From fac906aeb4493eb6508486c507a23487774578c0 Mon Sep 17 00:00:00 2001
From: Anna Larch <anna@nextcloud.com>
Date: Mon, 20 Dec 2021 16:17:26 +0100
Subject: [PATCH] Checnge how charsets are detected for emails

Signed-off-by: Anna Larch <anna@nextcloud.com>
---
 lib/Model/IMAPMessage.php | 20 +++++++++++++++++++-
 1 file changed, 19 insertions(+), 1 deletion(-)

diff --git a/lib/Model/IMAPMessage.php b/lib/Model/IMAPMessage.php
index 0b3c38f505..e0f3740648 100644
--- a/lib/Model/IMAPMessage.php
+++ b/lib/Model/IMAPMessage.php
@@ -48,10 +48,12 @@
 use OCA\Mail\Db\MailAccount;
 use OCA\Mail\Db\Message;
 use OCA\Mail\Db\Tag;
+use OCA\Mail\Exception\ServiceException;
 use OCA\Mail\Service\Html;
 use OCP\AppFramework\Db\DoesNotExistException;
 use OCP\Files\File;
 use OCP\Files\SimpleFS\ISimpleFile;
+use ValueError;
 use function in_array;
 use function mb_convert_encoding;
 use function mb_strcut;
@@ -588,6 +590,7 @@ private function handleHtmlMessage(Horde_Mime_Part $p, $partNo): void {
 	 * @return string
 	 * @throws DoesNotExistException
 	 * @throws Exception
+	 * @throws ServiceException
 	 */
 	private function loadBodyData(Horde_Mime_Part $p, $partNo): string {
 		// DECODE DATA
@@ -619,7 +622,22 @@ private function loadBodyData(Horde_Mime_Part $p, $partNo): string {
 		$p->setContents($data);
 		$data = $p->getContents();
 
-		$data = mb_convert_encoding($data, 'UTF-8', $p->getCharset());
+		$charset = mb_detect_encoding($data, 'UTF-8', true);
+
+		if (!$charset) {
+			$charset = $p->getCharset();
+		}
+
+		try {
+			$data = @mb_convert_encoding($data, 'UTF-8', $charset);
+		} catch (ValueError $e) {
+			throw new ServiceException('Could not detect charset for message ' . $e->getMessage(), $e->getCode());
+		}
+
+		if (!$data) {
+			throw new ServiceException('Could not detect charset for message');
+		}
+
 		return $data;
 	}