From 77e1ec6df2076183cf3643170a474278f368f968 Mon Sep 17 00:00:00 2001
From: Michael <heluecht@pirati.ca>
Date: Sun, 2 May 2021 17:33:32 +0000
Subject: [PATCH 1/2] Improved link detection

---
 src/Content/PageInfo.php |  9 +++++++--
 src/Model/Item.php       | 10 +++++++++-
 src/Model/Post/Media.php | 24 ++++++++++++++++++++----
 3 files changed, 36 insertions(+), 7 deletions(-)

diff --git a/src/Content/PageInfo.php b/src/Content/PageInfo.php
index 7226ff2a4..21f32cd3c 100644
--- a/src/Content/PageInfo.php
+++ b/src/Content/PageInfo.php
@@ -253,10 +253,15 @@ class PageInfo
 		// Fix for Mastodon where the mentions are in a different format
 		$body = preg_replace("~\[url=($URLSearchString)]([#!@])(.*?)\[/url]~is", '$2[url=$1]$3[/url]', $body);
 
-		preg_match("~(?<![!#@])\[url]($URLSearchString)\[/url]$~is", $body, $matches);
+		// Remove all hashtags and mentions
+		$body = preg_replace("/([#@!])\[url\=(.*?)\](.*?)\[\/url\]/ism", '', $body);
+
+		// Search for pure links
+		preg_match("/\[url\](.*?)\[\/url\]/ism", $body, $matches);
 
 		if (!$matches) {
-			preg_match("~(?<![!#@])\[url=($URLSearchString)].*\[/url]$~is", $body, $matches);
+			// Search for links with descriptions
+			preg_match("/\[url\=(.*?)\].*?\[\/url\]/ism", $body, $matches);
 		}
 
 		if (!$matches && $searchNakedUrls) {
diff --git a/src/Model/Item.php b/src/Model/Item.php
index 13da8ec75..2732330e0 100644
--- a/src/Model/Item.php
+++ b/src/Model/Item.php
@@ -971,6 +971,8 @@ class Item
 		$item['raw-body'] = Post\Media::insertFromBody($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']);
 
@@ -2646,7 +2648,7 @@ class Item
 		}
 
 		$body = $item['body'] ?? '';
-		$item['body'] = preg_replace("/\s*\[attachment .*?\].*?\[\/attachment\]\s*/ism", '', $item['body']);
+		$item['body'] = preg_replace("/\s*\[attachment .*?\].*?\[\/attachment\]\s*/ism", "\n", $item['body']);
 		self::putInCache($item);
 		$item['body'] = $body;
 		$s = $item["rendered-html"];
@@ -2722,6 +2724,12 @@ class Item
 	 */
 	public static function containsLink(string $body, string $url)
 	{
+		// Make sure that for example site parameters aren't used when testing if the link is contained in the body
+		$urlparts = parse_url($url);
+		unset($urlparts['query']);
+		unset($urlparts['fragment']);
+		$url = Network::unparseURL($urlparts);
+
 		if (strpos($body, $url)) {
 			return true;
 		}
diff --git a/src/Model/Post/Media.php b/src/Model/Post/Media.php
index fe362f735..63e349795 100644
--- a/src/Model/Post/Media.php
+++ b/src/Model/Post/Media.php
@@ -286,6 +286,8 @@ class Media
 		// Simplify image codes
 		$body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/ism", '[img]$3[/img]', $body);
 
+		$unshared_body = preg_replace("/\s*\[share .*?\].*?\[\/share\]\s*/ism", '', $body);
+
 		$attachments = [];
 		if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]\s*\[/url\]#ism", $body, $pictures, PREG_SET_ORDER)) {
 			foreach ($pictures as $picture) {
@@ -346,7 +348,10 @@ class Media
 		}
 
 		foreach ($attachments as $attachment) {
-			self::insert($attachment);
+			// Only store attachments that are part of the unshared body
+			if (strpos($unshared_body, $attachment['url']) !== false) {
+				self::insert($attachment);
+			}
 		}
 
 		return trim($body);
@@ -360,6 +365,9 @@ class Media
 	 */
 	public static function insertFromAttachmentData(int $uriid, string $body)
 	{
+		// Don't look at the shared content
+		$body = preg_replace("/\s*\[share .*?\].*?\[\/share\]\s*/ism", '', $body);
+
 		$data = BBCode::getAttachmentData($body);
 		if (empty($data))  {
 			return;
@@ -548,10 +556,18 @@ class Media
 			}
 
 			if ($media['type'] == self::IMAGE) {
-				if (!empty($media['description'])) {
-					$body .= "\n[img=" . $media['url'] . ']' . $media['description'] .'[/img]';
+				if (!empty($media['preview'])) {
+					if (!empty($media['description'])) {
+						$body .= "\n[url=" . $media['url'] . "][img=" . $media['preview'] . ']' . $media['description'] .'[/img][/url]';
+					} else {
+						$body .= "\n[url=" . $media['url'] . "][img]" . $media['preview'] .'[/img][/url]';
+					}
 				} else {
-					$body .= "\n[img]" . $media['url'] .'[/img]';
+					if (!empty($media['description'])) {
+						$body .= "\n[img=" . $media['url'] . ']' . $media['description'] .'[/img]';
+					} else {
+						$body .= "\n[img]" . $media['url'] .'[/img]';
+					}
 				}
 			} elseif ($media['type'] == self::AUDIO) {
 				$body .= "\n[audio]" . $media['url'] . "[/audio]\n";

From 0616f5230caa9e5c8b7c1fcd9552dcae1c5882ab Mon Sep 17 00:00:00 2001
From: Michael <heluecht@pirati.ca>
Date: Sun, 2 May 2021 17:53:11 +0000
Subject: [PATCH 2/2] Only detect http links

---
 src/Content/PageInfo.php | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/Content/PageInfo.php b/src/Content/PageInfo.php
index 21f32cd3c..d3b38336b 100644
--- a/src/Content/PageInfo.php
+++ b/src/Content/PageInfo.php
@@ -257,11 +257,11 @@ class PageInfo
 		$body = preg_replace("/([#@!])\[url\=(.*?)\](.*?)\[\/url\]/ism", '', $body);
 
 		// Search for pure links
-		preg_match("/\[url\](.*?)\[\/url\]/ism", $body, $matches);
+		preg_match("/\[url\](https?:.*?)\[\/url\]/ism", $body, $matches);
 
 		if (!$matches) {
 			// Search for links with descriptions
-			preg_match("/\[url\=(.*?)\].*?\[\/url\]/ism", $body, $matches);
+			preg_match("/\[url\=(https?:.*?)\].*?\[\/url\]/ism", $body, $matches);
 		}
 
 		if (!$matches && $searchNakedUrls) {