@hzttt/multimodal-rag 0.1.1 → 0.1.2

package/README.md

@@ -30,7 +30,7 @@ ollama pull qwen3-embedding:latest
 ### 方式一：从 npm 安装（推荐）
 
 ```bash
-openclaw plugins install @hzttt/multimodal-rag
+openclaw plugins install @hzttt/multimodal-rag@latest
 ```
 
 插件会自动安装到 `~/.openclaw/extensions/multimodal-rag/`，并自动安装所有运行时依赖。
@@ -64,11 +64,55 @@ openclaw multimodal-rag setup
 - **视觉模型**: `qwen3-vl:2b` (图像描述)
 - **嵌入模型**: `qwen3-embedding:latest` (向量生成)
 - **嵌入提供者**: `ollama` (本地)
-- **数据库路径**: `/home/lucy/.openclaw/multimodal-rag.lance`
+- **数据库路径**: `~/.openclaw/multimodal-rag.lance`
 - **启动时索引**: `true` (自动索引已有文件)
 
 你只需要指定要监听的文件夹路径即可。
 
+### 非交互式配置（适合脚本/远程部署）
+
+通过命令行参数一次性完成配置，无需交互输入，适用于 SSH 远程部署、CI/CD 脚本等场景：
+
+```bash
+# 最简用法：只指定监听路径（其他使用默认值）
+openclaw multimodal-rag setup --non-interactive --watch ~/photos --watch ~/audio
+
+# 简写形式
+openclaw multimodal-rag setup -n -w ~/photos -w ~/audio
+
+# 逗号分隔多个路径
+openclaw multimodal-rag setup -n --watch ~/photos,~/mic-recordings,~/usb_data
+
+# 自定义所有参数
+openclaw multimodal-rag setup -n \
+  --watch ~/photos --watch ~/audio \
+  --ollama-url http://192.168.0.100:11434 \
+  --vision-model qwen3-vl:2b \
+  --embed-model qwen3-embedding:latest \
+  --db-path ~/.openclaw/my-rag.lance \
+  --no-index-on-start
+
+# 使用 OpenAI 嵌入
+openclaw multimodal-rag setup -n \
+  --watch ~/photos \
+  --embedding-provider openai \
+  --openai-api-key sk-xxx \
+  --openai-model text-embedding-3-small
+```
+
+| 选项 | 简写 | 说明 | 默认值 |
+| --- | --- | --- | --- |
+| `--non-interactive` | `-n` | 启用非交互式模式 | - |
+| `--watch <paths...>` | `-w` | 监听路径（可多次指定或逗号分隔） | 必填 |
+| `--ollama-url <url>` | - | Ollama 服务地址 | `http://127.0.0.1:11434` |
+| `--vision-model <model>` | - | 视觉模型 | `qwen3-vl:2b` |
+| `--embed-model <model>` | - | 嵌入模型 | `qwen3-embedding:latest` |
+| `--embedding-provider <p>` | - | 嵌入提供者: `ollama` / `openai` | `ollama` |
+| `--openai-api-key <key>` | - | OpenAI API Key | - |
+| `--openai-model <model>` | - | OpenAI 嵌入模型 | `text-embedding-3-small` |
+| `--db-path <path>` | - | LanceDB 数据库路径 | `~/.openclaw/multimodal-rag.lance` |
+| `--no-index-on-start` | - | 启动时不索引已有文件 | `false` |
+
 ### 手动配置
 
 如需自定义配置，编辑 `~/.openclaw/openclaw.json`：
@@ -100,18 +144,20 @@ openclaw multimodal-rag setup
 
 ### 配置项说明
 
-| 配置项 | 类型 | 默认值 | 说明 |
-|--------|------|--------|------|
-| `watchPaths` | string[] | `[]` | 监听的文件夹路径（支持 `~` 展开） |
-| `ollama.baseUrl` | string | `http://127.0.0.1:11434` | Ollama 服务地址 |
-| `ollama.visionModel` | string | `qwen3-vl:2b` | 用于图像描述的视觉模型 |
-| `ollama.embedModel` | string | `qwen3-embedding:latest` | 用于生成嵌入向量的模型 |
-| `embedding.provider` | string | `ollama` | 嵌入提供者: `ollama` 或 `openai` |
-| `embedding.openaiApiKey` | string | - | OpenAI API Key（仅 openai 时需要） |
-| `embedding.openaiModel` | string | `text-embedding-3-small` | OpenAI 嵌入模型 |
-| `dbPath` | string | `~/.openclaw/multimodal-rag.lance` | LanceDB 数据库路径 |
-| `watchDebounceMs` | number | `1000` | 文件监听去抖延迟（毫秒） |
-| `indexExistingOnStart` | boolean | `true` | 启动时是否索引已有文件 |
+
+| 配置项                      | 类型       | 默认值                                | 说明                           |
+| ------------------------ | -------- | ---------------------------------- | ---------------------------- |
+| `watchPaths`             | string[] | `[]`                               | 监听的文件夹路径（支持 `~` 展开）          |
+| `ollama.baseUrl`         | string   | `http://127.0.0.1:11434`           | Ollama 服务地址                  |
+| `ollama.visionModel`     | string   | `qwen3-vl:2b`                      | 用于图像描述的视觉模型                  |
+| `ollama.embedModel`      | string   | `qwen3-embedding:latest`           | 用于生成嵌入向量的模型                  |
+| `embedding.provider`     | string   | `ollama`                           | 嵌入提供者: `ollama` 或 `openai`   |
+| `embedding.openaiApiKey` | string   | -                                  | OpenAI API Key（仅 openai 时需要） |
+| `embedding.openaiModel`  | string   | `text-embedding-3-small`           | OpenAI 嵌入模型                  |
+| `dbPath`                 | string   | `~/.openclaw/multimodal-rag.lance` | LanceDB 数据库路径                |
+| `watchDebounceMs`        | number   | `1000`                             | 文件监听去抖延迟（毫秒）                 |
+| `indexExistingOnStart`   | boolean  | `true`                             | 启动时是否索引已有文件                  |
+
 
 配置完成后，重启 OpenClaw Gateway 使配置生效。
 
@@ -155,6 +201,9 @@ Agent：[调用 media_stats] → 总计 120 个文件，图片 80，音频 40
 # 交互式配置
 openclaw multimodal-rag setup
 
+# 非交互式配置
+openclaw multimodal-rag setup -n --watch ~/photos --watch ~/audio
+
 # 手动索引文件或文件夹
 openclaw multimodal-rag index ~/Pictures/photo.jpg
 
@@ -205,4 +254,4 @@ openclaw plugins list | grep multimodal-rag
 
 ## 许可证
 
-MIT
+MIT

package/index.ts

@@ -15,7 +15,7 @@ import {
   createMediaListTool,
   createMediaStatsTool,
 } from "./src/tools.js";
-import { runSetup } from "./src/setup.js";
+import { runSetup, runNonInteractiveSetup } from "./src/setup.js";
 import type { PluginConfig } from "./src/types.js";
 
 const multimodalRagPlugin = {
@@ -288,11 +288,51 @@ const multimodalRagPlugin = {
         });
 
       // openclaw multimodal-rag setup
+      // 支持交互式和非交互式两种模式：
+      //   交互式: openclaw multimodal-rag setup
+      //   非交互式: openclaw multimodal-rag setup -n --watch ~/photos --watch ~/audio
       rag
         .command("setup")
-        .description("交互式引导配置插件")
-        .action(async () => {
-          await runSetup();
+        .description("配置插件（支持交互式和非交互式模式）")
+        .option("-n, --non-interactive", "非交互式模式（需配合 --watch 使用）")
+        .option("-w, --watch <paths...>", "监听路径（可多次指定或逗号分隔）")
+        .option("--ollama-url <url>", "Ollama 服务地址", "http://127.0.0.1:11434")
+        .option("--vision-model <model>", "视觉模型名称", "qwen3-vl:2b")
+        .option("--embed-model <model>", "嵌入模型名称", "qwen3-embedding:latest")
+        .option("--embedding-provider <provider>", "嵌入提供者: ollama 或 openai", "ollama")
+        .option("--openai-api-key <key>", "OpenAI API Key（仅 openai 时需要）")
+        .option("--openai-model <model>", "OpenAI 嵌入模型")
+        .option("--db-path <path>", "LanceDB 数据库路径")
+        .option("--no-index-on-start", "启动时不索引已有文件")
+        .action(async (opts: {
+          nonInteractive?: boolean;
+          watch?: string[];
+          ollamaUrl?: string;
+          visionModel?: string;
+          embedModel?: string;
+          embeddingProvider?: string;
+          openaiApiKey?: string;
+          openaiModel?: string;
+          dbPath?: string;
+          noIndexOnStart?: boolean;
+        }) => {
+          if (opts.nonInteractive) {
+            // 非交互式模式：展开逗号分隔的路径
+            const watchPaths = (opts.watch || []).flatMap((p) => p.split(",").map((s) => s.trim()).filter(Boolean));
+            await runNonInteractiveSetup({
+              watch: watchPaths,
+              ollamaUrl: opts.ollamaUrl,
+              visionModel: opts.visionModel,
+              embedModel: opts.embedModel,
+              embeddingProvider: opts.embeddingProvider as "ollama" | "openai" | undefined,
+              openaiApiKey: opts.openaiApiKey,
+              openaiModel: opts.openaiModel,
+              dbPath: opts.dbPath,
+              noIndexOnStart: opts.noIndexOnStart,
+            });
+          } else {
+            await runSetup();
+          }
         });
     }, { commands: ["multimodal-rag"] });
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hzttt/multimodal-rag",
-  "version": "0.1.1",
+  "version": "0.1.2",
   "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/setup.ts

@@ -1,8 +1,8 @@
 /**
- * 交互式引导配置
+ * 引导配置（交互式 & 非交互式）
  *
- * 运行 `openclaw multimodal-rag setup` 时调用，
- * 引导用户完成必要配置并写入 openclaw 配置文件。
+ * 交互式: `openclaw multimodal-rag setup`
+ * 非交互式: `openclaw multimodal-rag setup --watch ~/photos,~/audio --non-interactive`
  */
 
 import * as readline from "node:readline/promises";
@@ -31,6 +31,19 @@ type PluginConfigPartial = {
   indexExistingOnStart?: boolean;
 };
 
+/** 非交互式 setup 的选项 */
+export type NonInteractiveSetupOpts = {
+  watch: string[];
+  ollamaUrl?: string;
+  visionModel?: string;
+  embedModel?: string;
+  embeddingProvider?: "ollama" | "openai";
+  openaiApiKey?: string;
+  openaiModel?: string;
+  dbPath?: string;
+  noIndexOnStart?: boolean;
+};
+
 function loadOpenClawConfig(): Record<string, unknown> {
   if (!fs.existsSync(CONFIG_FILE)) {
     return {};
@@ -54,6 +67,75 @@ function getExistingPluginConfig(config: Record<string, unknown>): PluginConfigP
   return (entry?.config as PluginConfigPartial) || {};
 }
 
+/**
+ * 将插件配置写入 openclaw.json
+ */
+function writePluginConfig(pluginConfig: PluginConfigPartial): void {
+  const config = loadOpenClawConfig();
+  const plugins = (config.plugins || {}) as Record<string, unknown>;
+  const entries = (plugins.entries || {}) as Record<string, unknown>;
+  const pluginEntry = (entries["multimodal-rag"] || {}) as Record<string, unknown>;
+
+  pluginEntry.enabled = true;
+  pluginEntry.config = pluginConfig;
+
+  entries["multimodal-rag"] = pluginEntry;
+  plugins.entries = entries;
+  config.plugins = plugins;
+
+  saveOpenClawConfig(config);
+}
+
+/**
+ * 非交互式配置
+ *
+ * 适用于脚本自动化、SSH 远程部署等场景。
+ * 所有参数通过命令行选项传入，不读取 stdin。
+ *
+ * 用法:
+ *   openclaw multimodal-rag setup --non-interactive --watch ~/photos,~/audio
+ *   openclaw multimodal-rag setup -n -w ~/photos -w ~/audio --ollama-url http://host:11434
+ */
+export async function runNonInteractiveSetup(opts: NonInteractiveSetupOpts): Promise<void> {
+  if (opts.watch.length === 0) {
+    console.error("✗ 非交互式模式必须通过 --watch 指定至少一个监听路径");
+    console.error("  示例: openclaw multimodal-rag setup --non-interactive --watch ~/photos");
+    process.exit(1);
+  }
+
+  const existing = getExistingPluginConfig(loadOpenClawConfig());
+
+  const pluginConfig: PluginConfigPartial = {
+    watchPaths: opts.watch,
+    ollama: {
+      baseUrl: opts.ollamaUrl || existing.ollama?.baseUrl || "http://127.0.0.1:11434",
+      visionModel: opts.visionModel || existing.ollama?.visionModel || "qwen3-vl:2b",
+      embedModel: opts.embedModel || existing.ollama?.embedModel || "qwen3-embedding:latest",
+    },
+    embedding: {
+      provider: opts.embeddingProvider || existing.embedding?.provider || "ollama",
+      ...(opts.openaiApiKey && { openaiApiKey: opts.openaiApiKey }),
+      ...(opts.openaiModel && { openaiModel: opts.openaiModel }),
+    },
+    dbPath: opts.dbPath || existing.dbPath || "~/.openclaw/multimodal-rag.lance",
+    indexExistingOnStart: opts.noIndexOnStart ? false : (existing.indexExistingOnStart !== false),
+  };
+
+  writePluginConfig(pluginConfig);
+
+  console.log("✓ 配置已保存到 ~/.openclaw/openclaw.json\n");
+  console.log("配置摘要:");
+  console.log(`  监听路径:     ${pluginConfig.watchPaths!.join(", ")}`);
+  console.log(`  Ollama 地址:  ${pluginConfig.ollama!.baseUrl}`);
+  console.log(`  视觉模型:     ${pluginConfig.ollama!.visionModel}`);
+  console.log(`  嵌入模型:     ${pluginConfig.ollama!.embedModel}`);
+  console.log(`  嵌入提供者:   ${pluginConfig.embedding!.provider}`);
+  console.log(`  数据库路径:   ${pluginConfig.dbPath}`);
+  console.log(`  启动时索引:   ${pluginConfig.indexExistingOnStart ? "是" : "否"}`);
+  console.log();
+  console.log("提示: 重启 OpenClaw Gateway 以加载新配置");
+}
+
 async function checkOllamaHealth(baseUrl: string): Promise<boolean> {
   try {
     const res = await fetch(`${baseUrl}/api/tags`, { signal: AbortSignal.timeout(5000) });
@@ -148,19 +230,7 @@ export async function runSetup(): Promise<void> {
   // ================================================================
   console.log("\n── 保存配置 ──\n");
 
-  // 深合并到 openclaw 配置
-  const plugins = (config.plugins || {}) as Record<string, unknown>;
-  const entries = (plugins.entries || {}) as Record<string, unknown>;
-  const pluginEntry = (entries["multimodal-rag"] || {}) as Record<string, unknown>;
-
-  pluginEntry.enabled = true;
-  pluginEntry.config = pluginConfig;
-
-  entries["multimodal-rag"] = pluginEntry;
-  plugins.entries = entries;
-  config.plugins = plugins;
-
-  saveOpenClawConfig(config);
+  writePluginConfig(pluginConfig);
 
   console.log("✓ 配置已保存到 ~/.openclaw/openclaw.json\n");