支持doc、docx文件上传

llm-back/ruoyi-agent/pom.xml

@@ -50,14 +50,20 @@
             </exclusions>
         </dependency>
 
+        <dependency>
+            <groupId>org.apache.poi</groupId>
+            <artifactId>poi</artifactId>
+            <version>5.2.5</version>
+        </dependency>
         <dependency>
             <groupId>org.apache.poi</groupId>
             <artifactId>poi-ooxml</artifactId>
+            <version>5.2.5</version>
         </dependency>
         <dependency>
             <groupId>org.apache.poi</groupId>
             <artifactId>poi-scratchpad</artifactId>
-            <version>5.2.3</version>
+            <version>5.2.5</version>
         </dependency>
         <dependency>
             <groupId>com.alibaba.fastjson2</groupId>

llm-back/ruoyi-agent/src/main/java/com/ruoyi/agent/service/impl/McpServiceImpl.java

@@ -24,11 +24,10 @@ import io.milvus.param.collection.LoadCollectionParam;
 import io.milvus.param.collection.ReleaseCollectionParam;
 import io.milvus.param.dml.SearchParam;
 import io.milvus.response.SearchResultsWrapper;
-import org.apache.poi.hwpf.HWPFDocument;
+import org.apache.poi.extractor.POITextExtractor;
+import org.apache.poi.extractor.ExtractorFactory;
 import org.apache.poi.xwpf.extractor.XWPFWordExtractor;
-import org.apache.poi.xwpf.usermodel.XWPFDocument;
-import org.apache.poi.xwpf.usermodel.XWPFParagraph;
-import org.apache.poi.xwpf.usermodel.XWPFRun;
+import org.apache.poi.xwpf.usermodel.*;
 import org.noear.solon.Solon;
 import org.noear.solon.ai.annotation.ToolMapping;
 import org.noear.solon.ai.chat.ChatModel;
@@ -65,9 +64,13 @@ public class McpServiceImpl implements IMcpService {
                                            @Param(description = "章节名称") String title,
                                            @Param(description = "技术文件地址") String templatePath) throws IOException
     {
-        title = String.join(",", extractSubTitles( "/upload/agent/template/technical.docx", title));
-        List<JSONObject> contexts = retrieveFromMilvus(milvusClient, embeddingModel, collectionName, title, 10);
-        return generateAnswerWithDocumentAndCollection(embeddingModel, agentName, templatePath, title, contexts, "http://192.168.28.188:8000/v1/chat/completions");
+            try {
+                title = String.join(",", extractSubTitles( "/upload/agent/template/technical.docx", title));
+                List<JSONObject> contexts = retrieveFromMilvus(milvusClient, embeddingModel, collectionName, title, 10);
+                return generateAnswerWithDocumentAndCollection(embeddingModel, agentName, templatePath, title, contexts, "http://192.168.28.188:8000/v1/chat/completions");
+            } catch (IOException e) {
+                throw new RuntimeException(e);
+            }
     }
 
     /**
@@ -91,7 +94,12 @@ public class McpServiceImpl implements IMcpService {
      */
     public AssistantMessage generateAnswerWithDocumentAndCollection(EmbeddingModel embeddingModel, String agentName, String templatePath, String question, List<JSONObject> contexts, String llmServiceUrl) throws IOException {
         StringBuilder sb = new StringBuilder("招标文件内容：\n\n");
-        File profilePath = new File(templatePath.replace("/dev-api/profile", Solon.cfg().getProperty("cmc.profile")).replace("_" + agentName, ""));
+        String filename = templatePath.replace("/dev-api/profile", Solon.cfg().getProperty("cmc.profile")).replace("_" + agentName, "");
+        File profilePath = new File(filename);
+        if (!profilePath.exists()) {
+            filename = filename.replace(".docx", ".doc");
+            profilePath = new File(filename);
+        }
         List<TextSegment> segments = splitDocument(profilePath);
         List<Embedding> embeddings = embeddingModel.embedAll(segments).content();
         InMemoryEmbeddingStore<TextSegment> embeddingStore = new InMemoryEmbeddingStore<>();
@@ -131,7 +139,7 @@ public class McpServiceImpl implements IMcpService {
 
         ChatResponse response = chatModel.prompt(chatSession).call();
         String content = response.lastChoice().getMessage().getContent() + "\n\n" +
-                "招标文件分析完成，章节内容已写入【<a href='" + templatePath.replace("/dev-api", "") + "'> 技术文件" + "</a>】，请查阅";
+                "招标文件分析完成，章节内容已写入【<a href='" + templatePath + "'> 技术文件" + "</a>】，请查阅";
         String absolutePath = templatePath.replace("/dev-api/profile", Solon.cfg().getProperty("cmc.profile"));
         writeContent(response.lastChoice().getMessage().getContent(), question, absolutePath);
         return ChatMessage.ofAssistant(content);
@@ -144,47 +152,60 @@ public class McpServiceImpl implements IMcpService {
     public void writeContent(String content, String question, String absolutePath) throws IOException {
         File file = new File(absolutePath);
         FileInputStream fileInputStream = new FileInputStream(file);
-        XWPFDocument document = new XWPFDocument(fileInputStream);
-        String[] contentLines = content.split("\n");
-        Map<String, String> map = new HashMap<>();
-        String[] titles = question.split(",");
-        for (int i = 0; i < titles.length; i ++) {
-            int startIndex = Arrays.asList(contentLines).indexOf(titles[i]);
-            StringBuilder text = new StringBuilder("");
-            if (i < titles.length - 1) {
-                int endIndex = Arrays.asList(contentLines).indexOf(titles[i + 1]);
-                for (int c = startIndex + 1; c < endIndex; c++)
-                    text.append(contentLines[c]);
-            }
-            else {
-                if (startIndex + 1 < contentLines.length)
-                    for (int c = startIndex + 1; c < contentLines.length; c++)
+        try (XWPFDocument document = new XWPFDocument(fileInputStream)) {
+            String[] contentLines = content.split("\n");
+            Map<String, String> map = new HashMap<>();
+            String[] titles = question.split(",");
+            for (int i = 0; i < titles.length; i++) {
+                int startIndex = Arrays.asList(contentLines).indexOf(titles[i]);
+                StringBuilder text = new StringBuilder("");
+                if (i < titles.length - 1) {
+                    int endIndex = Arrays.asList(contentLines).indexOf(titles[i + 1]);
+                    for (int c = startIndex + 1; c < endIndex; c++)
                         text.append(contentLines[c]);
+                } else {
+                    if (startIndex + 1 < contentLines.length)
+                        for (int c = startIndex + 1; c < contentLines.length; c++)
+                            text.append(contentLines[c]);
+                }
+                map.put(titles[i], text.toString());
             }
-            map.put(titles[i], text.toString());
-        }
-        int count = 0;
-        for (int i = 0; i < document.getParagraphs().size(); i++) {
-            XWPFParagraph paragraph = document.getParagraphs().get(i);
-            for (String title : titles) {
-                if (paragraph.getText().equals(title)) {
-                    int pos = document.getBodyElements().indexOf(paragraph) + 1;
-                    XWPFParagraph contentParagraph = document.createParagraph();
-                    contentParagraph.setStyle("1");
-                    XWPFRun run = contentParagraph.createRun();
-                    run.setText(map.get(title));
-                    document.setParagraph(contentParagraph, pos);
-                    count++;
+            int count = 0;
+            int position = 0;
+            List<Integer> positions = new ArrayList<>();
+            List<String> contents = new ArrayList<>();
+            for (int i = 0; i < document.getBodyElements().size(); i++) {
+                IBodyElement element = document.getBodyElements().get(i);
+                if (element instanceof XWPFParagraph) {
+                    XWPFParagraph paragraph = (XWPFParagraph) element;
+                    for (String title : titles) {
+                        if (paragraph.getText().equals(title)) {
+                            int pos = position + 1;
+                            positions.add(pos);
+                            contents.add(map.get(title));
+                            count++;
+                        }
+                    }
+                    position++;
+                    if (count == titles.length)
+                        break;
                 }
+                else if (element instanceof XWPFTable) {
+                    XWPFTable table = (XWPFTable) element;
+                    position += table.getNumberOfRows();
+                }
+            }
+            for (int i = positions.size() - 1; i >= 0; i--) {
+                XWPFParagraph contentParagraph = document.createParagraph();
+                contentParagraph.setStyle("1");
+                document.setParagraph(contentParagraph, positions.get(i));
+                XWPFRun run = contentParagraph.createRun();
+                run.setText(contents.get(i));
+            }
+            try (FileOutputStream out = new FileOutputStream(absolutePath)) {
+                document.write(out);
             }
-            if (count == titles.length)
-                break;
         }
-        FileOutputStream out = new FileOutputStream(absolutePath);
-        document.write(out);
-        // 关闭文档
-        out.close();
-        document.close();
     }
 
     /**
@@ -195,25 +216,26 @@ public class McpServiceImpl implements IMcpService {
         boolean inTargetSection = false;
         filename = Solon.cfg().getProperty("cmc.profile") + filename;
         InputStream fileInputStream = new FileInputStream(filename);
-        XWPFDocument document = new XWPFDocument(fileInputStream);
-        for (XWPFParagraph paragraph : document.getParagraphs()) {
-            String text = paragraph.getText().trim();
-            if (paragraph.getStyle() != null) {
-                // 判断主标题
-                if (paragraph.getStyle().equals("3") &&
-                        text.contains(question)) {
-                    inTargetSection = true;
-                    continue;
-                }
-
-                // 如果已经在目标节中，收集标题3级别的子标题
-                if (inTargetSection) {
-                    if (paragraph.getStyle().equals("4")) {
-                        subTitles.add(text);
+        try (XWPFDocument document = new XWPFDocument(fileInputStream)) {
+            for (XWPFParagraph paragraph : document.getParagraphs()) {
+                String text = paragraph.getText().trim();
+                if (paragraph.getStyle() != null) {
+                    // 判断主标题
+                    if (paragraph.getStyle().equals("3") &&
+                            text.contains(question)) {
+                        inTargetSection = true;
+                        continue;
                     }
-                    // 遇到下一个Heading1则退出
-                    else if (paragraph.getStyle().equals("3")) {
-                        break;
+
+                    // 如果已经在目标节中，收集标题3级别的子标题
+                    if (inTargetSection) {
+                        if (paragraph.getStyle().equals("4")) {
+                            subTitles.add(text);
+                        }
+                        // 遇到下一个Heading1则退出
+                        else if (paragraph.getStyle().equals("3")) {
+                            break;
+                        }
                     }
                 }
             }
@@ -271,15 +293,14 @@ public class McpServiceImpl implements IMcpService {
         Document document;
         InputStream fileInputStream = new FileInputStream(transferFile);
         String filename = transferFile.getName().toLowerCase();
-        if (filename.endsWith(".doc")) {
-            HWPFDocument doc = new HWPFDocument(fileInputStream);
-            document = Document.from(doc.getDocumentText());
-        }
-        else if (filename.endsWith(".docx")) {
-            XWPFDocument docx = new XWPFDocument(fileInputStream);
-            XWPFWordExtractor extractor = new XWPFWordExtractor(docx);
-            String text = extractor.getText();
-            document = Document.from(text);
+        if (filename.endsWith(".doc") || filename.endsWith(".docx")) {
+            try (POITextExtractor extractor = ExtractorFactory.createExtractor(fileInputStream)) {
+                String text = extractor.getText();
+                document = Document.from(text);
+            }
+            catch (IOException e) {
+                throw new RuntimeException(e);
+            }
         }
         else if (filename.endsWith(".pdf")) {
             document = new ApachePdfBoxDocumentParser().parse(fileInputStream);

llm-back/ruoyi-common/pom.xml

@@ -72,9 +72,20 @@
         </dependency>
 
         <!-- excel工具 -->
+        <dependency>
+            <groupId>org.apache.poi</groupId>
+            <artifactId>poi</artifactId>
+            <version>5.2.5</version>
+        </dependency>
         <dependency>
             <groupId>org.apache.poi</groupId>
             <artifactId>poi-ooxml</artifactId>
+            <version>5.2.5</version>
+        </dependency>
+        <dependency>
+            <groupId>org.apache.poi</groupId>
+            <artifactId>poi-scratchpad</artifactId>
+            <version>5.2.5</version>
         </dependency>
 
         <!-- yml解析器 -->

llm-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/service/impl/LangChainMilvusServiceImpl.java

@@ -28,8 +28,8 @@ import io.milvus.param.collection.ReleaseCollectionParam;
 import io.milvus.param.dml.InsertParam;
 import io.milvus.param.dml.SearchParam;
 import io.milvus.response.SearchResultsWrapper;
-import org.apache.poi.hwpf.HWPFDocument;
-import org.apache.poi.xwpf.extractor.XWPFWordExtractor;
+import org.apache.poi.extractor.POITextExtractor;
+import org.apache.poi.extractor.ExtractorFactory;
 import org.apache.poi.xwpf.usermodel.XWPFDocument;
 import org.apache.poi.xwpf.usermodel.XWPFParagraph;
 import org.noear.solon.ai.chat.ChatModel;
@@ -69,7 +69,7 @@ public class LangChainMilvusServiceImpl implements ILangChainMilvusService
         File transferFile = new File( profilePath + "/" + file.getOriginalFilename());
         if (!transferFile.exists())
             file.transferTo(transferFile);
-        List<TextSegment> segments = splitDocument(file.getOriginalFilename(), transferFile);
+        List<TextSegment> segments = splitDocument(transferFile);
 
         // 提取文本和生成嵌入
         List<String> fileNames = new ArrayList<>();
@@ -200,7 +200,7 @@ public class LangChainMilvusServiceImpl implements ILangChainMilvusService
         StringBuilder sb = new StringBuilder("问题: " + question + "\n\n").append("根据以下上下文回答问题：\n\n");
         for (CmcDocument document : documentList) {
             File profilePath = new File(RuoYiConfig.getProfile() + document.getPath());
-            List<TextSegment> segments = splitDocument(document.getPath(), profilePath);
+            List<TextSegment> segments = splitDocument(profilePath);
             List<Embedding> embeddings = embeddingModel.embedAll(segments).content();
             InMemoryEmbeddingStore<TextSegment> embeddingStore = new InMemoryEmbeddingStore<>();
             embeddingStore.addAll(embeddings, segments);
@@ -236,7 +236,7 @@ public class LangChainMilvusServiceImpl implements ILangChainMilvusService
             if (documentList.size() == 1) {
                 for (CmcDocument document : documentList) {
                     File profilePath = new File(RuoYiConfig.getProfile() + document.getPath());
-                    List<TextSegment> segments = splitDocument(document.getPath(), profilePath);
+                    List<TextSegment> segments = splitDocument(profilePath);
                     List<Embedding> embeddings = embeddingModel.embedAll(segments).content();
                     InMemoryEmbeddingStore<TextSegment> embeddingStore = new InMemoryEmbeddingStore<>();
                     embeddingStore.addAll(embeddings, segments);
@@ -271,25 +271,26 @@ public class LangChainMilvusServiceImpl implements ILangChainMilvusService
         boolean inTargetSection = false;
 
         InputStream fileInputStream = new FileInputStream(filename);
-        XWPFDocument document = new XWPFDocument(fileInputStream);
-        for (XWPFParagraph paragraph : document.getParagraphs()) {
-            String text = paragraph.getText().trim();
-            if (paragraph.getStyle() != null) {
-                // 判断主标题
-                if (paragraph.getStyle().equals("3") &&
-                        text.contains(question)) {
-                    inTargetSection = true;
-                    continue;
-                }
-
-                // 如果已经在目标节中，收集标题3级别的子标题
-                if (inTargetSection) {
-                    if (paragraph.getStyle().equals("4")) {
-                        subTitles.add(text);
+        try (XWPFDocument document = new XWPFDocument(fileInputStream)) {
+            for (XWPFParagraph paragraph : document.getParagraphs()) {
+                String text = paragraph.getText().trim();
+                if (paragraph.getStyle() != null) {
+                    // 判断主标题
+                    if (paragraph.getStyle().equals("3") &&
+                            text.contains(question)) {
+                        inTargetSection = true;
+                        continue;
                     }
-                    // 遇到下一个Heading1则退出
-                    else if (paragraph.getStyle().equals("3")) {
-                        break;
+
+                    // 如果已经在目标节中，收集标题3级别的子标题
+                    if (inTargetSection) {
+                        if (paragraph.getStyle().equals("4")) {
+                            subTitles.add(text);
+                        }
+                        // 遇到下一个Heading1则退出
+                        else if (paragraph.getStyle().equals("3")) {
+                            break;
+                        }
                     }
                 }
             }
@@ -342,20 +343,19 @@ public class LangChainMilvusServiceImpl implements ILangChainMilvusService
     /**
      * 检索知识库
      */
-    private List<TextSegment> splitDocument(String filename, File transferFile) throws IOException {
+    private List<TextSegment> splitDocument(File transferFile) throws IOException {
         // 加载文档
         Document document;
         InputStream fileInputStream = new FileInputStream(transferFile);
-        filename = filename.toLowerCase();
-        if (filename.endsWith(".doc")) {
-            HWPFDocument doc = new HWPFDocument(fileInputStream);
-            document = Document.from(doc.getDocumentText());
-        }
-        else if (filename.endsWith(".docx")) {
-            XWPFDocument docx = new XWPFDocument(fileInputStream);
-            XWPFWordExtractor extractor = new XWPFWordExtractor(docx);
-            String text = extractor.getText();
-            document = Document.from(text);
+        String filename = transferFile.getName().toLowerCase();
+        if (filename.endsWith(".doc") || filename.endsWith(".docx")) {
+            try (POITextExtractor extractor = ExtractorFactory.createExtractor(fileInputStream)) {
+                String text = extractor.getText();
+                document = Document.from(text);
+            }
+            catch (IOException e) {
+                throw new RuntimeException(e);
+            }
         }
         else if (filename.endsWith(".pdf")) {
             document = new ApachePdfBoxDocumentParser().parse(fileInputStream);

llm-back/ruoyi-system/src/main/java/com/ruoyi/llm/service/impl/CmcAgentServiceImpl.java

@@ -1,9 +1,6 @@
 package com.ruoyi.llm.service.impl;
 
-import java.io.File;
-import java.io.FileInputStream;
-import java.io.FileOutputStream;
-import java.io.IOException;
+import java.io.*;
 import java.util.Date;
 import java.util.List;
 
@@ -101,7 +98,7 @@ public class CmcAgentServiceImpl implements ICmcAgentService
         String outputFilename = "/upload/agent/" + agentName + "/" + file.getOriginalFilename()
                 .replace(filenameSplit[filenameSplit.length - 2], filenameSplit[filenameSplit.length - 2] + "_" + agentName);
         if (file.getOriginalFilename().endsWith(".doc"))
-            outputFilename = outputFilename.replace(".doc", "docx");
+            outputFilename = outputFilename.replace(".doc", ".docx");
         CmcDocument cmcDocument = new CmcDocument();
         cmcDocument.setDocumentId(new SnowFlake().generateId());
         cmcDocument.setChatId(chatId);
@@ -115,14 +112,13 @@ public class CmcAgentServiceImpl implements ICmcAgentService
             cmcChat.setInput("招标文件地址：" + "/upload/agent/" + agentName + "/" + file.getOriginalFilename());
             cmcChat.setUserId(SecurityUtils.getUserId());
             cmcChatMapper.insertCmcChat(cmcChat);
-            XWPFDocument doc = new XWPFDocument(new FileInputStream(RuoYiConfig.getProfile() + "/upload/agent/template/technical.docx"));
-            // 保存文档到本地文件系统
-            FileOutputStream out = new FileOutputStream(RuoYiConfig.getProfile() + outputFilename);
-            doc.write(out);
-            // 关闭文档
-            out.close();
-            doc.close();
-            
+            InputStream fileInputStream = new FileInputStream(RuoYiConfig.getProfile() + "/upload/agent/template/technical.docx");
+            try (XWPFDocument doc = new XWPFDocument(fileInputStream)) {
+                // 保存文档到本地文件系统
+                try (FileOutputStream out = new FileOutputStream(RuoYiConfig.getProfile() + outputFilename)) {
+                    doc.write(out);
+                }
+            }
             message = "好的，我已经收到您上传的招标文件，我将给您提供技术文件模板，您可点击进行预览：" +
                     "【<a href='/profile" + outputFilename + "'> 模版 " + "</a>】\n\n" +
                     "技术文件涉及多个章节：\n" +