2 require_once("include/html2bbcode.php");
3 require_once("include/items.php");
6 * @brief Read a RSS/RDF/Atom feed and create an item entry for it
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 $simulate If enabled, no data is imported
14 * @return array In simulation mode it returns the header and the first item
16 function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
21 logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
23 logger("Test Atom/RSS feed", LOGGER_DEBUG);
26 logger('XML is empty.', LOGGER_DEBUG);
30 $doc = new DOMDocument();
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 $xpath->registerNamespace('poco', NAMESPACE_POCO);
44 if ($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"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
48 if ($author["author-name"] == "") {
49 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
51 $entries = $xpath->query('/rdf:RDF/rss:item');
55 if ($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"] = $attributes->textContent;
65 if ($author["author-link"] == "") {
66 $author["author-link"] = $author["author-id"];
68 if ($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"] = $attributes->textContent;
79 if ($author["author-link"] == "") {
80 $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
82 $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
84 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
86 if ($author["author-name"] == "") {
87 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
89 if ($author["author-name"] == "") {
90 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
92 $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
94 $author["author-name"] = $value;
97 $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
99 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
101 $author["author-nick"] = $value;
103 $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
105 $author["author-location"] = $value;
107 $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
109 $author["author-about"] = $value;
111 $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
112 if (is_object($avatar)) {
113 foreach ($avatar AS $attributes) {
114 if ($attributes->name == "href") {
115 $author["author-avatar"] = $attributes->textContent;
121 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
123 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
125 $entries = $xpath->query('/atom:feed/atom:entry');
129 if ($xpath->query('/rss/channel')->length > 0) {
130 $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
132 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
133 $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
135 if ($author["author-name"] == "") {
136 $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
138 if ($author["author-name"] == "") {
139 $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
141 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
143 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
145 $entries = $xpath->query('/rss/channel/item');
149 $author["author-link"] = $contact["url"];
151 if ($author["author-name"] == "") {
152 $author["author-name"] = $contact["name"];
154 $author["author-avatar"] = $contact["thumb"];
156 $author["owner-link"] = $contact["url"];
157 $author["owner-name"] = $contact["name"];
158 $author["owner-avatar"] = $contact["thumb"];
162 $header["uid"] = $importer["uid"];
163 $header["network"] = NETWORK_FEED;
164 $header["type"] = "remote";
166 $header["origin"] = 0;
167 $header["gravity"] = GRAVITY_PARENT;
168 $header["private"] = 2;
169 $header["verb"] = ACTIVITY_POST;
170 $header["object-type"] = ACTIVITY_OBJ_NOTE;
172 $header["contact-id"] = $contact["id"];
174 if (!strlen($contact["notify"])) {
175 // one way feed - no remote comment ability
176 $header["last-child"] = 0;
179 if (!is_object($entries)) {
180 logger("There are no entries in this feed.", LOGGER_DEBUG);
186 $entrylist = array();
188 foreach ($entries AS $entry) {
189 $entrylist[] = $entry;
191 foreach (array_reverse($entrylist) AS $entry) {
192 $item = array_merge($header, $author);
194 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
195 if (!is_object($alternate)) {
196 $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
198 if (is_object($alternate)) {
199 foreach ($alternate AS $attributes) {
200 if ($attributes->name == "href") {
201 $item["plink"] = $attributes->textContent;
205 if ($item["plink"] == "") {
206 $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
208 if ($item["plink"] == "") {
209 $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
212 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
214 if ($item["uri"] == "") {
215 $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
217 if ($item["uri"] == "") {
218 $item["uri"] = $item["plink"];
221 $orig_plink = $item["plink"];
223 $item["plink"] = original_url($item["plink"]);
225 $item["parent-uri"] = $item["uri"];
228 $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
229 intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
230 if (dbm::is_result($r)) {
231 logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
236 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
238 if ($item["title"] == "") {
239 $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
241 if ($item["title"] == "") {
242 $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
244 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
246 if ($published == "") {
247 $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
249 if ($published == "") {
250 $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
252 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
254 if ($updated == "") {
255 $updated = $published;
257 if ($published != "") {
258 $item["created"] = $published;
260 if ($updated != "") {
261 $item["edited"] = $updated;
263 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
265 if ($creator == "") {
266 $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
268 if ($creator == "") {
269 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
271 if ($creator != "") {
272 $item["author-name"] = $creator;
274 if ($pubDate != "") {
275 $item["edited"] = $item["created"] = $pubDate;
277 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
279 if ($creator != "") {
280 $item["author-name"] = $creator;
284 // <category>Ausland</category>
285 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
287 $attachments = array();
289 $enclosures = $xpath->query("enclosure", $entry);
290 foreach ($enclosures AS $enclosure) {
296 foreach ($enclosure->attributes AS $attributes) {
297 if ($attributes->name == "url") {
298 $href = $attributes->textContent;
299 } elseif ($attributes->name == "length") {
300 $length = $attributes->textContent;
301 } elseif ($attributes->name == "type") {
302 $type = $attributes->textContent;
305 if (strlen($item["attach"])) {
306 $item["attach"] .= ',';
309 $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
311 $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
314 if ($contact["fetch_further_information"]) {
317 // Handle enclosures and treat them as preview picture
318 foreach ($attachments AS $attachment) {
319 if ($attachment["type"] == "image/jpeg") {
320 $preview = $attachment["link"];
324 $item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
325 $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
327 $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
328 unset($item["attach"]);
330 $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
333 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
336 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
339 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
342 // remove the content of the title if it is identically to the body
343 // This helps with auto generated titles e.g. from tumblr
344 if (title_is_body($item["title"], $body)) {
347 $item["body"] = html2bbcode($body);
349 if (($item["body"] == '') && ($item["title"] != '')) {
350 $item["body"] = $item["title"];
354 if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
355 $item["body"] .= "[hr][url]".$item['plink']."[/url]";
360 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
362 $notify = item_is_remote_self($contact, $item);
364 // Distributed items should have a well formatted URI.
365 // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
367 $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
369 unset($item['parent-uri']);
372 $id = item_store($item, false, $notify);
374 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
384 return array("header" => $author, "items" => $items);