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/dba.php';
  16 require_once 'include/html2bbcode.php';
  17 require_once 'include/items.php';
  18
  19 /**
  20  * @brief This class contain functions to import feeds
  21  *
  22  */
  23 class Feed {
  24         /**
  25          * @brief Read a RSS/RDF/Atom feed and create an item entry for it
  26          *
  27          * @param string $xml The feed data
  28          * @param array $importer The user record of the importer
  29          * @param array $contact The contact record of the feed
  30          * @param string $hub Unused dummy value for compatibility reasons
  31          * @param bool $simulate If enabled, no data is imported
  32          *
  33          * @return array In simulation mode it returns the header and the first item
  34          */
  35         function import($xml, $importer, &$contact, &$hub, $simulate = false) {
  36
  37                 $a = get_app();
  38
  39                 if (!$simulate) {
  40                         logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
  41                 } else {
  42                         logger("Test Atom/RSS feed", LOGGER_DEBUG);
  43                 }
  44                 if ($xml == "") {
  45                         logger('XML is empty.', LOGGER_DEBUG);
  46                         return;
  47                 }
  48
  49                 if (!empty($contact['poll'])) {
  50                         $basepath = $contact['poll'];
  51                 } elseif (!empty($contact['url'])) {
  52                         $basepath = $contact['url'];
  53                 } else {
  54                         $basepath = '';
  55                 }
  56
  57                 $doc = new DOMDocument();
  58                 @$doc->loadXML(trim($xml));
  59                 $xpath = new DOMXPath($doc);
  60                 $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  61                 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  62                 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  63                 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  64                 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  65                 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  66                 $xpath->registerNamespace('poco', NAMESPACE_POCO);
  67
  68                 $author = array();
  69
  70                 // Is it RDF?
  71                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  72                         $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  73                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  74
  75                         if ($author["author-name"] == "") {
  76                                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  77                         }
  78                         $entries = $xpath->query('/rdf:RDF/rss:item');
  79                 }
  80
  81                 // Is it Atom?
  82                 if ($xpath->query('/atom:feed')->length > 0) {
  83                         $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  84                         if (is_object($alternate)) {
  85                                 foreach ($alternate AS $attributes) {
  86                                         if ($attributes->name == "href") {
  87                                                 $author["author-link"] = $attributes->textContent;
  88                                         }
  89                                 }
  90                         }
  91
  92                         if ($author["author-link"] == "") {
  93                                 $author["author-link"] = $author["author-id"];
  94                         }
  95                         if ($author["author-link"] == "") {
  96                                 $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  97                                 if (is_object($self)) {
  98                                         foreach ($self AS $attributes) {
  99                                                 if ($attributes->name == "href") {
 100                                                         $author["author-link"] = $attributes->textContent;
 101                                                 }
 102                                         }
 103                                 }
 104                         }
 105
 106                         if ($author["author-link"] == "") {
 107                                 $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
 108                         }
 109                         $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
 110
 111                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
 112
 113                         if ($author["author-name"] == "") {
 114                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
 115                         }
 116                         if ($author["author-name"] == "") {
 117                                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
 118                         }
 119                         $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
 120                         if ($value != "") {
 121                                 $author["author-name"] = $value;
 122                         }
 123                         if ($simulate) {
 124                                 $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
 125
 126                                 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 127                                 if ($value != "") {
 128                                         $author["author-nick"] = $value;
 129                                 }
 130                                 $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
 131                                 if ($value != "") {
 132                                         $author["author-location"] = $value;
 133                                 }
 134                                 $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 135                                 if ($value != "") {
 136                                         $author["author-about"] = $value;
 137                                 }
 138                                 $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
 139                                 if (is_object($avatar)) {
 140                                         foreach ($avatar AS $attributes) {
 141                                                 if ($attributes->name == "href") {
 142                                                         $author["author-avatar"] = $attributes->textContent;
 143                                                 }
 144                                         }
 145                                 }
 146                         }
 147
 148                         $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 149
 150                         $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 151
 152                         $entries = $xpath->query('/atom:feed/atom:entry');
 153                 }
 154
 155                 // Is it RSS?
 156                 if ($xpath->query('/rss/channel')->length > 0) {
 157                         $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 158
 159                         $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 160                         $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 161
 162                         if ($author["author-name"] == "") {
 163                                 $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 164                         }
 165                         if ($author["author-name"] == "") {
 166                                 $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 167                         }
 168                         $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 169
 170                         $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 171
 172                         $entries = $xpath->query('/rss/channel/item');
 173                 }
 174
 175                 if (!$simulate) {
 176                         $author["author-link"] = $contact["url"];
 177
 178                         if ($author["author-name"] == "") {
 179                                 $author["author-name"] = $contact["name"];
 180                         }
 181                         $author["author-avatar"] = $contact["thumb"];
 182
 183                         $author["owner-link"] = $contact["url"];
 184                         $author["owner-name"] = $contact["name"];
 185                         $author["owner-avatar"] = $contact["thumb"];
 186                 }
 187
 188                 $header = array();
 189                 $header["uid"] = $importer["uid"];
 190                 $header["network"] = NETWORK_FEED;
 191                 $header["type"] = "remote";
 192                 $header["wall"] = 0;
 193                 $header["origin"] = 0;
 194                 $header["gravity"] = GRAVITY_PARENT;
 195                 $header["private"] = 2;
 196                 $header["verb"] = ACTIVITY_POST;
 197                 $header["object-type"] = ACTIVITY_OBJ_NOTE;
 198
 199                 $header["contact-id"] = $contact["id"];
 200
 201                 if (!strlen($contact["notify"])) {
 202                         // one way feed - no remote comment ability
 203                         $header["last-child"] = 0;
 204                 }
 205
 206                 if (!is_object($entries)) {
 207                         logger("There are no entries in this feed.", LOGGER_DEBUG);
 208                         return;
 209                 }
 210
 211                 $items = array();
 212
 213                 $entrylist = array();
 214
 215                 foreach ($entries AS $entry) {
 216                         $entrylist[] = $entry;
 217                 }
 218                 foreach (array_reverse($entrylist) AS $entry) {
 219                         $item = array_merge($header, $author);
 220
 221                         $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 222                         if (!is_object($alternate)) {
 223                                 $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 224                         }
 225                         if (is_object($alternate)) {
 226                                 foreach ($alternate AS $attributes) {
 227                                         if ($attributes->name == "href") {
 228                                                 $item["plink"] = $attributes->textContent;
 229                                         }
 230                                 }
 231                         }
 232                         if ($item["plink"] == "") {
 233                                 $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 234                         }
 235                         if ($item["plink"] == "") {
 236                                 $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 237                         }
 238
 239                         $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 240
 241                         if ($item["uri"] == "") {
 242                                 $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 243                         }
 244                         if ($item["uri"] == "") {
 245                                 $item["uri"] = $item["plink"];
 246                         }
 247
 248                         $orig_plink = $item["plink"];
 249
 250                         $item["plink"] = original_url($item["plink"]);
 251
 252                         $item["parent-uri"] = $item["uri"];
 253
 254                         if (!$simulate) {
 255                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 256                                         $importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
 257                                 $previous = dba::select('item', ['id'], $condition, ['limit' => 1]);
 258                                 if (DBM::is_result($previous)) {
 259                                         logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG);
 260                                         continue;
 261                                 }
 262                         }
 263
 264                         $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 265
 266                         if ($item["title"] == "") {
 267                                 $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 268                         }
 269                         if ($item["title"] == "") {
 270                                 $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 271                         }
 272                         $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 273
 274                         if ($published == "") {
 275                                 $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 276                         }
 277                         if ($published == "") {
 278                                 $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 279                         }
 280                         $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 281
 282                         if ($updated == "") {
 283                                 $updated = $published;
 284                         }
 285                         if ($published != "") {
 286                                 $item["created"] = $published;
 287                         }
 288                         if ($updated != "") {
 289                                 $item["edited"] = $updated;
 290                         }
 291                         $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 292
 293                         if ($creator == "") {
 294                                 $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 295                         }
 296                         if ($creator == "") {
 297                                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 298                         }
 299                         if ($creator != "") {
 300                                 $item["author-name"] = $creator;
 301                         }
 302                         if ($pubDate != "") {
 303                                 $item["edited"] = $item["created"] = $pubDate;
 304                         }
 305                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 306
 307                         if ($creator != "") {
 308                                 $item["author-name"] = $creator;
 309                         }
 310
 311                         /// @TODO ?
 312                         // <category>Ausland</category>
 313                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 314
 315                         $attachments = array();
 316
 317                         $enclosures = $xpath->query("enclosure", $entry);
 318                         foreach ($enclosures AS $enclosure) {
 319                                 $href = "";
 320                                 $length = "";
 321                                 $type = "";
 322                                 $title = "";
 323
 324                                 foreach ($enclosure->attributes AS $attributes) {
 325                                         if ($attributes->name == "url") {
 326                                                 $href = $attributes->textContent;
 327                                         } elseif ($attributes->name == "length") {
 328                                                 $length = $attributes->textContent;
 329                                         } elseif ($attributes->name == "type") {
 330                                                 $type = $attributes->textContent;
 331                                         }
 332                                 }
 333                                 if (strlen($item["attach"])) {
 334                                         $item["attach"] .= ',';
 335                                 }
 336
 337                                 $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 338
 339                                 $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 340                         }
 341
 342                         $tags = '';
 343                         $categories = $xpath->query("category", $entry);
 344                         foreach ($categories AS $category) {
 345                                 $hashtag = $category->nodeValue;
 346                                 if ($tags != '') {
 347                                         $tags .= ', ';
 348                                 }
 349
 350                                 $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
 351                                 $tags .= $taglink;
 352                         }
 353
 354                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 355
 356                         if ($body == "") {
 357                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 358                         }
 359                         if ($body == "") {
 360                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 361                         }
 362                         if ($body == "") {
 363                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 364                         }
 365
 366                         // remove the content of the title if it is identically to the body
 367                         // This helps with auto generated titles e.g. from tumblr
 368                         if (title_is_body($item["title"], $body)) {
 369                                 $item["title"] = "";
 370                         }
 371                         $item["body"] = html2bbcode($body, $basepath);
 372
 373                         if (($item["body"] == '') && ($item["title"] != '')) {
 374                                 $item["body"] = $item["title"];
 375                                 $item["title"] = '';
 376                         }
 377
 378                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 379                                 $preview = "";
 380
 381                                 // Handle enclosures and treat them as preview picture
 382                                 foreach ($attachments AS $attachment) {
 383                                         if ($attachment["type"] == "image/jpeg") {
 384                                                 $preview = $attachment["link"];
 385                                         }
 386                                 }
 387
 388                                 // Remove a possible link to the item itself
 389                                 $item["body"] = str_replace($item["plink"], '', $item["body"]);
 390                                 $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
 391
 392                                 // Replace the content when the title is longer than the body
 393                                 $replace = (strlen($item["title"]) > strlen($item["body"]));
 394
 395                                 // Replace it, when there is an image in the body
 396                                 if (strstr($item["body"], '[/img]')) {
 397                                         $replace = true;
 398                                 }
 399
 400                                 // Replace it, when there is a link in the body
 401                                 if (strstr($item["body"], '[/url]')) {
 402                                         $replace = true;
 403                                 }
 404
 405                                 if ($replace) {
 406                                         $item["body"] = $item["title"];
 407                                 }
 408                                 // We always strip the title since it will be added in the page information
 409                                 $item["title"] = "";
 410                                 $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 411                                 $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 412                                 $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 413                                 unset($item["attach"]);
 414                         } else {
 415                                 if ($contact["fetch_further_information"] == 3) {
 416                                         if (!empty($tags)) {
 417                                                 $item["tag"] = $tags;
 418                                         } else {
 419                                                 $item["tag"] = add_page_keywords($item["plink"], false, $preview, true, $contact["ffi_keyword_blacklist"]);
 420                                         }
 421                                         $item["body"] .= "\n".$item['tag'];
 422                                 }
 423                                 if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
 424                                         $item["body"] .= "[hr][url]".$item['plink']."[/url]";
 425                                 }
 426                         }
 427
 428                         if (!$simulate) {
 429                                 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 430
 431                                 $notify = item_is_remote_self($contact, $item);
 432
 433                                 // Distributed items should have a well formatted URI.
 434                                 // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 435                                 if ($notify) {
 436                                         $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
 437                                         unset($item['uri']);
 438                                         unset($item['parent-uri']);
 439                                 }
 440
 441                                 $id = item_store($item, false, $notify);
 442
 443                                 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 444                         } else {
 445                                 $items[] = $item;
 446                         }
 447                         if ($simulate) {
 448                                 break;
 449                         }
 450                 }
 451
 452                 if ($simulate) {
 453                         return array("header" => $author, "items" => $items);
 454                 }
 455         }
 456 }