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                 $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
  63
  64                 if ($author["author-link"] == "")
  65                         $author["author-link"] = $author["author-id"];
  66
  67                 if ($author["author-link"] == "") {
  68                         $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  69                         if (is_object($self))
  70                                 foreach($self AS $attributes)
  71                                         if ($attributes->name == "href")
  72                                                 $author["author-link"] = $attributes->textContent;
  73                 }
  74
  75                 if ($author["author-link"] == "")
  76                         $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
  77
  78                 $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
  79
  80                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
  81
  82                 if ($author["author-name"] == "")
  83                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
  84
  85                 if ($author["author-name"] == "")
  86                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
  87
  88                 $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
  89                 if ($value != "")
  90                         $author["author-name"] = $value;
  91
  92                 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
  93                 if ($value != "")
  94                         $author["author-nick"] = $value;
  95
  96                 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
  97
  98                 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
  99
 100                 $entries = $xpath->query('/atom:feed/atom:entry');
 101         }
 102
 103         // Is it RSS?
 104         if ($xpath->query('/rss/channel')->length > 0) {
 105                 $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
 106
 107                 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
 108                 $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 109
 110                 if ($author["author-name"] == "")
 111                         $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
 112
 113                 if ($author["author-name"] == "")
 114                         $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
 115
 116                 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
 117
 118                 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 119
 120                 $entries = $xpath->query('/rss/channel/item');
 121         }
 122
 123         if (!$simulate) {
 124                 $author["author-link"] = $contact["url"];
 125
 126                 if ($author["author-name"] == "")
 127                         $author["author-name"] = $contact["name"];
 128
 129                 $author["author-avatar"] = $contact["thumb"];
 130
 131                 $author["owner-link"] = $contact["url"];
 132                 $author["owner-name"] = $contact["name"];
 133                 $author["owner-avatar"] = $contact["thumb"];
 134
 135                 // This is no field in the item table. So we have to unset it.
 136                 unset($author["author-nick"]);
 137                 unset($author["author-id"]);
 138         }
 139
 140         $header = array();
 141         $header["uid"] = $importer["uid"];
 142         $header["network"] = NETWORK_FEED;
 143         $header["type"] = "remote";
 144         $header["wall"] = 0;
 145         $header["origin"] = 0;
 146         $header["gravity"] = GRAVITY_PARENT;
 147         $header["private"] = 2;
 148         $header["verb"] = ACTIVITY_POST;
 149         $header["object-type"] = ACTIVITY_OBJ_NOTE;
 150
 151         $header["contact-id"] = $contact["id"];
 152
 153         if(!strlen($contact["notify"])) {
 154                 // one way feed - no remote comment ability
 155                 $header["last-child"] = 0;
 156         }
 157
 158         if (!is_object($entries)) {
 159                 logger("There are no entries in this feed.", LOGGER_DEBUG);
 160                 return;
 161         }
 162
 163         $items = array();
 164
 165         $entrylist = array();
 166
 167         foreach ($entries AS $entry)
 168                 $entrylist[] = $entry;
 169
 170         foreach (array_reverse($entrylist) AS $entry) {
 171                 $item = array_merge($header, $author);
 172
 173                 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 174
 175                 if ($item["title"] == "")
 176                         $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 177
 178                 if ($item["title"] == "")
 179                         $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 180
 181                 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 182                 if (!is_object($alternate))
 183                         $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 184
 185                 if (is_object($alternate))
 186                         foreach($alternate AS $attributes)
 187                                 if ($attributes->name == "href")
 188                                         $item["plink"] = $attributes->textContent;
 189
 190                 if ($item["plink"] == "")
 191                         $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 192
 193                 if ($item["plink"] == "")
 194                         $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 195
 196                 $item["plink"] = original_url($item["plink"]);
 197
 198                 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 199
 200                 if ($item["uri"] == "")
 201                         $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 202
 203                 if ($item["uri"] == "")
 204                         $item["uri"] = $item["plink"];
 205
 206                 $item["parent-uri"] = $item["uri"];
 207
 208                 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 209
 210                 if ($published == "")
 211                         $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 212
 213                 if ($published == "")
 214                         $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 215
 216                 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 217
 218                 if ($updated == "")
 219                         $updated = $published;
 220
 221                 if ($published != "")
 222                         $item["created"] = $published;
 223
 224                 if ($updated != "")
 225                         $item["edited"] = $updated;
 226
 227                 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 228
 229                 if ($creator == "")
 230                         $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 231
 232                 if ($creator == "")
 233                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 234
 235                 if ($creator != "")
 236                         $item["author-name"] = $creator;
 237
 238                 if ($pubDate != "")
 239                         $item["edited"] = $item["created"] = $pubDate;
 240
 241                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 242
 243                 if ($creator != "")
 244                         $item["author-name"] = $creator;
 245
 246                 if (!$simulate) {
 247                         $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
 248                                 intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
 249                         if ($r) {
 250                                 logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
 251                                 continue;
 252                         }
 253                 }
 254
 255                 /// @TODO ?
 256                 // <category>Ausland</category>
 257                 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 258
 259                 $attachments = array();
 260
 261                 $enclosures = $xpath->query("enclosure", $entry);
 262                 foreach ($enclosures AS $enclosure) {
 263                         $href = "";
 264                         $length = "";
 265                         $type = "";
 266                         $title = "";
 267
 268                         foreach($enclosure->attributes AS $attributes) {
 269                                 if ($attributes->name == "url")
 270                                         $href = $attributes->textContent;
 271                                 elseif ($attributes->name == "length")
 272                                         $length = $attributes->textContent;
 273                                 elseif ($attributes->name == "type")
 274                                         $type = $attributes->textContent;
 275                         }
 276                         if(strlen($item["attach"]))
 277                                 $item["attach"] .= ',';
 278
 279                         $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 280
 281                         $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 282                 }
 283
 284                 if ($contact["fetch_further_information"]) {
 285                         $preview = "";
 286
 287                         // Handle enclosures and treat them as preview picture
 288                         foreach ($attachments AS $attachment)
 289                                 if ($attachment["type"] == "image/jpeg")
 290                                         $preview = $attachment["link"];
 291
 292                         $item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 293                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 294                         $item["title"] = "";
 295                         $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 296                         unset($item["attach"]);
 297                 } else {
 298                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 299
 300                         if ($body == "")
 301                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 302
 303                         if ($body == "")
 304                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 305
 306                         if ($body == "")
 307                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 308
 309                         // remove the content of the title if it is identically to the body
 310                         // This helps with auto generated titles e.g. from tumblr
 311                         if (title_is_body($item["title"], $body))
 312                                 $item["title"] = "";
 313
 314                         $item["body"] = html2bbcode($body);
 315                 }
 316
 317                 if (!$simulate) {
 318                         logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 319
 320                         $notify = item_is_remote_self($contact, $item);
 321                         $id = item_store($item, false, $notify);
 322
 323                         logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 324                 } else
 325                         $items[] = $item;
 326
 327                 if ($simulate)
 328                         break;
 329         }
 330
 331         if ($simulate)
 332                 return array("header" => $author, "items" => $items);
 333 }
 334 ?>