]> git.mxchange.org Git - friendica.git/blobdiff - include/feed.php
API: use regex instead of SimpleXMLElement
[friendica.git] / include / feed.php
index d7bbb24cbfcd5a6bcb35222defcfe5713eb37276..5adb2294db0248e92b4b9ac3e81a9288c3866f11 100644 (file)
@@ -2,30 +2,47 @@
 require_once("include/html2bbcode.php");
 require_once("include/items.php");
 
-function feed_import($xml,$importer,&$contact, &$hub) {
+/**
+ * @brief Read a RSS/RDF/Atom feed and create an item entry for it
+ *
+ * @param string $xml The feed data
+ * @param array $importer The user record of the importer
+ * @param array $contact The contact record of the feed
+ * @param string $hub Unused dummy value for compatibility reasons
+ * @param bool $simulate If enabled, no data is imported
+ *
+ * @return array In simulation mode it returns the header and the first item
+ */
+function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
 
        $a = get_app();
 
-       logger("Import Atom/RSS feed", LOGGER_DEBUG);
+       if (!$simulate)
+               logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
+       else
+               logger("Test Atom/RSS feed", LOGGER_DEBUG);
 
-       if ($xml == "")
+       if ($xml == "") {
+               logger('XML is empty.', LOGGER_DEBUG);
                return;
+       }
 
        $doc = new DOMDocument();
        @$doc->loadXML($xml);
        $xpath = new DomXPath($doc);
-       $xpath->registerNamespace('atom', "http://www.w3.org/2005/Atom");
+       $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
        $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
        $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
        $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
        $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
        $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
+       $xpath->registerNamespace('poco', NAMESPACE_POCO);
 
        $author = array();
 
        // Is it RDF?
        if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
-               //$author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
+               $author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
                $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
 
                if ($author["author-name"] == "")
@@ -35,20 +52,28 @@ function feed_import($xml,$importer,&$contact, &$hub) {
        }
 
        // Is it Atom?
-       if ($xpath->query('/atom:feed/atom:entry')->length > 0) {
-               //$self = $xpath->query("/atom:feed/atom:link[@rel='self']")->item(0)->attributes;
-               //if (is_object($self))
-               //      foreach($self AS $attributes)
-               //              if ($attributes->name == "href")
-               //                      $author["author-link"] = $attributes->textContent;
-
-               //if ($author["author-link"] == "") {
-               //      $alternate = $xpath->query("/atom:feed/atom:link[@rel='alternate']")->item(0)->attributes;
-               //      if (is_object($alternate))
-               //              foreach($alternate AS $attributes)
-               //                      if ($attributes->name == "href")
-               //                              $author["author-link"] = $attributes->textContent;
-               //}
+       if ($xpath->query('/atom:feed')->length > 0) {
+               $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
+               if (is_object($alternate))
+                       foreach($alternate AS $attributes)
+                               if ($attributes->name == "href")
+                                       $author["author-link"] = $attributes->textContent;
+
+               if ($author["author-link"] == "")
+                       $author["author-link"] = $author["author-id"];
+
+               if ($author["author-link"] == "") {
+                       $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
+                       if (is_object($self))
+                               foreach($self AS $attributes)
+                                       if ($attributes->name == "href")
+                                               $author["author-link"] = $attributes->textContent;
+               }
+
+               if ($author["author-link"] == "")
+                       $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
+
+               $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
 
                $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
 
@@ -58,7 +83,26 @@ function feed_import($xml,$importer,&$contact, &$hub) {
                if ($author["author-name"] == "")
                        $author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
 
-               //$author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
+               $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
+               if ($value != "")
+                       $author["author-name"] = $value;
+
+               if ($simulate) {
+                       $author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
+
+                       $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
+                       if ($value != "")
+                               $author["author-nick"] = $value;
+
+                       $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
+                       if ($value != "")
+                               $author["author-location"] = $value;
+
+                       $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
+                       if ($value != "")
+                               $author["author-about"] = $value;
+
+               }
 
                $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
 
@@ -69,9 +113,10 @@ function feed_import($xml,$importer,&$contact, &$hub) {
 
        // Is it RSS?
        if ($xpath->query('/rss/channel')->length > 0) {
-               //$author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
+               $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
+
                $author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
-               //$author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
+               $author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
 
                if ($author["author-name"] == "")
                        $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
@@ -86,18 +131,18 @@ function feed_import($xml,$importer,&$contact, &$hub) {
                $entries = $xpath->query('/rss/channel/item');
        }
 
-       //if ($author["author-link"] == "")
+       if (!$simulate) {
                $author["author-link"] = $contact["url"];
 
-       if ($author["author-name"] == "")
-               $author["author-name"] = $contact["name"];
+               if ($author["author-name"] == "")
+                       $author["author-name"] = $contact["name"];
 
-       //if ($author["author-avatar"] == "")
                $author["author-avatar"] = $contact["thumb"];
 
-       $author["owner-link"] = $contact["url"];
-       $author["owner-name"] = $contact["name"];
-       $author["owner-avatar"] = $contact["thumb"];
+               $author["owner-link"] = $contact["url"];
+               $author["owner-name"] = $contact["name"];
+               $author["owner-avatar"] = $contact["thumb"];
+       }
 
        $header = array();
        $header["uid"] = $importer["uid"];
@@ -117,8 +162,12 @@ function feed_import($xml,$importer,&$contact, &$hub) {
                $header["last-child"] = 0;
        }
 
-       if (!is_object($entries))
+       if (!is_object($entries)) {
+               logger("There are no entries in this feed.", LOGGER_DEBUG);
                return;
+       }
+
+       $items = array();
 
        $entrylist = array();
 
@@ -201,16 +250,16 @@ function feed_import($xml,$importer,&$contact, &$hub) {
                if ($creator != "")
                        $item["author-name"] = $creator;
 
-               //$item["object"] = $xml;
-
-               $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
-                       intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
-               if ($r) {
-                       logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
-                       continue;
+               if (!$simulate) {
+                       $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
+                               intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
+                       if ($r) {
+                               logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
+                               continue;
+                       }
                }
 
-               // To-Do?
+               /// @TODO ?
                // <category>Ausland</category>
                // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 
@@ -272,14 +321,21 @@ function feed_import($xml,$importer,&$contact, &$hub) {
                        $item["body"] = html2bbcode($body);
                }
 
-               logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
+               if (!$simulate) {
+                       logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
 
-               $notify = item_is_remote_self($contact, $item);
-               $id = item_store($item, false, $notify);
+                       $notify = item_is_remote_self($contact, $item);
+                       $id = item_store($item, false, $notify);
 
-               //print_r($item);
+                       logger("Feed for contact ".$contact["url"]." stored under id ".$id);
+               } else
+                       $items[] = $item;
 
-               logger("Feed for contact ".$contact["url"]." stored under id ".$id);
+               if ($simulate)
+                       break;
        }
+
+       if ($simulate)
+               return array("header" => $author, "items" => $items);
 }
 ?>