From bd0818b7e123cebb57768f652c503d0833a1a373 Mon Sep 17 00:00:00 2001
From: ralf <rb@egroupware.org>
Date: Thu, 20 Jun 2024 12:29:13 +0200
Subject: [PATCH] * Filemanager: automatic shorten too long filenames and
 replace 4-byte utf-8 characters, which gave an error when trying to store VFS

---
 api/src/Vfs.php                              | 13 +++++++++++++
 api/src/Vfs/Sqlfs/StreamWrapper.php          | 12 +++++++++++-
 filemanager/inc/class.filemanager_ui.inc.php |  2 +-
 3 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/api/src/Vfs.php b/api/src/Vfs.php
index 1809cf7b08..86b193f934 100644
--- a/api/src/Vfs.php
+++ b/api/src/Vfs.php
@@ -1795,6 +1795,19 @@ class Vfs extends Vfs\Base
 		return str_replace(array_keys(self::$encode),array_values(self::$encode),$component);
 	}
 
+	/**
+	 * Limit filename to precision of column while keeping the extension
+	 *
+	 * Also takes care to replace 4-byte utf-8 chars e.g. used in some Emojis with a replacement character.
+	 *
+	 * @param string $name
+	 * @return string
+	 */
+	static public function limitFilename($name)
+	{
+		return Vfs\Sqlfs\StreamWrapper::limit_filename($name);
+	}
+
 	/**
 	 * Encode a path: replacing certain chars with their urlencoded counterparts
 	 *
diff --git a/api/src/Vfs/Sqlfs/StreamWrapper.php b/api/src/Vfs/Sqlfs/StreamWrapper.php
index 326549c9a7..6dc83c2e86 100644
--- a/api/src/Vfs/Sqlfs/StreamWrapper.php
+++ b/api/src/Vfs/Sqlfs/StreamWrapper.php
@@ -1767,16 +1767,26 @@ GROUP BY A.fs_id';
 	/**
 	 * Limit filename to precision of column while keeping the extension
 	 *
+	 * Also takes care to replace 4-byte utf-8 chars e.g. used in some Emojis with a replacement character.
+	 *
 	 * @param string $name
 	 * @return string
 	 */
-	static protected function limit_filename($name)
+	static public function limit_filename($name)
 	{
 		static $fs_name_precision = null;
 		if (!isset($fs_name_precision))
 		{
 			$fs_name_precision = $GLOBALS['egw']->db->get_column_attribute('fs_name', self::TABLE, 'phpgwapi', 'precision');
 		}
+		// MySQL and MariaDB not 10.1 need 4-byte utf8 chars replaced with our default utf8 charset
+		// (MariaDB 10.1 does the replacement automatic, 10.0 cuts everything off behind and MySQL gives an error)
+		// (MariaDB 10.3 gives an error too: Incorrect string value: '\xF0\x9F\x98\x8A\x0AW...')
+		// Changing charset to utf8mb4 requires schema update, shortening of some indexes and probably have negative impact on performance!
+		if (isset($name) && substr($GLOBALS['egw']->db->Type, 0, 5) === 'mysql')
+		{
+			$name = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $name);
+		}
 		if (mb_strlen($name) > $fs_name_precision)
 		{
 			$parts = explode('.', $name);
diff --git a/filemanager/inc/class.filemanager_ui.inc.php b/filemanager/inc/class.filemanager_ui.inc.php
index 8e5d36e7b2..f169e5fdf3 100644
--- a/filemanager/inc/class.filemanager_ui.inc.php
+++ b/filemanager/inc/class.filemanager_ui.inc.php
@@ -1755,7 +1755,7 @@ class filemanager_ui
 
 		foreach($selected as $tmp_name => &$data)
 		{
-			$path = Vfs::concat($dir, Vfs::encodePathComponent($data['name']));
+			$path = Vfs::concat($dir, Vfs::limitFilename(Vfs::encodePathComponent($data['name'])));
 
 			if(Vfs::deny_script($path))
 			{