moooc_review/src/main/java/com/mooctest/nlp/DistanceMatrix.java

package com.mooctest.nlp;

import com.hankcs.hanlp.mining.word2vec.DocVectorModel;
import com.mooctest.image.FingerPrint;
import com.mooctest.data.BugDTO;
import com.mooctest.data.DiffImg;
import com.mooctest.util.Doc2VecUtil;
import com.mooctest.util.ImageUtil;

import java.io.File;
import java.util.List;

import static com.mooctest.util.ImageUtil.IMAGE_PATH;
import static com.mooctest.util.ImageUtil.checkFileExist;

public class DistanceMatrix {

    public static double[][] genWord2VecDist(List<BugDTO> bugDTOs) {

        int n = bugDTOs.size();
        double[][] proximity = new double[n][];
        DocVectorModel docVectorModel = Doc2VecUtil.loadModel();

        for (int i = 0; i < n; i++) {
            proximity[i] = new double[i + 1];
            for (int j = 0; j < i; j++) {
                proximity[i][j] = 1 - docVectorModel.similarity(bugDTOs.get(i).getDescription(), bugDTOs.get(j).getDescription());
            }
        }
        return proximity;
    }

    public static double[][] genImgDist(List<DiffImg> diffImgs) {
        int n = diffImgs.size();
        double[][] distMatrix = new double[n][];

        for (int i = 0; i < n; i++) {
            distMatrix[i] = new double[i + 1];
            for (int j = 0; j < i; j++) {

                DiffImg diffImg1 = diffImgs.get(i);
                DiffImg diffImg2 = diffImgs.get(j);
                File file1 = checkFileExist(IMAGE_PATH + diffImg1.getBugId() + "_" + i);
                File file2 = checkFileExist(IMAGE_PATH + diffImg2.getBugId() + "_" + j);
                if (file1 == null || file2 == null) {
                    continue;
                }
                FingerPrint fp1 = ImageUtil.readImgFingerPrint(file1);
                FingerPrint fp2 = ImageUtil.readImgFingerPrint(file2);
                float sim = fp1.compare(fp2);
                distMatrix[i][j] = 1 - sim;
            }
        }
        return distMatrix;
    }
    public static double[][] genHybridDist(List<BugDTO> bugDTOs) {

        int n = bugDTOs.size();
        double[][] proximity = new double[n][];
        DocVectorModel docVectorModel = Doc2VecUtil.loadModel();

        for (int i = 0; i < n; i++) {
            proximity[i] = new double[i + 1];
            for (int j = 0; j < i; j++) {
                double txtDist = 1 - docVectorModel.similarity(bugDTOs.get(i).getDescription(), bugDTOs.get(j).getDescription());
                proximity[i][j] = txtDist;
                if (checkBugImageNotNull(bugDTOs.get(i)) && checkBugImageNotNull(bugDTOs.get(j))) {
                    double imgDist = calBugImgDist(bugDTOs.get(i), bugDTOs.get(j));
                    if (imgDist <= 0.15) {
                        proximity[i][j] = proximity[i][j] * 0.9;
                    }
                }

            }
        }
        return proximity;
    }

    private static double calBugImgDist(BugDTO bug1, BugDTO bug2) {
        float max_sim = 0;
        for (int i = 0; i < bug1.getImgUrls().length; i++) {
            for (int j = 0; j < bug2.getImgUrls().length; j++) {

                File file1 = checkFileExist(IMAGE_PATH + bug1.getId() + "_" + i);
                File file2 = checkFileExist(IMAGE_PATH + bug2.getId() + "_" + i);
                if (file1 == null || file2 == null) {
                    continue;
                }
                FingerPrint fp1 = ImageUtil.readImgFingerPrint(file1);
                FingerPrint fp2 = ImageUtil.readImgFingerPrint(file2);
                float sim = fp1.compare(fp2);
                max_sim = sim > max_sim ? sim : max_sim;
            }
        }

        return 1 - max_sim;
    }

    private static boolean checkBugImageNotNull(BugDTO bug) {
        return ((bug.getImgUrls() != null) && bug.getImgUrls().length > 0);
    }


}