No link guessing for DFRN / Don't show redundant data

src/Model/Item.php

@@ -172,7 +172,7 @@ class Item
 
 		Logger::info('Updating per single row method', ['fields' => $fields, 'condition' => $condition]);
 
-		$items = Post::select(['id', 'origin', 'uri-id', 'uid'], $condition);
+		$items = Post::select(['id', 'origin', 'uri-id', 'uid', 'author-network'], $condition);
 
 		$notify_items = [];
 
@@ -180,6 +180,10 @@ class Item
 			if (!empty($fields['body'])) {
 				Post\Media::insertFromAttachmentData($item['uri-id'], $fields['body']);
 
+				if ($item['author-network'] != Protocol::DFRN) {
+					Post\Media::insertFromRelevantUrl($item['uri-id'], $fields['body']);
+				}
+
 				$content_fields = ['raw-body' => trim($fields['raw-body'] ?? $fields['body'])];
 
 				// Remove all media attachments from the body and store them in the post-media table
@@ -967,12 +971,16 @@ class Item
 			unset($item['attachments']);
 		}
 
+		Post\Media::insertFromAttachmentData($item['uri-id'], $item['body']);
+
+		if (!DBA::exists('contact', ['id' => $item['author-id'], 'network' => Protocol::DFRN])) {
+			Post\Media::insertFromRelevantUrl($item['uri-id'], $item['body']);
+		}
+
 		// Remove all media attachments from the body and store them in the post-media table
 		$item['raw-body'] = Post\Media::insertFromBody($item['uri-id'], $item['raw-body']);
 		$item['raw-body'] = self::setHashtags($item['raw-body']);
 
-		Post\Media::insertFromAttachmentData($item['uri-id'], $item['body']);
-
 		// Check for hashtags in the body and repair or add hashtag links
 		$item['body'] = self::setHashtags($item['body']);
 
@@ -2891,28 +2899,48 @@ class Item
 					$data['preview'] = $attachment['preview'] ?? '';
 				}
 			}
+
+			if (!empty($data['description']) && !empty($content)) {
+				similar_text($data['description'], $content, $percent);
+			} else {
+				$percent = 0;
+			}
+
+			if (!empty($data['description']) && (($data['title'] == $data['description']) || ($percent > 95) || (strpos($content, $data['description']) !== false))) {
+				$data['description'] = '';
+			}
 		} elseif (preg_match("/.*(\[attachment.*?\].*?\[\/attachment\]).*/ism", $body, $match)) {
 			$data = BBCode::getAttachmentData($match[1]);
 		}
 		DI::profiler()->saveTimestamp($stamp1, 'rendering');
 
 		if (isset($data['url']) && !in_array($data['url'], $ignore_links)) {
-			$parts = parse_url($data['url']);
+			if (!empty($data['description']) || !empty($data['image'] || !empty($data['preview']))) {
-			if (!empty($parts['scheme']) && !empty($parts['host'])) {
+				$parts = parse_url($data['url']);
-				if (empty($data['provider_name'])) {
+				if (!empty($parts['scheme']) && !empty($parts['host'])) {
-					$data['provider_name'] = $parts['host'];
+					if (empty($data['provider_name'])) {
-				}
+						$data['provider_name'] = $parts['host'];
-				if (empty($data['provider_url']) || empty(parse_url($data['provider_url'], PHP_URL_SCHEME))) {
+					}
-					$data['provider_url'] = $parts['scheme'] . '://' . $parts['host'];
+					if (empty($data['provider_url']) || empty(parse_url($data['provider_url'], PHP_URL_SCHEME))) {
+						$data['provider_url'] = $parts['scheme'] . '://' . $parts['host'];
 
-					if (!empty($parts['port'])) {
+						if (!empty($parts['port'])) {
-						$data['provider_url'] .= ':' . $parts['port'];
+							$data['provider_url'] .= ':' . $parts['port'];
+						}
 					}
 				}
+
+				// @todo Use a template
+				$rendered = BBCode::convertAttachment('', BBCode::INTERNAL, false, $data);
+			} elseif (!self::containsLink($content, $data['url'])) {
+				$rendered = Renderer::replaceMacros(Renderer::getMarkupTemplate('content/link.tpl'), [
+					'$url'  => $data['url'],
+					'$title' => $data['title'],
+				]);
+			} else {
+				return $content;
 			}
 
-			// @todo Use a template
-			$rendered = BBCode::convertAttachment('', BBCode::INTERNAL, false, $data);
 			if ($shared) {
 				return str_replace(BBCode::BOTTOM_ANCHOR, BBCode::BOTTOM_ANCHOR . $rendered, $content);
 			} else {

src/Model/Post/Media.php

@@ -341,12 +341,6 @@ class Media
 			}
 		}
 
-		$url = PageInfo::getRelevantUrlFromBody($body);
-		if (!empty($url)) {
-			Logger::debug('Got page url', ['url' => $url]);
-			$attachments[$url] = ['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url];
-		}
-
 		foreach ($attachments as $attachment) {
 			// Only store attachments that are part of the unshared body
 			if (strpos($unshared_body, $attachment['url']) !== false) {
@@ -357,6 +351,37 @@ class Media
 		return trim($body);
 	}
 
+	/**
+	 * Add media links from a relevant url in the body
+	 *
+	 * @param integer $uriid
+	 * @param string $body
+	 */
+	public static function insertFromRelevantUrl(int $uriid, string $body)
+	{
+		// Don't look at the shared content
+		$body = preg_replace("/\s*\[share .*?\].*?\[\/share\]\s*/ism", '', $body);
+
+		// Remove all hashtags and mentions
+		$body = preg_replace("/([#@!])\[url\=(.*?)\](.*?)\[\/url\]/ism", '', $body);
+
+		// Search for pure links
+		if (preg_match_all("/\[url\](https?:.*?)\[\/url\]/ism", $body, $matches)) {
+			foreach ($matches[1] as $url) {
+				Logger::info('Got page url (link without description)', ['uri-id' => $uriid, 'url' => $url]);
+				self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url]);
+			}
+		}
+
+		// Search for links with descriptions
+		if (preg_match_all("/\[url\=(https?:.*?)\].*?\[\/url\]/ism", $body, $matches)) {
+			foreach ($matches[1] as $url) {
+				Logger::info('Got page url (link with description)', ['uri-id' => $uriid, 'url' => $url]);
+				self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url]);
+			}
+		}
+	}
+
 	/**
 	 * Add media links from the attachment field
 	 *

view/templates/content/link.tpl

@@ -0,0 +1 @@
+<p><a href="{{$url}}">{{if $title}}{{$title}}{{else}}{{$url}}{{/if}}</a></p>