compose-analysis/src/main/java/com/keyware/composeanalysis/task/LineAnalysisTask.java

package com.keyware.composeanalysis.task;


import cn.hutool.core.collection.CollectionUtil;
import cn.hutool.core.lang.Pair;
import cn.hutool.core.util.StrUtil;
import com.keyware.common.constant.enums.AnalysisStatusEnum;
import com.keyware.composeanalysis.constant.FixedValue;
import com.keyware.composeanalysis.constant.FunctionAndAnalysisAssemblyConst;
import com.keyware.composeanalysis.constant.RedisConst;
import com.keyware.composeanalysis.constant.SolrDBConst;
import com.keyware.composeanalysis.constant.enums.AnalysisLevelEnum;
import com.keyware.composeanalysis.constant.enums.FileAnalysisStatusEnum;
import com.keyware.composeanalysis.entity.AnalysisTask;
import com.keyware.composeanalysis.mongo.FileDataMongoDto;
import com.keyware.composeanalysis.mongo.LineDataMongoDto;
import com.keyware.composeanalysis.mongo.MatchOpenFile;
import com.keyware.composeanalysis.mongo.MatchOpenFileMongoDto;
import com.keyware.composeanalysis.solr.VersionTree;
import com.keyware.composeanalysis.util.*;
import com.keyware.keyswan.anaysis.Analysis;
import com.keyware.keyswan.anaysis.AnalysisFactory;
import com.keyware.keyswan.common.CodeFile;
import com.keyware.utils.IdGenerator;
import lombok.extern.log4j.Log4j2;
import org.apache.commons.lang3.StringUtils;
import org.apache.solr.common.SolrDocument;
import org.apache.solr.common.SolrDocumentList;
import org.springframework.data.mongodb.core.MongoTemplate;
import org.springframework.data.mongodb.core.query.Update;

import java.math.BigDecimal;
import java.math.RoundingMode;
import java.util.*;
import java.util.concurrent.CountDownLatch;
import java.util.stream.Collectors;

import static org.springframework.data.mongodb.core.query.Criteria.where;

/**
 * @author liuzongren
 * @ClassName LineAnalysisTask
 * @description: 行级别 特征提取定时任务
 * @datetime 2024年 07月 25日 16:19
 * @version: 1.0
 */

@Log4j2
public class LineAnalysisTask extends IAnalysisTask {

    private MongoTemplate mongoTemplate;
    private AnalysisTask analysisTask;
    //被测件的文件信息
    private FileDataMongoDto analysisFile;

    private SolrUtils solrUtils;

    private RedisUtil redisUtil;

    private CountDownLatch countDownLatch;

    public LineAnalysisTask(AnalysisTask analysisTask, FileDataMongoDto analysisFile, MongoTemplate mongoTemplate, CountDownLatch countDownLatch) {
        this.mongoTemplate = mongoTemplate;
        this.analysisTask = analysisTask;
        this.analysisFile = analysisFile;
        this.countDownLatch = countDownLatch;
        this.solrUtils = SpringContextUtils.getBean(SolrUtils.class);
        this.redisUtil = SpringContextUtils.getBean(RedisUtil.class);
    }

    /**
     * 行级别 源代码溯源
     * 当前任务 需要在 文件级分析完成后 进行
     */

    @Override
    public void run() {
        //执行任务前,判断一下任务执行的状态
        Object status = redisUtil.get(String.format(RedisConst.TASK_RUNNING_STATUS_KEY_PREFIX, analysisTask.getId()));
        if (status != null && (status.equals(AnalysisStatusEnum.STOP_ANALYSIS.getCode()) || status.equals(AnalysisStatusEnum.PAUSE_ANALYSIS.getCode()))) {
            log.info("任务已取消,fileName:{}", analysisFile.getName());
            countDownLatch.countDown();
            return;
        }

        //获取文件地址
        String filePath = analysisFile.getFileUrl();
        //获取文件名称
        String fileName = analysisFile.getName();

        AnalysisLogUtil.insert(mongoTemplate, "【行级特征提取】正在提取" + fileName);
        try {

            Analysis analysis = AnalysisFactory.getAnalysis(filePath);

            //获取文件行级特征md5
            CodeFile codeFile = analysis.analysisFile(filePath, FunctionAndAnalysisAssemblyConst.LINE_EXTRACT, FunctionAndAnalysisAssemblyConst.LINE_EXTRACT);

            //从solr中获取特征相似的 文件
            SolrDocumentList featureSimilarityFromSolr = getFeatureSimilarityFromSolr(codeFile);

            //计算文件的开源率
            doAnalysis(featureSimilarityFromSolr, codeFile);

            //更新文件表的分析状态为3 行级特征以分析完毕
            analysisFile.setFileAnalysisStatus(FileAnalysisStatusEnum.ANALYSIS_DONE.getCode());
            mongoTemplate.update(FileDataMongoDto.class)
                    .matching(where("_id").is(analysisFile.getId()))
                    .replaceWith(analysisFile)
                    .findAndReplace();

            //插入日志
            AnalysisLogUtil.insert(mongoTemplate, "【行级分析】完成" + fileName);
            log.info("文件" + fileName + ":行级分析完成");
        } catch (Exception e) {
            AnalysisLogUtil.insertErrorInfo(mongoTemplate, "【行级分析】失败" + fileName, e);
            log.error("文件:{}行级别分析失败!", fileName,e);
            //修改当前文件分析状态未失败
            mongoTemplate.update(FileDataMongoDto.class)
                    .matching(where("_id").is(analysisFile.getId()))
                    .apply(new Update().set("fileAnalysisStatus", FileAnalysisStatusEnum.FAILED_ANALYSIS.getCode()))
                    .first();
        } finally {
            countDownLatch.countDown();
        }
    }


    /**
     * 计算开源率 被测件的开源率
     *
     * @param matcheOpenSourceFiles
     * @param codeFile              文件解析结果
     */
    private void doAnalysis(SolrDocumentList matcheOpenSourceFiles, CodeFile codeFile) {

        //根据文件后缀判断需要查询的文件版本库名称
        String versionIdCoreName = FixedValue.SUFFIX_SOLR_VERSION.get(analysisFile.getSuffix());

        if (CollectionUtil.isEmpty(matcheOpenSourceFiles)) {
            //因为行的特征库较少，这里补充一个对比逻辑,如果当前文件解析失败，或者没有通过代码块匹配到数据，则直接通过文件的md5 再次查询一次solr库
            checkByOriginalFileMd5(versionIdCoreName, analysisFile.getMd5());
            return;
        }


        //保存所有匹配的行数信息,方便统计总的匹配行数
        Set<String> matchedFeatureMd5 = new HashSet<>();

        //保存所有匹配的行数信息,方便统计总的匹配行数
        Set<Integer> matchLineRowsNum = new HashSet<>();

        //获取文件总特征行数
        String traitFileLineMd5 = codeFile.getTraitFileLineMd5();
        List<String> lineFeatureList = Arrays.asList(traitFileLineMd5.split(","));

        //统计每个文件的开源率
        List<MatchOpenFile> matchOpenFilesRes = calculateSimilarityAndOpenRate(matcheOpenSourceFiles, codeFile, versionIdCoreName, matchLineRowsNum, matchedFeatureMd5);

        //计算文件的总体特征相似度
        BigDecimal featureSimilarity = new BigDecimal(matchedFeatureMd5.size()).divide(new BigDecimal(lineFeatureList.size()), 4, RoundingMode.HALF_UP).multiply(new BigDecimal(100)).setScale(2);

        //计算文件的总体开源率
        BigDecimal openRate = new BigDecimal(matchLineRowsNum.size()).divide(new BigDecimal(analysisFile.getCodeRowNum()), 4, RoundingMode.HALF_UP).multiply(new BigDecimal(100)).setScale(2);


        //获取开源率的阈值
        Integer openRateThreshold = analysisTask.getOpenRateThreshold();

        //如果开源率大于阈值，则将当前文件设置成开源
        if (openRate.compareTo(new BigDecimal(openRateThreshold)) >= 0) {
            analysisFile.setOpenType(true);
        }

        //保存当前文件的开源信息到mongo库中
        MatchOpenFileMongoDto matchOpenFileMongo = new MatchOpenFileMongoDto();
        matchOpenFileMongo.setId(IdGenerator.uuid32())
                .setFilePath(analysisFile.getFileUrl())
                .setFileName(analysisFile.getName())
                .setFeatureSimilarity(featureSimilarity.floatValue())
                .setOpenRate(openRate.floatValue())
                .setOpenType(analysisFile.getOpenType())
                .setMatchOpenFile(matchOpenFilesRes);
        log.info("文件" + analysisFile.getName() + ":开源率:" + openRate.floatValue() + ",特征相似度:" + featureSimilarity.floatValue());
        mongoTemplate.save(matchOpenFileMongo);
    }


    /**
     * 计算当前文件的特征相似度 和 开源率
     *
     * @param matchOpenFiles         通过MD5 匹配到的所有开源文件
     * @param sourceFileBaseCoreName 当前文件特征文件的 solr coreName
     * @param matchLineRowsNum       所有开源文件匹配到的开源行号列表
     * @param matchFeatureLineMd5s   所有开源文件匹配到的特征行MD5
     * @return 匹配的开源文件解析后的结果集
     */
    private List<MatchOpenFile> calculateSimilarityAndOpenRate(SolrDocumentList matchOpenFiles, CodeFile fileAnalysisRes, String sourceFileBaseCoreName, Set<Integer> matchLineRowsNum, Set<String> matchFeatureLineMd5s) {

        //匹配的开源文件列表
        List<MatchOpenFile> matchOpenFilesRes = new ArrayList<>();

        //首先根据文件的MD5查询开源文件的版本ID,和路径信息
        Set<String> openSourceFileMd5s = matchOpenFiles.stream().map(doc -> (String) doc.get("sourceMd5")).collect(Collectors.toSet());
        Map<String, SolrDocument> md5VersionInfoMap = solrUtils.batchQueryVersionIdFromSourceFileBaseBySourceMd5(sourceFileBaseCoreName, openSourceFileMd5s);

        //根据版本ID查询版本的详细信息
        //todo 这里 查询一个版本的信息 需要检索 两个 solr 库 而且还需要检索 versioinTree 后面需要优化
        Set<String> openSourceFileVersionIds = md5VersionInfoMap.values().stream().map(doc -> (String) doc.get("versionId")).collect(Collectors.toSet());
        List<VersionTree> versionTrees = solrUtils.queryBatchVersionInfoByVersionIds(openSourceFileVersionIds);
        Map<String, VersionTree> versionIdVersionInfoMap = versionTrees.stream().collect(Collectors.toMap(VersionTree::getVersionId, java.util.function.Function.identity()));

        String traitFileLineMd5 = fileAnalysisRes.getTraitFileLineMd5();
        List<String> lineFeatureList = Arrays.asList(traitFileLineMd5.split(","));

        for (SolrDocument openSourceFile : matchOpenFiles) {

            //开源文件MD5
            String openSourceFileMd5 = openSourceFile.getFieldValue("sourceMd5").toString();

            //解析文件的特征行
            String lineFeatureMd5s = (String) openSourceFile.get("tz_line_hay");

            List<String> openFileLineFeatures = Arrays.asList(lineFeatureMd5s.split(","));

            //获取开源文件的文本信息
            String openSourceContent = solrUtils.getOpenFileContentByMd5(openSourceFileMd5);

            //当前文件匹配特征行总行数
            int currentFileMatchFeatureLineCount = 0;

            //遍历函数特征MD5
            for (String lineFeatureMd5 : lineFeatureList) {
                //源文件的特征行列表
                for (String openFileLineFeature : openFileLineFeatures) {
                    if (lineFeatureMd5.equals(openFileLineFeature)) {
                        matchFeatureLineMd5s.add(lineFeatureMd5);
                        currentFileMatchFeatureLineCount++;
                    }
                }
            }

            //当前文件的开源率
            Pair<Float, HashSet<Integer>> openRateAndSaveRowNum = SimilarityUtil.getOpenRateAndSaveRowNum(fileAnalysisRes.getSourceFileContent(), openSourceContent);

            //将当前文件匹配的行号，存储到缓存中，方便统计整体的开源率
            matchLineRowsNum.addAll(openRateAndSaveRowNum.getValue());

            //统计当前文件的特征相似度
            BigDecimal featureSimilarity = new BigDecimal(currentFileMatchFeatureLineCount).divide(new BigDecimal(lineFeatureList.size()), 4, RoundingMode.HALF_UP).multiply(new BigDecimal(100)).setScale(2);

            SolrDocument openEntries = md5VersionInfoMap.get(openSourceFileMd5);
            VersionTree versionInfo = versionIdVersionInfoMap.get(openEntries.get("versionId"));
            if (versionInfo == null) {
                log.error("根据版本ID,未查询到相关的版本信息。versionId:{}", openEntries.get("versionId"));
                continue;
            }

            //组装当前开源文件的开源项目信息
            MatchOpenFile matchOpenFileInfo = new MatchOpenFile();
            matchOpenFileInfo.setPId(versionInfo.getProId())
                    .setPName(versionInfo.getProName())
                    .setSourceUrl((String) openEntries.get("fullPath"))
                    .setFeatureSimilarity(featureSimilarity.floatValue())
                    .setOpenRate(openRateAndSaveRowNum.getKey())
                    .setVersion(versionInfo.getVersionName())
                    .setLicenseType(versionInfo.getLicenseType())
                    .setAnalyzeType(AnalysisLevelEnum.FUNCTION_LEVEL.getCode());
            matchOpenFilesRes.add(matchOpenFileInfo);
        }
        return matchOpenFilesRes;
    }



    /**
     * 防止代码块特征库不全，再次根据文件MD5查询开源文件信息, 做二次校验
     *
     * @param originalFileMd5
     * @param versionIdCoreName
     */
    private void checkByOriginalFileMd5(String versionIdCoreName, String originalFileMd5) {

        //根据文件的MD5,查询特征库，看当前文件是否在开源代码库中
        SolrDocument versionIdAndPath = solrUtils.queryOne(versionIdCoreName, "sourceFileMd5:" + originalFileMd5, "versionId,fullPath,sourceFileMd5");

        if (versionIdAndPath != null) {
            //根据版本ID查询版本的详细信息
            VersionTree versionInfo = solrUtils.queryVersionInfoByVersionId((String) versionIdAndPath.get("versionId"));
            if (versionInfo != null) {
                //当前开源文件的开源项目信息
                MatchOpenFile matchOpenFileInfo = new MatchOpenFile();
                matchOpenFileInfo.setPId(versionInfo.getProId())
                        .setPName(versionInfo.getProName())
                        .setSourceUrl(versionInfo.getDownUrl())
                        .setFeatureSimilarity(100.00f)
                        .setOpenRate(100.00f)
                        .setAnalyzeType(AnalysisLevelEnum.FILE_LEVEL.getCode());

                //保存当前文件的开源信息到mongo库中
                MatchOpenFileMongoDto matchOpenFileMongo = new MatchOpenFileMongoDto();
                matchOpenFileMongo.setId(IdGenerator.uuid32())
                        .setFilePath(analysisFile.getFileUrl())
                        .setFileName(analysisFile.getName())
                        .setOpenRate(100.00f)
                        .setOpenType(analysisFile.getOpenType())
                        .setMatchOpenFile(Arrays.asList(matchOpenFileInfo));

                mongoTemplate.save(matchOpenFileMongo);
            }
        }
    }


    /**
     * 将特征值插入到mongo库中
     * @param features         特征集合
     * @param lineDataMongoDto 当前分析任务 ，特征信息存储
     *  todo 后期 看看有没有插入的必要
     * @param
     */
    @Deprecated
    private void insertFeatureValue(String features, LineDataMongoDto lineDataMongoDto) {
        String[] featureMd5Arr = {};
        if (StringUtils.isNotBlank(features)) {
            featureMd5Arr = features.split(",");
        }
        List<String> lineFeatures = Arrays.asList(featureMd5Arr);
        List<String> batchInsertList = new ArrayList<>();
        if (CollectionUtil.isNotEmpty(lineFeatures)) {
            //这里的批量插入逻辑可以进行校验
            //每10条存一次,解析的数据量如果过大，可能会超过MongoDB数据限制
            int batchInsertStpe = 5000;
            int total = 0;
            for (int i = 0; i < lineFeatures.size(); i++) {
                if (total != batchInsertStpe) {
                    batchInsertList.add(lineFeatures.get(i));
                    total++;
                }
                if (i == lineFeatures.size() - 1 && total != batchInsertStpe) {
                    total = 0;
                    lineDataMongoDto.setId(IdGenerator.uuid32())
                            .setLineFeatueMd5s(batchInsertList);
                    mongoTemplate.insert(lineDataMongoDto);
                }
                if (total == batchInsertStpe) {
                    total = 0;
                    lineDataMongoDto.setId(IdGenerator.uuid32())
                            .setLineFeatueMd5s(batchInsertList);
                    mongoTemplate.insert(lineDataMongoDto);
                    batchInsertList.clear();
                }
            }
        } else {
            lineDataMongoDto.setId(IdGenerator.uuid32());
            mongoTemplate.insert(lineDataMongoDto);
        }
    }


    /**
     * 根据 特征值 从特征库中检索 具有特征相似的
     *
     * @param codeFile 行特征信息
     * @return
     */
    private SolrDocumentList getFeatureSimilarityFromSolr(CodeFile codeFile) {
        Set<String> queryMd5Set = new HashSet<>();
        //每一行原内容MD5值集合
        String cutFileLineMd5 = codeFile.getCutFileLineMd5();
        if (StrUtil.isNotBlank(cutFileLineMd5)) {
            List<String> lineCutList = Arrays.asList(cutFileLineMd5.split(","));
            queryMd5Set.addAll(lineCutList);
        }
        //每一行特征内容MD5值集合
        String traitFileLineMd5 = codeFile.getTraitFileLineMd5();
        if (StrUtil.isNotBlank(traitFileLineMd5)) {
            List<String> lineFeatureList = Arrays.asList(traitFileLineMd5.split(","));
            queryMd5Set.addAll(lineFeatureList);
        }
        if (CollectionUtil.isEmpty(queryMd5Set)) {
            log.error("特征相似度检索失败，特征为空:{}", analysisFile.getName());
            return new SolrDocumentList();
        }
        //拼接行特征查询条件
        String queryStr = "tz_line_hay:(" + StringUtils.join(queryMd5Set, " OR ") + ")";
        log.info("查询条件: solrCoreName:{},queryStr:{}", SolrDBConst.CORE_NAME_SOURCE_FILE_INFO_TEMP, queryStr);
        SolrDocumentList result = solrUtils.query(SolrDBConst.CORE_NAME_SOURCE_FILE_INFO_TEMP, queryStr, "sourceMd5,tz_line_hay");
        return result;
    }


}