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 dba;
  12 use DOMDocument;
  13 use DomXPath;
  14
  15 require_once("include/html2bbcode.php");
  16 require_once("include/items.php");
  17
  18 /**
  19  * @brief This class contain functions to import feeds
  20  *
  21  */
  22 class Feed {
  23         /**
  24          * @brief Read a RSS/RDF/Atom feed and create an item entry for it
  25          *
  26          * @param string $xml The feed data
  27          * @param array $importer The user record of the importer
  28          * @param array $contact The contact record of the feed
  29          * @param string $hub Unused dummy value for compatibility reasons
  30          * @param bool $simulate If enabled, no data is imported
  31          *
  32          * @return array In simulation mode it returns the header and the first item
  33          */
  34         function import($xml, $importer, &$contact, &$hub, $simulate = false) {
  35
  36                 $a = get_app();
  37
  38                 if (!$simulate) {
  39                         logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
  40                 } else {
  41                         logger("Test Atom/RSS feed", LOGGER_DEBUG);
  42                 }
  43                 if ($xml == "") {
  44                         logger('XML is empty.', LOGGER_DEBUG);
  45                         return;
  46                 }
  47
  48                 if (!empty($contact['poll'])) {
  49                         $basepath = $contact['poll'];
  50                 } elseif (!empty($contact['url'])) {
  51                         $basepath = $contact['url'];
  52                 } else {
  53                         $basepath = '';
  54                 }
  55
  56                 $doc = new DOMDocument();
  57                 @$doc->loadXML(trim($xml));
  58                 $xpath = new DomXPath($doc);
  59                 $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  60                 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  61                 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  62                 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  63                 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  64                 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  65                 $xpath->registerNamespace('poco', NAMESPACE_POCO);
  66
  67                 $author = array();
  68
  69                 // Is it RDF?
  70                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  71                         $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  72                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  73
  74                         if ($author["author-name"] == "") {
  75                                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  76                         }
  77                         $entries = $xpath->query('/rdf:RDF/rss:item');
  78                 }
  79
  80                 // Is it Atom?
  81                 if ($xpath->query('/atom:feed')->length > 0) {
  82                         $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  83                         if (is_object($alternate)) {
  84                                 foreach ($alternate AS $attributes) {
  85                                         if ($attributes->name == "href") {
  86                                                 $author["author-link"] = $attributes->textContent;
  87                                         }
  88                                 }
  89                         }
  90
  91                         if ($author["author-link"] == "") {
  92                                 $author["author-link"] = $author["author-id"];
  93                         }
  94                         if ($author["author-link"] == "") {
  95                                 $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  96                                 if (is_object($self)) {
  97                                         foreach ($self AS $attributes) {
  98                                                 if ($attributes->name == "href") {
  99                                                         $author["author-link"] = $attributes->textContent;
 100                                                 }
 101                                         }
 102                                 }
 103                         }
 104
 105                         if ($author["author-link"] == "") {
 106                                 $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
 107                         }
 108                         $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
 109
 110                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
 111
 112                         if ($author["author-name"] == "") {
 113                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
 114                         }
 115                         if ($author["author-name"] == "") {
 116                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
 117                         }
 118                         $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
 119                         if ($value != "") {
 120                                 $author["author-name"] = $value;
 121                         }
 122                         if ($simulate) {
 123                                 $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
 124
 125                                 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 126                                 if ($value != "") {
 127                                         $author["author-nick"] = $value;
 128                                 }
 129                                 $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
 130                                 if ($value != "") {
 131                                         $author["author-location"] = $value;
 132                                 }
 133                                 $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 134                                 if ($value != "") {
 135                                         $author["author-about"] = $value;
 136                                 }
 137                                 $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
 138                                 if (is_object($avatar)) {
 139                                         foreach ($avatar AS $attributes) {
 140                                                 if ($attributes->name == "href") {
 141                                                         $author["author-avatar"] = $attributes->textContent;
 142                                                 }
 143                                         }
 144                                 }
 145                         }
 146
 147                         $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 148
 149                         $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 150
 151                         $entries = $xpath->query('/atom:feed/atom:entry');
 152                 }
 153
 154                 // Is it RSS?
 155                 if ($xpath->query('/rss/channel')->length > 0) {
 156                         $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 157
 158                         $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 159                         $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 160
 161                         if ($author["author-name"] == "") {
 162                                 $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 163                         }
 164                         if ($author["author-name"] == "") {
 165                                 $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 166                         }
 167                         $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 168
 169                         $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 170
 171                         $entries = $xpath->query('/rss/channel/item');
 172                 }
 173
 174                 if (!$simulate) {
 175                         $author["author-link"] = $contact["url"];
 176
 177                         if ($author["author-name"] == "") {
 178                                 $author["author-name"] = $contact["name"];
 179                         }
 180                         $author["author-avatar"] = $contact["thumb"];
 181
 182                         $author["owner-link"] = $contact["url"];
 183                         $author["owner-name"] = $contact["name"];
 184                         $author["owner-avatar"] = $contact["thumb"];
 185                 }
 186
 187                 $header = array();
 188                 $header["uid"] = $importer["uid"];
 189                 $header["network"] = NETWORK_FEED;
 190                 $header["type"] = "remote";
 191                 $header["wall"] = 0;
 192                 $header["origin"] = 0;
 193                 $header["gravity"] = GRAVITY_PARENT;
 194                 $header["private"] = 2;
 195                 $header["verb"] = ACTIVITY_POST;
 196                 $header["object-type"] = ACTIVITY_OBJ_NOTE;
 197
 198                 $header["contact-id"] = $contact["id"];
 199
 200                 if (!strlen($contact["notify"])) {
 201                         // one way feed - no remote comment ability
 202                         $header["last-child"] = 0;
 203                 }
 204
 205                 if (!is_object($entries)) {
 206                         logger("There are no entries in this feed.", LOGGER_DEBUG);
 207                         return;
 208                 }
 209
 210                 $items = array();
 211
 212                 $entrylist = array();
 213
 214                 foreach ($entries AS $entry) {
 215                         $entrylist[] = $entry;
 216                 }
 217                 foreach (array_reverse($entrylist) AS $entry) {
 218                         $item = array_merge($header, $author);
 219
 220                         $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 221                         if (!is_object($alternate)) {
 222                                 $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 223                         }
 224                         if (is_object($alternate)) {
 225                                 foreach ($alternate AS $attributes) {
 226                                         if ($attributes->name == "href") {
 227                                                 $item["plink"] = $attributes->textContent;
 228                                         }
 229                                 }
 230                         }
 231                         if ($item["plink"] == "") {
 232                                 $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 233                         }
 234                         if ($item["plink"] == "") {
 235                                 $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 236                         }
 237
 238                         $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 239
 240                         if ($item["uri"] == "") {
 241                                 $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 242                         }
 243                         if ($item["uri"] == "") {
 244                                 $item["uri"] = $item["plink"];
 245                         }
 246
 247                         $orig_plink = $item["plink"];
 248
 249                         $item["plink"] = original_url($item["plink"]);
 250
 251                         $item["parent-uri"] = $item["uri"];
 252
 253                         if (!$simulate) {
 254                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 255                                         $importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
 256                                 $previous = dba::select('item', ['id'], $condition, ['limit' => 1]);
 257                                 if (DBM::is_result($previous)) {
 258                                         logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG);
 259                                         continue;
 260                                 }
 261                         }
 262
 263                         $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 264
 265                         if ($item["title"] == "") {
 266                                 $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 267                         }
 268                         if ($item["title"] == "") {
 269                                 $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 270                         }
 271                         $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 272
 273                         if ($published == "") {
 274                                 $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 275                         }
 276                         if ($published == "") {
 277                                 $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 278                         }
 279                         $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 280
 281                         if ($updated == "") {
 282                                 $updated = $published;
 283                         }
 284                         if ($published != "") {
 285                                 $item["created"] = $published;
 286                         }
 287                         if ($updated != "") {
 288                                 $item["edited"] = $updated;
 289                         }
 290                         $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 291
 292                         if ($creator == "") {
 293                                 $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 294                         }
 295                         if ($creator == "") {
 296                                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 297                         }
 298                         if ($creator != "") {
 299                                 $item["author-name"] = $creator;
 300                         }
 301                         if ($pubDate != "") {
 302                                 $item["edited"] = $item["created"] = $pubDate;
 303                         }
 304                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 305
 306                         if ($creator != "") {
 307                                 $item["author-name"] = $creator;
 308                         }
 309
 310                         /// @TODO ?
 311                         // <category>Ausland</category>
 312                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 313
 314                         $attachments = array();
 315
 316                         $enclosures = $xpath->query("enclosure", $entry);
 317                         foreach ($enclosures AS $enclosure) {
 318                                 $href = "";
 319                                 $length = "";
 320                                 $type = "";
 321                                 $title = "";
 322
 323                                 foreach ($enclosure->attributes AS $attributes) {
 324                                         if ($attributes->name == "url") {
 325                                                 $href = $attributes->textContent;
 326                                         } elseif ($attributes->name == "length") {
 327                                                 $length = $attributes->textContent;
 328                                         } elseif ($attributes->name == "type") {
 329                                                 $type = $attributes->textContent;
 330                                         }
 331                                 }
 332                                 if (strlen($item["attach"])) {
 333                                         $item["attach"] .= ',';
 334                                 }
 335
 336                                 $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 337
 338                                 $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 339                         }
 340
 341                         $tags = '';
 342                         $categories = $xpath->query("category", $entry);
 343                         foreach ($categories AS $category) {
 344                                 $hashtag = $category->nodeValue;
 345                                 if ($tags != '') {
 346                                         $tags .= ', ';
 347                                 }
 348
 349                                 $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
 350                                 $tags .= $taglink;
 351                         }
 352
 353                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 354
 355                         if ($body == "") {
 356                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 357                         }
 358                         if ($body == "") {
 359                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 360                         }
 361                         if ($body == "") {
 362                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 363                         }
 364
 365                         // remove the content of the title if it is identically to the body
 366                         // This helps with auto generated titles e.g. from tumblr
 367                         if (title_is_body($item["title"], $body)) {
 368                                 $item["title"] = "";
 369                         }
 370                         $item["body"] = html2bbcode($body, $basepath);
 371
 372                         if (($item["body"] == '') && ($item["title"] != '')) {
 373                                 $item["body"] = $item["title"];
 374                                 $item["title"] = '';
 375                         }
 376
 377                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 378                                 $preview = "";
 379
 380                                 // Handle enclosures and treat them as preview picture
 381                                 foreach ($attachments AS $attachment) {
 382                                         if ($attachment["type"] == "image/jpeg") {
 383                                                 $preview = $attachment["link"];
 384                                         }
 385                                 }
 386
 387                                 // Remove a possible link to the item itself
 388                                 $item["body"] = str_replace($item["plink"], '', $item["body"]);
 389                                 $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
 390
 391                                 // Replace the content when the title is longer than the body
 392                                 $replace = (strlen($item["title"]) > strlen($item["body"]));
 393
 394                                 // Replace it, when there is an image in the body
 395                                 if (strstr($item["body"], '[/img]')) {
 396                                         $replace = true;
 397                                 }
 398
 399                                 // Replace it, when there is a link in the body
 400                                 if (strstr($item["body"], '[/url]')) {
 401                                         $replace = true;
 402                                 }
 403
 404                                 if ($replace) {
 405                                         $item["body"] = $item["title"];
 406                                 }
 407                                 // We always strip the title since it will be added in the page information
 408                                 $item["title"] = "";
 409                                 $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 410                                 $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 411                                 $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 412                                 unset($item["attach"]);
 413                         } else {
 414                                 if ($contact["fetch_further_information"] == 3) {
 415                                         if (!empty($tags)) {
 416                                                 $item["tag"] = $tags;
 417                                         } else {
 418                                                 $item["tag"] = add_page_keywords($item["plink"], false, $preview, true, $contact["ffi_keyword_blacklist"]);
 419                                         }
 420                                         $item["body"] .= "\n".$item['tag'];
 421                                 }
 422                                 if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
 423                                         $item["body"] .= "[hr][url]".$item['plink']."[/url]";
 424                                 }
 425                         }
 426
 427                         if (!$simulate) {
 428                                 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 429
 430                                 $notify = item_is_remote_self($contact, $item);
 431
 432                                 // Distributed items should have a well formatted URI.
 433                                 // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 434                                 if ($notify) {
 435                                         $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
 436                                         unset($item['uri']);
 437                                         unset($item['parent-uri']);
 438                                 }
 439
 440                                 $id = item_store($item, false, $notify);
 441
 442                                 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 443                         } else {
 444                                 $items[] = $item;
 445                         }
 446                         if ($simulate) {
 447                                 break;
 448                         }
 449                 }
 450
 451                 if ($simulate) {
 452                         return array("header" => $author, "items" => $items);
 453                 }
 454         }
 455 }