266 lines
8.9 KiB
JavaScript
266 lines
8.9 KiB
JavaScript
/**
|
|
* Copyright (c) 2024-2025 Тарабанов Александр Викторович
|
|
* All rights reserved.
|
|
*
|
|
* This software is proprietary and confidential.
|
|
* Unauthorized copying, modification, or distribution is prohibited.
|
|
*
|
|
* For licensing inquiries: info@hb3-accelerator.com
|
|
* Website: https://hb3-accelerator.com
|
|
* GitHub: https://github.com/HB3-ACCELERATOR
|
|
*/
|
|
|
|
const encryptedDb = require('./encryptedDatabaseService');
|
|
const OpenAI = require('openai');
|
|
const Anthropic = require('@anthropic-ai/sdk');
|
|
|
|
const TABLE = 'ai_providers_settings';
|
|
|
|
async function getProviderSettings(provider) {
|
|
const settings = await encryptedDb.getData(TABLE, { provider: provider }, 1);
|
|
return settings[0] || null;
|
|
}
|
|
|
|
async function upsertProviderSettings({ provider, api_key, base_url, selected_model, embedding_model }) {
|
|
const data = {
|
|
provider: provider,
|
|
api_key: api_key,
|
|
base_url: base_url,
|
|
selected_model: selected_model,
|
|
embedding_model: embedding_model,
|
|
updated_at: new Date()
|
|
};
|
|
|
|
// Проверяем, существует ли запись
|
|
const existing = await encryptedDb.getData(TABLE, { provider: provider }, 1);
|
|
|
|
if (existing.length > 0) {
|
|
// Обновляем существующую запись по ID
|
|
return await encryptedDb.saveData(TABLE, data, { id: existing[0].id });
|
|
} else {
|
|
// Создаем новую запись
|
|
return await encryptedDb.saveData(TABLE, data);
|
|
}
|
|
}
|
|
|
|
async function deleteProviderSettings(provider) {
|
|
await encryptedDb.deleteData(TABLE, { provider: provider });
|
|
}
|
|
|
|
async function getProviderModels(provider, { api_key, base_url } = {}) {
|
|
try {
|
|
if (provider === 'openai') {
|
|
const client = new OpenAI({ apiKey: api_key, baseURL: base_url });
|
|
const res = await client.models.list();
|
|
return res.data ? res.data.map(m => ({ id: m.id, ...m })) : [];
|
|
}
|
|
if (provider === 'anthropic') {
|
|
const client = new Anthropic({ apiKey: api_key, baseURL: base_url });
|
|
const res = await client.models.list();
|
|
return res.data ? res.data.map(m => ({ id: m.id, ...m })) : [];
|
|
}
|
|
if (provider === 'google') {
|
|
const { GoogleGenAI } = await import('@google/genai');
|
|
const ai = new GoogleGenAI({ apiKey: api_key, baseUrl: base_url });
|
|
const pager = await ai.models.list();
|
|
const models = [];
|
|
for await (const model of pager) {
|
|
models.push(model);
|
|
}
|
|
return models;
|
|
}
|
|
if (provider === 'ollama') {
|
|
// Для Ollama — через ai-assistant.js
|
|
return [];
|
|
}
|
|
return [];
|
|
} catch (error) {
|
|
return [];
|
|
}
|
|
}
|
|
|
|
async function verifyProviderKey(provider, { api_key, base_url } = {}) {
|
|
try {
|
|
if (provider === 'openai') {
|
|
const client = new OpenAI({ apiKey: api_key, baseURL: base_url });
|
|
await client.models.list();
|
|
return { success: true };
|
|
}
|
|
if (provider === 'anthropic') {
|
|
const client = new Anthropic({ apiKey: api_key, baseURL: base_url });
|
|
await client.models.list();
|
|
return { success: true };
|
|
}
|
|
if (provider === 'google') {
|
|
const { GoogleGenAI } = await import('@google/genai');
|
|
const ai = new GoogleGenAI({ apiKey: api_key, baseUrl: base_url });
|
|
const pager = await ai.models.list();
|
|
for await (const _ of pager) {
|
|
break;
|
|
}
|
|
return { success: true };
|
|
}
|
|
if (provider === 'ollama') {
|
|
// Для Ollama — всегда true (локальный)
|
|
return { success: true };
|
|
}
|
|
return { success: false, error: 'Unknown provider' };
|
|
} catch (error) {
|
|
return { success: false, error: error.message };
|
|
}
|
|
}
|
|
|
|
async function getAllLLMModels() {
|
|
try {
|
|
// Получаем все настройки провайдеров
|
|
const providers = await encryptedDb.getData(TABLE, {});
|
|
|
|
// Собираем все модели из всех провайдеров
|
|
const allModels = [];
|
|
|
|
for (const provider of providers) {
|
|
if (provider.selected_model) {
|
|
// Фильтруем embedding модели - они не должны быть в списке LLM
|
|
const modelName = provider.selected_model.toLowerCase();
|
|
const isEmbeddingModel = modelName.includes('embed') ||
|
|
modelName.includes('embedding') ||
|
|
modelName.includes('bge') ||
|
|
modelName.includes('nomic') ||
|
|
modelName.includes('text-embedding') ||
|
|
modelName.includes('mxbai') ||
|
|
modelName.includes('sentence') ||
|
|
modelName.includes('ada-002') ||
|
|
modelName.includes('text-embedding-ada') ||
|
|
modelName.includes('text-embedding-3');
|
|
|
|
if (!isEmbeddingModel) {
|
|
allModels.push({
|
|
id: provider.selected_model,
|
|
provider: provider.provider
|
|
});
|
|
}
|
|
}
|
|
}
|
|
|
|
// Для Ollama проверяем реально установленные модели через HTTP API
|
|
try {
|
|
const axios = require('axios');
|
|
const ollamaUrl = process.env.OLLAMA_BASE_URL || 'http://ollama:11434';
|
|
|
|
const response = await axios.get(`${ollamaUrl}/api/tags`, {
|
|
timeout: 5000
|
|
});
|
|
|
|
const models = response.data.models || [];
|
|
for (const model of models) {
|
|
// Фильтруем embedding модели из Ollama
|
|
const modelName = model.name.toLowerCase();
|
|
const isEmbeddingModel = modelName.includes('embed') ||
|
|
modelName.includes('embedding') ||
|
|
modelName.includes('bge') ||
|
|
modelName.includes('nomic') ||
|
|
modelName.includes('mxbai') ||
|
|
modelName.includes('sentence');
|
|
|
|
if (!isEmbeddingModel) {
|
|
allModels.push({
|
|
id: model.name,
|
|
provider: 'ollama'
|
|
});
|
|
}
|
|
}
|
|
} catch (ollamaError) {
|
|
// console.error('Error checking Ollama models:', ollamaError);
|
|
// Если не удалось проверить Ollama, добавляем базовые модели
|
|
allModels.push({ id: 'qwen2.5:7b', provider: 'ollama' });
|
|
}
|
|
|
|
// Убираем дубликаты
|
|
const uniqueModels = [];
|
|
const seen = new Set();
|
|
|
|
for (const model of allModels) {
|
|
const key = `${model.id}-${model.provider}`;
|
|
if (!seen.has(key)) {
|
|
seen.add(key);
|
|
uniqueModels.push(model);
|
|
}
|
|
}
|
|
|
|
return uniqueModels;
|
|
} catch (error) {
|
|
// console.error('Error getting LLM models:', error);
|
|
return [];
|
|
}
|
|
}
|
|
|
|
async function getAllEmbeddingModels() {
|
|
try {
|
|
// Получаем все настройки провайдеров
|
|
const providers = await encryptedDb.getData(TABLE, {});
|
|
|
|
// Собираем все embedding модели из всех провайдеров
|
|
const allModels = [];
|
|
|
|
for (const provider of providers) {
|
|
if (provider.embedding_model) {
|
|
allModels.push({
|
|
id: provider.embedding_model,
|
|
provider: provider.provider
|
|
});
|
|
}
|
|
}
|
|
|
|
// Для Ollama проверяем реально установленные embedding модели через HTTP API
|
|
try {
|
|
const axios = require('axios');
|
|
const ollamaUrl = process.env.OLLAMA_BASE_URL || 'http://ollama:11434';
|
|
|
|
const response = await axios.get(`${ollamaUrl}/api/tags`, {
|
|
timeout: 5000
|
|
});
|
|
|
|
const models = response.data.models || [];
|
|
for (const model of models) {
|
|
// Проверяем, что это embedding модель
|
|
if (model.name.includes('embed') || model.name.includes('bge') || model.name.includes('nomic')) {
|
|
allModels.push({
|
|
id: model.name,
|
|
provider: 'ollama'
|
|
});
|
|
}
|
|
}
|
|
} catch (ollamaError) {
|
|
// console.error('Error checking Ollama embedding models:', ollamaError);
|
|
// Если не удалось проверить Ollama, добавляем базовые embedding модели
|
|
allModels.push({ id: 'mxbai-embed-large:latest', provider: 'ollama' });
|
|
}
|
|
|
|
// Убираем дубликаты
|
|
const uniqueModels = [];
|
|
const seen = new Set();
|
|
|
|
for (const model of allModels) {
|
|
const key = `${model.id}-${model.provider}`;
|
|
if (!seen.has(key)) {
|
|
seen.add(key);
|
|
uniqueModels.push(model);
|
|
}
|
|
}
|
|
|
|
return uniqueModels;
|
|
} catch (error) {
|
|
// console.error('Error getting embedding models:', error);
|
|
return [];
|
|
}
|
|
}
|
|
|
|
module.exports = {
|
|
getProviderSettings,
|
|
upsertProviderSettings,
|
|
deleteProviderSettings,
|
|
getProviderModels,
|
|
verifyProviderKey,
|
|
getAllLLMModels,
|
|
getAllEmbeddingModels,
|
|
};
|