src/Protocol/Feed.php

   1 <?php
   2 use Friendica\Database\DBM;
   3 use Friendica\Core\System;
   4
   5 require_once("include/html2bbcode.php");
   6 require_once("include/items.php");
   7
   8 /**
   9  * @brief Read a RSS/RDF/Atom feed and create an item entry for it
  10  *
  11  * @param string $xml The feed data
  12  * @param array $importer The user record of the importer
  13  * @param array $contact The contact record of the feed
  14  * @param string $hub Unused dummy value for compatibility reasons
  15  * @param bool $simulate If enabled, no data is imported
  16  *
  17  * @return array In simulation mode it returns the header and the first item
  18  */
  19 function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
  20
  21         $a = get_app();
  22
  23         if (!$simulate) {
  24                 logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
  25         } else {
  26                 logger("Test Atom/RSS feed", LOGGER_DEBUG);
  27         }
  28         if ($xml == "") {
  29                 logger('XML is empty.', LOGGER_DEBUG);
  30                 return;
  31         }
  32
  33         if (!empty($contact['poll'])) {
  34                 $basepath = $contact['poll'];
  35         } elseif (!empty($contact['url'])) {
  36                 $basepath = $contact['url'];
  37         } else {
  38                 $basepath = '';
  39         }
  40
  41         $doc = new DOMDocument();
  42         @$doc->loadXML(trim($xml));
  43         $xpath = new DomXPath($doc);
  44         $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  45         $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  46         $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  47         $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  48         $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  49         $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  50         $xpath->registerNamespace('poco', NAMESPACE_POCO);
  51
  52         $author = array();
  53
  54         // Is it RDF?
  55         if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  56                 $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  57                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  58
  59                 if ($author["author-name"] == "") {
  60                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  61                 }
  62                 $entries = $xpath->query('/rdf:RDF/rss:item');
  63         }
  64
  65         // Is it Atom?
  66         if ($xpath->query('/atom:feed')->length > 0) {
  67                 $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  68                 if (is_object($alternate)) {
  69                         foreach ($alternate AS $attributes) {
  70                                 if ($attributes->name == "href") {
  71                                         $author["author-link"] = $attributes->textContent;
  72                                 }
  73                         }
  74                 }
  75
  76                 if ($author["author-link"] == "") {
  77                         $author["author-link"] = $author["author-id"];
  78                 }
  79                 if ($author["author-link"] == "") {
  80                         $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  81                         if (is_object($self)) {
  82                                 foreach ($self AS $attributes) {
  83                                         if ($attributes->name == "href") {
  84                                                 $author["author-link"] = $attributes->textContent;
  85                                         }
  86                                 }
  87                         }
  88                 }
  89
  90                 if ($author["author-link"] == "") {
  91                         $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
  92                 }
  93                 $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
  94
  95                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
  96
  97                 if ($author["author-name"] == "") {
  98                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
  99                 }
 100                 if ($author["author-name"] == "") {
 101                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
 102                 }
 103                 $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
 104                 if ($value != "") {
 105                         $author["author-name"] = $value;
 106                 }
 107                 if ($simulate) {
 108                         $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
 109
 110                         $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 111                         if ($value != "") {
 112                                 $author["author-nick"] = $value;
 113                         }
 114                         $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
 115                         if ($value != "") {
 116                                 $author["author-location"] = $value;
 117                         }
 118                         $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 119                         if ($value != "") {
 120                                 $author["author-about"] = $value;
 121                         }
 122                         $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
 123                         if (is_object($avatar)) {
 124                                 foreach ($avatar AS $attributes) {
 125                                         if ($attributes->name == "href") {
 126                                                 $author["author-avatar"] = $attributes->textContent;
 127                                         }
 128                                 }
 129                         }
 130                 }
 131
 132                 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 133
 134                 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 135
 136                 $entries = $xpath->query('/atom:feed/atom:entry');
 137         }
 138
 139         // Is it RSS?
 140         if ($xpath->query('/rss/channel')->length > 0) {
 141                 $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 142
 143                 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 144                 $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 145
 146                 if ($author["author-name"] == "") {
 147                         $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 148                 }
 149                 if ($author["author-name"] == "") {
 150                         $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 151                 }
 152                 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 153
 154                 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 155
 156                 $entries = $xpath->query('/rss/channel/item');
 157         }
 158
 159         if (!$simulate) {
 160                 $author["author-link"] = $contact["url"];
 161
 162                 if ($author["author-name"] == "") {
 163                         $author["author-name"] = $contact["name"];
 164                 }
 165                 $author["author-avatar"] = $contact["thumb"];
 166
 167                 $author["owner-link"] = $contact["url"];
 168                 $author["owner-name"] = $contact["name"];
 169                 $author["owner-avatar"] = $contact["thumb"];
 170         }
 171
 172         $header = array();
 173         $header["uid"] = $importer["uid"];
 174         $header["network"] = NETWORK_FEED;
 175         $header["type"] = "remote";
 176         $header["wall"] = 0;
 177         $header["origin"] = 0;
 178         $header["gravity"] = GRAVITY_PARENT;
 179         $header["private"] = 2;
 180         $header["verb"] = ACTIVITY_POST;
 181         $header["object-type"] = ACTIVITY_OBJ_NOTE;
 182
 183         $header["contact-id"] = $contact["id"];
 184
 185         if (!strlen($contact["notify"])) {
 186                 // one way feed - no remote comment ability
 187                 $header["last-child"] = 0;
 188         }
 189
 190         if (!is_object($entries)) {
 191                 logger("There are no entries in this feed.", LOGGER_DEBUG);
 192                 return;
 193         }
 194
 195         $items = array();
 196
 197         $entrylist = array();
 198
 199         foreach ($entries AS $entry) {
 200                 $entrylist[] = $entry;
 201         }
 202         foreach (array_reverse($entrylist) AS $entry) {
 203                 $item = array_merge($header, $author);
 204
 205                 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 206                 if (!is_object($alternate)) {
 207                         $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 208                 }
 209                 if (is_object($alternate)) {
 210                         foreach ($alternate AS $attributes) {
 211                                 if ($attributes->name == "href") {
 212                                         $item["plink"] = $attributes->textContent;
 213                                 }
 214                         }
 215                 }
 216                 if ($item["plink"] == "") {
 217                         $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 218                 }
 219                 if ($item["plink"] == "") {
 220                         $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 221                 }
 222
 223                 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 224
 225                 if ($item["uri"] == "") {
 226                         $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 227                 }
 228                 if ($item["uri"] == "") {
 229                         $item["uri"] = $item["plink"];
 230                 }
 231
 232                 $orig_plink = $item["plink"];
 233
 234                 $item["plink"] = original_url($item["plink"]);
 235
 236                 $item["parent-uri"] = $item["uri"];
 237
 238                 if (!$simulate) {
 239                         $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
 240                                 intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
 241                         if (DBM::is_result($r)) {
 242                                 logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
 243                                 continue;
 244                         }
 245                 }
 246
 247                 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 248
 249                 if ($item["title"] == "") {
 250                         $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 251                 }
 252                 if ($item["title"] == "") {
 253                         $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 254                 }
 255                 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 256
 257                 if ($published == "") {
 258                         $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 259                 }
 260                 if ($published == "") {
 261                         $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 262                 }
 263                 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 264
 265                 if ($updated == "") {
 266                         $updated = $published;
 267                 }
 268                 if ($published != "") {
 269                         $item["created"] = $published;
 270                 }
 271                 if ($updated != "") {
 272                         $item["edited"] = $updated;
 273                 }
 274                 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 275
 276                 if ($creator == "") {
 277                         $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 278                 }
 279                 if ($creator == "") {
 280                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 281                 }
 282                 if ($creator != "") {
 283                         $item["author-name"] = $creator;
 284                 }
 285                 if ($pubDate != "") {
 286                         $item["edited"] = $item["created"] = $pubDate;
 287                 }
 288                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 289
 290                 if ($creator != "") {
 291                         $item["author-name"] = $creator;
 292                 }
 293
 294                 /// @TODO ?
 295                 // <category>Ausland</category>
 296                 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 297
 298                 $attachments = array();
 299
 300                 $enclosures = $xpath->query("enclosure", $entry);
 301                 foreach ($enclosures AS $enclosure) {
 302                         $href = "";
 303                         $length = "";
 304                         $type = "";
 305                         $title = "";
 306
 307                         foreach ($enclosure->attributes AS $attributes) {
 308                                 if ($attributes->name == "url") {
 309                                         $href = $attributes->textContent;
 310                                 } elseif ($attributes->name == "length") {
 311                                         $length = $attributes->textContent;
 312                                 } elseif ($attributes->name == "type") {
 313                                         $type = $attributes->textContent;
 314                                 }
 315                         }
 316                         if (strlen($item["attach"])) {
 317                                 $item["attach"] .= ',';
 318                         }
 319
 320                         $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 321
 322                         $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 323                 }
 324
 325                 $tags = '';
 326                 $categories = $xpath->query("category", $entry);
 327                 foreach ($categories AS $category) {
 328                         $hashtag = $category->nodeValue;
 329                         if ($tags != '') {
 330                                 $tags .= ', ';
 331                         }
 332
 333                         $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
 334                         $tags .= $taglink;
 335                 }
 336
 337                 $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 338
 339                 if ($body == "") {
 340                         $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 341                 }
 342                 if ($body == "") {
 343                         $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 344                 }
 345                 if ($body == "") {
 346                         $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 347                 }
 348
 349                 // remove the content of the title if it is identically to the body
 350                 // This helps with auto generated titles e.g. from tumblr
 351                 if (title_is_body($item["title"], $body)) {
 352                         $item["title"] = "";
 353                 }
 354                 $item["body"] = html2bbcode($body, $basepath);
 355
 356                 if (($item["body"] == '') && ($item["title"] != '')) {
 357                         $item["body"] = $item["title"];
 358                         $item["title"] = '';
 359                 }
 360
 361                 if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 362                         $preview = "";
 363
 364                         // Handle enclosures and treat them as preview picture
 365                         foreach ($attachments AS $attachment) {
 366                                 if ($attachment["type"] == "image/jpeg") {
 367                                         $preview = $attachment["link"];
 368                                 }
 369                         }
 370
 371                         // Remove a possible link to the item itself
 372                         $item["body"] = str_replace($item["plink"], '', $item["body"]);
 373                         $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
 374
 375                         // Replace the content when the title is longer than the body
 376                         $replace = (strlen($item["title"]) > strlen($item["body"]));
 377
 378                         // Replace it, when there is an image in the body
 379                         if (strstr($item["body"], '[/img]')) {
 380                                 $replace = true;
 381                         }
 382
 383                         // Replace it, when there is a link in the body
 384                         if (strstr($item["body"], '[/url]')) {
 385                                 $replace = true;
 386                         }
 387
 388                         if ($replace) {
 389                                 $item["body"] = $item["title"];
 390                         }
 391                         // We always strip the title since it will be added in the page information
 392                         $item["title"] = "";
 393                         $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 394                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 395                         $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 396                         unset($item["attach"]);
 397                 } else {
 398                         if ($contact["fetch_further_information"] == 3) {
 399                                 if (!empty($tags)) {
 400                                         $item["tag"] = $tags;
 401                                 } else {
 402                                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, true, $contact["ffi_keyword_blacklist"]);
 403                                 }
 404                                 $item["body"] .= "\n".$item['tag'];
 405                         }
 406                         if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
 407                                 $item["body"] .= "[hr][url]".$item['plink']."[/url]";
 408                         }
 409                 }
 410
 411                 if (!$simulate) {
 412                         logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 413
 414                         $notify = item_is_remote_self($contact, $item);
 415
 416                         // Distributed items should have a well formatted URI.
 417                         // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 418                         if ($notify) {
 419                                 $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
 420                                 unset($item['uri']);
 421                                 unset($item['parent-uri']);
 422                         }
 423
 424                         $id = item_store($item, false, $notify);
 425
 426                         logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 427                 } else {
 428                         $items[] = $item;
 429                 }
 430                 if ($simulate) {
 431                         break;
 432                 }
 433         }
 434
 435         if ($simulate) {
 436                 return array("header" => $author, "items" => $items);
 437         }
 438 }