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 dba;
  13 use DOMDocument;
  14 use DOMXPath;
  15
  16 require_once 'include/dba.php';
  17 require_once 'include/html2bbcode.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                 $author = [];
  70
  71                 // Is it RDF?
  72                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  73                         $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  74                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  75
  76                         if ($author["author-name"] == "") {
  77                                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  78                         }
  79                         $entries = $xpath->query('/rdf:RDF/rss:item');
  80                 }
  81
  82                 // Is it Atom?
  83                 if ($xpath->query('/atom:feed')->length > 0) {
  84                         $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  85                         if (is_object($alternate)) {
  86                                 foreach ($alternate AS $attributes) {
  87                                         if ($attributes->name == "href") {
  88                                                 $author["author-link"] = $attributes->textContent;
  89                                         }
  90                                 }
  91                         }
  92
  93                         if ($author["author-link"] == "") {
  94                                 $author["author-link"] = $author["author-id"];
  95                         }
  96                         if ($author["author-link"] == "") {
  97                                 $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  98                                 if (is_object($self)) {
  99                                         foreach ($self AS $attributes) {
 100                                                 if ($attributes->name == "href") {
 101                                                         $author["author-link"] = $attributes->textContent;
 102                                                 }
 103                                         }
 104                                 }
 105                         }
 106
 107                         if ($author["author-link"] == "") {
 108                                 $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
 109                         }
 110                         $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
 111
 112                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
 113
 114                         if ($author["author-name"] == "") {
 115                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
 116                         }
 117                         if ($author["author-name"] == "") {
 118                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
 119                         }
 120                         $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
 121                         if ($value != "") {
 122                                 $author["author-name"] = $value;
 123                         }
 124                         if ($simulate) {
 125                                 $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
 126
 127                                 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 128                                 if ($value != "") {
 129                                         $author["author-nick"] = $value;
 130                                 }
 131                                 $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()')->item(0)->nodeValue;
 132                                 if ($value != "") {
 133                                         $author["author-location"] = $value;
 134                                 }
 135                                 $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 136                                 if ($value != "") {
 137                                         $author["author-about"] = $value;
 138                                 }
 139                                 $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
 140                                 if (is_object($avatar)) {
 141                                         foreach ($avatar AS $attributes) {
 142                                                 if ($attributes->name == "href") {
 143                                                         $author["author-avatar"] = $attributes->textContent;
 144                                                 }
 145                                         }
 146                                 }
 147                         }
 148
 149                         $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 150
 151                         $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 152
 153                         $entries = $xpath->query('/atom:feed/atom:entry');
 154                 }
 155
 156                 // Is it RSS?
 157                 if ($xpath->query('/rss/channel')->length > 0) {
 158                         $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 159
 160                         $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 161                         $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 162
 163                         if ($author["author-name"] == "") {
 164                                 $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 165                         }
 166                         if ($author["author-name"] == "") {
 167                                 $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 168                         }
 169                         $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 170
 171                         $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 172
 173                         $entries = $xpath->query('/rss/channel/item');
 174                 }
 175
 176                 if (!$simulate) {
 177                         $author["author-link"] = $contact["url"];
 178
 179                         if ($author["author-name"] == "") {
 180                                 $author["author-name"] = $contact["name"];
 181                         }
 182                         $author["author-avatar"] = $contact["thumb"];
 183
 184                         $author["owner-link"] = $contact["url"];
 185                         $author["owner-name"] = $contact["name"];
 186                         $author["owner-avatar"] = $contact["thumb"];
 187                 }
 188
 189                 $header = [];
 190                 $header["uid"] = $importer["uid"];
 191                 $header["network"] = NETWORK_FEED;
 192                 $header["type"] = "remote";
 193                 $header["wall"] = 0;
 194                 $header["origin"] = 0;
 195                 $header["gravity"] = GRAVITY_PARENT;
 196                 $header["private"] = 2;
 197                 $header["verb"] = ACTIVITY_POST;
 198                 $header["object-type"] = ACTIVITY_OBJ_NOTE;
 199
 200                 $header["contact-id"] = $contact["id"];
 201
 202                 if (!is_object($entries)) {
 203                         logger("There are no entries in this feed.", LOGGER_DEBUG);
 204                         return;
 205                 }
 206
 207                 $items = [];
 208
 209                 $entrylist = [];
 210
 211                 foreach ($entries AS $entry) {
 212                         $entrylist[] = $entry;
 213                 }
 214                 foreach (array_reverse($entrylist) AS $entry) {
 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 $attributes) {
 223                                         if ($attributes->name == "href") {
 224                                                 $item["plink"] = $attributes->textContent;
 225                                         }
 226                                 }
 227                         }
 228                         if ($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                         $item["plink"] = original_url($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 = dba::selectFirst('item', ['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", $entry);
 311                         foreach ($enclosures AS $enclosure) {
 312                                 $href = "";
 313                                 $length = "";
 314                                 $type = "";
 315                                 $title = "";
 316
 317                                 foreach ($enclosure->attributes AS $attributes) {
 318                                         if ($attributes->name == "url") {
 319                                                 $href = $attributes->textContent;
 320                                         } elseif ($attributes->name == "length") {
 321                                                 $length = $attributes->textContent;
 322                                         } elseif ($attributes->name == "type") {
 323                                                 $type = $attributes->textContent;
 324                                         }
 325                                 }
 326                                 if (strlen($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                         $item["body"] = html2bbcode($body, $basepath);
 365
 366                         if (($item["body"] == '') && ($item["title"] != '')) {
 367                                 $item["body"] = $item["title"];
 368                                 $item["title"] = '';
 369                         }
 370
 371                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 372                                 $preview = "";
 373
 374                                 // Handle enclosures and treat them as preview picture
 375                                 foreach ($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"], false, $preview, ($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                                                 $item["tag"] = add_page_keywords($item["plink"], false, $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 (!strstr($item["body"], $item['plink']) && ($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                                 $notify = item_is_remote_self($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
 435                                 $id = item_store($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 }