From 538e212a84873dfa290c594e66a8c038fbf5b382 Mon Sep 17 00:00:00 2001
From: Michael <heluecht@pirati.ca>
Date: Sun, 19 Apr 2020 16:33:06 +0000
Subject: [PATCH] Hashtag handling with Diaspora improved

---
 src/Content/Text/BBCode.php |  2 +-
 src/Model/Item.php          |  5 ++++-
 src/Model/Tag.php           | 34 ++++++++++++++++++++++++++++++++++
 src/Protocol/Diaspora.php   |  4 ++--
 4 files changed, 41 insertions(+), 4 deletions(-)

diff --git a/src/Content/Text/BBCode.php b/src/Content/Text/BBCode.php
index 334efb32a7..601bf63b25 100644
--- a/src/Content/Text/BBCode.php
+++ b/src/Content/Text/BBCode.php
@@ -2100,7 +2100,7 @@ class BBCode
 		$ret = [];
 
 		// Convert hashtag links to hashtags
-		$string = preg_replace('/#\[url\=([^\[\]]*)\](.*?)\[\/url\]/ism', '#$2', $string);
+		$string = preg_replace('/#\[url\=([^\[\]]*)\](.*?)\[\/url\]/ism', '#$2 ', $string);
 
 		// ignore anything in a code block
 		$string = preg_replace('/\[code.*?\].*?\[\/code\]/sm', '', $string);
diff --git a/src/Model/Item.php b/src/Model/Item.php
index 1f2faa2698..0e82189298 100644
--- a/src/Model/Item.php
+++ b/src/Model/Item.php
@@ -2610,7 +2610,10 @@ class Item
 
 		// This sorting is important when there are hashtags that are part of other hashtags
 		// Otherwise there could be problems with hashtags like #test and #test2
-		rsort($tags);
+		// Because of this we are sorting from the longest to the shortest tag.
+		usort($rawtags, function($a, $b) {
+			return strlen($b) <=> strlen($a);
+		});
 
 		$URLSearchString = "^\[\]";
 
diff --git a/src/Model/Tag.php b/src/Model/Tag.php
index bd1945c480..ac26009033 100644
--- a/src/Model/Tag.php
+++ b/src/Model/Tag.php
@@ -21,7 +21,9 @@
 
 namespace Friendica\Model;
 
+use Friendica\Content\Text\BBCode;
 use Friendica\Core\Logger;
+use Friendica\Core\System;
 use Friendica\Database\DBA;
 use Friendica\Util\Strings;
 
@@ -87,6 +89,7 @@ class Tag
 			} else {
 				// The contact wasn't found in the system (most likely some dead account)
 				// We ensure that we only store a single entry by overwriting the previous name
+				Logger::info('Update tag', ['url' => $url, 'name' => $name]);
 				DBA::update('tag', ['name' => substr($name, 0, 96)], ['url' => $url]);
 			}
 		}
@@ -148,15 +151,46 @@ class Tag
 			$tags =  self::TAG_CHARACTER[self::HASHTAG] . self::TAG_CHARACTER[self::MENTION] . self::TAG_CHARACTER[self::EXCLUSIVE_MENTION];
 		}
 
+		Logger::info('Check for tags', ['uri-id' => $uriid, 'hash' => $tags, 'callstack' => System::callstack()]);
+
 		if (!preg_match_all("/([" . $tags . "])\[url\=([^\[\]]*)\]([^\[\]]*)\[\/url\]/ism", $body, $result, PREG_SET_ORDER)) {
 			return;
 		}
 
+		Logger::info('Found tags', ['uri-id' => $uriid, 'hash' => $tags, 'result' => $result]);
+
 		foreach ($result as $tag) {
 			self::storeByHash($uriid, $tag[1], $tag[3], $tag[2]);
 		}
 	}
 
+	/**
+	 * Store raw tags (not encapsulated in links) from the body
+	 * This function is needed in the intermediate phase.
+	 * Later we can call item::setHashtags in advance to have all tags converted.
+	 * 
+	 * @param integer $uriid URI-Id
+	 * @param string  $body   Body of the post
+	 */
+	public static function storeRawTagsFromBody(int $uriid, string $body)
+	{
+		Logger::info('Check for tags', ['uri-id' => $uriid, 'callstack' => System::callstack()]);
+
+		$result = BBCode::getTags($body);
+		if (empty($result)) {
+			return;
+		}
+
+		Logger::info('Found tags', ['uri-id' => $uriid, 'result' => $result]);
+
+		foreach ($result as $tag) {
+			if (substr($tag, 0, 1) != self::TAG_CHARACTER[self::HASHTAG]) {
+				continue;
+			}
+			self::storeByHash($uriid, substr($tag, 0, 1), substr($tag, 1));
+		}
+	}
+
 	/**
 	 * Remove tag/mention
 	 *
diff --git a/src/Protocol/Diaspora.php b/src/Protocol/Diaspora.php
index d7e8f60a2e..513cd0bcc0 100644
--- a/src/Protocol/Diaspora.php
+++ b/src/Protocol/Diaspora.php
@@ -1938,7 +1938,7 @@ class Diaspora
 		$datarray["body"] = self::replacePeopleGuid($body, $person["url"]);
 
 		self::storeMentions($datarray['uri-id'], $text);
-		Tag::storeFromBody($datarray['uri-id'], $datarray["body"], '#');
+		Tag::storeRawTagsFromBody($datarray['uri-id'], $datarray["body"]);
 
 		self::fetchGuid($datarray);
 
@@ -3015,7 +3015,7 @@ class Diaspora
 		$datarray["body"] = self::replacePeopleGuid($body, $contact["url"]);
 
 		self::storeMentions($datarray['uri-id'], $text);
-		Tag::storeFromBody($datarray['uri-id'], $datarray["body"], '#');
+		Tag::storeRawTagsFromBody($datarray['uri-id'], $datarray["body"]);
 
 		if ($provider_display_name != "") {
 			$datarray["app"] = $provider_display_name;