SpringAI 本地调用 Ollama

SpringAI 本地调用 Ollama

  • [1 Ollama 软件下载](#1 Ollama 软件下载)
  • [2 Ollama 模型下载](#2 Ollama 模型下载)
    • [2.1 PowerShell 下载脚本](#2.1 PowerShell 下载脚本)
    • [2.2 Ollama模型运行](#2.2 Ollama模型运行)
  • [3 SpringAI](#3 SpringAI)
    • [1 依赖](#1 依赖)
    • [2 配置](#2 配置)
    • [3 MemConf.java](#3 MemConf.java)
    • [4 OllamaController.java](#4 OllamaController.java)

LangChain4j 详细介绍

LangChain4j 是一个专为 Java 生态系统设计的开源框架,旨在简化大语言模型(LLM)与应用程序的集成,提供了构建基于 LLM 的复杂应用所需的核心组件和工具。其设计理念借鉴了 Python 生态的 LangChain,但针对 Java 开发者的习惯和需求进行了优化。

LangChain4j 核心特点

  1. 模块化设计:提供了清晰的组件划分,包括模型接口、提示模板、记忆机制、工具调用、链(Chains)等,开发者可按需组合。
  2. 多模型支持:兼容主流 LLM,如 OpenAI、GPT-4、Claude、Llama 等,同时支持本地模型(如通过 Ollama 部署的开源模型)。
  3. 强大的记忆机制:支持对话历史管理,包括临时记忆、持久化记忆(如 Redis、数据库存储),方便构建多轮对话应用。
  4. 工具调用能力:允许 LLM 调用外部工具(如搜索引擎、数据库、API),扩展模型的能力边界。
  5. 与 Java 生态融合:无缝集成 Spring、Jakarta EE 等主流框架,支持依赖注入、异步处理等 Java 开发模式。
  6. 轻量级:核心库体积小,无过多依赖,易于嵌入现有 Java 项目。

LangChain4j 应用场景

  • 构建智能客服、聊天机器人
  • 实现文档问答(基于向量数据库的检索增强生成,RAG)
  • 开发自动化工作流(结合工具调用)
  • 集成 LLM 到企业级 Java 应用中

LangChain4j VS Spring AI

维度 LangChain4j Spring AI
定位 专注于 LLM 应用开发的工具链,类似 Java 版 LangChain Spring 生态的官方 AI 扩展,提供标准化集成
生态依赖 独立框架,可与任何 Java 框架集成(包括 Spring) 深度依赖 Spring 生态(如 Spring Boot、Spring Cloud)
核心能力 提供完整的 LLM 应用组件(链、记忆、工具等) 聚焦于模型接口标准化、数据处理、集成能力
模型支持 支持主流闭源/开源模型,配置灵活 支持主流模型,通过统一接口抽象,符合 Spring 风格
易用性 需手动组合组件,灵活性高,但学习成本略高 遵循 Spring 约定优于配置,开箱即用,适合 Spring 开发者
企业级特性 需自行集成安全、监控等特性 天然支持 Spring 生态的企业级特性(如安全、审计、监控)
社区与更新 社区活跃,更新频率中等 官方背书,更新频繁,与 Spring 版本同步演进
适用场景 复杂 LLM 应用(如多链协作、自定义记忆策略) 快速集成 LLM 到 Spring 应用,追求标准化和稳定性

选择建议

  • 若项目已基于 Spring 生态(如 Spring Boot),且希望快速集成 LLM 并利用 Spring 的企业级特性,优先选择 Spring AI
  • 若需要更灵活的 LLM 应用构建能力,或项目不依赖 Spring 框架,LangChain4j 是更合适的选择。
  • 若需实现复杂的 RAG、多轮对话记忆或工具调用逻辑,LangChain4j 的组件化设计可能更具优势。

1 Ollama 软件下载

Ollama

2 Ollama 模型下载

2.1 PowerShell 下载脚本

复制内容到ollama_download.ps1文件中

powershell 复制代码
# 定义模型名称避免硬编码
$modelName = "gemma3:27b"

# 设置重试参数
$maxRetries = 50
$retryInterval = 3  # 秒
$downloadTimeout = 80  # 秒

for ($retry = 1; $retry -le $maxRetries; $retry++) {
    # 精确检查模型是否存在
    $modelExists = ollama list | Where-Object { $_ -match "\b$modelName\b" }
    
    if ($modelExists) {
        Write-Host "[$(Get-Date)] model is already downloaded!"
        exit 0
    }

    Write-Host "[$(Get-Date)] start $retry times downloading..."
    
    # 启动进程并显示实时输出
    $process = Start-Process -FilePath "ollama" `
        -ArgumentList "run", $modelName `
        -PassThru `
        -NoNewWindow

    # 等待下载完成或超时
    try {
        $process | Wait-Process -Timeout $downloadTimeout -ErrorAction Stop
    } catch {
        # 超时处理
        Write-Host "download timeout, safe terminate process..."
        $process | Stop-Process -Force
    }

    if (-not $process.HasExited) {
        $process | Stop-Process -Force
        Write-Host "process terminated due to timeout."
    } else {
        # 检查退出代码
        if ($process.ExitCode -eq 0) {
            Write-Host "download success!"
            exit 0
        }
        Write-Host "download failed, exit code: $($process.ExitCode)"
    }

    Start-Sleep -Seconds $retryInterval
}

Write-Host "exceeded maximum retries ($maxRetries), download failed."
exit 1
powershell 复制代码
PS C:\Users\xuhya\Desktop> .\ollama_download.ps1
[10/09/2025 21:42:20] start 1 times downloading...
pulling manifest
pulling e796792eba26:  98% ▕████████████████████████████████████████████████████████  ▏  17 GB/ 17 GB  1.7 MB/s   3m20sdownload timeout, safe terminate process...
download failed, exit code: -1
[10/09/2025 21:43:43] start 2 times downloading...
pulling manifest
pulling e796792eba26: 100% ▕██████████████████████████████████████████████████████████▏  17 GB
pulling e0a42594d802: 100% ▕██████████████████████████████████████████████████████████▏  358 B
pulling dd084c7d92a3: 100% ▕██████████████████████████████████████████████████████████▏ 8.4 KB
pulling 3116c5225075: 100% ▕██████████████████████████████████████████████████████████▏   77 B
pulling f838f048d368: 100% ▕██████████████████████████████████████████████████████████▏  490 B
verifying sha256 digest ⠸ download timeout, safe terminate process...
download failed, exit code: -1
[10/09/2025 21:45:06] start 3 times downloading...
pulling manifest
pulling e796792eba26: 100% ▕██████████████████████████████████████████████████████████▏  17 GB
pulling e0a42594d802: 100% ▕██████████████████████████████████████████████████████████▏  358 B
pulling dd084c7d92a3: 100% ▕██████████████████████████████████████████████████████████▏ 8.4 KB
pulling 3116c5225075: 100% ▕██████████████████████████████████████████████████████████▏   77 B
pulling f838f048d368: 100% ▕██████████████████████████████████████████████████████████▏  490 B
verifying sha256 digest
writing manifest
success
⠏ download timeout, safe terminate process...
download failed, exit code: -1
[10/09/2025 21:46:29] model is already downloaded!
PS C:\Users\xuhya\Desktop>

2.2 Ollama模型运行

powershell 复制代码
PS C:\Users\xuhya\Desktop> ollama list
NAME             ID              SIZE      MODIFIED
gemma3:27b       a418f5838eaf    17 GB     2 minutes ago
gemma3:latest    a2af6cc3eb7f    3.3 GB    7 minutes ago
PS C:\Users\xuhya\Desktop>

3 SpringAI

1 依赖

xml 复制代码
<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
	xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 https://maven.apache.org/xsd/maven-4.0.0.xsd">

	<modelVersion>4.0.0</modelVersion>

	<parent>
		<groupId>org.springframework.boot</groupId>
		<artifactId>spring-boot-starter-parent</artifactId>
		<version>3.5.6</version>
		<relativePath/> <!-- lookup parent from repository -->
	</parent>

	<groupId>com.xu</groupId>
	<artifactId>spring-ai</artifactId>
	<version>0.0.1-SNAPSHOT</version>
	<name>spring-ai</name>
	<description>Demo project for Spring Boot</description>

	<properties>
		<java.version>25</java.version>
	</properties>

    <dependencyManagement>
        <dependencies>
            <dependency>
                <groupId>org.springframework.ai</groupId>
                <artifactId>spring-ai-bom</artifactId>
                <version>1.0.3</version>
                <type>pom</type>
                <scope>import</scope>
            </dependency>
        </dependencies>
    </dependencyManagement>


    <dependencies>

		<dependency>
			<groupId>org.springframework.boot</groupId>
			<artifactId>spring-boot-starter-web</artifactId>
		</dependency>

        <dependency>
            <groupId>org.springframework.ai</groupId>
            <artifactId>spring-ai-starter-model-ollama</artifactId>
        </dependency>

		<dependency>
			<groupId>org.springframework.boot</groupId>
			<artifactId>spring-boot-devtools</artifactId>
			<scope>runtime</scope>
			<optional>true</optional>
		</dependency>

		<dependency>
			<groupId>org.projectlombok</groupId>
			<artifactId>lombok</artifactId>
			<optional>true</optional>
		</dependency>

		<dependency>
			<groupId>org.springframework.boot</groupId>
			<artifactId>spring-boot-starter-test</artifactId>
			<scope>test</scope>
		</dependency>

	</dependencies>

	<build>
		<plugins>
			<plugin>
				<groupId>org.apache.maven.plugins</groupId>
				<artifactId>maven-compiler-plugin</artifactId>
				<configuration>
					<annotationProcessorPaths>
						<path>
							<groupId>org.projectlombok</groupId>
							<artifactId>lombok</artifactId>
						</path>
					</annotationProcessorPaths>
				</configuration>
			</plugin>
			<plugin>
				<groupId>org.springframework.boot</groupId>
				<artifactId>spring-boot-maven-plugin</artifactId>
				<configuration>
					<excludes>
						<exclude>
							<groupId>org.projectlombok</groupId>
							<artifactId>lombok</artifactId>
						</exclude>
					</excludes>
				</configuration>
			</plugin>
		</plugins>
	</build>

</project>

2 配置

yml 复制代码
spring:
  application:
    name: spring-ai
  ai:
    ollama:
      base-url: http://localhost:11434
      chat:
        model: gemma3:latest

3 MemConf.java

java 复制代码
package com.xu.conf;

import org.springframework.ai.chat.memory.ChatMemory;
import org.springframework.ai.chat.memory.InMemoryChatMemoryRepository;
import org.springframework.ai.chat.memory.MessageWindowChatMemory;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;

@Configuration
public class MemConf {

	@Bean
	public ChatMemory chatMemory() {
		return MessageWindowChatMemory.builder()
				.chatMemoryRepository(new InMemoryChatMemoryRepository())
				.maxMessages(20)
				.build();
	}

}

4 OllamaController.java

java 复制代码
package com.xu.ollama;

import lombok.AllArgsConstructor;
import org.springframework.ai.chat.memory.ChatMemory;
import org.springframework.ai.chat.messages.UserMessage;
import org.springframework.ai.chat.model.ChatResponse;
import org.springframework.ai.chat.prompt.Prompt;
import org.springframework.ai.content.Media;
import org.springframework.ai.ollama.OllamaChatModel;
import org.springframework.http.MediaType;
import org.springframework.util.MimeType;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RequestParam;
import org.springframework.web.bind.annotation.RestController;
import org.springframework.web.multipart.MultipartFile;
import reactor.core.publisher.Flux;

@RestController
@AllArgsConstructor
@RequestMapping("/ollama")
public class OllamaController {

	private final ChatMemory chatMemory;

	private final OllamaChatModel model;

	/**
	 * 聊天
	 *
	 * @param content 聊天内容
	 * @return 聊天结果
	 */
	@RequestMapping("/common/chat")
	public Object chat(String content) {
		return model.call(content);
	}

	/**
	 * 聊天记忆
	 *
	 * @param content 聊天内容
	 * @return 聊天结果
	 */
	@RequestMapping("/common/memory")
	public Object memory(String content, String msgId) {

		UserMessage message = UserMessage.builder().text(content).build();
		chatMemory.add(msgId, message);
		ChatResponse response = model.call(new Prompt(message));
		chatMemory.add(msgId, response.getResult().getOutput());

		return response.getResult().getOutput().getText();
	}

	/**
	 * 聊天图片
	 *
	 * @param content 聊天内容
	 * @return 聊天结果
	 */
	@RequestMapping("/common/image")
	public Object image(@RequestParam("content") String content, @RequestParam("file") MultipartFile file) {
		if (file.isEmpty()) {
			return model.call(content);
		}
		var media = new Media(new MimeType("image", "png"), file.getResource());
		return model.call(new Prompt(UserMessage.builder().media(media).text(content).build()));
	}

	/**
	 * 聊天流式
	 *
	 * @param content 聊天内容
	 * @return 聊天结果
	 */
	@RequestMapping(value = "/stream/chat", produces = MediaType.TEXT_EVENT_STREAM_VALUE)
	public Flux<String> stream(String content) {
		return model.stream(content);
	}

}
相关推荐
大模型教程9 小时前
AI 新玩法:GraphRAG × Ollama 打造更聪明的智能体
程序员·llm·ollama
whltaoin1 天前
AI 超级智能体全栈项目阶段五:RAG 四大流程详解、最佳实践与调优(基于 Spring AI 实现)
java·人工智能·spring·rag·springai
wxl7812272 天前
Milvus部署在T4 GPU上,Dify检索性能可以提升多少?
知识库·dify·rag·milvus gpu
有点不太正常4 天前
Differentially Private Synthetic Text Generation for RAG——论文阅读
论文阅读·大模型·llm·rag
tongsound4 天前
ollama部署本地模型(deepseek,qwen,gemma3,...)
ollama·deepseek
laizi_laizi4 天前
ollama+openwebui本地部署及踩坑记录
ollama·openwebui
老顾聊技术4 天前
【AI课程上线了哦,打造类FastGPT产品】
langchain·rag
mit6.8245 天前
[VoiceRAG] 集成向量化 | Azure AI Search中建立自动化系统
llm·rag
深色風信子7 天前
SpringBoot 集成 LangChain4j 本地调用 Ollama
java·spring boot·spring·ollama·langchain4j