Merge remote-tracking branch 'origin/dev' into dev

kernel/model/transaction.go

@@ -343,6 +343,10 @@ func (tx *Transaction) processUndoInsertWithFoldedHeading() {
 	// 删除折叠标题后撤销，需要调整 insert 顺序和 previousID
 	// https://github.com/siyuan-note/siyuan/issues/16120
 
+	if 1 > len(tx.DoOperations) {
+		return
+	}
+
 	// 所有操作均为 insert 才处理
 	for _, op := range tx.DoOperations {
 		if "insert" != op.Action {
@@ -350,6 +354,8 @@ func (tx *Transaction) processUndoInsertWithFoldedHeading() {
 		}
 	}
 
+	// 找到 ignoreProcess=true 的区间 [j, k]
+	var j, k int
 	for i := 0; i < len(tx.DoOperations); i++ {
 		op := tx.DoOperations[i]
 		ignoreProcess := false
@@ -361,38 +367,26 @@ func (tx *Transaction) processUndoInsertWithFoldedHeading() {
 				return
 			}
 		}
-		if ignoreProcess {
-			// 找到从当前 i 到下个 ignoreProcess=false 的区间，整体反转
-			j := i + 1
-			if j >= len(tx.DoOperations) {
-				return
-			}
 
-			for ; j < len(tx.DoOperations); j++ {
-				nextOp := tx.DoOperations[j]
-				nextIgnoreProcess := false
-				if nil != nextOp.Context["ignoreProcess"] {
-					var convErr error
-					nextIgnoreProcess, convErr = strconv.ParseBool(nextOp.Context["ignoreProcess"].(string))
-					if nil != convErr {
-						logging.LogErrorf("parse ignoreProcess failed: %s", convErr)
-						return
-					}
-				}
-				if !nextIgnoreProcess {
+		if !ignoreProcess {
+			if 0 != j && 0 != k {
 				break
 			}
+			continue
 		}
-			for _, nextOp := range tx.DoOperations[i:j] {
-				nextOp.PreviousID = tx.DoOperations[j].ID
+
+		if 0 == j && 0 == k {
+			j = i
 		}
-			slices.Reverse(tx.DoOperations[i : j+1])
-			i = j
-		}
-		if i >= len(tx.DoOperations) {
-			return
+		k = i
 	}
+
+	// 调整 [j, k] 区间内的操作顺序和 previousID
+	for x := j; x <= k; x++ {
+		opx := tx.DoOperations[x]
+		opx.PreviousID = tx.DoOperations[k].PreviousID
 	}
+	slices.Reverse(tx.DoOperations[j : k+1])
 }
 
 func (tx *Transaction) processLargeInsert() bool {

kernel/util/ocr.go

@@ -19,6 +19,7 @@ package util
 import (
 	"bytes"
 	"context"
+	"errors"
 	"fmt"
 	"os"
 	"os/exec"
@@ -182,9 +183,29 @@ 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
@@ -214,14 +235,23 @@ func Tesseract(imgAbsPath string) (ret []map[string]interface{}) {
 
 	defer logging.Recover()
 
-	ctx, cancel := context.WithTimeout(context.Background(), 7*time.Second)
+	timeout := 7000
+	timeoutEnv := os.Getenv("SIYUAN_TESSERACT_TIMEOUT")
+	if "" != timeoutEnv {
+		if timeoutParsed, parseErr := strconv.Atoi(timeoutEnv); nil == parseErr {
+			timeout = timeoutParsed
+		} else {
+			logging.LogWarnf("parse tesseract timeout [%s] failed: %s", timeoutEnv, parseErr)
+		}
+	}
+	ctx, cancel := context.WithTimeout(context.Background(), time.Duration(timeout)*time.Millisecond)
 	defer cancel()
 
 	cmd := exec.CommandContext(ctx, TesseractBin, "-c", "debug_file=/dev/null", imgAbsPath, "stdout", "-l", strings.Join(TesseractLangs, "+"), "tsv")
 	gulu.CmdAttr(cmd)
 	output, err := cmd.CombinedOutput()
-	if ctx.Err() == context.DeadlineExceeded {
-		logging.LogWarnf("tesseract [path=%s, size=%d] timeout", imgAbsPath, info.Size())
+	if errors.Is(ctx.Err(), context.DeadlineExceeded) {
+		logging.LogWarnf("tesseract [path=%s, size=%d] timeout [%dms]", imgAbsPath, info.Size(), timeout)
 		return
 	}
 
@@ -300,7 +330,7 @@ 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