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