3 * @file src/Protocol/Feed.php
4 * @brief Imports RSS/RDF/Atom feeds
7 namespace Friendica\Protocol;
9 use Friendica\Database\DBM;
10 use Friendica\Core\System;
13 require_once("include/html2bbcode.php");
14 require_once("include/items.php");
17 * @brief This class contain functions to import feeds
22 * @brief Read a RSS/RDF/Atom feed and create an item entry for it
24 * @param string $xml The feed data
25 * @param array $importer The user record of the importer
26 * @param array $contact The contact record of the feed
27 * @param string $hub Unused dummy value for compatibility reasons
28 * @param bool $simulate If enabled, no data is imported
30 * @return array In simulation mode it returns the header and the first item
32 function import($xml, $importer, &$contact, &$hub, $simulate = false) {
37 logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
39 logger("Test Atom/RSS feed", LOGGER_DEBUG);
42 logger('XML is empty.', LOGGER_DEBUG);
46 if (!empty($contact['poll'])) {
47 $basepath = $contact['poll'];
48 } elseif (!empty($contact['url'])) {
49 $basepath = $contact['url'];
54 $doc = new DOMDocument();
55 @$doc->loadXML(trim($xml));
56 $xpath = new DomXPath($doc);
57 $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
58 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
59 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
60 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
61 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
62 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
63 $xpath->registerNamespace('poco', NAMESPACE_POCO);
68 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
69 $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
70 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
72 if ($author["author-name"] == "") {
73 $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
75 $entries = $xpath->query('/rdf:RDF/rss:item');
79 if ($xpath->query('/atom:feed')->length > 0) {
80 $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
81 if (is_object($alternate)) {
82 foreach ($alternate AS $attributes) {
83 if ($attributes->name == "href") {
84 $author["author-link"] = $attributes->textContent;
89 if ($author["author-link"] == "") {
90 $author["author-link"] = $author["author-id"];
92 if ($author["author-link"] == "") {
93 $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
94 if (is_object($self)) {
95 foreach ($self AS $attributes) {
96 if ($attributes->name == "href") {
97 $author["author-link"] = $attributes->textContent;
103 if ($author["author-link"] == "") {
104 $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
106 $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
108 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
110 if ($author["author-name"] == "") {
111 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
113 if ($author["author-name"] == "") {
114 $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
116 $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
118 $author["author-name"] = $value;
121 $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
123 $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
125 $author["author-nick"] = $value;
127 $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
129 $author["author-location"] = $value;
131 $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
133 $author["author-about"] = $value;
135 $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
136 if (is_object($avatar)) {
137 foreach ($avatar AS $attributes) {
138 if ($attributes->name == "href") {
139 $author["author-avatar"] = $attributes->textContent;
145 $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
147 $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
149 $entries = $xpath->query('/atom:feed/atom:entry');
153 if ($xpath->query('/rss/channel')->length > 0) {
154 $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
156 $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
157 $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
159 if ($author["author-name"] == "") {
160 $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
162 if ($author["author-name"] == "") {
163 $author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
165 $author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
167 $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
169 $entries = $xpath->query('/rss/channel/item');
173 $author["author-link"] = $contact["url"];
175 if ($author["author-name"] == "") {
176 $author["author-name"] = $contact["name"];
178 $author["author-avatar"] = $contact["thumb"];
180 $author["owner-link"] = $contact["url"];
181 $author["owner-name"] = $contact["name"];
182 $author["owner-avatar"] = $contact["thumb"];
186 $header["uid"] = $importer["uid"];
187 $header["network"] = NETWORK_FEED;
188 $header["type"] = "remote";
190 $header["origin"] = 0;
191 $header["gravity"] = GRAVITY_PARENT;
192 $header["private"] = 2;
193 $header["verb"] = ACTIVITY_POST;
194 $header["object-type"] = ACTIVITY_OBJ_NOTE;
196 $header["contact-id"] = $contact["id"];
198 if (!strlen($contact["notify"])) {
199 // one way feed - no remote comment ability
200 $header["last-child"] = 0;
203 if (!is_object($entries)) {
204 logger("There are no entries in this feed.", LOGGER_DEBUG);
210 $entrylist = array();
212 foreach ($entries AS $entry) {
213 $entrylist[] = $entry;
215 foreach (array_reverse($entrylist) AS $entry) {
216 $item = array_merge($header, $author);
218 $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
219 if (!is_object($alternate)) {
220 $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
222 if (is_object($alternate)) {
223 foreach ($alternate AS $attributes) {
224 if ($attributes->name == "href") {
225 $item["plink"] = $attributes->textContent;
229 if ($item["plink"] == "") {
230 $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
232 if ($item["plink"] == "") {
233 $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
236 $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
238 if ($item["uri"] == "") {
239 $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
241 if ($item["uri"] == "") {
242 $item["uri"] = $item["plink"];
245 $orig_plink = $item["plink"];
247 $item["plink"] = original_url($item["plink"]);
249 $item["parent-uri"] = $item["uri"];
252 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
253 $importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
254 $previous = dba::select('item', ['id'], $condition, ['limit' => 1]);
255 if (DBM::is_result($previous)) {
256 logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG);
261 $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
263 if ($item["title"] == "") {
264 $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
266 if ($item["title"] == "") {
267 $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
269 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
271 if ($published == "") {
272 $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
274 if ($published == "") {
275 $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
277 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
279 if ($updated == "") {
280 $updated = $published;
282 if ($published != "") {
283 $item["created"] = $published;
285 if ($updated != "") {
286 $item["edited"] = $updated;
288 $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
290 if ($creator == "") {
291 $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
293 if ($creator == "") {
294 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
296 if ($creator != "") {
297 $item["author-name"] = $creator;
299 if ($pubDate != "") {
300 $item["edited"] = $item["created"] = $pubDate;
302 $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
304 if ($creator != "") {
305 $item["author-name"] = $creator;
309 // <category>Ausland</category>
310 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
312 $attachments = array();
314 $enclosures = $xpath->query("enclosure", $entry);
315 foreach ($enclosures AS $enclosure) {
321 foreach ($enclosure->attributes AS $attributes) {
322 if ($attributes->name == "url") {
323 $href = $attributes->textContent;
324 } elseif ($attributes->name == "length") {
325 $length = $attributes->textContent;
326 } elseif ($attributes->name == "type") {
327 $type = $attributes->textContent;
330 if (strlen($item["attach"])) {
331 $item["attach"] .= ',';
334 $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
336 $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
340 $categories = $xpath->query("category", $entry);
341 foreach ($categories AS $category) {
342 $hashtag = $category->nodeValue;
347 $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
351 $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
354 $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
357 $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
360 $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
363 // remove the content of the title if it is identically to the body
364 // This helps with auto generated titles e.g. from tumblr
365 if (title_is_body($item["title"], $body)) {
368 $item["body"] = html2bbcode($body, $basepath);
370 if (($item["body"] == '') && ($item["title"] != '')) {
371 $item["body"] = $item["title"];
375 if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
378 // Handle enclosures and treat them as preview picture
379 foreach ($attachments AS $attachment) {
380 if ($attachment["type"] == "image/jpeg") {
381 $preview = $attachment["link"];
385 // Remove a possible link to the item itself
386 $item["body"] = str_replace($item["plink"], '', $item["body"]);
387 $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
389 // Replace the content when the title is longer than the body
390 $replace = (strlen($item["title"]) > strlen($item["body"]));
392 // Replace it, when there is an image in the body
393 if (strstr($item["body"], '[/img]')) {
397 // Replace it, when there is a link in the body
398 if (strstr($item["body"], '[/url]')) {
403 $item["body"] = $item["title"];
405 // We always strip the title since it will be added in the page information
407 $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
408 $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
409 $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
410 unset($item["attach"]);
412 if ($contact["fetch_further_information"] == 3) {
414 $item["tag"] = $tags;
416 $item["tag"] = add_page_keywords($item["plink"], false, $preview, true, $contact["ffi_keyword_blacklist"]);
418 $item["body"] .= "\n".$item['tag'];
420 if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
421 $item["body"] .= "[hr][url]".$item['plink']."[/url]";
426 logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
428 $notify = item_is_remote_self($contact, $item);
430 // Distributed items should have a well formatted URI.
431 // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
433 $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
435 unset($item['parent-uri']);
438 $id = item_store($item, false, $notify);
440 logger("Feed for contact ".$contact["url"]." stored under id ".$id);
450 return array("header" => $author, "items" => $items);