SpringAI(GA)：RAG下的ETL快速上手

原文链接：SpringAI(GA)：RAG下的ETL快速上手

教程说明

说明：本教程将采用2025年5月20日正式的GA版，给出如下内容

1. 核心功能模块的快速上手教程
2. 核心功能模块的源码级解读
3. Spring ai alibaba增强的快速上手教程 + 源码级解读

版本：JDK21 + SpringBoot3.4.5 + SpringAI 1.0.0 + SpringAI Alibaba 1.0.0.2

将陆续完成如下章节教程。本章是第六章（Rag增强问答质量）下的ETL Pipeline快速上手篇

代码开源如下：https://github.com/GTyingzi/spring-ai-tutorial

往届解读可参考：

第一章内容

SpringAI(GA)的chat:快速上手+自动注入源码解读

SpringAI(GA):ChatClient调用链路解读

第二章内容

SpringAI的Advisor：快速上手+源码解读

SpringAI(GA)：Sqlite、Mysql、Redis消息存储快速上手

第三章内容

SpringAI(GA)：Tool工具整合—快速上手

第五章内容

SpringAI(GA)：内存、Redis、ES的向量数据库存储—快速上手

SpringAI(GA)：向量数据库理论源码解读+Redis、Es接入源码

第六章内容

SpringAI(GA)：RAG快速上手+模块化解读

获取更好的观赏体验，可付费获取飞书云文档Spring AI最新教程权限，目前49.9，随着内容不断完善，会逐步涨价。

注：M6版快速上手教程+源码解读飞书云文档已免费提供

RAG 的 ETL Pipeline 快速上手

[!TIP]
提取（Extract）、转换（Transform）和加载（Load）框架是《第六章：Rag 增强问答质量》中数据处理的链路，将原始数据源导入到向量化存储的流程，确保数据处于最佳格式，以便 AI 模型进行检索

实战代码可见：https://github.com/GTyingzi/spring-ai-tutorial 下的 rag/rag-etl-pipeline

源码解读可见：《ETL Pipeline 源码解析》

pom 文件

<dependencies><dependency><groupId>org.springframework.boot</groupId><artifactId>spring-boot-starter-web</artifactId></dependency><dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-autoconfigure-model-openai</artifactId></dependency><dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-commons</artifactId></dependency><dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-rag</artifactId></dependency><dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-jsoup-document-reader</artifactId></dependency><dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-markdown-document-reader</artifactId></dependency><dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-pdf-document-reader</artifactId></dependency><dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-tika-document-reader</artifactId></dependency></dependencies>

application.yml

server:port: 8080spring:application:name: rag-etl-pipelineai:openai:api-key: ${DASHSCOPEAPIKEY}base-url: https://dashscope.aliyuncs.com/compatible-modechat:options:model: qwen-maxembedding:options:model: text-embedding-v1

提取文档

Constant

package com.spring.ai.tutorial.rag.model;public class Constant {public static final String PREFIX = "classpath:data/";public static final String TEXTFILEPATH = PREFIX + "/text.txt";public static final String JSONFILEPATH = PREFIX + "/text.json";public static final String MARKDOWNFILEPATH = PREFIX + "/text.md";public static final String PDFFILEPATH = PREFIX + "/google-ai-agents-whitepaper.pdf";;public static final String HTMLFILEPATH = PREFIX + "/spring-ai.html";
}

ReaderController

package com.spring.ai.tutorial.rag.controller;import com.spring.ai.tutorial.rag.model.Constant;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.ai.document.Document;
import org.springframework.ai.reader.JsonReader;
import org.springframework.ai.reader.TextReader;
import org.springframework.ai.reader.jsoup.JsoupDocumentReader;
import org.springframework.ai.reader.markdown.MarkdownDocumentReader;
import org.springframework.ai.reader.pdf.PagePdfDocumentReader;
import org.springframework.ai.reader.pdf.ParagraphPdfDocumentReader;
import org.springframework.ai.reader.tika.TikaDocumentReader;
import org.springframework.core.io.DefaultResourceLoader;
import org.springframework.core.io.Resource;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;import java.util.List;@RestController
@RequestMapping("/reader")
public class ReaderController {private static final Logger logger = LoggerFactory.getLogger(ReaderController.class);@GetMapping("/text")public List<Document> readText() {logger.info("start read text file");Resource resource = new DefaultResourceLoader().getResource(Constant.TEXTFILEPATH);TextReader textReader = new TextReader(resource); // 适用于文本数据return textReader.read();}@GetMapping("/json")public List<Document> readJson() {logger.info("start read json file");Resource resource = new DefaultResourceLoader().getResource(Constant.JSONFILEPATH);JsonReader jsonReader = new JsonReader(resource); // 只可以传json格式文件return jsonReader.read();}@GetMapping("/pdf-page")public List<Document> readPdfPage() {logger.info("start read pdf file by page");Resource resource = new DefaultResourceLoader().getResource(Constant.PDFFILEPATH);PagePdfDocumentReader pagePdfDocumentReader = new PagePdfDocumentReader(resource); // 只可以传pdf格式文件return pagePdfDocumentReader.read();}@GetMapping("/pdf-paragraph")public List<Document> readPdfParagraph() {logger.info("start read pdf file by paragraph");Resource resource = new DefaultResourceLoader().getResource(Constant.PDFFILEPATH);ParagraphPdfDocumentReader paragraphPdfDocumentReader = new ParagraphPdfDocumentReader(resource); // 有目录的pdf文件return paragraphPdfDocumentReader.read();}@GetMapping("/markdown")public List<Document> readMarkdown() {logger.info("start read markdown file");MarkdownDocumentReader markdownDocumentReader = new MarkdownDocumentReader(Constant.MARKDOWNFILEPATH); // 只可以传markdown格式文件return markdownDocumentReader.read();}@GetMapping("/html")public List<Document> readHtml() {logger.info("start read html file");Resource resource = new DefaultResourceLoader().getResource(Constant.HTMLFILEPATH);JsoupDocumentReader jsoupDocumentReader = new JsoupDocumentReader(resource); // 只可以传html格式文件return jsoupDocumentReader.read();}@GetMapping("/tika")public List<Document> readTika() {logger.info("start read file with Tika");Resource resource = new DefaultResourceLoader().getResource(Constant.HTMLFILEPATH);TikaDocumentReader tikaDocumentReader = new TikaDocumentReader(resource); // 可以传多种文档格式return tikaDocumentReader.read();}
}

效果

读取文本文件

读取 json 文件

读取 pdf 文件

读取带目录的 pdf 文件

读取 markdown 文件

读取 html 文件

利用 tika 读取任意文档格式

转换文档

TransformerController

package com.spring.ai.tutorial.rag.controller;import com.spring.ai.tutorial.rag.model.Constant;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.ai.chat.model.ChatModel;
import org.springframework.ai.document.DefaultContentFormatter;
import org.springframework.ai.document.Document;
import org.springframework.ai.model.transformer.KeywordMetadataEnricher;
import org.springframework.ai.model.transformer.SummaryMetadataEnricher;
import org.springframework.ai.reader.pdf.PagePdfDocumentReader;
import org.springframework.ai.transformer.ContentFormatTransformer;
import org.springframework.ai.transformer.splitter.TokenTextSplitter;
import org.springframework.core.io.DefaultResourceLoader;
import org.springframework.core.io.Resource;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;import java.util.List;@RestController
@RequestMapping("/transformer")
public class TransformerController {private static final Logger logger = LoggerFactory.getLogger(TransformerController.class);private final List<Document> documents;private final ChatModel chatModel;public TransformerController(ChatModel chatModel) {logger.info("start read pdf file by page");Resource resource = new DefaultResourceLoader().getResource(Constant.PDFFILEPATH);PagePdfDocumentReader pagePdfDocumentReader = new PagePdfDocumentReader(resource); // 只可以传pdf格式文件this.documents = pagePdfDocumentReader.read();this.chatModel = chatModel;}@GetMapping("/token-text-splitter")public List<Document> tokenTextSplitter() {logger.info("start token text splitter");TokenTextSplitter tokenTextSplitter = TokenTextSplitter.builder()// 每个文本块的目标token数量.withChunkSize(800)// 每个文本块的最小字符数.withMinChunkSizeChars(350)// 丢弃小于此长度的文本块.withMinChunkLengthToEmbed(5)// 文本中生成的最大块数.withMaxNumChunks(10000)// 是否保留分隔符.withKeepSeparator(true).build();return tokenTextSplitter.split(this.documents);}@GetMapping("/content-format-transformer")public List<Document> contentFormatTransformer() {logger.info("start content format transformer");DefaultContentFormatter defaultContentFormatter = DefaultContentFormatter.defaultConfig();ContentFormatTransformer contentFormatTransformer = new ContentFormatTransformer(defaultContentFormatter);return contentFormatTransformer.apply(this.documents);}@GetMapping("/keyword-metadata-enricher")public List<Document> keywordMetadataEnricher() {logger.info("start keyword metadata enricher");KeywordMetadataEnricher keywordMetadataEnricher = new KeywordMetadataEnricher(this.chatModel, 3);return keywordMetadataEnricher.apply(this.documents);}@GetMapping("/summary-metadata-enricher")public List<Document> summaryMetadataEnricher() {logger.info("start summary metadata enricher");List<SummaryMetadataEnricher.SummaryType> summaryTypes = List.of(SummaryMetadataEnricher.SummaryType.NEXT,SummaryMetadataEnricher.SummaryType.CURRENT,SummaryMetadataEnricher.SummaryType.PREVIOUS);SummaryMetadataEnricher summaryMetadataEnricher = new SummaryMetadataEnricher(this.chatModel, summaryTypes);return summaryMetadataEnricher.apply(this.documents);}
}

效果

TokenTextSplitter 切分

DefaultContentFormatter 格式化

KeywordMetadataEnricher 提取关键字

SummaryMetadataEnricher 提取摘要

写出文档

WriterController

package com.spring.ai.tutorial.rag.controller;import com.spring.ai.tutorial.rag.model.Constant;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.ai.document.Document;
import org.springframework.ai.embedding.EmbeddingModel;
import org.springframework.ai.reader.pdf.PagePdfDocumentReader;
import org.springframework.ai.vectorstore.SearchRequest;
import org.springframework.ai.vectorstore.SimpleVectorStore;
import org.springframework.ai.writer.FileDocumentWriter;
import org.springframework.core.io.DefaultResourceLoader;
import org.springframework.core.io.Resource;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;import java.util.List;@RestController
@RequestMapping("/writer")
public class WriterController {private static final Logger logger = LoggerFactory.getLogger(WriterController.class);private final List<Document> documents;private final SimpleVectorStore simpleVectorStore;public WriterController(EmbeddingModel embeddingModel) {logger.info("start read pdf file by page");Resource resource = new DefaultResourceLoader().getResource(Constant.PDFFILEPATH);PagePdfDocumentReader pagePdfDocumentReader = new PagePdfDocumentReader(resource); // 只可以传pdf格式文件this.documents = pagePdfDocumentReader.read();this.simpleVectorStore = SimpleVectorStore.builder(embeddingModel).build();}@GetMapping("/file")public void writeFile() {logger.info("Writing file...");String fileName = "output.txt";FileDocumentWriter fileDocumentWriter = new FileDocumentWriter(fileName, true);fileDocumentWriter.accept(this.documents);}@GetMapping("/vector")public void writeVector() {logger.info("Writing vector...");simpleVectorStore.add(documents);}@GetMapping("/search")public List<Document> search() {logger.info("start search data");return simpleVectorStore.similaritySearch(SearchRequest.builder().query("Spring").topK(2).build());}
}

效果

Document 写出文本文件

写入 vector

从 vector 中查找