@xagent-ai/cli 1.2.2 → 1.3.1

package/src/ai-client/providers/remote.ts

@@ -0,0 +1,439 @@
+import axios, { AxiosInstance } from 'axios';
+import https from 'https';
+import type {
+  AIProvider,
+  RemoteAIProvider,
+  RemoteConfig,
+  Message,
+  CompletionOptions,
+  CompletionResponse,
+  StreamEvent,
+  Model,
+  RemoteModelsResponse,
+} from '../types';
+import { DEFAULT_RETRY_CONFIG } from '../types.js';
+import { withRetry, RetryConfig } from '../../retry.js';
+
+// ============================================================================
+// Remote Provider (xAgent Web Service)
+// ============================================================================
+
+/**
+ * Remote Provider - communicates with xAgent Web Service
+ * Used for cloud-based AI processing
+ */
+export class RemoteProvider implements AIProvider {
+  readonly type = 'remote';
+
+  private authToken: string;
+  private webBaseUrl: string;
+  private agentApi: string;
+  private vlmApi: string;
+  private showDebugInfo: boolean;
+  private client: AxiosInstance;
+
+  constructor(config: RemoteConfig) {
+    this.authToken = config.authToken || '';
+    this.webBaseUrl = (config.baseUrl || '').replace(/\/$/, '');
+    this.agentApi = `${this.webBaseUrl}/api/agent`;
+    this.vlmApi = `${this.webBaseUrl}/api/agent/vlm`;
+    this.showDebugInfo = config.showDebugInfo ?? false;
+
+    this.client = axios.create({
+      timeout: 300000,
+      httpsAgent: new https.Agent({ rejectUnauthorized: false }),
+    });
+  }
+
+  /**
+   * Get available models from remote service
+   * Implements AIProvider interface - returns LLM models only
+   */
+  async getModels(): Promise<Model[]> {
+    try {
+      const response = await this.client.get(`${this.webBaseUrl}/api/models`, {
+        headers: this.getHeaders(),
+        timeout: 10000,
+      });
+
+      const data = response.data as RemoteModelsResponse;
+      return data.llm || [];
+    } catch {
+      // Return default models if fetch fails
+      return REMOTE_MODELS;
+    }
+  }
+
+  /**
+   * Get available models from remote service including both LLM and VLM
+   * Extended method for RemoteAIProvider interface
+   */
+  async getRemoteModels(): Promise<RemoteModelsResponse> {
+    try {
+      const response = await this.client.get(`${this.webBaseUrl}/api/models`, {
+        headers: this.getHeaders(),
+        timeout: 10000,
+      });
+
+      const data = response.data as RemoteModelsResponse;
+      return {
+        llm: data.llm || [],
+        vlm: data.vlm || [],
+      };
+    } catch {
+      // Return default models if fetch fails
+      return {
+        llm: REMOTE_MODELS,
+        vlm: [],
+      };
+    }
+  }
+
+  /**
+   * Non-streaming completion
+   */
+  async complete(
+    messages: Message[],
+    options?: CompletionOptions
+  ): Promise<CompletionResponse> {
+    const showDebug = this.showDebugInfo;
+
+    if (showDebug) {
+      this.debugRequest(messages);
+    }
+
+    // Build request body matching original implementation
+    const requestBody = {
+      messages,
+      taskId: options?.taskId,
+      status: options?.status || 'begin',
+      conversationId: options?.conversationId,
+      context: options?.context,
+      toolResults: options?.toolResults,
+      tools: options?.tools,
+      options: {
+        llmModelName: options?.llmModelName || options?.model,
+        vlmModelName: options?.vlmModelName,
+      },
+    };
+
+    const retryConfig: RetryConfig = { ...DEFAULT_RETRY_CONFIG };
+
+    const result = await withRetry(async () => {
+      const response = await this.client.post(
+        `${this.agentApi}/chat`,
+        requestBody,
+        {
+          headers: this.getHeaders(),
+        }
+      );
+
+      const data = response.data;
+
+      if (showDebug) {
+        this.debugResponse(data);
+      }
+
+      const completionResponse: CompletionResponse = {
+        id: data?.id || `remote-${Date.now()}`,
+        object: 'chat.completion',
+        created: Date.now(),
+        model: options?.model || 'remote-llm',
+        choices: [{
+          index: 0,
+          message: {
+            role: 'assistant',
+            content: data?.message || data?.content || '',
+            reasoning_content: data?.reasoning_content || data?.reasoningContent || '',
+            tool_calls: data?.tool_calls || data?.toolCalls,
+          },
+          finish_reason: 'stop',
+        }],
+        usage: undefined,
+      };
+      return completionResponse;
+    }, retryConfig);
+
+    if (result.success) {
+      return result.data!;
+    }
+    throw result.error || new Error('Remote API request failed after retries');
+  }
+
+  /**
+   * Streaming completion - remote provider doesn't support streaming yet
+   * Falls back to non-streaming with incremental yield
+   */
+  async *stream(
+    messages: Message[],
+    _options?: CompletionOptions
+  ): AsyncIterable<StreamEvent> {
+    // Remote provider doesn't support streaming, yield full response
+    const response = await this.complete(messages);
+
+    const content = response.choices[0]?.message.content;
+    const textContent = typeof content === 'string' ? content : '';
+
+    for (const char of textContent) {
+      yield { type: 'text_delta', delta: char };
+    }
+
+    const reasoningContent = response.choices[0]?.message.reasoning_content;
+    if (reasoningContent) {
+      for (const char of reasoningContent) {
+        yield { type: 'reasoning_delta', delta: char };
+      }
+    }
+
+    yield { type: 'done', reason: 'stop' };
+  }
+
+  /**
+   * Abort ongoing request - not implemented for remote
+   */
+  abort(): void {
+    // Remote requests cannot be aborted from client side
+  }
+
+  /**
+   * Close provider
+   */
+  async close(): Promise<void> {
+    // No resources to close
+  }
+
+  // --------------------------------------------------------------------------
+  // Remote Task Management Methods
+  // --------------------------------------------------------------------------
+
+  /**
+   * Complete a remote task
+   */
+  async completeTask(taskId: string): Promise<void> {
+    await this.makeTaskRequest('end', taskId);
+  }
+
+  /**
+   * Cancel a remote task
+   */
+  async cancelTask(taskId: string): Promise<void> {
+    await this.makeTaskRequest('cancel', taskId);
+  }
+
+  /**
+   * Fail a remote task
+   */
+  async failTask(taskId: string, reason: 'timeout' | 'failure'): Promise<void> {
+    await this.makeTaskRequest(reason, taskId);
+  }
+
+  /**
+   * Helper for task management requests - uses /chat endpoint like original implementation
+   */
+  private async makeTaskRequest(status: string, taskId: string): Promise<void> {
+    const url = `${this.agentApi}/chat`;
+    const requestBody = {
+      taskId,
+      status,
+      messages: [],
+      options: {}
+    };
+
+    const httpsAgent = new https.Agent({ rejectUnauthorized: false });
+    try {
+      await this.client.post(url, requestBody, {
+        headers: {
+          ...this.getHeaders(), // Include xagent-cli-version and other headers
+        },
+        httpsAgent
+      });
+    } catch (error: any) {
+      throw this.handleError(error);
+    }
+  }
+
+  // --------------------------------------------------------------------------
+
+  /**
+   * Get request headers
+   */
+  private getHeaders(): Record<string, string> {
+    return {
+      'Content-Type': 'application/json',
+      'Authorization': `Bearer ${this.authToken}`,
+      'xagent-cli-version': '1.0',
+    };
+  }
+
+  /**
+   * Debug request output
+   */
+  private debugRequest(messages: Message[]): void {
+    console.log('\n╔══════════════════════════════════════════════════════════╗');
+    console.log('║              AI REQUEST DEBUG (REMOTE)                   ║');
+    console.log('╚══════════════════════════════════════════════════════════╝');
+    console.log(`🌐 Base URL: ${this.webBaseUrl}`);
+    console.log(`💬 Messages: ${messages.length}`);
+    console.log('─'.repeat(60));
+    console.log('\n📤 Sending request to Remote API...\n');
+  }
+
+  /**
+   * Debug response output
+   */
+  private debugResponse(data: any): void {
+    console.log('\n╔══════════════════════════════════════════════════════════╗');
+    console.log('║             AI RESPONSE DEBUG (REMOTE)                   ║');
+    console.log('╚══════════════════════════════════════════════════════════╝');
+    console.log(`🆔 ID: remote-${Date.now()}`);
+    console.log(`📏 Content length: ${data.content?.length || 0}`);
+    if (data.tool_calls?.length) {
+      console.log(`🔧 Tool calls: ${data.tool_calls.length}`);
+    }
+    console.log('\n📥 Response received.\n');
+  }
+
+  /**
+   * Generic request helper for task management
+   */
+  private async makeRequest(endpoint: string, method: 'GET' | 'POST' | 'PUT' | 'DELETE', data?: any): Promise<any> {
+    const url = `${this.agentApi}${endpoint}`;
+    try {
+      const response = await this.client.request({
+        method,
+        url,
+        data,
+        headers: this.getHeaders(),
+      });
+      return response.data;
+    } catch (error: any) {
+      throw this.handleError(error);
+    }
+  }
+
+  /**
+   * Handle errors
+   */
+  private handleError(error: any): Error {
+    if (error.response) {
+      const status = error.response.status;
+      let message = `Remote API Error: ${status}`;
+
+      if (status === 401) {
+        message = 'Remote API Error: Authentication token is invalid or expired. Please log in again.';
+      } else if (status === 429) {
+        message = 'Remote API Error: Rate limit exceeded. Please try again later.';
+      } else if (status === 500) {
+        message = 'Remote API Error: Server error. Please try again later.';
+      } else if (status === 503) {
+        message = 'Remote API Error: Service temporarily unavailable. Please try again.';
+      }
+
+      return new Error(message);
+    }
+
+    if (error.request) {
+      return new Error('Remote API Error: Network error - No response received');
+    }
+
+    return new Error(`Remote API Error: ${error.message}`);
+  }
+
+  /**
+   * Invoke VLM (Vision Language Model) for image analysis
+   * @param messages - Full messages array (consistent with local mode)
+   * @param _systemPrompt - System prompt (optional, for reference)
+   * @param remoteChatOptions - Other options including AbortSignal, taskId
+   */
+  async invokeVLM(
+    messages: Message[],
+    _systemPrompt: string,
+    remoteChatOptions: {
+      taskId?: string;
+      status?: 'begin' | 'continue';
+      context?: { cwd?: string; workspace?: string };
+      signal?: AbortSignal;
+    } = {}
+  ): Promise<string> {
+    // Forward complete messages to backend (same format as local mode)
+    const requestBody = {
+      messages,
+      taskId: remoteChatOptions.taskId,
+      status: remoteChatOptions.status || 'begin',
+      context: remoteChatOptions.context,
+    };
+
+    // Handle abort signal
+    const controller = remoteChatOptions.signal ? new AbortController() : undefined;
+    const abortSignal = remoteChatOptions.signal || controller?.signal;
+    if (remoteChatOptions.signal) {
+      remoteChatOptions.signal.addEventListener?.('abort', () => controller?.abort());
+    }
+
+    try {
+      const response = await axios.post(this.vlmApi, requestBody, {
+        headers: {
+          'Content-Type': 'application/json',
+          'Authorization': `Bearer ${this.authToken}`,
+        },
+        signal: abortSignal,
+        httpsAgent: new https.Agent({ rejectUnauthorized: false }),
+        timeout: 120000,
+      });
+
+      const data = response.data as { content?: string };
+      return data.content || '';
+    } catch (error: any) {
+      throw this.handleError(error);
+    }
+  }
+}
+
+// ============================================================================
+// Remote Models
+// ============================================================================
+
+const REMOTE_MODELS: Model[] = [
+  { id: 'remote-llm', name: 'Remote LLM', provider: 'remote', supportsStreaming: false },
+];
+
+// ============================================================================
+// Factory Function
+// ============================================================================
+
+/**
+ * Create Remote provider instance
+ */
+export function createRemoteProvider(config: RemoteConfig): RemoteAIProvider {
+  return new RemoteProvider(config);
+}
+
+// ============================================================================
+// Static Methods
+// ============================================================================
+
+export interface RemoteModelInfo {
+  name: string;
+  displayName: string;
+}
+
+export interface RemoteDefaultModels {
+  llm: RemoteModelInfo;
+  vlm: RemoteModelInfo;
+}
+
+/**
+ * Fetch default models from remote service (static method for initialization)
+ */
+export async function fetchDefaultModels(authToken: string, baseUrl: string): Promise<RemoteDefaultModels> {
+  const url = `${baseUrl}/api/models/default`;
+  const httpsAgent = new https.Agent({ rejectUnauthorized: false });
+
+  const response = await axios.get(url, {
+    headers: { 'Authorization': `Bearer ${authToken}` },
+    httpsAgent,
+    timeout: 10000
+  });
+
+  return response.data;
+}

package/src/ai-client/registry.ts

@@ -0,0 +1,97 @@
+import type { AIProvider, AIConfig, Model, ProviderRegistryEntry } from './types.js';
+
+// ============================================================================
+// Provider Registry
+// ============================================================================
+
+/**
+ * Global registry for AI providers
+ */
+class ProviderRegistry {
+  private providers = new Map<string, ProviderRegistryEntry>();
+
+  /**
+   * Register a provider
+   */
+  register(type: string, entry: ProviderRegistryEntry): void {
+    this.providers.set(type, entry);
+  }
+
+  /**
+   * Get provider entry
+   */
+  get(type: string): ProviderRegistryEntry | undefined {
+    return this.providers.get(type);
+  }
+
+  /**
+   * Check if provider is registered
+   */
+  has(type: string): boolean {
+    return this.providers.has(type);
+  }
+
+  /**
+   * Get all registered types
+   */
+  keys(): string[] {
+    return Array.from(this.providers.keys());
+  }
+
+  /**
+   * Get all models from all registered providers
+   */
+  getAllModels(): Model[] {
+    const allModels: Model[] = [];
+    for (const entry of this.providers.values()) {
+      allModels.push(...entry.models);
+    }
+    return allModels;
+  }
+}
+
+/**
+ * Global registry instance
+ */
+export const registry = new ProviderRegistry();
+
+// ============================================================================
+// Registration Helpers
+// ============================================================================
+
+/**
+ * Decorator to register a provider
+ */
+export function registerProvider(
+  type: string,
+  models: Model[]
+) {
+  return function <T extends { new (...args: any[]): AIProvider }>(constructor: T) {
+    registry.register(type, {
+      create: (config: AIConfig) => new constructor(config),
+      models,
+    });
+    return constructor;
+  };
+}
+
+/**
+ * Get provider entry
+ */
+export function getProviderEntry(type: string): ProviderRegistryEntry | undefined {
+  return registry.get(type);
+}
+
+/**
+ * List all registered provider types
+ */
+export function listProviderTypes(): string[] {
+  return registry.keys();
+}
+
+/**
+ * Get all available models
+ */
+export function getAllModels(): Model[] {
+  return registry.getAllModels();
+}