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         $doc = new DOMDocument();
  31         @$doc->loadXML($xml);
  32         $xpath = new DomXPath($doc);
  33         $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  34         $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  35         $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  36         $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  37         $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  38         $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  39         $xpath->registerNamespace('poco', NAMESPACE_POCO);
  40
  41         $author = array();
  42
  43         // Is it RDF?
  44         if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  45                 $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  46                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  47
  48                 if ($author["author-name"] == "") {
  49                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  50                 }
  51                 $entries = $xpath->query('/rdf:RDF/rss:item');
  52         }
  53
  54         // Is it Atom?
  55         if ($xpath->query('/atom:feed')->length > 0) {
  56                 $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  57                 if (is_object($alternate)) {
  58                         foreach ($alternate AS $attributes) {
  59                                 if ($attributes->name == "href") {
  60                                         $author["author-link"] = $attributes->textContent;
  61                                 }
  62                         }
  63                 }
  64
  65                 if ($author["author-link"] == "") {
  66                         $author["author-link"] = $author["author-id"];
  67                 }
  68                 if ($author["author-link"] == "") {
  69                         $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  70                         if (is_object($self)) {
  71                                 foreach ($self AS $attributes) {
  72                                         if ($attributes->name == "href") {
  73                                                 $author["author-link"] = $attributes->textContent;
  74                                         }
  75                                 }
  76                         }
  77                 }
  78
  79                 if ($author["author-link"] == "") {
  80                         $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
  81                 }
  82                 $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
  83
  84                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
  85
  86                 if ($author["author-name"] == "") {
  87                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
  88                 }
  89                 if ($author["author-name"] == "") {
  90                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
  91                 }
  92                 $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
  93                 if ($value != "") {
  94                         $author["author-name"] = $value;
  95                 }
  96                 if ($simulate) {
  97                         $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
  98
  99                         $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
 100                         if ($value != "") {
 101                                 $author["author-nick"] = $value;
 102                         }
 103                         $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
 104                         if ($value != "") {
 105                                 $author["author-location"] = $value;
 106                         }
 107                         $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
 108                         if ($value != "") {
 109                                 $author["author-about"] = $value;
 110                         }
 111                 }
 112
 113                 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 114
 115                 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
 116
 117                 $entries = $xpath->query('/atom:feed/atom:entry');
 118         }
 119
 120         // Is it RSS?
 121         if ($xpath->query('/rss/channel')->length > 0) {
 122                 $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 123
 124                 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 125                 $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 126
 127                 if ($author["author-name"] == "") {
 128                         $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 129                 }
 130                 if ($author["author-name"] == "") {
 131                         $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 132                 }
 133                 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 134
 135                 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 136
 137                 $entries = $xpath->query('/rss/channel/item');
 138         }
 139
 140         if (!$simulate) {
 141                 $author["author-link"] = $contact["url"];
 142
 143                 if ($author["author-name"] == "") {
 144                         $author["author-name"] = $contact["name"];
 145                 }
 146                 $author["author-avatar"] = $contact["thumb"];
 147
 148                 $author["owner-link"] = $contact["url"];
 149                 $author["owner-name"] = $contact["name"];
 150                 $author["owner-avatar"] = $contact["thumb"];
 151         }
 152
 153         $header = array();
 154         $header["uid"] = $importer["uid"];
 155         $header["network"] = NETWORK_FEED;
 156         $header["type"] = "remote";
 157         $header["wall"] = 0;
 158         $header["origin"] = 0;
 159         $header["gravity"] = GRAVITY_PARENT;
 160         $header["private"] = 2;
 161         $header["verb"] = ACTIVITY_POST;
 162         $header["object-type"] = ACTIVITY_OBJ_NOTE;
 163
 164         $header["contact-id"] = $contact["id"];
 165
 166         if (!strlen($contact["notify"])) {
 167                 // one way feed - no remote comment ability
 168                 $header["last-child"] = 0;
 169         }
 170
 171         if (!is_object($entries)) {
 172                 logger("There are no entries in this feed.", LOGGER_DEBUG);
 173                 return;
 174         }
 175
 176         $items = array();
 177
 178         $entrylist = array();
 179
 180         foreach ($entries AS $entry) {
 181                 $entrylist[] = $entry;
 182         }
 183         foreach (array_reverse($entrylist) AS $entry) {
 184                 $item = array_merge($header, $author);
 185
 186                 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 187                 if (!is_object($alternate)) {
 188                         $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 189                 }
 190                 if (is_object($alternate)) {
 191                         foreach ($alternate AS $attributes) {
 192                                 if ($attributes->name == "href") {
 193                                         $item["plink"] = $attributes->textContent;
 194                                 }
 195                         }
 196                 }
 197                 if ($item["plink"] == "") {
 198                         $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 199                 }
 200                 if ($item["plink"] == "") {
 201                         $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 202                 }
 203
 204                 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 205
 206                 if ($item["uri"] == "") {
 207                         $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 208                 }
 209                 if ($item["uri"] == "") {
 210                         $item["uri"] = $item["plink"];
 211                 }
 212
 213                 $orig_plink = $item["plink"];
 214
 215                 $item["plink"] = original_url($item["plink"]);
 216
 217                 $item["parent-uri"] = $item["uri"];
 218
 219                 if (!$simulate) {
 220                         $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
 221                                 intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
 222                         if (dbm::is_result($r)) {
 223                                 logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
 224                                 continue;
 225                         }
 226                 }
 227
 228                 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 229
 230                 if ($item["title"] == "") {
 231                         $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 232                 }
 233                 if ($item["title"] == "") {
 234                         $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 235                 }
 236                 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 237
 238                 if ($published == "") {
 239                         $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 240                 }
 241                 if ($published == "") {
 242                         $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 243                 }
 244                 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 245
 246                 if ($updated == "") {
 247                         $updated = $published;
 248                 }
 249                 if ($published != "") {
 250                         $item["created"] = $published;
 251                 }
 252                 if ($updated != "") {
 253                         $item["edited"] = $updated;
 254                 }
 255                 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 256
 257                 if ($creator == "") {
 258                         $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 259                 }
 260                 if ($creator == "") {
 261                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 262                 }
 263                 if ($creator != "") {
 264                         $item["author-name"] = $creator;
 265                 }
 266                 if ($pubDate != "") {
 267                         $item["edited"] = $item["created"] = $pubDate;
 268                 }
 269                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 270
 271                 if ($creator != "") {
 272                         $item["author-name"] = $creator;
 273                 }
 274
 275                 /// @TODO ?
 276                 // <category>Ausland</category>
 277                 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 278
 279                 $attachments = array();
 280
 281                 $enclosures = $xpath->query("enclosure", $entry);
 282                 foreach ($enclosures AS $enclosure) {
 283                         $href = "";
 284                         $length = "";
 285                         $type = "";
 286                         $title = "";
 287
 288                         foreach ($enclosure->attributes AS $attributes) {
 289                                 if ($attributes->name == "url") {
 290                                         $href = $attributes->textContent;
 291                                 } elseif ($attributes->name == "length") {
 292                                         $length = $attributes->textContent;
 293                                 } elseif ($attributes->name == "type") {
 294                                         $type = $attributes->textContent;
 295                                 }
 296                         }
 297                         if (strlen($item["attach"])) {
 298                                 $item["attach"] .= ',';
 299                         }
 300
 301                         $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 302
 303                         $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 304                 }
 305
 306                 if ($contact["fetch_further_information"]) {
 307                         $preview = "";
 308
 309                         // Handle enclosures and treat them as preview picture
 310                         foreach ($attachments AS $attachment) {
 311                                 if ($attachment["type"] == "image/jpeg") {
 312                                         $preview = $attachment["link"];
 313                                 }
 314                         }
 315
 316                         $item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 317                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 318                         $item["title"] = "";
 319                         $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 320                         unset($item["attach"]);
 321                 } else {
 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()', $entry)->item(0)->nodeValue);
 332                         }
 333                         // remove the content of the title if it is identically to the body
 334                         // This helps with auto generated titles e.g. from tumblr
 335                         if (title_is_body($item["title"], $body)) {
 336                                 $item["title"] = "";
 337                         }
 338                         $item["body"] = html2bbcode($body);
 339                 }
 340
 341                 if (!$simulate) {
 342                         logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 343
 344                         $notify = item_is_remote_self($contact, $item);
 345
 346                         // Distributed items should have a well formatted URI.
 347                         // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 348                         if ($notify) {
 349                                 $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
 350                                 unset($item['uri']);
 351                                 unset($item['parent-uri']);
 352                         }
 353
 354                         $id = item_store($item, false, $notify);
 355
 356                         logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 357                 } else {
 358                         $items[] = $item;
 359                 }
 360                 if ($simulate) {
 361                         break;
 362                 }
 363         }
 364
 365         if ($simulate) {
 366                 return array("header" => $author, "items" => $items);
 367         }
 368 }