From d06b37007782e456f3a8d0c97c8681f79db30d23 Mon Sep 17 00:00:00 2001
From: Daniel <845765@qq.com>
Date: Sat, 22 Nov 2025 10:55:08 +0800
Subject: [PATCH] :art: Image OCR supports more formats
 https://github.com/siyuan-note/siyuan/issues/16418

Signed-off-by: Daniel <845765@qq.com>
---
 kernel/util/ocr.go | 24 ++++++++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

diff --git a/kernel/util/ocr.go b/kernel/util/ocr.go
index a9370071e..040468b54 100644
--- a/kernel/util/ocr.go
+++ b/kernel/util/ocr.go
@@ -182,9 +182,29 @@ func RemoveAssetText(asset string) {
 	assetsTextsChanged.Store(true)
 }
 
+var tesseractExts = []string{
+	".png",
+	".jpg",
+	".jpeg",
+	".tif",
+	".tiff",
+	".bmp",
+	".gif",
+	".webp",
+	".pbm",
+	".pgm",
+	".ppm",
+	".pnm",
+}
+
 func IsTesseractExtractable(p string) bool {
 	lowerName := strings.ToLower(p)
-	return strings.HasSuffix(lowerName, ".png") || strings.HasSuffix(lowerName, ".jpg") || strings.HasSuffix(lowerName, ".jpeg")
+	for _, ext := range tesseractExts {
+		if strings.HasSuffix(lowerName, ext) {
+			return true
+		}
+	}
+	return false
 }
 
 // tesseractOCRLock 用于 Tesseract OCR 加锁串行执行提升稳定性 https://github.com/siyuan-note/siyuan/issues/7265
@@ -300,7 +320,7 @@ func InitTesseract() {
 
 	langs := getTesseractLangs()
 	if 1 > len(langs) {
-		logging.LogWarnf("no tesseract langs found")
+		logging.LogWarnf("no tesseract langs found, disabling tesseract-ocr")
 		TesseractEnabled = false
 		tesseractInited.Store(true)
 		return