huxy-llm-api 1.0.0-beta.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 yiru
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,212 @@
+# huxy-llm-api
+
+一个简洁、易用的用于简化 Ollama 和 OpenAI API 调用的 Node.js 库。支持流式响应和自定义配置，适用于快速集成大语言模型服务。
+
+## 特性
+
+- **统一接口**：提供一致的 API 调用方式，支持 Ollama 和 OpenAI 两种服务
+- **流式支持**：内置流式响应处理，支持实时回调
+- **灵活配置**：支持自定义模型参数和 API 配置
+- **错误处理**：内置参数验证和错误处理机制
+- **多功能支持**：支持聊天、文本生成、响应处理等多种 AI 功能
+- **自定义 Fetch**：使用 `Undici` 实现高性能 HTTP 请求
+
+## 安装
+
+```bash
+npm install huxy-llm-api
+# 或
+pnpm add huxy-llm-api
+# 或
+yarn add huxy-llm-api
+```
+
+## 快速开始
+
+### 基本用法
+
+```javascript
+import startApi from 'huxy-llm-api';
+
+// 初始化 Ollama API
+const ollamaApi = startApi('ollama', {
+  apiKey: 'your-api-key',
+  host: 'http://localhost:11434',
+});
+
+// 初始化 OpenAI API
+const openaiApi = startApi('openai', {
+  apiKey: 'your-api-key',
+  baseURL: 'https://api.openai.com/v1',
+});
+```
+
+### 调用示例
+
+#### Ollama - 生成文本
+
+```javascript
+const result = await ollamaApi.generate('你好', {
+  model: 'qwen3-vl',
+  stream: false,
+  options: {
+    temperature: 0.15,
+    top_p: 0.9,
+  },
+}, (message) => {
+  console.log('实时响应:', message);
+});
+
+console.log('最终结果:', result);
+```
+
+#### OpenAI - 聊天对话
+
+```javascript
+const response = await openaiApi.chat('你是谁', {
+  model: 'gpt-3.5-turbo',
+  temperature: 0.7,
+  stream: true,
+}, (message, rawResponse) => {
+  console.log('实时消息:', message);
+  console.log('原始响应:', rawResponse);
+});
+
+console.log('对话结果:', response);
+```
+
+## API 文档
+
+### `startApi(apiType, userConfig)`
+
+初始化 LLM API 客户端。
+
+**参数:**
+
+- `apiType`: `'ollama'` 或 `'openai'` - 指定要使用的 API 类型
+- `userConfig`: 对象 - 自定义配置，会覆盖默认配置
+
+**返回:** API 客户端实例，包含以下方法：
+
+### Ollama 方法
+
+- `generate(prompt, configs, callback)`: 文本生成
+- `chat(prompt, configs, callback)`: 聊天对话
+- `responses(prompt, configs, callback)`: 结构化响应
+
+### OpenAI 方法
+
+- `chat(prompt, configs, callback)`: 聊天对话
+- `responses(prompt, configs, callback)`: 结构化响应
+
+### 通用参数
+
+- `prompt`: 字符串或消息数组 - 输入提示
+- `configs`: 对象 - 模型参数配置
+  - `model`: 模型名称
+  - `stream`: 是否流式响应（默认: false）
+  - `system`: 系统提示（聊天模式）
+  - `options`: 其他模型参数
+     - `temperature`: 生成温度（0-1）
+     - `top_p`: 核采样概率
+- `callback`: 函数 - 流式响应回调
+
+## 配置
+
+### 默认配置
+
+项目提供了默认配置，可以通过环境变量或参数覆盖：
+
+**Ollama 默认配置:**
+
+```javascript
+{
+  apiKey: process.env.OLLM_API_KEY || 'huxy@zys',
+  host: process.env.OLLM_API_HOST || 'http://192.168.0.111:11434',
+  params: {
+    model: 'qwen3-vl:latest',
+    keep_alive: -1,
+  },
+  options: {
+    temperature: 0.6,
+    num_ctx: 4096,
+    top_k: 20,
+    top_p: 0.9,
+    repeat_penalty: 1.15,
+  }
+}
+```
+
+**OpenAI 默认配置:**
+
+```javascript
+{
+  apiKey: process.env.LLM_API_KEY || 'huxy@zys',
+  baseURL: process.env.LLM_API_BASEURL || 'http://192.168.0.111:11434/v1',
+  timeout: 3 * 60 * 60 * 1000,
+  maxRetries: 3,
+  params: {
+    model: 'qwen3-vl:latest',
+    temperature: 0.15,
+    max_tokens: 4096,
+    top_p: 0.9,
+    presence_penalty: 0.5,
+    frequency_penalty: 0.5,
+  },
+  options: {
+    top_k: 20,
+    repeat_penalty: 1.15,
+    thinking: true,
+  }
+}
+```
+
+## 高级用法
+
+### 自定义 Fetch
+
+项目使用 Undici 实现高性能 HTTP 请求，并支持自定义超时：
+
+```javascript
+import customFetch from 'huxy-llm-api/customFetch';
+
+// 自定义请求
+const response = await customFetch('https://api.example.com', {
+  method: 'POST',
+  headers: {
+    'Content-Type': 'application/json',
+  },
+  body: JSON.stringify({prompt: 'Hello'}),
+});
+```
+
+### 环境变量
+
+支持通过环境变量配置 API 密钥和地址：
+
+```bash
+# Ollama
+export OLLM_API_KEY="your-key"
+export OLLM_API_HOST="http://localhost:11434"
+
+# OpenAI
+export LLM_API_KEY="your-key"
+export LLM_API_BASEURL="https://api.openai.com/v1"
+```
+
+## 示例
+
+查看 [example.js](./example.js) 了解完整用法示例。
+
+## 贡献
+
+欢迎提交 Issue 和 Pull Request。
+
+## 许可证
+
+MIT License © [ahyiru](https://github.com/ahyiru)
+
+## 联系
+
+- GitHub: [https://github.com/ahyiru/huxy-llm-api](https://github.com/ahyiru/huxy-llm-api)
+- Issues: [https://github.com/ahyiru/huxy-llm-api/issues](https://github.com/ahyiru/huxy-llm-api/issues)

package/example.js

@@ -0,0 +1,43 @@
+import startApi from './index.js';
+
+const ollamaApi = startApi('ollama', {
+  apiKey: '123',
+  host: 'http://192.168.0.111:11434',
+  // headers
+  // fetch
+});
+
+const openaiApi = startApi('openai', {
+  apiKey: '123',
+  baseURL: 'http://192.168.0.111:11434/v1',
+  // headers
+  // fetch
+});
+
+const demo = async () => {
+  const ollamaResult = await ollamaApi.generate('你好', {
+    model: 'devstral-small-2',
+    stream: false,
+    options: {
+      temperature: 0.15,
+      top_p: 0.9,
+    },
+  }, (mesg) => {
+    console.log(mesg);
+  });
+  console.log(ollamaResult);
+
+  const openaiResult = await openaiApi.chat('你是谁', {
+    model: 'qwen3-vl',
+    temperature: 0.15,
+    stream: true,
+    extra_body: {
+      top_k: 20,
+    },
+  }, (mesg, resp) => {
+    console.log(mesg, resp);
+  });
+  console.log(openaiResult);
+};
+
+demo();

package/package.json

@@ -0,0 +1,44 @@
+{
+  "name": "huxy-llm-api",
+  "version": "1.0.0-beta.0",
+  "description": "一个简洁、易用的用于简化 Ollama 和 OpenAI API 调用的 Node.js 库。",
+  "type": "module",
+  "module": "./src/index.js",
+  "exports": {
+    ".": {
+      "import": "./src/index.js"
+    },
+    "./*": {
+      "import": "./src/*.js"
+    }
+  },
+  "scripts": {
+    "start": "node ./example.js"
+  },
+  "keywords": [
+    "nodejs",
+    "esm",
+    "ollama",
+    "openai",
+    "llm",
+    "api"
+  ],
+  "author": "ahyiru",
+  "license": "MIT",
+  "bugs": {
+    "url": "https://github.com/ahyiru/huxy-llm-api/issues"
+  },
+  "homepage": "https://github.com/ahyiru/huxy-llm-apir#readme",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/ahyiru/huxy-llm-api.git"
+  },
+  "dependencies": {
+    "ollama": "^0.6.3",
+    "openai": "^6.15.0",
+    "undici": "^7.16.0"
+  },
+  "engines": {
+    "node": ">=20.0.0"
+  }
+}

package/src/index.js

@@ -0,0 +1,164 @@
+import {Ollama as T} from 'ollama';
+import E from 'openai';
+import {fetch as I, Agent as L} from 'undici';
+var P = 300 * 60 * 1e3,
+  k = (t, a) => I(t, {...a, dispatcher: new L({headersTimeout: P})}),
+  x = k;
+var v = {
+    config: {
+      apiKey: process.env.LLM_API_KEY || 'ah.yiru@gmail.com',
+      baseURL: process.env.LLM_API_BASEURL || 'http://192.168.0.111:11434/v1',
+      timeout: process.env.LLM_API_TIMEOUT || 108e5,
+      maxRetries: 3,
+    },
+    params: {model: 'qwen3-vl:latest', temperature: 0.15, max_tokens: 4096, top_p: 0.9, presence_penalty: 0.5, frequency_penalty: 0.5},
+    options: {top_k: 20, repeat_penalty: 1.15, thinking: !0},
+  },
+  h = v;
+var C = {
+    config: {apiKey: process.env.OLLM_API_KEY || 'ah.yiru@gmail.com', host: process.env.OLLM_API_HOST || 'http://192.168.0.111:11434'},
+    params: {model: 'qwen3-vl:latest', keep_alive: -1},
+    options: {temperature: 0.6, num_ctx: 4096, top_k: 20, top_p: 0.9, repeat_penalty: 1.15},
+  },
+  g = C;
+var _ = async (t, a, o) => {
+    if (a) {
+      let r = '',
+        e = '';
+      for await (let n of t) {
+        let {type: c, delta: i} = n;
+        (c === 'response.reasoning_text.delta' && (e += i), c === 'response.output_text.delta' && (r += i), o?.({content: r, reasoning: e}, n));
+      }
+      return {content: r, reasoning: e};
+    }
+    let [s = {}] = t.output ?? [];
+    return (o?.(t), {reasoning: s.content?.[0]?.text, content: t.output_text});
+  },
+  R = async (t, a, o) => {
+    if (a) {
+      let n = '',
+        c = '';
+      for await (let i of t) {
+        let {delta: p} = i.choices?.[0] ?? {},
+          {reasoning: u, content: f} = p ?? {};
+        (u && (c += u), f && (n += f), o?.({content: n, reasoning: c}, i));
+      }
+      return {content: n, reasoning: c};
+    }
+    let {message: s} = t.choices?.[0] ?? {},
+      {content: r, reasoning: e} = s;
+    return (o?.(t), {content: r, reasoning: e});
+  };
+var d = async (t, a, o) => {
+    if (a) {
+      let r = '',
+        e = '';
+      for await (let n of t) {
+        let {type: c, delta: i} = n;
+        (c === 'response.reasoning_text.delta' && (e += i), c === 'response.output_text.delta' && (r += i), o?.({content: r, reasoning: e}, n));
+      }
+      return {content: r, reasoning: e};
+    }
+    let [s = {}] = t.output ?? [];
+    return (o?.(t), {reasoning: s.content?.[0]?.text, content: t.output_text});
+  },
+  A = async (t, a, o) => {
+    if (a) {
+      let e = '',
+        n = '';
+      for await (let c of t) {
+        let i = c.reasoning ?? c.thinking,
+          p = c.content ?? c.response;
+        (i && (n += i), p && (e += p), o?.({content: e, reasoning: n}, c));
+      }
+      return {content: e, reasoning: n};
+    }
+    let s = t.reasoning ?? t.thinking,
+      r = t.content ?? t.response;
+    return (o?.(t), {content: r, reasoning: s});
+  },
+  w = async (t, a, o) => {
+    if (a) {
+      let n = '',
+        c = '';
+      for await (let i of t) {
+        let {message: p} = i,
+          u = p.reasoning ?? p.thinking,
+          f = p.content ?? p.response;
+        (u && (c += u), f && (n += f), o?.({content: n, reasoning: c}, i));
+      }
+      return {content: n, reasoning: c};
+    }
+    let {message: s} = t,
+      r = s.reasoning ?? s.thinking,
+      e = s.content ?? s.response;
+    return (o?.(t), {content: e, reasoning: r});
+  };
+var m = (t, a = {}, o = 'chat') => {
+    if (!t) throw Error('\u8BF7\u4F20\u5165\u4F60\u7684 prompt !');
+    if (o === 'chat') {
+      let n = Array.isArray(t) ? t : [{role: 'user', content: t}],
+        {system: c, ...i} = a;
+      return (c && (n = [{role: 'system', content: c}, ...n]), {messages: n, ...i});
+    }
+    let s = Array.isArray(t) ? t.slice(-1)[0]?.content : t,
+      {instructions: r, ...e} = a;
+    return (r || (e.instructions = e.system), {prompt: s, ...e});
+  },
+  l = ({options: t, extra_body: a, ...o}, s, r) => {
+    let e = {...s.params, ...o},
+      n = {...s.options, ...t};
+    return (r === 'openai' ? (e.extra_body = {...n, ...a}) : (e.options = n), e);
+  };
+var y = {
+  openai: {
+    API: t => new E({...t, fetch: x}),
+    config: h,
+    llm: t => ({
+      chat: async (a, o = {}, s) => {
+        let r = R,
+          e = m(a, o, 'chat'),
+          n = await t.chat.completions.create(l(e, h, 'openai'));
+        return r(n, e.stream, s);
+      },
+      responses: async (a, o = {}, s) => {
+        let r = _,
+          e = m(a, o, 'responses'),
+          n = await t.responses.create(l(e, h, 'openai'));
+        return r(n, e.stream, s);
+      },
+    }),
+  },
+  ollama: {
+    API: t => new T({...t, fetch: x}),
+    config: g,
+    llm: t => ({
+      chat: async (a, o = {}, s) => {
+        let r = w,
+          e = m(a, o, 'chat'),
+          n = await t.chat(l(e, g, 'ollama'));
+        return r(n, e.stream, s);
+      },
+      generate: async (a, o = {}, s) => {
+        let r = A,
+          e = m(a, o, 'generate'),
+          n = await t.generate(l(e, g, 'ollama'));
+        return r(n, e.stream, s);
+      },
+      responses: async (a, o = {}, s) => {
+        let r = d,
+          e = m(a, o, 'responses'),
+          n = await t.responses(l(e, g, 'ollama'));
+        return r(n, e.stream, s);
+      },
+    }),
+  },
+};
+var K = (t = 'ollama', a) => {
+    let o = y[t] ?? y.ollama,
+      {API: s, config: r, llm: e} = o,
+      n = s({...r.config, ...a});
+    return e(n);
+  },
+  Q = K;
+export {Q as default, K as startApi};