@hzttt/multimodal-rag 0.2.9 → 0.3.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hzttt/multimodal-rag",
-  "version": "0.2.9",
+  "version": "0.3.0",
   "description": "OpenClaw plugin for multimodal RAG - semantic indexing and time-aware search for images and audio using local AI models",
   "type": "module",
   "repository": {

package/src/notifier.ts

@@ -5,7 +5,7 @@
 import type { PluginRuntime } from "openclaw/plugin-sdk";
 import type { MediaType, NotificationConfig, IndexEventCallbacks } from "./types.js";
 
-type BatchFileStatus = "queued" | "indexed" | "failed";
+type BatchFileStatus = "queued" | "indexed" | "skipped" | "failed";
 type BatchFile = { status: BatchFileStatus; fileType?: MediaType; error?: string };
 type MainSessionConfig = Parameters<PluginRuntime["system"]["resolveMainSessionKey"]>[0];
 type RuntimeSystemCompat = {
@@ -114,6 +114,17 @@ export class IndexNotifier implements IndexEventCallbacks {
     this.resetQuietTimer();
   }
 
+  /**
+   * 文件跳过事件（例如已存在的重复内容）
+   */
+  onFileSkipped(filePath: string, fileType: MediaType): void {
+    if (this.state === "idle" && !this.batch.has(filePath)) {
+      return;
+    }
+    this.batch.set(filePath, { status: "skipped", fileType });
+    this.resetQuietTimer();
+  }
+
   /**
    * 文件索引失败事件
    */
@@ -676,8 +687,7 @@ export class IndexNotifier implements IndexEventCallbacks {
    * 构建"开始索引"消息
    */
   private buildStartMessage(): string {
-    const count = this.batch.size;
-    return `[Multimodal RAG] 新文件索引通知: 检测到 ${count} 个新的媒体文件正在被索引，请通知用户。`;
+    return "[Multimodal RAG] 新文件索引通知: 已开始处理本轮新增媒体文件，请通知用户。";
   }
 
   /**
@@ -685,9 +695,10 @@ export class IndexNotifier implements IndexEventCallbacks {
    */
   private buildSummaryMessage(): string {
     const files = [...this.batch.values()];
-    const total = files.length;
     const succeeded = files.filter((f) => f.status === "indexed");
+    const skipped = files.filter((f) => f.status === "skipped");
     const failed = files.filter((f) => f.status === "failed");
+    const total = succeeded.length + failed.length;
 
     // 统计成功文件的类型
     const images = succeeded.filter((f) => f.fileType === "image").length;
@@ -701,8 +712,15 @@ export class IndexNotifier implements IndexEventCallbacks {
     const durationStr =
       minutes > 0 ? `${minutes} 分 ${seconds} 秒` : `${seconds} 秒`;
 
-    // 构建消息
-    let message = `[Multimodal RAG] 索引完成通知: 共处理 ${total} 个文件，`;
+    // 构建消息（仅统计本轮真正处理的文件：成功 + 失败，不把 skipped 算进处理总数）
+    if (total === 0) {
+      if (skipped.length > 0) {
+        return `[Multimodal RAG] 索引完成通知: 本轮没有新增文件需要索引（跳过 ${skipped.length} 个已存在文件）。耗时 ${durationStr}。请通知用户。`;
+      }
+      return `[Multimodal RAG] 索引完成通知: 本轮没有可汇总的处理结果。耗时 ${durationStr}。请通知用户。`;
+    }
+
+    let message = `[Multimodal RAG] 索引完成通知: 本轮共处理 ${total} 个文件，`;
     message += `成功 ${succeeded.length} 个`;
 
     if (images > 0 || audios > 0) {
@@ -716,6 +734,10 @@ export class IndexNotifier implements IndexEventCallbacks {
       message += `，失败 ${failed.length} 个`;
     }
 
+    if (skipped.length > 0) {
+      message += `（另跳过 ${skipped.length} 个已存在文件）`;
+    }
+
     message += `。耗时 ${durationStr}。请发送索引完成总结通知给用户。`;
 
     return message;

package/src/types.ts

@@ -54,6 +54,7 @@ export type NotificationConfig = {
 export type IndexEventCallbacks = {
   onFileQueued: (filePath: string) => void;
   onFileIndexed: (filePath: string, fileType: MediaType) => void;
+  onFileSkipped?: (filePath: string, fileType: MediaType, reason?: string) => void;
   onFileFailed: (filePath: string, error: string) => void;
   dispose?: () => void;
 };

package/src/watcher.ts

@@ -3,7 +3,7 @@
  */
 
 import chokidar from "chokidar";
-import { stat, readdir } from "node:fs/promises";
+import { stat, readdir, realpath } from "node:fs/promises";
 import { basename, extname, resolve, join } from "node:path";
 import { createHash } from "node:crypto";
 import { readFile } from "node:fs/promises";
@@ -253,9 +253,9 @@ export class MediaWatcher {
       const existing = await this.storage.findByHash(fileHash);
       if (existing) {
         this.logger.info?.(`Skipping duplicate: ${fileName}`);
-        // 索引成功（已存在），清除失败记录并通知回调
+        // 已存在内容：标记为 skipped，不计入“本次新索引成功数”
         this.failedFiles.delete(filePath);
-        this.callbacks?.onFileIndexed(filePath, fileType);
+        this.callbacks?.onFileSkipped?.(filePath, fileType, "duplicate");
         return true;
       }
 
@@ -375,12 +375,27 @@ export class MediaWatcher {
     const { entries: indexedFiles } = await this.storage.list({ 
       limit: 10000  // 获取所有文件
     });
-    const indexedPathsSet = new Set(indexedFiles.map(f => f.filePath));
+    const indexedPathsSet = new Set(indexedFiles.map((f) => f.filePath));
+
+    // 用 realpath 归一化，避免因为软链/路径别名导致“已索引文件被误判为缺失”
+    const normalizedIndexedPaths = await Promise.all(
+      indexedFiles.map(async (file) => await this.normalizeComparablePath(file.filePath)),
+    );
+    for (const normalized of normalizedIndexedPaths) {
+      indexedPathsSet.add(normalized);
+    }
+
+    const comparableAllFiles = await Promise.all(
+      allFiles.map(async (filePath) => ({
+        filePath,
+        comparablePath: await this.normalizeComparablePath(filePath),
+      })),
+    );
 
     // 找出缺失的文件
     let missingFiles = 0;
-    for (const filePath of allFiles) {
-      if (!indexedPathsSet.has(filePath)) {
+    for (const { filePath, comparablePath } of comparableAllFiles) {
+      if (!indexedPathsSet.has(filePath) && !indexedPathsSet.has(comparablePath)) {
         missingFiles++;
         this.enqueueFile(filePath);
       }
@@ -395,6 +410,15 @@ export class MediaWatcher {
     }
   }
 
+  private async normalizeComparablePath(filePath: string): Promise<string> {
+    const resolved = resolve(filePath);
+    try {
+      return await realpath(resolved);
+    } catch {
+      return resolved;
+    }
+  }
+
   /**
    * 递归扫描目录获取所有支持的文件
    */