include/feed.php

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