Merge pull request #3868 from rabuzarus/20171104_-_use_best_link_url_for_events

[friendica.git] / include / feed.php

diff --git a/include/feed.php b/include/feed.php
index 317f6bee7d7259ba753cc188684255401b35a7b1..fb97c14ba126aa1875d1202df05a5caa8aad64e7 100644 (file)
--- a/include/feed.php
+++ b/include/feed.php
@@ -27,8 +27,16 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                return;
        }
 
+       if (!empty($contact['poll'])) {
+               $basepath = $contact['poll'];
+       } elseif (!empty($contact['url'])) {
+               $basepath = $contact['url'];
+       } else {
+               $basepath = '';
+       }
+
        $doc = new DOMDocument();
-       @$doc->loadXML($xml);
+       @$doc->loadXML(trim($xml));
        $xpath = new DomXPath($doc);
        $xpath->registerNamespace('atom', NAMESPACE_ATOM1);
        $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
@@ -55,7 +63,7 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
        if ($xpath->query('/atom:feed')->length > 0) {
                $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
                if (is_object($alternate)) {
-                       foreach($alternate AS $attributes) {
+                       foreach ($alternate AS $attributes) {
                                if ($attributes->name == "href") {
                                        $author["author-link"] = $attributes->textContent;
                                }
@@ -68,7 +76,7 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                if ($author["author-link"] == "") {
                        $self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
                        if (is_object($self)) {
-                               foreach($self AS $attributes) {
+                               foreach ($self AS $attributes) {
                                        if ($attributes->name == "href") {
                                                $author["author-link"] = $attributes->textContent;
                                        }
@@ -108,6 +116,14 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                        if ($value != "") {
                                $author["author-about"] = $value;
                        }
+                       $avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
+                       if (is_object($avatar)) {
+                               foreach ($avatar AS $attributes) {
+                                       if ($attributes->name == "href") {
+                                               $author["author-avatar"] = $attributes->textContent;
+                                       }
+                               }
+                       }
                }
 
                $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
@@ -188,7 +204,7 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                        $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
                }
                if (is_object($alternate)) {
-                       foreach($alternate AS $attributes) {
+                       foreach ($alternate AS $attributes) {
                                if ($attributes->name == "href") {
                                        $item["plink"] = $attributes->textContent;
                                }
@@ -200,7 +216,6 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                if ($item["plink"] == "") {
                        $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
                }
-               $item["plink"] = original_url($item["plink"]);
 
                $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
 
@@ -210,12 +225,17 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                if ($item["uri"] == "") {
                        $item["uri"] = $item["plink"];
                }
+
+               $orig_plink = $item["plink"];
+
+               $item["plink"] = original_url($item["plink"]);
+
                $item["parent-uri"] = $item["uri"];
 
                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) {
+                       if (dbm::is_result($r)) {
                                logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
                                continue;
                        }
@@ -267,6 +287,7 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                if ($creator != "") {
                        $item["author-name"] = $creator;
                }
+
                /// @TODO ?
                // <category>Ausland</category>
                // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
@@ -298,6 +319,30 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                        $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
                }
 
+               $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
+
+               if ($body == "") {
+                       $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
+               }
+               if ($body == "") {
+                       $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
+               }
+               if ($body == "") {
+                       $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
+               }
+
+               // remove the content of the title if it is identically to the body
+               // This helps with auto generated titles e.g. from tumblr
+               if (title_is_body($item["title"], $body)) {
+                       $item["title"] = "";
+               }
+               $item["body"] = html2bbcode($body, $basepath);
+
+               if (($item["body"] == '') && ($item["title"] != '')) {
+                       $item["body"] = $item["title"];
+                       $item["title"] = '';
+               }
+
                if ($contact["fetch_further_information"]) {
                        $preview = "";
 
@@ -308,29 +353,36 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                                }
                        }
 
-                       $item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
-                       $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
-                       $item["title"] = "";
-                       $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
-                       unset($item["attach"]);
-               } else {
-                       $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
+                       // Remove a possible link to the item itself
+                       $item["body"] = str_replace($item["plink"], '', $item["body"]);
+                       $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
+
+                       // Replace the content when the title is longer than the body
+                       $replace = (strlen($item["title"]) > strlen($item["body"]));
 
-                       if ($body == "") {
-                               $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
+                       // Replace it, when there is an image in the body
+                       if (strstr($item["body"], '[/img]')) {
+                               $replace = true;
                        }
-                       if ($body == "") {
-                               $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
+
+                       // Replace it, when there is a link in the body
+                       if (strstr($item["body"], '[/url]')) {
+                               $replace = true;
                        }
-                       if ($body == "") {
-                               $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
+
+                       if ($replace) {
+                               $item["body"] = $item["title"];
                        }
-                       // remove the content of the title if it is identically to the body
-                       // This helps with auto generated titles e.g. from tumblr
-                       if (title_is_body($item["title"], $body)) {
-                               $item["title"] = "";
+                       // We always strip the title since it will be added in the page information
+                       $item["title"] = "";
+                       $item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
+                       $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
+                       $item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
+                       unset($item["attach"]);
+               } else {
+                       if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
+                               $item["body"] .= "[hr][url]".$item['plink']."[/url]";
                        }
-                       $item["body"] = html2bbcode($body);
                }
 
                if (!$simulate) {
@@ -341,6 +393,7 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                        // Distributed items should have a well formatted URI.
                        // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
                        if ($notify) {
+                               $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
                                unset($item['uri']);
                                unset($item['parent-uri']);
                        }
@@ -360,4 +413,3 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
                return array("header" => $author, "items" => $items);
        }
 }
-?>