akerbladkernel.java

A Java port of LDC s Champollion sentence aligner (http://champollion.sourceforge.net). Intended aud

package net.sourceforge.akerblad;

import info.jonclark.log.LogUtils;
import info.jonclark.util.ArrayUtils;
import info.jonclark.util.StringUtils;

import java.util.ArrayList;
import java.util.Arrays;
import java.util.HashMap;
import java.util.logging.Logger;

public class AkerbladKernel {

    private final Akerblad.Config config;
    private final Akerblad.DocPair dp;

    private static final double MINS = -10;

    private static final double penalty01 = 0.8;
    private static final double penalty21 = 0.95;
    private static final double penalty12 = 0.95;
    private static final double penalty22 = 0.85;
    private static final double penalty31 = 0.94;
    private static final double penalty13 = 0.94;
    private static final double penalty41 = 0.92; // 0.85
    private static final double penalty14 = 0.92; // 0.85

    // assumed contants -- not defined in perl version
    private static final double penalty23 = 0.0;
    private static final double penalty32 = 0.0;
    private static final double penalty33 = 0.0;

    private final double[][] score;
    private final double[][] stScores;
    private final int[][] pathX;
    private final int[][] pathY;

    private static final Logger log = LogUtils.getLogger();

    public AkerbladKernel(Akerblad.Config config, Akerblad.DocPair dp) {
	this.config = config;
	this.dp = dp;

	this.pathX = new int[dp.nx + 1][dp.ny + 1];
	this.pathY = new int[dp.nx + 1][dp.ny + 1];
	this.score = new double[dp.nx + 1][dp.ny + 1];

	// this.stScores = new double[dp.nx][dp.ny];
	this.stScores = null;
	for (int i = 0; i < score.length; i++) {
	    for (int j = 0; j < score[i].length; j++) {
		score[i][j] = Double.NEGATIVE_INFINITY;
	    }
	}
    }

    private void setScore(int x, int y, double d) {
	// System.err.println(x + " " + y + " " + d);
	score[x][y] = d;
    }

    private double getScore(int x, int y) {
	if (score[x][y] != Double.NEGATIVE_INFINITY) {
	    return score[x][y];
	} else {
	    return 0;
	}
    }

    public int align(ArrayList<Integer> x, ArrayList<Integer> y, int nx, int ny) {

	// System.err.println("ALIGN: " + nx + " " + ny);
	
	// s13 not declared in perl version
	double s1 = 0.0, s2 = 0.0, s3 = 0.0, s4 = 0.0, s5 = 0.0, s6 = 0.0, s7 = 0.0, s8 = 0.0, s9 = 0.0, s10 = 0.0, s11 = 0.0, s12 = 0.0, s13 = 0.0;
	int i, j, oi, oj;
	double si, sj, smax;
	int im1, im2, im3, im4, jm1, jm2, jm3, jm4;

	double xyratio = (double) nx / (double) ny;
	for (j = 0; j <= ny; j++) {

	    int center = (int) (j * xyratio);
	    int windowStart = center - dp.windowSize > 0 ? center - dp.windowSize : 0;
	    int windowEnd = center + dp.windowSize < nx ? center + dp.windowSize : nx;
	    // System.err.println(windowStart + " " + windowEnd);
	    for (i = windowStart; i <= windowEnd; i++) {
		// s1=s2=s3=s4=s5=s6=s7=s8=s9=s10=s11=s12=-100;
		im1 = i - 1;
		im2 = i - 2;
		im3 = i - 3;
		im4 = i - 4;
		jm1 = j - 1;
		jm2 = j - 2;
		jm3 = j - 3;
		jm4 = j - 4;

		s1 = i > 0 && j > 0 ? // 1-1
		getScore(i - 1, j - 1) + matchSentences(new int[] { im1 }, new int[] { jm1 })
			: MINS;		
		s2 = i > 0 ? // 1-0
		getScore(i - 1, j) + matchSentences(new int[] { im1 }, new int[0])
			: MINS;
		s3 = j > 0 ? // 0-1
		getScore(i, j - 1) + matchSentences(new int[0], new int[] { jm1 })
			: MINS;
		s4 = i > 1 && j > 0 ? // 2-1
		getScore(i - 2, j - 1) + matchSentences(new int[] { im2, im1 }, new int[] { jm1 })
			: MINS;
		s5 = i > 0 && j > 1 ? // 1-2
		getScore(i - 1, j - 2) + matchSentences(new int[] { im1 }, new int[] { jm2, jm1 })
			: MINS;
		s6 = i > 1 && j > 1 ? // 2-2
		getScore(i - 2, j - 2)
			+ matchSentences(new int[] { im2, im1 }, new int[] { jm2, jm1 })
			: MINS;

		if (!config.disallow3) {
		    s7 = i > 0 && j > 2 ? // 1-3
		    getScore(i - 1, j - 3)
			    + matchSentences(new int[] { im1 }, new int[] { jm3, jm2, jm1 })
			    : MINS;
		    s8 = i > 2 && j > 0 ? // 3-1
		    getScore(i - 3, j - 1)
			    + matchSentences(new int[] { im3, im2, im1 }, new int[] { jm1 })
			    : MINS;
		    s9 = i > 1 && j > 2 ? // 2-3
		    getScore(i - 2, j - 3)
			    + matchSentences(new int[] { im2, im1 }, new int[] { jm3, jm2, jm1 })
			    : MINS;
		    s10 = i > 2 && j > 1 ? // 3-2
		    getScore(i - 3, j - 2)
			    + matchSentences(new int[] { im3, im2, im1 }, new int[] { jm2, jm1 })
			    : MINS;
		    // s11 = i>2 && j>2 ? //// 3-3
		    // getScore(i-3, j-3) + matchSentences("im3 im2 im1 - jm3
		    // jm2 jm1")
		    // : MINS;
		    s12 = i > 0 && j > 3 ? // 1-4
		    getScore(i - 1, j - 4)
			    + matchSentences(new int[] { im1 }, new int[] { jm4, jm3, jm2, jm1 })
			    : MINS;
		    s13 = i > 3 && j > 0 ? // 4-1
		    getScore(i - 4, j - 1)
			    + matchSentences(new int[] { im4, im3, im2, im1 }, new int[] { jm1 })
			    : MINS;
		}

		smax = s1;
		if (s2 > smax) {
		    smax = s2;
		}
		if (s3 > smax) {
		    smax = s3;
		}
		if (s4 > smax) {
		    smax = s4;
		}
		if (s5 > smax) {
		    smax = s5;
		}
		if (s6 > smax) {
		    smax = s6;
		}
		if (s7 > smax) {
		    smax = s7;
		}
		if (s8 > smax) {
		    smax = s8;
		}
		if (s9 > smax) {
		    smax = s9;
		}
		if (s10 > smax) {
		    smax = s10;
		}
		// if(s11>smax) { smax=s11; }
		if (s12 > smax) {
		    smax = s12;
		}
		if (s13 > smax) {
		    smax = s13;
		}

//		 System.err.println("s: " + s1 + " " + s2 + " " + s3 + " " + s4 + " " + s5 + " "
//			+ s6 + " " + s7 + " " + s8 + " " + s9 + " " + s10 + " " + s11 + " " + s12
//			+ " " + s13);
		// System.err.println("MAX: " + smax);

		if (smax == MINS) {
		    setScore(i, j, 0);
		} else if (smax == s1) { // 1-1
		    setScore(i, j, s1);
		    pathX[i][j] = i - 1;
		    pathY[i][j] = j - 1;
		} else if (smax == s2) { // 1-0
		    setScore(i, j, s2);
		    pathX[i][j] = i - 1;
		    pathY[i][j] = j;
		} else if (smax == s3) { // 0-1
		    setScore(i, j, s3);
		    pathX[i][j] = i;
		    pathY[i][j] = j - 1;
		} else if (smax == s4) { // 2-1
		    setScore(i, j, s4);
		    pathX[i][j] = i - 2;
		    pathY[i][j] = j - 1;
		} else if (smax == s5) { // 1-2
		    setScore(i, j, s5);
		    pathX[i][j] = i - 1;
		    pathY[i][j] = j - 2;
		} else if (smax == s6) { // 2-2
		    setScore(i, j, s6);
		    pathX[i][j] = i - 2;
		    pathY[i][j] = j - 2;
		} else if (smax == s7) { // 1-3
		    setScore(i, j, s7);
		    pathX[i][j] = i - 1;
		    pathY[i][j] = j - 3;
		} else if (smax == s8) { // 3-1
		    setScore(i, j, s8);
		    pathX[i][j] = i - 3;
		    pathY[i][j] = j - 1;
		} else if (smax == s9) { // 2-3
		    setScore(i, j, s9);
		    pathX[i][j] = i - 2;
		    pathY[i][j] = j - 3;
		} else if (smax == s10) { // 3-2
		    setScore(i, j, s10);
		    pathX[i][j] = i - 3;
		    pathY[i][j] = j - 2;
		    // } else if (smax == s11) { // 3-3
		    // setScore(i,j,s11);
		    // pathX[i][j] = i-3;
		    // pathY[i][j] = j-3;
		} else if (smax == s12) { // 1-4
		    setScore(i, j, s12);
		    pathX[i][j] = i - 1;
		    pathY[i][j] = j - 4;
		} else if (smax == s13) { // 4-1
		    setScore(i, j, s13);
		    pathX[i][j] = i - 4;
		    pathY[i][j] = j - 1;
		}
		
//		System.err.println("smax: " + smax);
//		System.err.println("i,j: " + i + "," + j);
//		 System.err.println("BLEH X: " + pathX[i][j]);
//		 System.err.println("BLEH Y: " + pathY[i][j]);
	    }
//	    System.err.println("Finished iteration over: " + i);
	}
	
//	 System.err.println("nx=" + nx + " ny=" + ny);

	int n = 0;
	for (i = nx, j = ny; i > 0 || j > 0; i = oi, j = oj, n++) {
	    oi = pathX[i][j];
	    oj = pathY[i][j];
	    si = i - oi;
	    sj = j - oj;

//	     System.err.println("oi=" + oi + " oj=" + oj);

	    im1 = i - 1;
	    im2 = i - 2;
	    im3 = i - 3;
	    jm1 = j - 1;
	    jm2 = j - 2;
	    jm3 = j - 3;

	    if (si == 1 && sj == 1) { // 1-1
		dp.rAlign.add(i + " <=> " + j);
	    } else if (si == 1 && sj == 0) { // 1-0
		dp.rAlign.add(i + " <=> omitted");