multi-llm
Version:
A unified TypeScript/JavaScript package to use LLMs across ALL platforms with support for 17 major providers, streaming, MCP tools, and intelligent response parsing
151 lines • 6 kB
JavaScript
;
var __importDefault = (this && this.__importDefault) || function (mod) {
return (mod && mod.__esModule) ? mod : { "default": mod };
};
Object.defineProperty(exports, "__esModule", { value: true });
exports.AzureProvider = void 0;
const axios_1 = __importDefault(require("axios"));
const provider_1 = require("../provider");
const llm_1 = require("../llm");
const parser_1 = require("../utils/parser");
class AzureProvider extends provider_1.Provider {
constructor(apiKey, baseUrl, apiVersion = '2024-02-15-preview') {
super(apiKey, baseUrl);
if (!baseUrl) {
throw new Error('Azure baseUrl is required (e.g., https://your-resource.openai.azure.com)');
}
this.baseUrl = baseUrl;
this.apiVersion = apiVersion;
}
async getModels() {
try {
const response = await axios_1.default.get(`${this.baseUrl}/openai/models?api-version=${this.apiVersion}`, {
headers: {
'api-key': this.apiKey,
'Content-Type': 'application/json'
}
});
return response.data.data.map((model) => ({
id: model.id,
name: model.id,
contextWindow: this.getContextWindow(model.id),
maxOutputTokens: this.getMaxOutputTokens(model.id),
pricing: this.getPricing(model.id)
}));
}
catch (error) {
throw new Error(`Failed to fetch Azure OpenAI models: ${error}`);
}
}
createLLM(modelId) {
return new llm_1.LLM(this, modelId);
}
async chat(modelId, messages, options, streamCallback) {
try {
const payload = {
messages: messages.map(msg => ({ role: msg.role, content: msg.content })),
temperature: options.temperature,
max_tokens: options.maxTokens,
top_p: options.topP,
stream: !!streamCallback
};
const url = `${this.baseUrl}/openai/deployments/${modelId}/chat/completions?api-version=${this.apiVersion}`;
if (streamCallback) {
return this.streamChat(url, payload, streamCallback);
}
else {
const response = await axios_1.default.post(url, payload, {
headers: {
'api-key': this.apiKey,
'Content-Type': 'application/json'
}
});
const content = response.data.choices[0].message.content;
const parsed = parser_1.ResponseParser.parseResponse(content);
return {
raw: response.data,
parsed,
usage: {
inputTokens: response.data.usage?.prompt_tokens || 0,
outputTokens: response.data.usage?.completion_tokens || 0,
totalTokens: response.data.usage?.total_tokens || 0
}
};
}
}
catch (error) {
throw new Error(`Azure OpenAI chat failed: ${error}`);
}
}
async streamChat(url, payload, streamCallback) {
return new Promise((resolve, reject) => {
let fullContent = '';
let rawResponse = null;
const source = axios_1.default.post(url, payload, {
headers: {
'api-key': this.apiKey,
'Content-Type': 'application/json'
},
responseType: 'stream'
});
source.then(response => {
response.data.on('data', (chunk) => {
const lines = chunk.toString().split('\n');
for (const line of lines) {
if (line.startsWith('data: ')) {
const data = line.slice(6);
if (data === '[DONE]') {
const parsed = parser_1.ResponseParser.parseResponse(fullContent);
resolve({
raw: rawResponse,
parsed,
usage: undefined
});
return;
}
try {
const parsed = JSON.parse(data);
rawResponse = parsed;
const content = parsed.choices[0]?.delta?.content || '';
if (content) {
fullContent += content;
streamCallback(content);
}
}
catch (e) {
// Ignore parsing errors for partial chunks
}
}
}
});
response.data.on('error', reject);
}).catch(reject);
});
}
getContextWindow(modelId) {
const contextWindows = {
'gpt-4o': 128000,
'gpt-4o-mini': 128000,
'gpt-4-turbo': 128000,
'gpt-4': 8192,
'gpt-35-turbo': 16385
};
return contextWindows[modelId] || 8192;
}
getMaxOutputTokens(modelId) {
const maxOutputTokens = {
'gpt-4o': 4096,
'gpt-4o-mini': 16384,
'gpt-4-turbo': 4096,
'gpt-4': 4096,
'gpt-35-turbo': 4096
};
return maxOutputTokens[modelId] || 4096;
}
getPricing(modelId) {
// Azure pricing varies by region and deployment, so we return undefined
return undefined;
}
}
exports.AzureProvider = AzureProvider;
//# sourceMappingURL=azure.js.map