🎨 OCR 未启用时不保存 ocr-texts.json https://github.com/siyuan-note/siyuan/issues/11171

kernel/model/ocr.go

@@ -2,13 +2,9 @@ package model
 
 import (
 	"path/filepath"
-	"runtime/debug"
 	"strings"
 	"time"
 
-	"github.com/88250/go-humanize"
-	"github.com/88250/gulu"
-	"github.com/siyuan-note/filelock"
 	"github.com/siyuan-note/logging"
 	"github.com/siyuan-note/siyuan/kernel/cache"
 	"github.com/siyuan-note/siyuan/kernel/sql"
@@ -40,19 +36,14 @@ func autoOCRAssets() {
 			text := util.Tesseract(assetAbsPath)
 			p := strings.TrimPrefix(assetAbsPath, assetsPath)
 			p = "assets" + filepath.ToSlash(p)
-			util.AssetsTextsLock.Lock()
-			util.AssetsTexts[p] = text
-			util.AssetsTextsLock.Unlock()
-			if "" != text {
-				util.AssetsTextsChanged.Store(true)
-			}
+			util.SetAssetText(p, text)
 			if 7 <= i { // 一次任务中最多处理 7 张图片，防止长时间占用系统资源
 				break
 			}
 		}
 	}
 
-	cleanNotExistAssetsTexts()
+	util.CleanNotExistAssetsTexts()
 
 	// 刷新 OCR 结果到数据库
 	util.NodeOCRQueueLock.Lock()
@@ -63,27 +54,6 @@ func autoOCRAssets() {
 	util.NodeOCRQueue = nil
 }
 
-func cleanNotExistAssetsTexts() {
-	util.AssetsTextsLock.Lock()
-	defer util.AssetsTextsLock.Unlock()
-
-	assetsPath := util.GetDataAssetsAbsPath()
-	var toRemoves []string
-	for asset, _ := range util.AssetsTexts {
-		assetAbsPath := strings.TrimPrefix(asset, "assets")
-		assetAbsPath = filepath.Join(assetsPath, assetAbsPath)
-		if !filelock.IsExist(assetAbsPath) {
-			toRemoves = append(toRemoves, asset)
-		}
-	}
-
-	for _, asset := range toRemoves {
-		delete(util.AssetsTexts, asset)
-		util.AssetsTextsChanged.Store(true)
-	}
-	return
-}
-
 func getUnOCRAssetsAbsPaths() (ret []string) {
 	var assetsPaths []string
 	assets := cache.GetAssets()
@@ -95,9 +65,8 @@ func getUnOCRAssetsAbsPaths() (ret []string) {
 	}
 
 	assetsPath := util.GetDataAssetsAbsPath()
-	assetsTextsTmp := util.AssetsTexts
 	for _, assetPath := range assetsPaths {
-		if _, ok := assetsTextsTmp[assetPath]; ok {
+		if util.ExistsAssetText(assetPath) {
 			continue
 		}
 		absPath := filepath.Join(assetsPath, strings.TrimPrefix(assetPath, "assets"))
@@ -107,66 +76,5 @@ func getUnOCRAssetsAbsPaths() (ret []string) {
 }
 
 func FlushAssetsTextsJob() {
-	SaveAssetsTexts()
-}
-
-func LoadAssetsTexts() {
-	assetsPath := util.GetDataAssetsAbsPath()
-	assetsTextsPath := filepath.Join(assetsPath, "ocr-texts.json")
-	if !filelock.IsExist(assetsTextsPath) {
-		return
-	}
-
-	start := time.Now()
-	data, err := filelock.ReadFile(assetsTextsPath)
-	if nil != err {
-		logging.LogErrorf("read assets texts failed: %s", err)
-		return
-	}
-
-	util.AssetsTextsLock.Lock()
-	if err = gulu.JSON.UnmarshalJSON(data, &util.AssetsTexts); nil != err {
-		logging.LogErrorf("unmarshal assets texts failed: %s", err)
-		if err = filelock.Remove(assetsTextsPath); nil != err {
-			logging.LogErrorf("removed corrupted assets texts failed: %s", err)
-		}
-		return
-	}
-	util.AssetsTextsLock.Unlock()
-	debug.FreeOSMemory()
-
-	if elapsed := time.Since(start).Seconds(); 2 < elapsed {
-		logging.LogWarnf("read assets texts [%s] to [%s], elapsed [%.2fs]", humanize.BytesCustomCeil(uint64(len(data)), 2), assetsTextsPath, elapsed)
-	}
-	return
-}
-
-func SaveAssetsTexts() {
-	if !util.AssetsTextsChanged.Load() {
-		return
-	}
-
-	start := time.Now()
-
-	util.AssetsTextsLock.Lock()
-	data, err := gulu.JSON.MarshalIndentJSON(util.AssetsTexts, "", "  ")
-	if nil != err {
-		logging.LogErrorf("marshal assets texts failed: %s", err)
-		return
-	}
-	util.AssetsTextsLock.Unlock()
-
-	assetsPath := util.GetDataAssetsAbsPath()
-	assetsTextsPath := filepath.Join(assetsPath, "ocr-texts.json")
-	if err = filelock.WriteFile(assetsTextsPath, data); nil != err {
-		logging.LogErrorf("write assets texts failed: %s", err)
-		return
-	}
-	debug.FreeOSMemory()
-
-	if elapsed := time.Since(start).Seconds(); 2 < elapsed {
-		logging.LogWarnf("save assets texts [size=%s] to [%s], elapsed [%.2fs]", humanize.BytesCustomCeil(uint64(len(data)), 2), assetsTextsPath, elapsed)
-	}
-
-	util.AssetsTextsChanged.Store(false)
+	util.SaveAssetsTexts()
 }