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 Friendica\Database\DBM;
  10 use Friendica\Core\System;
  11 use Friendica\Model\Item;
  12 use Friendica\Util\Network;
  13 use dba;
  14 use DOMDocument;
  15 use DOMXPath;
  16
  17 require_once 'include/dba.php';
  18 require_once 'include/items.php';
  19
  20 /**
  21  * @brief This class contain functions to import feeds
  22  *
  23  */
  24 class Feed {
  25         /**
  26          * @brief Read a RSS/RDF/Atom feed and create an item entry for it
  27          *
  28          * @param string $xml The feed data
  29          * @param array $importer The user record of the importer
  30          * @param array $contact The contact record of the feed
  31          * @param string $hub Unused dummy value for compatibility reasons
  32          * @param bool $simulate If enabled, no data is imported
  33          *
  34          * @return array In simulation mode it returns the header and the first item
  35          */
  36         public static function import($xml, $importer, &$contact, &$hub, $simulate = false) {
  37
  38                 $a = get_app();
  39
  40                 if (!$simulate) {
  41                         logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
  42                 } else {
  43                         logger("Test Atom/RSS feed", LOGGER_DEBUG);
  44                 }
  45                 if ($xml == "") {
  46                         logger('XML is empty.', LOGGER_DEBUG);
  47                         return;
  48                 }
  49
  50                 if (!empty($contact['poll'])) {
  51                         $basepath = $contact['poll'];
  52                 } elseif (!empty($contact['url'])) {
  53                         $basepath = $contact['url'];
  54                 } else {
  55                         $basepath = '';
  56                 }
  57
  58                 $doc = new DOMDocument();
  59                 @$doc->loadXML(trim($xml));
  60                 $xpath = new DOMXPath($doc);
  61                 $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  62                 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  63                 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  64                 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  65                 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  66                 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  67                 $xpath->registerNamespace('poco', NAMESPACE_POCO);
  68
  69                 $author = [];
  70                 $entries = null;
  71
  72                 // Is it RDF?
  73                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  74                         $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  75                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  76
  77                         if ($author["author-name"] == "") {
  78                                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  79                         }
  80                         $entries = $xpath->query('/rdf:RDF/rss:item');
  81                 }
  82
  83                 // Is it Atom?
  84                 if ($xpath->query('/atom:feed')->length > 0) {
  85                         $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  86                         if (is_object($alternate)) {
  87                                 foreach ($alternate AS $attributes) {
  88                                         if ($attributes->name == "href") {
  89                                                 $author["author-link"] = $attributes->textContent;
  90                                         }
  91                                 }
  92                         }
  93
  94                         if ($author["author-link"] == "") {
  95                                 $author["author-link"] = $author["author-id"];
  96                         }
  97                         if ($author["author-link"] == "") {
  98                                 $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  99                                 if (is_object($self)) {
 100                                         foreach ($self AS $attributes) {
 101                                                 if ($attributes->name == "href") {
 102                                                         $author["author-link"] = $attributes->textContent;
 103                                                 }
 104                                         }
 105                                 }
 106                         }
 107
 108                         if ($author["author-link"] == "") {
 109                                 $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
 110                         }
 111                         $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
 112
 113                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
 114
 115                         if ($author["author-name"] == "") {
 116                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
 117                         }
 118                         if ($author["author-name"] == "") {
 119                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
 120                         }
 121                         $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
 122                         if ($value != "") {
 123                                 $author["author-name"] = $value;
 124                         }
 125                         if ($simulate) {
 126                                 $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
 127
 128                                 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 129                                 if ($value != "") {
 130                                         $author["author-nick"] = $value;
 131                                 }
 132                                 $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()')->item(0)->nodeValue;
 133                                 if ($value != "") {
 134                                         $author["author-location"] = $value;
 135                                 }
 136                                 $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 137                                 if ($value != "") {
 138                                         $author["author-about"] = $value;
 139                                 }
 140                                 $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
 141                                 if (is_object($avatar)) {
 142                                         foreach ($avatar AS $attributes) {
 143                                                 if ($attributes->name == "href") {
 144                                                         $author["author-avatar"] = $attributes->textContent;
 145                                                 }
 146                                         }
 147                                 }
 148                         }
 149
 150                         $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 151
 152                         $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 153
 154                         $entries = $xpath->query('/atom:feed/atom:entry');
 155                 }
 156
 157                 // Is it RSS?
 158                 if ($xpath->query('/rss/channel')->length > 0) {
 159                         $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 160
 161                         $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 162                         $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 163
 164                         if ($author["author-name"] == "") {
 165                                 $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 166                         }
 167                         if ($author["author-name"] == "") {
 168                                 $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 169                         }
 170                         $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 171
 172                         $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 173
 174                         $entries = $xpath->query('/rss/channel/item');
 175                 }
 176
 177                 if (!$simulate) {
 178                         $author["author-link"] = $contact["url"];
 179
 180                         if ($author["author-name"] == "") {
 181                                 $author["author-name"] = $contact["name"];
 182                         }
 183                         $author["author-avatar"] = $contact["thumb"];
 184
 185                         $author["owner-link"] = $contact["url"];
 186                         $author["owner-name"] = $contact["name"];
 187                         $author["owner-avatar"] = $contact["thumb"];
 188                 }
 189
 190                 $header = [];
 191                 $header["uid"] = $importer["uid"];
 192                 $header["network"] = NETWORK_FEED;
 193                 $header["type"] = "remote";
 194                 $header["wall"] = 0;
 195                 $header["origin"] = 0;
 196                 $header["gravity"] = GRAVITY_PARENT;
 197                 $header["private"] = 2;
 198                 $header["verb"] = ACTIVITY_POST;
 199                 $header["object-type"] = ACTIVITY_OBJ_NOTE;
 200
 201                 $header["contact-id"] = $contact["id"];
 202
 203                 if (!is_object($entries)) {
 204                         logger("There are no entries in this feed.", LOGGER_DEBUG);
 205                         return;
 206                 }
 207
 208                 $items = [];
 209
 210                 $entrylist = [];
 211
 212                 foreach ($entries AS $entry) {
 213                         $entrylist[] = $entry;
 214                 }
 215                 foreach (array_reverse($entrylist) AS $entry) {
 216                         $item = array_merge($header, $author);
 217
 218                         $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 219                         if (!is_object($alternate)) {
 220                                 $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 221                         }
 222                         if (is_object($alternate)) {
 223                                 foreach ($alternate AS $attributes) {
 224                                         if ($attributes->name == "href") {
 225                                                 $item["plink"] = $attributes->textContent;
 226                                         }
 227                                 }
 228                         }
 229                         if ($item["plink"] == "") {
 230                                 $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 231                         }
 232                         if ($item["plink"] == "") {
 233                                 $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 234                         }
 235
 236                         $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 237
 238                         if ($item["uri"] == "") {
 239                                 $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 240                         }
 241                         if ($item["uri"] == "") {
 242                                 $item["uri"] = $item["plink"];
 243                         }
 244
 245                         $orig_plink = $item["plink"];
 246
 247                         $item["plink"] = Network::finalUrl($item["plink"]);
 248
 249                         $item["parent-uri"] = $item["uri"];
 250
 251                         if (!$simulate) {
 252                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 253                                         $importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
 254                                 $previous = dba::selectFirst('item', ['id'], $condition);
 255                                 if (DBM::is_result($previous)) {
 256                                         logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG);
 257                                         continue;
 258                                 }
 259                         }
 260
 261                         $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 262
 263                         if ($item["title"] == "") {
 264                                 $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 265                         }
 266                         if ($item["title"] == "") {
 267                                 $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 268                         }
 269                         $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 270
 271                         if ($published == "") {
 272                                 $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 273                         }
 274                         if ($published == "") {
 275                                 $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 276                         }
 277                         $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 278
 279                         if ($updated == "") {
 280                                 $updated = $published;
 281                         }
 282                         if ($published != "") {
 283                                 $item["created"] = $published;
 284                         }
 285                         if ($updated != "") {
 286                                 $item["edited"] = $updated;
 287                         }
 288                         $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 289
 290                         if ($creator == "") {
 291                                 $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 292                         }
 293                         if ($creator == "") {
 294                                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 295                         }
 296                         if ($creator != "") {
 297                                 $item["author-name"] = $creator;
 298                         }
 299                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 300
 301                         if ($creator != "") {
 302                                 $item["author-name"] = $creator;
 303                         }
 304
 305                         /// @TODO ?
 306                         // <category>Ausland</category>
 307                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 308
 309                         $attachments = [];
 310
 311                         $enclosures = $xpath->query("enclosure", $entry);
 312                         foreach ($enclosures AS $enclosure) {
 313                                 $href = "";
 314                                 $length = "";
 315                                 $type = "";
 316                                 $title = "";
 317
 318                                 foreach ($enclosure->attributes AS $attributes) {
 319                                         if ($attributes->name == "url") {
 320                                                 $href = $attributes->textContent;
 321                                         } elseif ($attributes->name == "length") {
 322                                                 $length = $attributes->textContent;
 323                                         } elseif ($attributes->name == "type") {
 324                                                 $type = $attributes->textContent;
 325                                         }
 326                                 }
 327                                 if (strlen($item["attach"])) {
 328                                         $item["attach"] .= ',';
 329                                 }
 330
 331                                 $attachments[] = ["link" => $href, "type" => $type, "length" => $length];
 332
 333                                 $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 334                         }
 335
 336                         $tags = '';
 337                         $categories = $xpath->query("category", $entry);
 338                         foreach ($categories AS $category) {
 339                                 $hashtag = $category->nodeValue;
 340                                 if ($tags != '') {
 341                                         $tags .= ', ';
 342                                 }
 343
 344                                 $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
 345                                 $tags .= $taglink;
 346                         }
 347
 348                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 349
 350                         if ($body == "") {
 351                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 352                         }
 353                         if ($body == "") {
 354                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 355                         }
 356                         if ($body == "") {
 357                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 358                         }
 359
 360                         // remove the content of the title if it is identically to the body
 361                         // This helps with auto generated titles e.g. from tumblr
 362                         if (self::titleIsBody($item["title"], $body)) {
 363                                 $item["title"] = "";
 364                         }
 365                         $item["body"] = Friendica\Content\Text\HTML::toBBCode($body, $basepath);
 366
 367                         if (($item["body"] == '') && ($item["title"] != '')) {
 368                                 $item["body"] = $item["title"];
 369                                 $item["title"] = '';
 370                         }
 371
 372                         $preview = '';
 373                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 374                                 // Handle enclosures and treat them as preview picture
 375                                 foreach ($attachments AS $attachment) {
 376                                         if ($attachment["type"] == "image/jpeg") {
 377                                                 $preview = $attachment["link"];
 378                                         }
 379                                 }
 380
 381                                 // Remove a possible link to the item itself
 382                                 $item["body"] = str_replace($item["plink"], '', $item["body"]);
 383                                 $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
 384
 385                                 // Replace the content when the title is longer than the body
 386                                 $replace = (strlen($item["title"]) > strlen($item["body"]));
 387
 388                                 // Replace it, when there is an image in the body
 389                                 if (strstr($item["body"], '[/img]')) {
 390                                         $replace = true;
 391                                 }
 392
 393                                 // Replace it, when there is a link in the body
 394                                 if (strstr($item["body"], '[/url]')) {
 395                                         $replace = true;
 396                                 }
 397
 398                                 if ($replace) {
 399                                         $item["body"] = $item["title"];
 400                                 }
 401                                 // We always strip the title since it will be added in the page information
 402                                 $item["title"] = "";
 403                                 $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 404                                 $item["tag"] = add_page_keywords($item["plink"], $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 405                                 $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 406                                 unset($item["attach"]);
 407                         } else {
 408                                 if ($contact["fetch_further_information"] == 3) {
 409                                         if (!empty($tags)) {
 410                                                 $item["tag"] = $tags;
 411                                         } else {
 412                                                 // @todo $preview is never set in this case, is it intended? - @MrPetovan 2018-02-13
 413                                                 $item["tag"] = add_page_keywords($item["plink"], $preview, true, $contact["ffi_keyword_blacklist"]);
 414                                         }
 415                                         $item["body"] .= "\n".$item['tag'];
 416                                 }
 417                                 // Add the link to the original feed entry if not present in feed
 418                                 if (($item['plink'] != '') && !strstr($item["body"], $item['plink'])) {
 419                                         $item["body"] .= "[hr][url]".$item['plink']."[/url]";
 420                                 }
 421                         }
 422
 423                         if (!$simulate) {
 424                                 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 425
 426                                 $notify = Item::isRemoteSelf($contact, $item);
 427
 428                                 // Distributed items should have a well formatted URI.
 429                                 // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 430                                 if ($notify) {
 431                                         $item['guid'] = Item::guidFromUri($orig_plink, $a->get_hostname());
 432                                         unset($item['uri']);
 433                                         unset($item['parent-uri']);
 434                                 }
 435
 436                                 $id = Item::insert($item, false, $notify);
 437
 438                                 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 439                         } else {
 440                                 $items[] = $item;
 441                         }
 442                         if ($simulate) {
 443                                 break;
 444                         }
 445                 }
 446
 447                 if ($simulate) {
 448                         return ["header" => $author, "items" => $items];
 449                 }
 450         }
 451
 452         private static function titleIsBody($title, $body)
 453         {
 454                 $title = strip_tags($title);
 455                 $title = trim($title);
 456                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 457                 $title = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $title);
 458
 459                 $body = strip_tags($body);
 460                 $body = trim($body);
 461                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 462                 $body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
 463
 464                 if (strlen($title) < strlen($body)) {
 465                         $body = substr($body, 0, strlen($title));
 466                 }
 467
 468                 if (($title != $body) && (substr($title, -3) == "...")) {
 469                         $pos = strrpos($title, "...");
 470                         if ($pos > 0) {
 471                                 $title = substr($title, 0, $pos);
 472                                 $body = substr($body, 0, $pos);
 473                         }
 474                 }
 475                 return ($title == $body);
 476         }
 477 }