include/feed.php

   1 <?php
   2 require_once("include/html2bbcode.php");
   3 require_once("include/items.php");
   4
   5 function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
   6
   7         $a = get_app();
   8
   9         logger("Import Atom/RSS feed", LOGGER_DEBUG);
  10
  11         if ($xml == "")
  12                 return;
  13
  14         $doc = new DOMDocument();
  15         @$doc->loadXML($xml);
  16         $xpath = new DomXPath($doc);
  17         $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
  18         $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  19         $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  20         $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  21         $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  22         $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  23         $xpath->registerNamespace('poco', NAMESPACE_POCO);
  24
  25         $author = array();
  26
  27         // Is it RDF?
  28         if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
  29                 $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
  30                 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
  31
  32                 if ($author["author-name"] == "")
  33                         $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
  34
  35                 $entries = $xpath->query('/rdf:RDF/rss:item');
  36         }
  37
  38         // Is it Atom?
  39         if ($xpath->query('/atom:feed/atom:entry')->length > 0) {
  40                 $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
  41                 if (is_object($alternate))
  42                         foreach($alternate AS $attributes)
  43                                 if ($attributes->name == "href")
  44                                         $author["author-link"] = $attributes->textContent;
  45
  46                 if ($author["author-link"] == "")
  47                         $author["author-link"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
  48
  49                 if ($author["author-link"] == "") {
  50                         $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
  51                         if (is_object($self))
  52                                 foreach($self AS $attributes)
  53                                         if ($attributes->name == "href")
  54                                                 $author["author-link"] = $attributes->textContent;
  55                 }
  56
  57                 if ($author["author-link"] == "")
  58                         $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
  59
  60                 $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
  61
  62                 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
  63
  64                 if ($author["author-name"] == "")
  65                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
  66
  67                 if ($author["author-name"] == "")
  68                         $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
  69
  70                 $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
  71                 if ($value != "")
  72                         $author["author-name"] = $value;
  73
  74                 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
  75                 if ($value != "")
  76                         $author["author-nick"] = $value;
  77
  78                 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
  79
  80                 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
  81
  82                 $entries = $xpath->query('/atom:feed/atom:entry');
  83         }
  84
  85         // Is it RSS?
  86         if ($xpath->query('/rss/channel')->length > 0) {
  87                 $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
  88
  89                 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
  90                 $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
  91
  92                 if ($author["author-name"] == "")
  93                         $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
  94
  95                 if ($author["author-name"] == "")
  96                         $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
  97
  98                 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
  99
 100                 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
 101
 102                 $entries = $xpath->query('/rss/channel/item');
 103         }
 104
 105         if (is_array($contact)) {
 106                 $author["author-link"] = $contact["url"];
 107
 108                 if ($author["author-name"] == "")
 109                         $author["author-name"] = $contact["name"];
 110
 111                 $author["author-avatar"] = $contact["thumb"];
 112
 113                 $author["owner-link"] = $contact["url"];
 114                 $author["owner-name"] = $contact["name"];
 115                 $author["owner-avatar"] = $contact["thumb"];
 116         }
 117
 118         $header = array();
 119         $header["uid"] = $importer["uid"];
 120         $header["network"] = NETWORK_FEED;
 121         $header["type"] = "remote";
 122         $header["wall"] = 0;
 123         $header["origin"] = 0;
 124         $header["gravity"] = GRAVITY_PARENT;
 125         $header["private"] = 2;
 126         $header["verb"] = ACTIVITY_POST;
 127         $header["object-type"] = ACTIVITY_OBJ_NOTE;
 128
 129         $header["contact-id"] = $contact["id"];
 130
 131         if(!strlen($contact["notify"])) {
 132                 // one way feed - no remote comment ability
 133                 $header["last-child"] = 0;
 134         }
 135
 136         if (!is_object($entries))
 137                 return;
 138
 139         $items = array();
 140
 141         $entrylist = array();
 142
 143         foreach ($entries AS $entry)
 144                 $entrylist[] = $entry;
 145
 146         foreach (array_reverse($entrylist) AS $entry) {
 147                 $item = array_merge($header, $author);
 148
 149                 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
 150
 151                 if ($item["title"] == "")
 152                         $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
 153
 154                 if ($item["title"] == "")
 155                         $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
 156
 157                 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
 158                 if (!is_object($alternate))
 159                         $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
 160
 161                 if (is_object($alternate))
 162                         foreach($alternate AS $attributes)
 163                                 if ($attributes->name == "href")
 164                                         $item["plink"] = $attributes->textContent;
 165
 166                 if ($item["plink"] == "")
 167                         $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
 168
 169                 if ($item["plink"] == "")
 170                         $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
 171
 172                 $item["plink"] = original_url($item["plink"]);
 173
 174                 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 175
 176                 if ($item["uri"] == "")
 177                         $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
 178
 179                 if ($item["uri"] == "")
 180                         $item["uri"] = $item["plink"];
 181
 182                 $item["parent-uri"] = $item["uri"];
 183
 184                 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
 185
 186                 if ($published == "")
 187                         $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
 188
 189                 if ($published == "")
 190                         $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
 191
 192                 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
 193
 194                 if ($updated == "")
 195                         $updated = $published;
 196
 197                 if ($published != "")
 198                         $item["created"] = $published;
 199
 200                 if ($updated != "")
 201                         $item["edited"] = $updated;
 202
 203                 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
 204
 205                 if ($creator == "")
 206                         $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
 207
 208                 if ($creator == "")
 209                         $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 210
 211                 if ($creator != "")
 212                         $item["author-name"] = $creator;
 213
 214                 if ($pubDate != "")
 215                         $item["edited"] = $item["created"] = $pubDate;
 216
 217                 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
 218
 219                 if ($creator != "")
 220                         $item["author-name"] = $creator;
 221
 222                 if (!$simulate) {
 223                         $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
 224                                 intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
 225                         if ($r) {
 226                                 logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
 227                                 continue;
 228                         }
 229                 }
 230
 231                 /// @TODO ?
 232                 // <category>Ausland</category>
 233                 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 234
 235                 $attachments = array();
 236
 237                 $enclosures = $xpath->query("enclosure", $entry);
 238                 foreach ($enclosures AS $enclosure) {
 239                         $href = "";
 240                         $length = "";
 241                         $type = "";
 242                         $title = "";
 243
 244                         foreach($enclosure->attributes AS $attributes) {
 245                                 if ($attributes->name == "url")
 246                                         $href = $attributes->textContent;
 247                                 elseif ($attributes->name == "length")
 248                                         $length = $attributes->textContent;
 249                                 elseif ($attributes->name == "type")
 250                                         $type = $attributes->textContent;
 251                         }
 252                         if(strlen($item["attach"]))
 253                                 $item["attach"] .= ',';
 254
 255                         $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
 256
 257                         $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
 258                 }
 259
 260                 if ($contact["fetch_further_information"]) {
 261                         $preview = "";
 262
 263                         // Handle enclosures and treat them as preview picture
 264                         foreach ($attachments AS $attachment)
 265                                 if ($attachment["type"] == "image/jpeg")
 266                                         $preview = $attachment["link"];
 267
 268                         $item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 269                         $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 270                         $item["title"] = "";
 271                         $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
 272                         unset($item["attach"]);
 273                 } else {
 274                         $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
 275
 276                         if ($body == "")
 277                                 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
 278
 279                         if ($body == "")
 280                                 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
 281
 282                         if ($body == "")
 283                                 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
 284
 285                         // remove the content of the title if it is identically to the body
 286                         // This helps with auto generated titles e.g. from tumblr
 287                         if (title_is_body($item["title"], $body))
 288                                 $item["title"] = "";
 289
 290                         $item["body"] = html2bbcode($body);
 291                 }
 292
 293                 if (!$simulate) {
 294                         logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 295
 296                         $notify = item_is_remote_self($contact, $item);
 297                         $id = item_store($item, false, $notify);
 298
 299                         logger("Feed for contact ".$contact["url"]." stored under id ".$id);
 300                 } else
 301                         $items[] = $item;
 302
 303                 if ($simulate)
 304                         break;
 305         }
 306
 307         if ($simulate)
 308                 return array("header" => $author, "items" => $items);
 309 }
 310 ?>