⚡ Optimize asset cache lookups and add accessors (#16894)

* optimize asset cache lookups and add accessors.

* adopt cache.FilterAssets to optimize asset processing.

kernel/cache/asset.go

@@ -104,17 +104,53 @@ var (
 	assetsLock  = sync.Mutex{}
 )
 
+// GetAssets 返回所有资源的副本
 func GetAssets() (ret map[string]*Asset) {
 	assetsLock.Lock()
 	defer assetsLock.Unlock()
 
-	ret = map[string]*Asset{}
+	// 指定长度避免分配新内存时扩容和迁移
+	ret = make(map[string]*Asset, len(assetsCache))
 	for k, v := range assetsCache {
 		ret[k] = v
 	}
 	return
 }
 
+// IterateAssets 遍历所有资源，适合只读场景
+func IterateAssets(fn func(path string, asset *Asset) bool) {
+	assetsLock.Lock()
+	defer assetsLock.Unlock()
+
+	for path, asset := range assetsCache {
+		if !fn(path, asset) {
+			break
+		}
+	}
+}
+
+// FilterAssets 根据过滤函数返回符合条件的资源
+func FilterAssets(filter func(path string, asset *Asset) bool) (ret map[string]*Asset) {
+	assetsLock.Lock()
+	defer assetsLock.Unlock()
+
+	ret = make(map[string]*Asset)
+	for path, asset := range assetsCache {
+		if filter(path, asset) {
+			ret[path] = asset
+		}
+	}
+	return
+}
+
+// GetAssetByPath 根据路径获取资源
+func GetAssetByPath(path string) *Asset {
+	assetsLock.Lock()
+	defer assetsLock.Unlock()
+
+	return assetsCache[path]
+}
+
 func RemoveAsset(path string) {
 	assetsLock.Lock()
 	defer assetsLock.Unlock()

kernel/model/assets.go

@@ -448,7 +448,9 @@ func SearchAssetsByName(keyword string, exts []string) (ret []*cache.Asset) {
 	}
 	pathHitCount := map[string]int{}
 	filterByExt := 0 < len(exts)
-	for _, asset := range cache.GetAssets() {
+	matchedAssets := cache.FilterAssets(func(path string, asset *cache.Asset) bool {
+
+		// 扩展名过滤
 		if filterByExt {
 			ext := filepath.Ext(asset.HName)
 			includeExt := false
@@ -459,10 +461,11 @@ func SearchAssetsByName(keyword string, exts []string) (ret []*cache.Asset) {
 				}
 			}
 			if !includeExt {
-				continue
+				return false
 			}
 		}
 
+		// 关键字匹配
 		lowerHName := strings.ToLower(asset.HName)
 		lowerPath := strings.ToLower(asset.Path)
 		var hitNameCount, hitPathCount int
@@ -485,13 +488,21 @@ func SearchAssetsByName(keyword string, exts []string) (ret []*cache.Asset) {
 			}
 		}
 
+		// 只返回有匹配的资源
 		if 1 > hitNameCount+hitPathCount {
-			continue
+			return false
 		}
-		pathHitCount[asset.Path] += hitNameCount + hitPathCount
 
+		// 记录命中次数用于排序
+		pathHitCount[asset.Path] = hitNameCount + hitPathCount
+		return true
+	})
+
+	// 添加高亮
+	for _, asset := range matchedAssets {
+		hitCount := pathHitCount[asset.Path]
 		hName := asset.HName
-		if 0 < hitNameCount {
+		if hitCount > 0 {
 			_, hName = search.MarkText(asset.HName, strings.Join(keywords, search.TermSep), 64, Conf.Search.CaseSensitive)
 		}
 		ret = append(ret, &cache.Asset{

kernel/model/history.go

@@ -770,15 +770,21 @@ func (box *Box) recentModifiedDocs() (ret []string) {
 var assetsLatestHistoryTime = time.Now().Unix()
 
 func recentModifiedAssets() (ret []string) {
-	assets := cache.GetAssets()
-	for _, asset := range assets {
-		if asset.Updated > assetsLatestHistoryTime {
-			absPath := filepath.Join(util.DataDir, asset.Path)
-			if filelock.IsHidden(absPath) {
-				continue
-			}
-			ret = append(ret, absPath)
+	// 只获取最近修改的资源
+	recentAssets := cache.FilterAssets(func(path string, asset *cache.Asset) bool {
+		if asset.Updated <= assetsLatestHistoryTime {
+			return false
 		}
+		absPath := filepath.Join(util.DataDir, asset.Path)
+		if filelock.IsHidden(absPath) {
+			return false
+		}
+		return true
+	})
+
+	for _, asset := range recentAssets {
+		absPath := filepath.Join(util.DataDir, asset.Path)
+		ret = append(ret, absPath)
 	}
 	assetsLatestHistoryTime = time.Now().Unix()
 	return

kernel/model/ocr.go

@@ -55,21 +55,18 @@ func autoOCRAssets() {
 }
 
 func getUnOCRAssetsAbsPaths() (ret []string) {
-	var assetsPaths []string
-	assets := cache.GetAssets()
-	for _, asset := range assets {
-		if !util.IsTesseractExtractable(asset.Path) {
-			continue
-		}
-		assetsPaths = append(assetsPaths, asset.Path)
-	}
+	// 只获取需要 OCR 的资源
+	ocrAssets := cache.FilterAssets(func(path string, asset *cache.Asset) bool {
+		return util.IsTesseractExtractable(asset.Path)
+	})
 
 	assetsPath := util.GetDataAssetsAbsPath()
-	for _, assetPath := range assetsPaths {
-		if util.ExistsAssetText(assetPath) {
+	for _, asset := range ocrAssets {
+		// 跳过已经存在 OCR 文本的资源
+		if util.ExistsAssetText(asset.Path) {
 			continue
 		}
-		absPath := filepath.Join(assetsPath, strings.TrimPrefix(assetPath, "assets"))
+		absPath := filepath.Join(assetsPath, strings.TrimPrefix(asset.Path, "assets"))
 		ret = append(ret, absPath)
 	}
 	return