $title);
 457
 458                 $body = strip_tags($body);
 459                 $body = trim($body);
 460                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 461                 $body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
 462
 463                 if (strlen($title) < strlen($body)) {
 464                         $body = substr($body, 0, strlen($title));
 465                 }
 466
 467                 if (($title != $body) && (substr($title, -3) == "...")) {
 468                         $pos = strrpos($title, "...");
 469                         if ($pos > 0) {
 470                                 $title = substr($title, 0, $pos);
 471                                 $body = substr($body, 0, $pos);
 472                         }
 473                 }
 474                 return ($title == $body);
 475         }
 476 }