From 4baeeed1ecb7454e532fb5f4be72fca855b29585 Mon Sep 17 00:00:00 2001
From: Daniel <845765@qq.com>
Date: Sun, 27 Aug 2023 10:53:48 +0800
Subject: [PATCH] :art: Non-UTF-8 encoded text files are not included in asset
 file content searching Fix https://github.com/siyuan-note/siyuan/issues/9052

---
 kernel/model/asset_content.go | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/kernel/model/asset_content.go b/kernel/model/asset_content.go
index 2ba193b8e..c5ac3c7d9 100644
--- a/kernel/model/asset_content.go
+++ b/kernel/model/asset_content.go
@@ -26,6 +26,7 @@ import (
 	"strings"
 	"sync"
 	"time"
+	"unicode/utf8"
 
 	"code.sajari.com/docconv"
 	"github.com/88250/gulu"
@@ -516,6 +517,11 @@ func (parser *TxtAssetParser) Parse(absPath string) (ret *AssetParseResult) {
 		return
 	}
 
+	if !utf8.Valid(data) {
+		// Non-UTF-8 encoded text files are not included in asset file content searching https://github.com/siyuan-note/siyuan/issues/9052
+		return
+	}
+
 	content := string(data)
 	ret = &AssetParseResult{
 		Content: content,