添加图片库备用

ask-data-ai/ask-data-ai-api/src/main/java/com/ask/api/entity/AskImagesRecord.java

+package com.ask.api.entity;
+
+import com.baomidou.mybatisplus.annotation.TableName;
+import io.swagger.v3.oas.annotations.media.Schema;
+import lombok.Data;
+
+import java.time.LocalDateTime;
+
+@Data
+@TableName(value = "ask_images_record")
+@Schema(description = "图片表")
+public class AskImagesRecord {
+
+    // 路径
+    private Long id;
+    // 值
+    private String imageName;
+    // 时间戳
+    private byte[] imageData;
+}

ask-data-ai/ask-data-ai-biz/pom.xml

@@ -31,6 +31,16 @@
             <version>${project.version}</version>
         </dependency>
 
+        <dependency>
+            <groupId>commons-io</groupId>
+            <artifactId>commons-io</artifactId>
+            <version>2.17.0</version>
+        </dependency>
+<!--        <dependency>-->
+<!--            <groupId>org.apache.tika</groupId>-->
+<!--            <artifactId>tika-parsers-ocr</artifactId>-->
+<!--            <version>3.0.0</version>-->
+<!--        </dependency>-->
         <!-- API -->
         <dependency>
             <groupId>com.ask</groupId>

ask-data-ai/ask-data-ai-biz/src/main/java/com/ask/mapper/AskHistoryCollectDataMapper.java

@@ -3,9 +3,17 @@ package com.ask.mapper;
 import com.ask.api.entity.AskHistoryCollectData;
 import com.baomidou.mybatisplus.core.mapper.BaseMapper;
 import org.apache.ibatis.annotations.Mapper;
+import org.apache.ibatis.annotations.Param;
+import org.apache.ibatis.annotations.Select;
+
+import java.time.LocalDateTime;
+import java.util.List;
 
 @Mapper
 public interface AskHistoryCollectDataMapper extends BaseMapper<AskHistoryCollectData> {
 
 
+    List<AskHistoryCollectData> selectLastRecordForEachPath(@Param("paths") List<String> paths,
+                                                            @Param("startTime") LocalDateTime startTime,
+                                                            @Param("endTime") LocalDateTime endTime);
 }
\ No newline at end of file

ask-data-ai/ask-data-ai-biz/src/main/java/com/ask/mapper/AskImagesRecordMapper.java

+package com.ask.mapper;
+
+import com.ask.api.entity.AskHistoryCollectData;
+import com.ask.api.entity.AskImagesRecord;
+import com.baomidou.mybatisplus.core.mapper.BaseMapper;
+import org.apache.ibatis.annotations.Mapper;
+
+@Mapper
+public interface AskImagesRecordMapper extends BaseMapper<AskImagesRecord> {
+
+}

ask-data-ai/ask-data-ai-biz/src/main/java/com/ask/service/impl/DocumentParseService.java

+package com.ask.service.impl;
+
+import com.ask.api.entity.AskImagesRecord;
+import com.ask.mapper.AskImagesRecordMapper;
+import lombok.AllArgsConstructor;
+import org.apache.tika.exception.TikaException;
+import org.apache.tika.extractor.EmbeddedDocumentExtractor;
+import org.apache.tika.metadata.Metadata;
+import org.apache.tika.parser.AutoDetectParser;
+import org.apache.tika.parser.ParseContext;
+import org.apache.tika.parser.Parser;
+import org.apache.tika.parser.microsoft.OfficeParserConfig;
+import org.apache.tika.sax.ContentHandlerDecorator;
+import org.springframework.stereotype.Component;
+import org.xml.sax.Attributes;
+import org.xml.sax.ContentHandler;
+import org.xml.sax.SAXException;
+
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.LinkedHashMap;
+import java.util.Map;
+
+@Component
+@AllArgsConstructor
+public class DocumentParseService {
+
+    private final AskImagesRecordMapper askImagesRecordMapper;
+
+    public String extractText(InputStream inputStream) {
+        // 初始化解析器、元数据和上下文
+        Parser parser = new AutoDetectParser();
+        Metadata metadata = new Metadata();
+        ParseContext parseContext = new ParseContext();
+        OfficeParserConfig officeParserConfig = new OfficeParserConfig();
+        //忽略页眉页脚
+        officeParserConfig.setIncludeHeadersAndFooters(false);
+        parseContext.set(OfficeParserConfig.class, officeParserConfig);
+        Map<String, Long> imageMap = new LinkedHashMap<>();
+        // 自定义ContentHandler用于插入占位符
+        class MarkdownImageHandler extends ContentHandlerDecorator {
+            private final StringBuilder markdown = new StringBuilder();
+
+            private String localName = null;
+
+            @Override
+            public void characters(char[] ch, int start, int length) {
+                String text = new String(ch, start, length);
+                if (this.localName.equals("h1")) {
+                    markdown.append("# ").append(text);
+                } else if (this.localName.equals("p")) {
+                    markdown.append("\n").append(text);
+                } else {
+                    markdown.append(text);
+                }
+            }
+
+            @Override
+            public void startElement(String uri, String localName, String qName, Attributes attrs) {
+                this.localName = localName;
+                //  System.out.println("localName="+localName+"  qName="+qName+"  text="+text);
+                if ("img".equals(localName)) { // 捕获图片节点
+                    String src = attrs.getValue("src");
+                    if (src != null && src.startsWith("embedded:")) {
+                        String imageName = src.split(":")[1];
+                        //TODO 存储图片
+                        AskImagesRecord askImagesRecord = new AskImagesRecord();
+                        askImagesRecord.setImageName(imageName);
+                        askImagesRecordMapper.insert(askImagesRecord);
+                        imageMap.put(imageName, askImagesRecord.getId());
+                        markdown.append("![").append(imageName).append("](ImageId:").append(askImagesRecord.getId()).append(")\n");
+                    }
+                }
+            }
+
+            public String getMarkdown() {
+                return markdown.toString();
+            }
+
+        }
+        MarkdownImageHandler contentHandler = new MarkdownImageHandler();
+        EmbeddedDocumentExtractor extractor = new EmbeddedDocumentExtractor() {
+            @Override
+            public boolean shouldParseEmbedded(Metadata metadata) {
+                // 只处理图片类型
+                return metadata.get(Metadata.CONTENT_TYPE) != null &&
+                        metadata.get(Metadata.CONTENT_TYPE).startsWith("image/");
+            }
+
+            @Override
+            public void parseEmbedded(InputStream inputStream, ContentHandler embeddedHandler, Metadata metadata, boolean b) throws IOException, SAXException {
+                String fileName = metadata.get("resourceName");
+                Long imageId = imageMap.get(fileName);
+                AskImagesRecord askImagesRecord = askImagesRecordMapper.selectById(imageId);
+                askImagesRecord.setImageData(inputStream.readAllBytes());
+                askImagesRecordMapper.updateById(askImagesRecord);
+            }
+        };
+        parseContext.set(EmbeddedDocumentExtractor.class, extractor);
+        // 开始解析文档
+        try {
+            parser.parse(inputStream, contentHandler, metadata, parseContext);
+        } catch (IOException | SAXException | TikaException e) {
+            throw new RuntimeException(e);
+        }
+        //System.out.println("文件内容:" + contentHandler.getMarkdown());
+        return contentHandler.getMarkdown();
+    }
+}
\ No newline at end of file

ask-data-ai/ask-data-ai-biz/src/main/java/com/ask/service/impl/KnowledgeDocumentServiceImpl.java

@@ -45,7 +45,7 @@ public class KnowledgeDocumentServiceImpl extends ServiceImpl<KnowledgeDocumentM
     private final SysFileService sysFileService;
     private final AskVectorStoreService askVectorStoreService;
     private final AsyncVectorizationService asyncVectorizationService;
-
+    private final DocumentParseService documentParseService;
 
 
     /**
@@ -78,8 +78,8 @@ public class KnowledgeDocumentServiceImpl extends ServiceImpl<KnowledgeDocumentM
      * @param maxTokensPerSlice 每片最大token数（仅对CUSTOM策略有效）
      * @return 文档片段列表，每个Document就是一片
      */
-    public List<Document> slicePdfDocument(String bucketName,String fileName , SliceStrategy sliceStrategy) {
-        InputStreamResource resource = new InputStreamResource(sysFileService.getFileStream(bucketName,fileName));
+    public List<Document> slicePdfDocument(String bucketName, String fileName, SliceStrategy sliceStrategy) {
+        InputStreamResource resource = new InputStreamResource(sysFileService.getFileStream(bucketName, fileName));
         List<Document> documents = new ArrayList<>();
 
         try {
@@ -99,9 +99,9 @@ public class KnowledgeDocumentServiceImpl extends ServiceImpl<KnowledgeDocumentM
             }
             log.info("PDF切片完成，策略: {}, 切片数量: {}", sliceStrategy, documents.size());
 
-        }catch (Exception e) {
+        } catch (Exception e) {
             log.error("PDF切片失败，策略: {}, 错误: {}", sliceStrategy, e.getMessage(), e);
-            documents = sliceByTokens(new InputStreamResource(sysFileService.getFileStream(bucketName,fileName)));
+            documents = sliceByTokens(new InputStreamResource(sysFileService.getFileStream(bucketName, fileName)));
         }
 
         return documents;
@@ -146,7 +146,7 @@ public class KnowledgeDocumentServiceImpl extends ServiceImpl<KnowledgeDocumentM
                                 .build())
                         .withPagesPerDocument(0)
                         .build());
-        List<Document>  pageDocuments = pdfReader.read();
+        List<Document> pageDocuments = pdfReader.read();
 
 
         List<Document> paragraphDocuments = new ArrayList<>();
@@ -348,6 +348,28 @@ public class KnowledgeDocumentServiceImpl extends ServiceImpl<KnowledgeDocumentM
         return tokenDocuments;
     }
 
+    private List<Document> sliceByTokens(String text) {
+
+        // 使用TokenTextSplitter进行切片
+        TokenTextSplitter textSplitter = new TokenTextSplitter(
+                4096,  // 分片大小
+                50,   // 最小分片字符数
+                50,    // 设置最小需要嵌入的长度
+                1000,  // 最大片段大小
+                true   // 保持分隔符
+        );
+
+        Document fullDocument = new Document(text);
+        List<Document> tokenDocuments = textSplitter.apply(List.of(fullDocument));
+        for (Document subDoc : tokenDocuments) {
+            subDoc.getMetadata().put("size", Objects.requireNonNull(subDoc.getText()).length());
+            subDoc.getMetadata().put("title", "");
+        }
+
+        return tokenDocuments;
+    }
+
+
     /**
      * PDF切片策略枚举
      */
@@ -361,9 +383,12 @@ public class KnowledgeDocumentServiceImpl extends ServiceImpl<KnowledgeDocumentM
         List<DocumentSegmentResult> results = new ArrayList<>();
 
         for (SysFile file : request.getFiles()) {
+
+//            String docText = documentParseService.extractText(sysFileService.getFileStream(file.getBucketName(), file.getFileName()));
+//            List<Document> segments = sliceByTokens(docText);
             SliceStrategy sliceStrategy = SliceStrategy.CUSTOM;
-            if("pdf".equals(file.getType())){
-                sliceStrategy =SliceStrategy.PARAGRAPH;
+            if ("pdf".equals(file.getType())) {
+                sliceStrategy = SliceStrategy.PARAGRAPH;
             }
             // 读取文档内容 - 使用新的PDF切片函数
             List<Document> segments = slicePdfDocument(

ask-data-ai/ask-data-ai-biz/src/main/java/com/ask/tools/ExcelTools.java

@@ -21,11 +21,9 @@ import java.time.format.DateTimeFormatter;
 import java.time.format.DateTimeFormatterBuilder;
 import java.time.format.SignStyle;
 import java.time.temporal.ChronoField;
-import java.util.HashMap;
-import java.util.List;
-import java.util.Map;
-import java.util.Objects;
+import java.util.*;
 import java.util.concurrent.atomic.AtomicReference;
+import java.util.stream.Collectors;
 
 @Component
 @Slf4j
@@ -105,6 +103,28 @@ public class ExcelTools {
         List<AskReportDict> askReportDicts = askReportDictMapper.selectList(
                 Wrappers.lambdaQuery(AskReportDict.class).eq(AskReportDict::getType, 1)
         );
+        List<String> paramSer = new ArrayList<>();
+        for (AskReportDict askReportDict : askReportDicts) {
+            askReportDict.convertParamsToParamList();
+            askReportDict.getParamList().forEach(params -> {
+                if (params.getType() == 1) {
+                    if (!paramSer.contains(params.getParam())) {
+                        paramSer.add(params.getParam());
+                    }
+                }
+            });
+        }
+        if (CollectionUtils.isEmpty(paramSer)) {
+            return paramMap;
+        }
+        List<AskHistoryCollectData> askHistoryCollectDataList = askHistoryCollectDataMapper.selectLastRecordForEachPath(paramSer, startTime, endTime);
+
+        Map<String, Double> result = askHistoryCollectDataList.stream()
+                .collect(Collectors.toMap(
+                        AskHistoryCollectData::getPath, // 获取 path 作为键
+                        AskHistoryCollectData::getValue, // 获取 value 作为值
+                        (existingValue, newValue) -> existingValue // 如果有重复的键，这里决定如何处理。这里选择保留现有的值。
+                ));
 
         for (AskReportDict askReportDict : askReportDicts) {
             String key = askReportDict.getKey();
@@ -117,13 +137,13 @@ public class ExcelTools {
 
             switch (askReportDict.getType()) {
                 case 1:
-                    handleType1(askReportDict.getKey(), params, paramMap, startTime, endTime);
+                    handleType1(askReportDict.getKey(), params, paramMap, result);
                     break;
                 case 2:
-                    handleType2(askReportDict.getKey(), params, paramMap, startTime, endTime);
+                    handleType2(askReportDict.getKey(), params, paramMap, result);
                     break;
                 case 3:
-                    handleType3(askReportDict.getKey(), params, paramMap, startTime, endTime);
+                    handleType3(askReportDict.getKey(), params, paramMap, result);
                     break;
                 default:
                     // Handle other types if necessary
@@ -134,50 +154,50 @@ public class ExcelTools {
         return paramMap;
     }
 
-    private void handleType1(String key, List<AskReportDict.Params> params, Map<String, Object> paramMap, LocalDateTime startTime, LocalDateTime endTime) {
+    private void handleType1(String key, List<AskReportDict.Params> params, Map<String, Object> paramMap, Map<String, Double> result) {
 
         AskReportDict.Params param = params.get(0);
         if (param == null) {
             return;
         }
-        Double value = getLatestValue(param.getParam(), startTime, endTime);
+        Double value = result.get(param.getParam());
         if (value != null) {
             paramMap.put(key, value);
         }
     }
 
-    private void handleType2(String key, List<AskReportDict.Params> params, Map<String, Object> paramMap, LocalDateTime startTime, LocalDateTime endTime) {
+    private void handleType2(String key, List<AskReportDict.Params> params, Map<String, Object> paramMap, Map<String, Double> result) {
         if (params.size() < 2) {
             return;
         }
 
-        Double numerator = getParamValue(params.get(0), startTime, endTime);
-        Double denominator = getParamValue(params.get(1), startTime, endTime);
+        Double numerator = result.get(params.get(0).getParam());
+        Double denominator = result.get(params.get(1).getParam());
 
         if (numerator == null || denominator == null || denominator == 0) {
             return;
         }
 
-        double result = Math.round(numerator / denominator * 100.0) / 100.0; // 保留两位小数
-        paramMap.put(key, result);
+        Double value = Math.round(numerator / denominator * 100.0) / 100.0; // 保留两位小数
+        paramMap.put(key, value);
     }
 
-    private void handleType3(String key, List<AskReportDict.Params> params, Map<String, Object> paramMap, LocalDateTime startTime, LocalDateTime endTime) {
+    private void handleType3(String key, List<AskReportDict.Params> params, Map<String, Object> paramMap, Map<String, Double> result) {
         if (params.size() < 2) {
             return;
         }
 
-        Double numerator = getParamValue(params.get(0), startTime, endTime);
-        Double denominator = getParamValue(params.get(1), startTime, endTime);
+        Double numerator = result.get(params.get(0).getParam());
+        Double denominator = result.get(params.get(1).getParam());
 
         if (numerator == null || denominator == null || denominator == 0) {
             return;
         }
 
-        double result = (numerator / denominator) * 100; // 计算百分比
-        result = Math.round(result * 100.0) / 100.0; // 保留两位小数
+        Double value = (numerator / denominator) * 100; // 计算百分比
+        value = Math.round(value * 100.0) / 100.0; // 保留两位小数
 
-        String percentageResult = String.format("%.2f%%", result); // 格式化为百分比字符串
+        String percentageResult = String.format("%.2f%%", value); // 格式化为百分比字符串
         paramMap.put(key, percentageResult);
     }
 

ask-data-ai/ask-data-ai-biz/src/main/resources/mapper/AskHistoryCollectDataMapper.xml

+<?xml version="1.0" encoding="UTF-8"?>
+
+<!DOCTYPE mapper PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN" "http://mybatis.org/dtd/mybatis-3-mapper.dtd">
+
+<mapper namespace="com.ask.mapper.AskHistoryCollectDataMapper">
+
+    <select id="selectLastRecordForEachPath" resultType="com.ask.api.entity.AskHistoryCollectData">
+        SELECT *
+        FROM (
+        SELECT *,
+        ROW_NUMBER() OVER (PARTITION BY path ORDER BY datetime DESC) AS rn
+        FROM ask_history_collect_data
+        WHERE path IN
+        <foreach item="item" collection="paths" open="(" separator="," close=")">
+            #{item}
+        </foreach>
+        AND datetime BETWEEN #{startTime} AND #{endTime}
+        ) t
+        WHERE t.rn = 1
+    </select>
+
+</mapper>

ask-data-ai/ask-data-ai-boot/src/main/resources/application.yml

@@ -15,7 +15,7 @@ spring:
   datasource:
     url: jdbc:postgresql://8.152.98.45:5432/ask_data_ai_db?stringtype=unspecified
     username: postgres
-    password: postgres123
+    password: e5d039e4ba5246068
     driver-class-name: org.postgresql.Driver
   ai:
     vectorstore: