FileDocCategorySizeDatePackage
DiffMyers.javaAPI DocAndroid 1.5 API30781Wed May 06 22:41:22 BST 2009jdiff

DiffMyers

public class DiffMyers extends Object
A class to compare vectors of objects. The result of comparison is a list of change objects which form an edit script. The objects compared are traditionally lines of text from two files. Comparison options such as "ignore whitespace" are implemented by modifying the equals and hashcode methods for the objects compared.

The basic algorithm is described in:
"An O(ND) Difference Algorithm and its Variations", Eugene Myers, Algorithmica Vol. 1 No. 2, 1986, p 251.

This class outputs different results from GNU diff 1.15 on some inputs. Our results are actually better (smaller change list, smaller total size of changes), but it would be nice to know why. Perhaps there is a memory overwrite bug in GNU diff 1.15.

author
Stuart D. Gathman, translated from GNU diff 1.15 Copyright (C) 2000 Business Management Systems, Inc.

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 1, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

Fields Summary
private int
equiv_max
1 more than the maximum equivalence value used for this or its sibling file.
public boolean
heuristic
When set to true, the comparison uses a heuristic to speed it up. With this heuristic, for files with a constant small density of changes, the algorithm is linear in the file size.
public boolean
no_discards
When set to true, the algorithm returns a guarranteed minimal set of changes. This makes things slower, sometimes much slower.
private int[]
xvec
private int[]
yvec
private int[]
fdiag
private int[]
bdiag
private int
fdiagoff
private int
bdiagoff
private final file_data[]
filevec
private int
cost
private boolean
inhibit
Constructors Summary
public DiffMyers(Object[] a, Object[] b)
Prepare to find differences between two arrays. Each element of the arrays is translated to an "equivalence number" based on the result of equals. The original Object arrays are no longer needed for computing the differences. They will be needed again later to print the results of the comparison as an edit script, if desired.

    Hashtable h = new Hashtable(a.length + b.length);
    filevec[0] = new file_data(a,h);
    filevec[1] = new file_data(b,h);
  
Methods Summary
private jdiff.DiffMyers$changebuild_reverse_script()
Scan the tables of which lines are inserted and deleted, producing an edit script in reverse order.

    change script = null;
    final boolean[] changed0 = filevec[0].changed_flag;
    final boolean[] changed1 = filevec[1].changed_flag;
    final int len0 = filevec[0].buffered_lines;
    final int len1 = filevec[1].buffered_lines;

    /* Note that changedN[len0] does exist, and contains 0.  */

    int i0 = 0, i1 = 0;

    while (i0 < len0 || i1 < len1)
      {
        if (changed0[1+i0] || changed1[1+i1])
          {
            int line0 = i0, line1 = i1;

            /* Find # lines changed here in each file.  */
            while (changed0[1+i0]) ++i0;
            while (changed1[1+i1]) ++i1;

            /* Record this change.  */
            script = new change(line0, line1, i0 - line0, i1 - line1, script);
          }

        /* We have reached lines in the two files that match each other.  */
        i0++; i1++;
      }

    return script;
  
private jdiff.DiffMyers$changebuild_script()
Scan the tables of which lines are inserted and deleted, producing an edit script in forward order.

    change script = null;
    final boolean[] changed0 = filevec[0].changed_flag;
    final boolean[] changed1 = filevec[1].changed_flag;
    final int len0 = filevec[0].buffered_lines;
    final int len1 = filevec[1].buffered_lines;
    int i0 = len0, i1 = len1;

    /* Note that changedN[-1] does exist, and contains 0.  */

    while (i0 >= 0 || i1 >= 0)
      {
        if (changed0[i0] || changed1[i1])
          {
            int line0 = i0, line1 = i1;

            /* Find # lines changed here in each file.  */
            while (changed0[i0]) --i0;
            while (changed1[i1]) --i1;

            /* Record this change.  */
            script = new change(i0, i1, line0 - i0, line1 - i1, script);
          }

        /* We have reached lines in the two files that match each other.  */
        i0--; i1--;
      }

    return script;
  
private voidcompareseq(int xoff, int xlim, int yoff, int ylim)
Compare in detail contiguous subsequences of the two files which are known, as a whole, to match each other. The results are recorded in the vectors filevec[N].changed_flag, by storing a 1 in the element for each line that is an insertion or deletion. The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1. Note that XLIM, YLIM are exclusive bounds. All line numbers are origin-0 and discarded lines are not counted.

    /* Slide down the bottom initial diagonal. */
    while (xoff < xlim && yoff < ylim && xvec[xoff] == yvec[yoff]) {
      ++xoff; ++yoff;
    }
    /* Slide up the top initial diagonal. */
    while (xlim > xoff && ylim > yoff && xvec[xlim - 1] == yvec[ylim - 1]) {
      --xlim; --ylim;
    }
    
    /* Handle simple cases. */
    if (xoff == xlim)
      while (yoff < ylim)
        filevec[1].changed_flag[1+filevec[1].realindexes[yoff++]] = true;
    else if (yoff == ylim)
      while (xoff < xlim)
        filevec[0].changed_flag[1+filevec[0].realindexes[xoff++]] = true;
    else
      {
        /* Find a point of correspondence in the middle of the files.  */

        int d = diag (xoff, xlim, yoff, ylim);
        int c = cost;
        int f = fdiag[fdiagoff + d];
        int b = bdiag[bdiagoff + d];

        if (c == 1)
          {
            /* This should be impossible, because it implies that
               one of the two subsequences is empty,
               and that case was handled above without calling `diag'.
               Let's verify that this is true.  */
            throw new IllegalArgumentException("Empty subsequence");
          }
        else
          {
            /* Use that point to split this problem into two subproblems.  */
            compareseq (xoff, b, yoff, b - d);
            /* This used to use f instead of b,
               but that is incorrect!
               It is not necessarily the case that diagonal d
               has a snake from b to f.  */
            compareseq (b, xlim, b - d, ylim);
          }
      }
  
private intdiag(int xoff, int xlim, int yoff, int ylim)
Find the midpoint of the shortest edit script for a specified portion of the two files. We scan from the beginnings of the files, and simultaneously from the ends, doing a breadth-first search through the space of edit-sequence. When the two searches meet, we have found the midpoint of the shortest edit sequence. The value returned is the number of the diagonal on which the midpoint lies. The diagonal number equals the number of inserted lines minus the number of deleted lines (counting only lines before the midpoint). The edit cost is stored into COST; this is the total number of lines inserted or deleted (counting only lines before the midpoint). This function assumes that the first lines of the specified portions of the two files do not match, and likewise that the last lines do not match. The caller must trim matching lines from the beginning and end of the portions it is going to specify. Note that if we return the "wrong" diagonal value, or if the value of bdiag at that diagonal is "wrong", the worst this can do is cause suboptimal diff output. It cannot cause incorrect diff output.


                                                                                                                                                                                                                                                                              

            
  
    final int[] fd = fdiag;        // Give the compiler a chance.
    final int[] bd = bdiag;        // Additional help for the compiler.
    final int[] xv = xvec;                // Still more help for the compiler.
    final int[] yv = yvec;                // And more and more . . .
    final int dmin = xoff - ylim;        // Minimum valid diagonal.
    final int dmax = xlim - yoff;        // Maximum valid diagonal.
    final int fmid = xoff - yoff;        // Center diagonal of top-down search.
    final int bmid = xlim - ylim;        // Center diagonal of bottom-up search.
    int fmin = fmid, fmax = fmid;        // Limits of top-down search.
    int bmin = bmid, bmax = bmid;        // Limits of bottom-up search.
    /* True if southeast corner is on an odd
                                     diagonal with respect to the northwest. */
    final boolean odd = (fmid - bmid & 1) != 0;        

    fd[fdiagoff + fmid] = xoff;
    bd[bdiagoff + bmid] = xlim;

    for (int c = 1;; ++c)
      {
        int d;                        /* Active diagonal. */
        boolean big_snake = false;

        /* Extend the top-down search by an edit step in each diagonal. */
        if (fmin > dmin)
          fd[fdiagoff + --fmin - 1] = -1;
        else
          ++fmin;
        if (fmax < dmax)
          fd[fdiagoff + ++fmax + 1] = -1;
        else
          --fmax;
        for (d = fmax; d >= fmin; d -= 2)
          {
            int x, y, oldx, tlo = fd[fdiagoff + d - 1], thi = fd[fdiagoff + d + 1];

            if (tlo >= thi)
              x = tlo + 1;
            else
              x = thi;
            oldx = x;
            y = x - d;
            while (x < xlim && y < ylim && xv[x] == yv[y]) {
              ++x; ++y;
            }
            if (x - oldx > 20)
              big_snake = true;
            fd[fdiagoff + d] = x;
            if (odd && bmin <= d && d <= bmax && bd[bdiagoff + d] <= fd[fdiagoff + d])
              {
                cost = 2 * c - 1;
                return d;
              }
          }

        /* Similar extend the bottom-up search. */
        if (bmin > dmin)
          bd[bdiagoff + --bmin - 1] = Integer.MAX_VALUE;
        else
          ++bmin;
        if (bmax < dmax)
          bd[bdiagoff + ++bmax + 1] = Integer.MAX_VALUE;
        else
          --bmax;
        for (d = bmax; d >= bmin; d -= 2)
          {
            int x, y, oldx, tlo = bd[bdiagoff + d - 1], thi = bd[bdiagoff + d + 1];

            if (tlo < thi)
              x = tlo;
            else
              x = thi - 1;
            oldx = x;
            y = x - d;
            while (x > xoff && y > yoff && xv[x - 1] == yv[y - 1]) {
              --x; --y;
            }
            if (oldx - x > 20)
              big_snake = true;
            bd[bdiagoff + d] = x;
            if (!odd && fmin <= d && d <= fmax && bd[bdiagoff + d] <= fd[fdiagoff + d])
              {
                cost = 2 * c;
                return d;
              }
          }

        /* Heuristic: check occasionally for a diagonal that has made
           lots of progress compared with the edit distance.
           If we have any such, find the one that has made the most
           progress and return it as if it had succeeded.

           With this heuristic, for files with a constant small density
           of changes, the algorithm is linear in the file size.  */

        if (c > 200 && big_snake && heuristic)
          {
            int best = 0;
            int bestpos = -1;

            for (d = fmax; d >= fmin; d -= 2)
              {
                int dd = d - fmid;
                if ((fd[fdiagoff + d] - xoff)*2 - dd > 12 * (c + (dd > 0 ? dd : -dd)))
                  {
                    if (fd[fdiagoff + d] * 2 - dd > best
                        && fd[fdiagoff + d] - xoff > 20
                        && fd[fdiagoff + d] - d - yoff > 20)
                      {
                        int k;
                        int x = fd[fdiagoff + d];

                        /* We have a good enough best diagonal;
                           now insist that it end with a significant snake.  */
                        for (k = 1; k <= 20; k++)
                          if (xvec[x - k] != yvec[x - d - k])
                            break;

                        if (k == 21)
                          {
                            best = fd[fdiagoff + d] * 2 - dd;
                            bestpos = d;
                          }
                      }
                  }
              }
            if (best > 0)
              {
                cost = 2 * c - 1;
                return bestpos;
              }

            best = 0;
            for (d = bmax; d >= bmin; d -= 2)
              {
                int dd = d - bmid;
                if ((xlim - bd[bdiagoff + d])*2 + dd > 12 * (c + (dd > 0 ? dd : -dd)))
                  {
                    if ((xlim - bd[bdiagoff + d]) * 2 + dd > best
                        && xlim - bd[bdiagoff + d] > 20
                        && ylim - (bd[bdiagoff + d] - d) > 20)
                      {
                        /* We have a good enough best diagonal;
                           now insist that it end with a significant snake.  */
                        int k;
                        int x = bd[bdiagoff + d];

                        for (k = 0; k < 20; k++)
                          if (xvec[x + k] != yvec[x - d + k])
                            break;
                        if (k == 20)
                          {
                            best = (xlim - bd[bdiagoff + d]) * 2 + dd;
                            bestpos = d;
                          }
                      }
                  }
              }
            if (best > 0)
              {
                cost = 2 * c - 1;
                return bestpos;
              }
          }
      }
  
public jdiff.DiffMyers$changediff_2(boolean reverse)


    /* Some lines are obviously insertions or deletions
       because they don't match anything.  Detect them now,
       and avoid even thinking about them in the main comparison algorithm.  */

    discard_confusing_lines ();

    /* Now do the main comparison algorithm, considering just the
       undiscarded lines.  */

    xvec = filevec[0].undiscarded;
    yvec = filevec[1].undiscarded;

    int diags =
      filevec[0].nondiscarded_lines + filevec[1].nondiscarded_lines + 3;
    fdiag = new int[diags];
    fdiagoff = filevec[1].nondiscarded_lines + 1;
    bdiag = new int[diags];
    bdiagoff = filevec[1].nondiscarded_lines + 1;

    compareseq (0, filevec[0].nondiscarded_lines,
                0, filevec[1].nondiscarded_lines);
    fdiag = null;
    bdiag = null;

    /* Modify the results slightly to make them prettier
       in cases where that can validly be done.  */

    shift_boundaries ();

    /* Get the results of comparison in the form of a chain
       of `struct change's -- an edit script.  */

    if (reverse)
      return build_reverse_script();
    else
      return build_script();
  
private voiddiscard_confusing_lines()
Discard lines from one file that have no matches in the other file.

    filevec[0].discard_confusing_lines(filevec[1]);
    filevec[1].discard_confusing_lines(filevec[0]);
  
private voidshift_boundaries()
Adjust inserts/deletes of blank lines to join changes as much as possible.


                     

     
    if (inhibit)
      return;
    filevec[0].shift_boundaries(filevec[1]);
    filevec[1].shift_boundaries(filevec[0]);