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/html2bbcode.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 ".$importer["uid"], LOGGER_DEBUG);
  43                 } else {
  44                         logger("Test Atom/RSS feed", LOGGER_DEBUG);
  45                 }
  46                 if ($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"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  76                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  77
  78                         if ($author["author-name"] == "") {
  79                                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  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 = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  87                         if (is_object($alternate)) {
  88                                 foreach ($alternate AS $attributes) {
  89                                         if ($attributes->name == "href") {
  90                                                 $author["author-link"] = $attributes->textContent;
  91                                         }
  92                                 }
  93                         }
  94
  95                         if ($author["author-link"] == "") {
  96                                 $author["author-link"] = $author["author-id"];
  97                         }
  98                         if ($author["author-link"] == "") {
  99                                 $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
 100                                 if (is_object($self)) {
 101                                         foreach ($self AS $attributes) {
 102                                                 if ($attributes->name == "href") {
 103                                                         $author["author-link"] = $attributes->textContent;
 104                                                 }
 105                                         }
 106                                 }
 107                         }
 108
 109                         if ($author["author-link"] == "") {
 110                                 $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
 111                         }
 112                         $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
 113
 114                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
 115
 116                         if ($author["author-name"] == "") {
 117                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
 118                         }
 119                         if ($author["author-name"] == "") {
 120                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
 121                         }
 122                         $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
 123                         if ($value != "") {
 124                                 $author["author-name"] = $value;
 125                         }
 126                         if ($simulate) {
 127                                 $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
 128
 129                                 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 130                                 if ($value != "") {
 131                                         $author["author-nick"] = $value;
 132                                 }
 133                                 $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()')->item(0)->nodeValue;
 134                                 if ($value != "") {
 135                                         $author["author-location"] = $value;
 136                                 }
 137                                 $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 138                                 if ($value != "") {
 139                                         $author["author-about"] = $value;
 140                                 }
 141                                 $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
 142                                 if (is_object($avatar)) {
 143                                         foreach ($avatar AS $attributes) {
 144                                                 if ($attributes->name == "href") {
 145                                                         $author["author-avatar"] = $attributes->textContent;
 146                                                 }
 147                                         }
 148                                 }
 149                         }
 150
 151                         $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 152
 153                         $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 154
 155                         $entries = $xpath->query('/atom:feed/atom:entry');
 156                 }
 157
 158                 // Is it RSS?
 159                 if ($xpath->query('/rss/channel')->length > 0) {
 160                         $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 161
 162                         $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 163                         $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 164
 165                         if ($author["author-name"] == "") {
 166                                 $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 167                         }
 168                         if ($author["author-name"] == "") {
 169                                 $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 170                         }
 171                         $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 172
 173                         $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 174
 175                         $entries = $xpath->query('/rss/channel/item');
 176                 }
 177
 178                 if (!$simulate) {
 179                         $author["author-link"] = $contact["url"];
 180
 181                         if ($author["author-name"] == "") {
 182                                 $author["author-name"] = $contact["name"];
 183                         }
 184                         $author["author-avatar"] = $contact["thumb"];
 185
 186                         $author["owner-link"] = $contact["url"];
 187                         $author["owner-name"] = $contact["name"];
 188                         $author["owner-avatar"] = $contact["thumb"];
 189                 }
 190
 191                 $header = [];
 192                 $header["uid"] = $importer["uid"];
 193                 $header["network"] = NETWORK_FEED;
 194                 $header["type"] = "remote";
 195                 $header["wall"] = 0;
 196                 $header["origin"] = 0;
 197                 $header["gravity"] = GRAVITY_PARENT;
 198                 $header["private"] = 2;
 199                 $header["verb"] = ACTIVITY_POST;
 200                 $header["object-type"] = ACTIVITY_OBJ_NOTE;
 201
 202                 $header["contact-id"] = $contact["id"];
 203
 204                 if (!is_object($entries)) {
 205                         logger("There are no entries in this feed.", LOGGER_DEBUG);
 206                         return;
 207                 }
 208
 209                 $items = [];
 210
 211                 $entrylist = [];
 212
 213                 foreach ($entries AS $entry) {
 214                         $entrylist[] = $entry;
 215                 }
 216                 foreach (array_reverse($entrylist) AS $entry) {
 217                         $item = array_merge($header, $author);
 218
 219                         $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 220                         if (!is_object($alternate)) {
 221                                 $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 222                         }
 223                         if (is_object($alternate)) {
 224                                 foreach ($alternate AS $attributes) {
 225                                         if ($attributes->name == "href") {
 226                                                 $item["plink"] = $attributes->textContent;
 227                                         }
 228                                 }
 229                         }
 230                         if ($item["plink"] == "") {
 231                                 $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 232                         }
 233                         if ($item["plink"] == "") {
 234                                 $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 235                         }
 236
 237                         $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 238
 239                         if ($item["uri"] == "") {
 240                                 $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 241                         }
 242                         if ($item["uri"] == "") {
 243                                 $item["uri"] = $item["plink"];
 244                         }
 245
 246                         $orig_plink = $item["plink"];
 247
 248                         $item["plink"] = Network::finalUrl($item["plink"]);
 249
 250                         $item["parent-uri"] = $item["uri"];
 251
 252                         if (!$simulate) {
 253                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 254                                         $importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
 255                                 $previous = dba::selectFirst('item', ['id'], $condition);
 256                                 if (DBM::is_result($previous)) {
 257                                         logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG);
 258                                         continue;
 259                                 }
 260                         }
 261
 262                         $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 263
 264                         if ($item["title"] == "") {
 265                                 $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 266                         }
 267                         if ($item["title"] == "") {
 268                                 $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 269                         }
 270                         $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 271
 272                         if ($published == "") {
 273                                 $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 274                         }
 275                         if ($published == "") {
 276                                 $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 277                         }
 278                         $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 279
 280                         if ($updated == "") {
 281                                 $updated = $published;
 282                         }
 283                         if ($published != "") {
 284                                 $item["created"] = $published;
 285                         }
 286                         if ($updated != "") {
 287                                 $item["edited"] = $updated;
 288                         }
 289                         $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 290
 291                         if ($creator == "") {
 292                                 $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 293                         }
 294                         if ($creator == "") {
 295                                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 296                         }
 297                         if ($creator != "") {
 298                                 $item["author-name"] = $creator;
 299                         }
 300                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 301
 302                         if ($creator != "") {
 303                                 $item["author-name"] = $creator;
 304                         }
 305
 306                         /// @TODO ?
 307                         // <category>Ausland</category>
 308                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 309
 310                         $attachments = [];
 311
 312                         $enclosures = $xpath->query("enclosure", $entry);
 313                         foreach ($enclosures AS $enclosure) {
 314                                 $href = "";
 315                                 $length = "";
 316                                 $type = "";
 317                                 $title = "";
 318
 319                                 foreach ($enclosure->attributes AS $attributes) {
 320                                         if ($attributes->name == "url") {
 321                                                 $href = $attributes->textContent;
 322                                         } elseif ($attributes->name == "length") {
 323                                                 $length = $attributes->textContent;
 324                                         } elseif ($attributes->name == "type") {
 325                                                 $type = $attributes->textContent;
 326                                         }
 327                                 }
 328                                 if (strlen($item["attach"])) {
 329                                         $item["attach"] .= ',';
 330                                 }
 331
 332                                 $attachments[] = ["link" => $href, "type" => $type, "length" => $length];
 333
 334                                 $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 335                         }
 336
 337                         $tags = '';
 338                         $categories = $xpath->query("category", $entry);
 339                         foreach ($categories AS $category) {
 340                                 $hashtag = $category->nodeValue;
 341                                 if ($tags != '') {
 342                                         $tags .= ', ';
 343                                 }
 344
 345                                 $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
 346                                 $tags .= $taglink;
 347                         }
 348
 349                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 350
 351                         if ($body == "") {
 352                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 353                         }
 354                         if ($body == "") {
 355                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 356                         }
 357                         if ($body == "") {
 358                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 359                         }
 360
 361                         // remove the content of the title if it is identically to the body
 362                         // This helps with auto generated titles e.g. from tumblr
 363                         if (self::titleIsBody($item["title"], $body)) {
 364                                 $item["title"] = "";
 365                         }
 366                         $item["body"] = html2bbcode($body, $basepath);
 367
 368                         if (($item["body"] == '') && ($item["title"] != '')) {
 369                                 $item["body"] = $item["title"];
 370                                 $item["title"] = '';
 371                         }
 372
 373                         $preview = '';
 374                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 375                                 // Handle enclosures and treat them as preview picture
 376                                 foreach ($attachments AS $attachment) {
 377                                         if ($attachment["type"] == "image/jpeg") {
 378                                                 $preview = $attachment["link"];
 379                                         }
 380                                 }
 381
 382                                 // Remove a possible link to the item itself
 383                                 $item["body"] = str_replace($item["plink"], '', $item["body"]);
 384                                 $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
 385
 386                                 // Replace the content when the title is longer than the body
 387                                 $replace = (strlen($item["title"]) > strlen($item["body"]));
 388
 389                                 // Replace it, when there is an image in the body
 390                                 if (strstr($item["body"], '[/img]')) {
 391                                         $replace = true;
 392                                 }
 393
 394                                 // Replace it, when there is a link in the body
 395                                 if (strstr($item["body"], '[/url]')) {
 396                                         $replace = true;
 397                                 }
 398
 399                                 if ($replace) {
 400                                         $item["body"] = $item["title"];
 401                                 }
 402                                 // We always strip the title since it will be added in the page information
 403                                 $item["title"] = "";
 404                                 $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 405                                 $item["tag"] = add_page_keywords($item["plink"], $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 406                                 $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 407                                 unset($item["attach"]);
 408                         } else {
 409                                 if ($contact["fetch_further_information"] == 3) {
 410                                         if (!empty($tags)) {
 411                                                 $item["tag"] = $tags;
 412                                         } else {
 413                                                 // @todo $preview is never set in this case, is it intended? - @MrPetovan 2018-02-13
 414                                                 $item["tag"] = add_page_keywords($item["plink"], $preview, true, $contact["ffi_keyword_blacklist"]);
 415                                         }
 416                                         $item["body"] .= "\n".$item['tag'];
 417                                 }
 418                                 // Add the link to the original feed entry if not present in feed
 419                                 if (($item['plink'] != '') && !strstr($item["body"], $item['plink'])) {
 420                                         $item["body"] .= "[hr][url]".$item['plink']."[/url]";
 421                                 }
 422                         }
 423
 424                         if (!$simulate) {
 425                                 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 426
 427                                 $notify = Item::isRemoteSelf($contact, $item);
 428
 429                                 // Distributed items should have a well formatted URI.
 430                                 // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 431                                 if ($notify) {
 432                                         $item['guid'] = Item::guidFromUri($orig_plink, $a->get_hostname());
 433                                         unset($item['uri']);
 434                                         unset($item['parent-uri']);
 435                                 }
 436
 437                                 $id = Item::insert($item, false, $notify);
 438
 439                                 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 440                         } else {
 441                                 $items[] = $item;
 442                         }
 443                         if ($simulate) {
 444                                 break;
 445                         }
 446                 }
 447
 448                 if ($simulate) {
 449                         return ["header" => $author, "items" => $items];
 450                 }
 451         }
 452
 453         private static function titleIsBody($title, $body)
 454         {
 455                 $title = strip_tags($title);
 456                 $title = trim($title);
 457                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 458                 $title = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $title);
 459
 460                 $body = strip_tags($body);
 461                 $body = trim($body);
 462                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 463                 $body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
 464
 465                 if (strlen($title) < strlen($body)) {
 466                         $body = substr($body, 0, strlen($title));
 467                 }
 468
 469                 if (($title != $body) && (substr($title, -3) == "...")) {
 470                         $pos = strrpos($title, "...");
 471                         if ($pos > 0) {
 472                                 $title = substr($title, 0, $pos);
 473                                 $body = substr($body, 0, $pos);
 474                         }
 475                 }
 476                 return ($title == $body);
 477         }
 478 }