managarten/chat/apps/mobile/utils/backendApi.ts
Till-JS c638a7ffee feat(chat): integrate chat project into monorepo with full app structure
- Restructure chat as apps/mobile, apps/web, apps/landing, backend
- Add NestJS backend for secure Azure OpenAI API calls
- Remove exposed API key from mobile app (security fix)
- Add shared chat-types package
- Create SvelteKit web app scaffold
- Create Astro landing page scaffold
- Update pnpm workspace configuration
- Add project-level CLAUDE.md documentation

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-25 13:48:24 +01:00

132 lines
3 KiB
TypeScript

/**
* Backend API Client
* Handles all communication with the NestJS backend
* API keys are stored securely on the backend - not in the mobile app
*/
import { ChatMessage } from '../services/openai';
const BACKEND_URL = process.env.EXPO_PUBLIC_BACKEND_URL || 'http://localhost:3001';
// Token usage type
export type TokenUsage = {
prompt_tokens: number;
completion_tokens: number;
total_tokens: number;
};
// Chat completion response
export type ChatCompletionResponse = {
content: string;
usage: TokenUsage;
};
// AI Model type from backend
export type AIModel = {
id: string;
name: string;
description: string;
parameters: {
temperature: number;
max_tokens: number;
provider: string;
deployment: string;
endpoint: string;
api_version: string;
};
};
/**
* Fetches available AI models from the backend
*/
export async function getAvailableModels(): Promise<AIModel[]> {
try {
const response = await fetch(`${BACKEND_URL}/api/chat/models`, {
method: 'GET',
headers: {
'Content-Type': 'application/json',
},
});
if (!response.ok) {
throw new Error(`Failed to fetch models: ${response.status}`);
}
return await response.json();
} catch (error) {
console.error('Error fetching models from backend:', error);
throw error;
}
}
/**
* Sends a chat completion request to the backend
* The backend securely handles the Azure OpenAI API call
*/
export async function sendChatCompletion(
messages: ChatMessage[],
modelId: string,
temperature?: number,
maxTokens?: number
): Promise<ChatCompletionResponse> {
console.log('Sending chat request to backend:', {
messagesCount: messages.length,
modelId,
temperature,
maxTokens,
backendUrl: BACKEND_URL,
});
try {
const response = await fetch(`${BACKEND_URL}/api/chat/completions`, {
method: 'POST',
headers: {
'Content-Type': 'application/json',
},
body: JSON.stringify({
messages: messages.map((msg) => ({
role: msg.role,
content: msg.content,
})),
modelId,
temperature,
maxTokens,
}),
});
if (!response.ok) {
const errorText = await response.text();
console.error('Backend API error:', response.status, errorText);
throw new Error(`Backend API error: ${response.status} - ${errorText}`);
}
const data = await response.json();
console.log('Backend response received:', {
contentLength: data.content?.length,
usage: data.usage,
});
return {
content: data.content,
usage: data.usage,
};
} catch (error) {
console.error('Error sending chat request to backend:', error);
throw error;
}
}
/**
* Health check for the backend
*/
export async function checkBackendHealth(): Promise<boolean> {
try {
const response = await fetch(`${BACKEND_URL}/api/health`, {
method: 'GET',
});
return response.ok;
} catch (error) {
console.error('Backend health check failed:', error);
return false;
}
}