fix: 重复文件归类、图片探测与明暗分析健壮性

- organize/store: 按 MD5 去重，重复图片/视频/音频统一移入「重复文件」目录，不再覆盖已有记录
- probe/image: webp/bmp/psd/psb 专用解码失败后走 ffprobe 兜底，与通用路径一致
- probe/brightness: 大图内存缩放为缩略图再分析明暗，超4000万像素跳过分析防OOM
- README: 补充重复文件目录说明，修正 Go 版本为 1.25.0

internal/organize/organize.go

@@ -23,6 +23,7 @@ const (
 	dirVideoErr     = "视频_读取异常"
 	dirVideoIgnored = "视频_已忽略"
 	dirAudio        = "音频"
+	dirDuplicate    = "重复文件"
 )
 
 // Organizer 整理器：按目标目录收集文件，处理完成后统一建目录并移动
@@ -55,17 +56,19 @@ func (o *Organizer) add(f scanner.File, subDir string) {
 	o.buckets[dir] = append(o.buckets[dir], f.Path)
 }
 
-// record 将媒体信息写入数据库（大文件优先用指纹复用已入库的 MD5）
-func (o *Organizer) record(f scanner.File, mediaType string, res *probe.ProbeResult) {
+// calcFileInfo 读取文件大小并计算 MD5（大文件优先用指纹复用已入库的 MD5，跳过全量读取）。
+// store 为 nil 时直接返回空值（无需去重/入库）。
+func (o *Organizer) calcFileInfo(f scanner.File) (md5Str, headMd5, tailMd5 string, size int64) {
+	if o.store == nil {
+		return "", "", "", 0
+	}
 	st, err := os.Stat(f.Path)
 	if err != nil {
 		fmt.Printf("=== 读取文件信息失败: %s, 错误: %v\n", f.Path, err)
-		return
+		return "", "", "", 0
 	}
-	size := st.Size()
+	size = st.Size()
 
-	md5Str := ""
-	headMd5, tailMd5 := "", ""
 	if size > digest.FingerprintThreshold {
 		if head, tail, ferr := digest.CalcFileFingerprint(f.Path, size); ferr == nil {
 			headMd5, tailMd5 = head, tail
@@ -77,10 +80,26 @@ func (o *Organizer) record(f scanner.File, mediaType string, res *probe.ProbeRes
 	if md5Str == "" {
 		if md5Str, err = digest.CalcFileMd5(f.Path); err != nil {
 			fmt.Printf("=== 计算MD5失败: %s, 错误: %v\n", f.Path, err)
-			return
+			return "", "", "", size
 		}
 	}
+	return md5Str, headMd5, tailMd5, size
+}
 
+// isDuplicate 判断文件是否与已入库记录重复（按 MD5）
+func (o *Organizer) isDuplicate(md5Str string) bool {
+	if o.store == nil || md5Str == "" {
+		return false
+	}
+	_, ok := o.store.FindMediaByMd5(md5Str)
+	return ok
+}
+
+// record 将媒体信息写入数据库
+func (o *Organizer) record(f scanner.File, mediaType string, res *probe.ProbeResult, md5Str, headMd5, tailMd5 string, size int64) {
+	if o.store == nil || md5Str == "" {
+		return
+	}
 	codec := res.VideoCodec
 	if mediaType == "audio" {
 		codec = res.AudioCodec
@@ -107,20 +126,22 @@ func (o *Organizer) record(f scanner.File, mediaType string, res *probe.ProbeRes
 	}
 }
 
-// ProcessImages 处理全部图片：探测尺寸并归类
+// ProcessImages 处理全部图片：去重 → 探测尺寸并归类
 func (o *Organizer) ProcessImages(images []scanner.File) {
-	success, fail, ignore := 0, 0, 0
+	success, fail, ignore, dup := 0, 0, 0, 0
 	for _, f := range images {
 		switch {
 		case strings.EqualFold(f.Ext, ".psd"):
 			o.add(f, dirImagePSD)
 			success++
 			continue
-		case !probe.IsSupportImage(f.Ext):
+		case !probe.IsSupportImage(f.Ext) && o.ffprobePath == "":
+			// 非白名单扩展名且无 ffprobe 兜底时直接忽略
 			o.add(f, dirImageIgnored)
 			ignore++
 			continue
 		}
+
 		res, err := probe.ProbeImage(f.Path, o.ffprobePath)
 		if err != nil {
 			o.add(f, dirImageErr)
@@ -128,7 +149,15 @@ func (o *Organizer) ProcessImages(images []scanner.File) {
 			fmt.Printf("=== 异常图片: %s \n", f.Path)
 			continue
 		}
-		o.record(f, "image", res)
+
+		md5Str, headMd5, tailMd5, size := o.calcFileInfo(f)
+		if o.isDuplicate(md5Str) {
+			o.add(f, dirDuplicate)
+			dup++
+			fmt.Printf("=== 重复图片: %s \n", f.Path)
+			continue
+		}
+		o.record(f, "image", res, md5Str, headMd5, tailMd5, size)
 		darkScore, lightScore := 0.0, 0.0
 		if !strings.EqualFold(f.Ext, ".gif") {
 			dark, light, berr := probe.AnalyzeBrightness(f.Path)
@@ -141,12 +170,12 @@ func (o *Organizer) ProcessImages(images []scanner.File) {
 		o.add(f, classify.ImageDir(f.Ext, res.Width, res.Height, darkScore, lightScore))
 		success++
 	}
-	fmt.Printf("=== 图片处理完毕，成功: %d, 读取异常: %d, 忽略: %d \n\n", success, fail, ignore)
+	fmt.Printf("=== 图片处理完毕，成功: %d, 读取异常: %d, 忽略: %d, 重复: %d \n\n", success, fail, ignore, dup)
 }
 
-// ProcessMedia 处理媒体候选：ffprobe 判定视频/音频/不支持
+// ProcessMedia 处理媒体候选：去重 → ffprobe 判定视频/音频/不支持
 func (o *Organizer) ProcessMedia(media []scanner.File) {
-	video, audio, fail, ignore := 0, 0, 0, 0
+	video, audio, fail, ignore, dup := 0, 0, 0, 0, 0
 	for _, f := range media {
 		res, err := probe.ProbeMediaFile(o.ffprobePath, f.Path)
 		if err != nil {
@@ -167,12 +196,26 @@ func (o *Organizer) ProcessMedia(media []scanner.File) {
 		}
 		switch {
 		case res.HasVideo:
+			md5Str, headMd5, tailMd5, size := o.calcFileInfo(f)
+			if o.isDuplicate(md5Str) {
+				dup++
+				o.add(f, dirDuplicate)
+				fmt.Printf("=== 重复媒体: %s \n", f.Path)
+				continue
+			}
 			video++
-			o.record(f, "video", res)
+			o.record(f, "video", res, md5Str, headMd5, tailMd5, size)
 			o.add(f, classify.VideoDir(f.Ext, res.Width, res.Height))
 		case res.HasAudio:
+			md5Str, headMd5, tailMd5, size := o.calcFileInfo(f)
+			if o.isDuplicate(md5Str) {
+				dup++
+				o.add(f, dirDuplicate)
+				fmt.Printf("=== 重复媒体: %s \n", f.Path)
+				continue
+			}
 			audio++
-			o.record(f, "audio", res)
+			o.record(f, "audio", res, md5Str, headMd5, tailMd5, size)
 			o.add(f, dirAudio)
 		default:
 			// ffprobe 无音视频流：MIME 明确为媒体的归入忽略目录，其余原地不动
@@ -185,7 +228,7 @@ func (o *Organizer) ProcessMedia(media []scanner.File) {
 			}
 		}
 	}
-	fmt.Printf("=== 媒体处理完毕，视频: %d, 音频: %d, 读取异常: %d, 忽略: %d \n\n", video, audio, fail, ignore)
+	fmt.Printf("=== 媒体处理完毕，视频: %d, 音频: %d, 读取异常: %d, 忽略: %d, 重复: %d \n\n", video, audio, fail, ignore, dup)
 }
 
 // Flush 将所有收集到的文件移动到对应目录，并回填归类去向