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 Friendica\Content\Text\HTML;
  14
  15 use dba;
  16 use DOMDocument;
  17 use DOMXPath;
  18
  19 require_once 'include/dba.php';
  20 require_once 'include/items.php';
  21
  22 /**
  23  * @brief This class contain functions to import feeds
  24  *
  25  */
  26 class Feed {
  27         /**
  28          * @brief Read a RSS/RDF/Atom feed and create an item entry for it
  29          *
  30          * @param string $xml The feed data
  31          * @param array $importer The user record of the importer
  32          * @param array $contact The contact record of the feed
  33          * @param string $hub Unused dummy value for compatibility reasons
  34          * @param bool $simulate If enabled, no data is imported
  35          *
  36          * @return array In simulation mode it returns the header and the first item
  37          */
  38         public static function import($xml, $importer, &$contact, &$hub, $simulate = false) {
  39
  40                 $a = get_app();
  41
  42                 if (!$simulate) {
  43                         logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
  44                 } else {
  45                         logger("Test Atom/RSS feed", LOGGER_DEBUG);
  46                 }
  47                 if ($xml == "") {
  48                         logger('XML is empty.', LOGGER_DEBUG);
  49                         return;
  50                 }
  51
  52                 if (!empty($contact['poll'])) {
  53                         $basepath = $contact['poll'];
  54                 } elseif (!empty($contact['url'])) {
  55                         $basepath = $contact['url'];
  56                 } else {
  57                         $basepath = '';
  58                 }
  59
  60                 $doc = new DOMDocument();
  61                 @$doc->loadXML(trim($xml));
  62                 $xpath = new DOMXPath($doc);
  63                 $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  64                 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  65                 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  66                 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  67                 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  68                 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  69                 $xpath->registerNamespace('poco', NAMESPACE_POCO);
  70
  71                 $author = [];
  72                 $entries = null;
  73
  74                 // Is it RDF?
  75                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  76                         $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  77                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  78
  79                         if ($author["author-name"] == "") {
  80                                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  81                         }
  82                         $entries = $xpath->query('/rdf:RDF/rss:item');
  83                 }
  84
  85                 // Is it Atom?
  86                 if ($xpath->query('/atom:feed')->length > 0) {
  87                         $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  88                         if (is_object($alternate)) {
  89                                 foreach ($alternate AS $attribute) {
  90                                         if ($attribute->name == "href") {
  91                                                 $author["author-link"] = $attribute->textContent;
  92                                         }
  93                                 }
  94                         }
  95
  96                         if ($author["author-link"] == "") {
  97                                 $author["author-link"] = $author["author-id"];
  98                         }
  99                         if ($author["author-link"] == "") {
 100                                 $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
 101                                 if (is_object($self)) {
 102                                         foreach ($self AS $attribute) {
 103                                                 if ($attribute->name == "href") {
 104                                                         $author["author-link"] = $attribute->textContent;
 105                                                 }
 106                                         }
 107                                 }
 108                         }
 109
 110                         if ($author["author-link"] == "") {
 111                                 $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
 112                         }
 113                         $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
 114
 115                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
 116
 117                         if ($author["author-name"] == "") {
 118                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
 119                         }
 120                         if ($author["author-name"] == "") {
 121                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
 122                         }
 123                         $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
 124                         if ($value != "") {
 125                                 $author["author-name"] = $value;
 126                         }
 127                         if ($simulate) {
 128                                 $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
 129
 130                                 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 131                                 if ($value != "") {
 132                                         $author["author-nick"] = $value;
 133                                 }
 134                                 $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()')->item(0)->nodeValue;
 135                                 if ($value != "") {
 136                                         $author["author-location"] = $value;
 137                                 }
 138                                 $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 139                                 if ($value != "") {
 140                                         $author["author-about"] = $value;
 141                                 }
 142                                 $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
 143                                 if (is_object($avatar)) {
 144                                         foreach ($avatar AS $attribute) {
 145                                                 if ($attribute->name == "href") {
 146                                                         $author["author-avatar"] = $attribute->textContent;
 147                                                 }
 148                                         }
 149                                 }
 150                         }
 151
 152                         $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 153
 154                         $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 155
 156                         $entries = $xpath->query('/atom:feed/atom:entry');
 157                 }
 158
 159                 // Is it RSS?
 160                 if ($xpath->query('/rss/channel')->length > 0) {
 161                         $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 162
 163                         $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 164                         $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 165
 166                         if ($author["author-name"] == "") {
 167                                 $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 168                         }
 169                         if ($author["author-name"] == "") {
 170                                 $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 171                         }
 172                         $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 173
 174                         $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 175
 176                         $entries = $xpath->query('/rss/channel/item');
 177                 }
 178
 179                 if (!$simulate) {
 180                         $author["author-link"] = $contact["url"];
 181
 182                         if ($author["author-name"] == "") {
 183                                 $author["author-name"] = $contact["name"];
 184                         }
 185                         $author["author-avatar"] = $contact["thumb"];
 186
 187                         $author["owner-link"] = $contact["url"];
 188                         $author["owner-name"] = $contact["name"];
 189                         $author["owner-avatar"] = $contact["thumb"];
 190                 }
 191
 192                 $header = [];
 193                 $header["uid"] = $importer["uid"];
 194                 $header["network"] = NETWORK_FEED;
 195                 $header["type"] = "remote";
 196                 $header["wall"] = 0;
 197                 $header["origin"] = 0;
 198                 $header["gravity"] = GRAVITY_PARENT;
 199                 $header["private"] = 2;
 200                 $header["verb"] = ACTIVITY_POST;
 201                 $header["object-type"] = ACTIVITY_OBJ_NOTE;
 202
 203                 $header["contact-id"] = $contact["id"];
 204
 205                 if (!is_object($entries)) {
 206                         logger("There are no entries in this feed.", LOGGER_DEBUG);
 207                         return;
 208                 }
 209
 210                 $items = [];
 211                 // Importing older entries first
 212                 for($i = $entries->length - 1; $i >= 0;--$i) {
 213                         $entry = $entries->item($i);
 214
 215                         $item = array_merge($header, $author);
 216
 217                         $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 218                         if (!is_object($alternate)) {
 219                                 $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 220                         }
 221                         if (is_object($alternate)) {
 222                                 foreach ($alternate AS $attribute) {
 223                                         if ($attribute->name == "href") {
 224                                                 $item["plink"] = $attribute->textContent;
 225                                         }
 226                                 }
 227                         }
 228                         if ($item["plink"] == "") {
 229                                 $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 230                         }
 231                         if ($item["plink"] == "") {
 232                                 $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 233                         }
 234
 235                         $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 236
 237                         if ($item["uri"] == "") {
 238                                 $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 239                         }
 240                         if ($item["uri"] == "") {
 241                                 $item["uri"] = $item["plink"];
 242                         }
 243
 244                         $orig_plink = $item["plink"];
 245
 246                         $item["plink"] = Network::finalUrl($item["plink"]);
 247
 248                         $item["parent-uri"] = $item["uri"];
 249
 250                         if (!$simulate) {
 251                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 252                                         $importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
 253                                 $previous = Item::selectFirst(['id'], $condition);
 254                                 if (DBM::is_result($previous)) {
 255                                         logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG);
 256                                         continue;
 257                                 }
 258                         }
 259
 260                         $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 261
 262                         if ($item["title"] == "") {
 263                                 $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 264                         }
 265                         if ($item["title"] == "") {
 266                                 $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 267                         }
 268                         $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 269
 270                         if ($published == "") {
 271                                 $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 272                         }
 273                         if ($published == "") {
 274                                 $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 275                         }
 276                         $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 277
 278                         if ($updated == "") {
 279                                 $updated = $published;
 280                         }
 281                         if ($published != "") {
 282                                 $item["created"] = $published;
 283                         }
 284                         if ($updated != "") {
 285                                 $item["edited"] = $updated;
 286                         }
 287                         $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 288
 289                         if ($creator == "") {
 290                                 $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 291                         }
 292                         if ($creator == "") {
 293                                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 294                         }
 295                         if ($creator != "") {
 296                                 $item["author-name"] = $creator;
 297                         }
 298                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 299
 300                         if ($creator != "") {
 301                                 $item["author-name"] = $creator;
 302                         }
 303
 304                         /// @TODO ?
 305                         // <category>Ausland</category>
 306                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 307
 308                         $attachments = [];
 309
 310                         $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
 311                         foreach ($enclosures AS $enclosure) {
 312                                 $href = "";
 313                                 $length = "";
 314                                 $type = "";
 315                                 $title = "";
 316
 317                                 foreach ($enclosure->attributes AS $attribute) {
 318                                         if (in_array($attribute->name, ["url", "href"])) {
 319                                                 $href = $attribute->textContent;
 320                                         } elseif ($attribute->name == "length") {
 321                                                 $length = $attribute->textContent;
 322                                         } elseif ($attribute->name == "type") {
 323                                                 $type = $attribute->textContent;
 324                                         }
 325                                 }
 326                                 if (strlen($item["attach"])) {
 327                                         $item["attach"] .= ',';
 328                                 }
 329
 330                                 $attachments[] = ["link" => $href, "type" => $type, "length" => $length];
 331
 332                                 $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 333                         }
 334
 335                         $tags = '';
 336                         $categories = $xpath->query("category", $entry);
 337                         foreach ($categories AS $category) {
 338                                 $hashtag = $category->nodeValue;
 339                                 if ($tags != '') {
 340                                         $tags .= ', ';
 341                                 }
 342
 343                                 $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
 344                                 $tags .= $taglink;
 345                         }
 346
 347                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 348
 349                         if ($body == "") {
 350                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 351                         }
 352                         if ($body == "") {
 353                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 354                         }
 355                         if ($body == "") {
 356                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 357                         }
 358
 359                         // remove the content of the title if it is identically to the body
 360                         // This helps with auto generated titles e.g. from tumblr
 361                         if (self::titleIsBody($item["title"], $body)) {
 362                                 $item["title"] = "";
 363                         }
 364                         $item["body"] = HTML::toBBCode($body, $basepath);
 365
 366                         if (($item["body"] == '') && ($item["title"] != '')) {
 367                                 $item["body"] = $item["title"];
 368                                 $item["title"] = '';
 369                         }
 370
 371                         $preview = '';
 372                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 373                                 // Handle enclosures and treat them as preview picture
 374                                 foreach ($attachments AS $attachment) {
 375                                         if ($attachment["type"] == "image/jpeg") {
 376                                                 $preview = $attachment["link"];
 377                                         }
 378                                 }
 379
 380                                 // Remove a possible link to the item itself
 381                                 $item["body"] = str_replace($item["plink"], '', $item["body"]);
 382                                 $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
 383
 384                                 // Replace the content when the title is longer than the body
 385                                 $replace = (strlen($item["title"]) > strlen($item["body"]));
 386
 387                                 // Replace it, when there is an image in the body
 388                                 if (strstr($item["body"], '[/img]')) {
 389                                         $replace = true;
 390                                 }
 391
 392                                 // Replace it, when there is a link in the body
 393                                 if (strstr($item["body"], '[/url]')) {
 394                                         $replace = true;
 395                                 }
 396
 397                                 if ($replace) {
 398                                         $item["body"] = $item["title"];
 399                                 }
 400                                 // We always strip the title since it will be added in the page information
 401                                 $item["title"] = "";
 402                                 $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 403                                 $item["tag"] = add_page_keywords($item["plink"], $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 404                                 $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 405                                 unset($item["attach"]);
 406                         } else {
 407                                 if ($contact["fetch_further_information"] == 3) {
 408                                         if (!empty($tags)) {
 409                                                 $item["tag"] = $tags;
 410                                         } else {
 411                                                 // @todo $preview is never set in this case, is it intended? - @MrPetovan 2018-02-13
 412                                                 $item["tag"] = add_page_keywords($item["plink"], $preview, true, $contact["ffi_keyword_blacklist"]);
 413                                         }
 414                                         $item["body"] .= "\n".$item['tag'];
 415                                 }
 416                                 // Add the link to the original feed entry if not present in feed
 417                                 if (($item['plink'] != '') && !strstr($item["body"], $item['plink'])) {
 418                                         $item["body"] .= "[hr][url]".$item['plink']."[/url]";
 419                                 }
 420                         }
 421
 422                         if (!$simulate) {
 423                                 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 424
 425                                 $notify = Item::isRemoteSelf($contact, $item);
 426
 427                                 // Distributed items should have a well formatted URI.
 428                                 // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 429                                 if ($notify) {
 430                                         $item['guid'] = Item::guidFromUri($orig_plink, $a->get_hostname());
 431                                         unset($item['uri']);
 432                                         unset($item['parent-uri']);
 433
 434                                         // Set the delivery priority for "remote self" to "medium"
 435                                         $notify = PRIORITY_MEDIUM;
 436                                 }
 437
 438                                 $id = Item::insert($item, false, $notify);
 439
 440                                 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 441                         } else {
 442                                 $items[] = $item;
 443                         }
 444                         if ($simulate) {
 445                                 break;
 446                         }
 447                 }
 448
 449                 if ($simulate) {
 450                         return ["header" => $author, "items" => $items];
 451                 }
 452         }
 453
 454         private static function titleIsBody($title, $body)
 455         {
 456                 $title = strip_tags($title);
 457                 $title = trim($title);
 458                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 459                 $title = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $title);
 460
 461                 $body = strip_tags($body);
 462                 $body = trim($body);
 463                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 464                 $body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
 465
 466                 if (strlen($title) < strlen($body)) {
 467                         $body = substr($body, 0, strlen($title));
 468                 }
 469
 470                 if (($title != $body) && (substr($title, -3) == "...")) {
 471                         $pos = strrpos($title, "...");
 472                         if ($pos > 0) {
 473                                 $title = substr($title, 0, $pos);
 474                                 $body = substr($body, 0, $pos);
 475                         }
 476                 }
 477                 return ($title == $body);
 478         }
 479 }