1.修复代码块级别 特征相似度大于100%的bug

src/main/java/com/keyware/composeanalysis/config/GlobalExceptionHandler.java

@@ -13,20 +13,20 @@ import org.springframework.web.bind.annotation.RestControllerAdvice;
 @RestControllerAdvice
 public class GlobalExceptionHandler {
 
-	//全局异常处理
+    //全局异常处理
-	@ExceptionHandler(value = Exception.class)
+    @ExceptionHandler(value = Exception.class)
     public Result defaultErrorHandler(Exception e) {
-		log.error("全局异常信息,ex={}",e.getMessage(),e);
+        log.error("全局异常信息,ex={}", e.getMessage(), e);
-		return Result.fail(ResultCode.FAIL.getCode(), e.getMessage());
+        return Result.fail(ResultCode.FAIL.getCode(), e.getMessage());
     }
 
-	//自定义异常处理
+    //自定义异常处理
-	//业务异常
+    //业务异常
-	@ExceptionHandler(value = BusinessException.class)
+    @ExceptionHandler(value = BusinessException.class)
-	@ResponseBody
+    @ResponseBody
-	public Result businessExceptionHandler(BusinessException e) {
+    public Result businessExceptionHandler(BusinessException e) {
-		log.error("业务异常信息",e);
+        log.error("业务异常信息", e);
-		return Result.fail(e.getCode(), e.getMsg());
+        return Result.fail(e.getCode(), e.getMsg());
-	}
+    }
 
 }

src/main/java/com/keyware/composeanalysis/task/CodeBlockAnalysisTask.java

@@ -172,7 +172,6 @@ public class CodeBlockAnalysisTask extends IAnalysisTask {
             return;
         }
 
-
         //保存所有匹配的特征代码块MD5信息,方便统计总的匹配行数
         Set<String> matchingTraitLineSet = new HashSet<>();
 
@@ -190,8 +189,11 @@ public class CodeBlockAnalysisTask extends IAnalysisTask {
             matchCodeBlockLineCount += traitsFeatureMd5AndFeatureLineNumMap.get(matchFeatureMd5);
         }
 
+        //特征行总数
+        int totalFeatureLineCount = traitsFeatureMd5AndFeatureLineNumMap.values().stream().mapToInt(Integer::intValue).sum();
+
         //计算文件的总体特征相似度
-        BigDecimal featureSimilarity = new BigDecimal(matchCodeBlockLineCount).divide(new BigDecimal(analysisFile.getCodeRowNum()), 4, RoundingMode.HALF_UP).multiply(new BigDecimal(100)).setScale(2);
+        BigDecimal featureSimilarity = new BigDecimal(matchCodeBlockLineCount).divide(new BigDecimal(totalFeatureLineCount), 4, RoundingMode.HALF_UP).multiply(new BigDecimal(100)).setScale(2);
 
         //计算文件的总体开源率
         BigDecimal openRate = new BigDecimal(matchedLineRowsNum.size()).divide(new BigDecimal(analysisFile.getCodeRowNum()), 4, RoundingMode.HALF_UP).multiply(new BigDecimal(100)).setScale(2);
@@ -344,12 +346,13 @@ public class CodeBlockAnalysisTask extends IAnalysisTask {
     }
 
     /**
-     * 或者特征代码块的md5 和 当前md5包含的特征行数
+     *  获取每个特征代码块包含的特征行数
      *
      * @param codeBlockInfos
      * @return
      */
     private Map<String, Integer> getTraitsFeatureMd5AndFeatureLineNumMap(List<LineModel> codeBlockInfos) {
+        //按照特征代码块的md5进行分组
         Map<String, List<LineModel>> traitMd5GroupMap = codeBlockInfos.stream().collect(Collectors.groupingBy(LineModel::getTraitLineMd5));
         Map<String, Integer> resultMap = new HashMap<>();
         for (String traitMd5 : traitMd5GroupMap.keySet()) {

src/main/java/com/keyware/composeanalysis/task/FunctionAnalysisTask.java

@@ -106,15 +106,15 @@ public class FunctionAnalysisTask extends IAnalysisTask {
             //根据文件的名称获取函数解析器
             Analysis analysis = AnalysisFactory.getAnalysis(fileName);
 
-            if (analysis == null){
+            if (analysis == null) {
-                throw  new BusinessException("获取文件解析器失败,文件名称："+fileName);
+                throw new BusinessException("获取文件解析器失败,文件名称：" + fileName);
             }
 
             //解析文件
             CodeFile codeFile = analysis.analysisFile(new FileInputStream(filePath));
 
             //根据函数特征去匹配到开源文件
-            SolrDocumentList matchOpenFiles = getFeatureSimilarityFromSolr(featureCoreName,  codeFile.getFunctionList());
+            SolrDocumentList matchOpenFiles = getFeatureSimilarityFromSolr(featureCoreName, codeFile.getFunctionList());
 
             //计算开源率
             doAnalysis(matchOpenFiles, sourceFileBaseCoreName, codeFile);
@@ -238,7 +238,7 @@ public class FunctionAnalysisTask extends IAnalysisTask {
 
 
         //被测件文本内容
-        String sourcefileContent= FileUtil.readUtf8String(analysisFile.getFileUrl());
+        String sourcefileContent = FileUtil.readUtf8String(analysisFile.getFileUrl());
 
         //将文本内容解析成行信息,用于后续文件的开源率计算
         List<String> analysisFileLineInfo = SimilarityUtil.getSplitWords(sourcefileContent);
@@ -291,7 +291,9 @@ public class FunctionAnalysisTask extends IAnalysisTask {
             SolrDocument openEntries = md5VersionInfoMap.get(openSourceFileMd5);
             VersionTree versionInfo = versionIdVersionInfoMap.get(openEntries.get("versionId"));
             String openFilePath = (String) md5VersionInfoMap.get(openSourceFileMd5).getFieldValue("fullPath");
-
+            if (versionInfo == null){
+                throw new BusinessException("根据版本ID,未查询到相关的版本信息。versionId:" + openEntries.get("versionId"));
+            }
             //组装当前开源文件的开源项目信息
             MatchOpenFile matchOpenFileInfo = new MatchOpenFile();
             matchOpenFileInfo.setId(IdGenerator.uuid32())

src/main/java/com/keyware/composeanalysis/task/PorjectAnalysisTask.java

@@ -76,7 +76,7 @@ public class PorjectAnalysisTask {
             if (!matchedPrject) {
                 List<FileDataMongoDto> unMatchedFiles = matchByAllFilesMd5();
 
-                //剩余没有匹配文件，用文件的md5去匹配solr库的versionTree
+                //剩余没有匹配文件，用文件的md5去批量匹配solr库的versionTree
                 if (CollectionUtils.isNotEmpty(unMatchedFiles)) {
                     matchByFileMd5s(unMatchedFiles);
                 }

src/main/java/com/keyware/composeanalysis/util/SolrUtils.java

@@ -7,7 +7,6 @@ import lombok.Data;
 import lombok.extern.log4j.Log4j2;
 import org.apache.commons.collections.CollectionUtils;
 import org.apache.commons.lang3.StringUtils;
-import org.apache.solr.client.solrj.SolrClient;
 import org.apache.solr.client.solrj.SolrRequest;
 import org.apache.solr.client.solrj.SolrServerException;
 import org.apache.solr.client.solrj.impl.HttpSolrClient;

src/main/resources/application-dev.yaml

@@ -8,7 +8,7 @@ spring:
     mongodb:
       uri: mongodb://127.0.0.1:27017/KEYSWAN
     redis:
-      host: 127.0.0.1
+      host: 172.16.36.7
       port: 6379
       password: 123456
   datasource:
@@ -30,7 +30,7 @@ spring:
 
 #solr检索库地址
 solr:
-  solrUrl: http://172.16.36.7:8993/solr/
+  solrUrl: http://172.16.36.5:8993/solr/
   #批量匹配时,返回的匹配数量
   row: 5