rsalz@uunet.uu.net (Rich Salz) (02/26/91)
Submitted-by: Paul Eggert <eggert@twinsun.com>
Posting-number: Volume 24, Issue 20
Archive-name: gnudiff1.15/part05
#! /bin/sh
# This is a shell archive. Remove anything before this line, then unpack
# it by saving it into a file and typing "sh file". To overwrite existing
# files, type "sh file -c". You can also feed this as standard input via
# unshar, or by typing "sh <file", e.g.. If this archive is complete, you
# will see the following message at the end:
# "End of archive 5 (of 8)."
# Contents: analyze.c
# Wrapped by eggert@ata on Mon Jan 7 11:25:30 1991
PATH=/bin:/usr/bin:/usr/ucb ; export PATH
if test -f 'analyze.c' -a "${1}" != "-c" ; then
echo shar: Will not clobber existing file \"'analyze.c'\"
else
echo shar: Extracting \"'analyze.c'\" \(26295 characters\)
sed "s/^X//" >'analyze.c' <<'END_OF_FILE'
X/* Analyze file differences for GNU DIFF.
X Copyright (C) 1988, 1989 Free Software Foundation, Inc.
X
XThis file is part of GNU DIFF.
X
XGNU DIFF is free software; you can redistribute it and/or modify
Xit under the terms of the GNU General Public License as published by
Xthe Free Software Foundation; either version 1, or (at your option)
Xany later version.
X
XGNU DIFF is distributed in the hope that it will be useful,
Xbut WITHOUT ANY WARRANTY; without even the implied warranty of
XMERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
XGNU General Public License for more details.
X
XYou should have received a copy of the GNU General Public License
Xalong with GNU DIFF; see the file COPYING. If not, write to
Xthe Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
X
X/* The basic algorithm is described in:
X "An O(ND) Difference Algorithm and its Variations", Eugene Myers,
X Algorithmica Vol. 1 No. 2, 1986, p 251. */
X
X#include "diff.h"
X
Xstruct change *find_change ();
Xvoid finish_output ();
Xvoid print_context_header ();
Xvoid print_context_script ();
Xvoid print_ed_script ();
Xvoid print_ifdef_script ();
Xvoid print_normal_script ();
Xvoid print_rcs_script ();
Xvoid pr_forward_ed_script ();
Xvoid setup_output ();
X
Xextern int no_discards;
X
Xstatic int *xvec, *yvec; /* Vectors being compared. */
Xstatic int *fdiag; /* Vector, indexed by diagonal, containing
X the X coordinate of the point furthest
X along the given diagonal in the forward
X search of the edit matrix. */
Xstatic int *bdiag; /* Vector, indexed by diagonal, containing
X the X coordinate of the point furthest
X along the given diagonal in the backward
X search of the edit matrix. */
X
X/* Find the midpoint of the shortest edit script for a specified
X portion of the two files.
X
X We scan from the beginnings of the files, and simultaneously from the ends,
X doing a breadth-first search through the space of edit-sequence.
X When the two searches meet, we have found the midpoint of the shortest
X edit sequence.
X
X The value returned is the number of the diagonal on which the midpoint lies.
X The diagonal number equals the number of inserted lines minus the number
X of deleted lines (counting only lines before the midpoint).
X The edit cost is stored into *COST; this is the total number of
X lines inserted or deleted (counting only lines before the midpoint).
X
X This function assumes that the first lines of the specified portions
X of the two files do not match, and likewise that the last lines do not
X match. The caller must trim matching lines from the beginning and end
X of the portions it is going to specify.
X
X Note that if we return the "wrong" diagonal value, or if
X the value of bdiag at that diagonal is "wrong",
X the worst this can do is cause suboptimal diff output.
X It cannot cause incorrect diff output. */
X
Xstatic int
Xdiag (xoff, xlim, yoff, ylim, cost)
X int xoff, xlim, yoff, ylim;
X int *cost;
X{
X int *const fd = fdiag; /* Give the compiler a chance. */
X int *const bd = bdiag; /* Additional help for the compiler. */
X int *const xv = xvec; /* Still more help for the compiler. */
X int *const yv = yvec; /* And more and more . . . */
X const int dmin = xoff - ylim; /* Minimum valid diagonal. */
X const int dmax = xlim - yoff; /* Maximum valid diagonal. */
X const int fmid = xoff - yoff; /* Center diagonal of top-down search. */
X const int bmid = xlim - ylim; /* Center diagonal of bottom-up search. */
X int fmin = fmid, fmax = fmid; /* Limits of top-down search. */
X int bmin = bmid, bmax = bmid; /* Limits of bottom-up search. */
X int c; /* Cost. */
X int odd = fmid - bmid & 1; /* True if southeast corner is on an odd
X diagonal with respect to the northwest. */
X
X fd[fmid] = xoff;
X bd[bmid] = xlim;
X
X for (c = 1;; ++c)
X {
X int d; /* Active diagonal. */
X int big_snake = 0;
X
X /* Extend the top-down search by an edit step in each diagonal. */
X fmin > dmin ? fd[--fmin - 1] = -1 : ++fmin;
X fmax < dmax ? fd[++fmax + 1] = -1 : --fmax;
X for (d = fmax; d >= fmin; d -= 2)
X {
X int x, y, oldx, tlo = fd[d - 1], thi = fd[d + 1];
X
X if (tlo >= thi)
X x = tlo + 1;
X else
X x = thi;
X oldx = x;
X y = x - d;
X while (x < xlim && y < ylim && xv[x] == yv[y])
X ++x, ++y;
X if (x - oldx > 20)
X big_snake = 1;
X fd[d] = x;
X if (odd && bmin <= d && d <= bmax && bd[d] <= fd[d])
X {
X *cost = 2 * c - 1;
X return d;
X }
X }
X
X /* Similar extend the bottom-up search. */
X bmin > dmin ? bd[--bmin - 1] = INT_MAX : ++bmin;
X bmax < dmax ? bd[++bmax + 1] = INT_MAX : --bmax;
X for (d = bmax; d >= bmin; d -= 2)
X {
X int x, y, oldx, tlo = bd[d - 1], thi = bd[d + 1];
X
X if (tlo < thi)
X x = tlo;
X else
X x = thi - 1;
X oldx = x;
X y = x - d;
X while (x > xoff && y > yoff && xv[x - 1] == yv[y - 1])
X --x, --y;
X if (oldx - x > 20)
X big_snake = 1;
X bd[d] = x;
X if (!odd && fmin <= d && d <= fmax && bd[d] <= fd[d])
X {
X *cost = 2 * c;
X return d;
X }
X }
X
X /* Heuristic: check occasionally for a diagonal that has made
X lots of progress compared with the edit distance.
X If we have any such, find the one that has made the most
X progress and return it as if it had succeeded.
X
X With this heuristic, for files with a constant small density
X of changes, the algorithm is linear in the file size. */
X
X if (c > 200 && big_snake && heuristic)
X {
X int best;
X int bestpos;
X
X best = 0;
X for (d = fmax; d >= fmin; d -= 2)
X {
X int dd = d - fmid;
X if ((fd[d] - xoff)*2 - dd > 12 * (c + (dd > 0 ? dd : -dd)))
X {
X if (fd[d] * 2 - dd > best
X && fd[d] - xoff > 20
X && fd[d] - d - yoff > 20)
X {
X int k;
X int x = fd[d];
X
X /* We have a good enough best diagonal;
X now insist that it end with a significant snake. */
X for (k = 1; k <= 20; k++)
X if (xvec[x - k] != yvec[x - d - k])
X break;
X
X if (k == 21)
X {
X best = fd[d] * 2 - dd;
X bestpos = d;
X }
X }
X }
X }
X if (best > 0)
X {
X *cost = 2 * c - 1;
X return bestpos;
X }
X
X best = 0;
X for (d = bmax; d >= bmin; d -= 2)
X {
X int dd = d - bmid;
X if ((xlim - bd[d])*2 + dd > 12 * (c + (dd > 0 ? dd : -dd)))
X {
X if ((xlim - bd[d]) * 2 + dd > best
X && xlim - bd[d] > 20
X && ylim - (bd[d] - d) > 20)
X {
X /* We have a good enough best diagonal;
X now insist that it end with a significant snake. */
X int k;
X int x = bd[d];
X
X for (k = 0; k < 20; k++)
X if (xvec[x + k] != yvec[x - d + k])
X break;
X if (k == 20)
X {
X best = (xlim - bd[d]) * 2 + dd;
X bestpos = d;
X }
X }
X }
X }
X if (best > 0)
X {
X *cost = 2 * c - 1;
X return bestpos;
X }
X }
X }
X}
X
X/* Compare in detail contiguous subsequences of the two files
X which are known, as a whole, to match each other.
X
X The results are recorded in the vectors files[N].changed_flag, by
X storing a 1 in the element for each line that is an insertion or deletion.
X
X The subsequence of file 0 is [XOFF, XLIM) and likewise for file 1.
X
X Note that XLIM, YLIM are exclusive bounds.
X All line numbers are origin-0 and discarded lines are not counted. */
X
Xstatic void
Xcompareseq (xoff, xlim, yoff, ylim)
X int xoff, xlim, yoff, ylim;
X{
X /* Slide down the bottom initial diagonal. */
X while (xoff < xlim && yoff < ylim && xvec[xoff] == yvec[yoff])
X ++xoff, ++yoff;
X /* Slide up the top initial diagonal. */
X while (xlim > xoff && ylim > yoff && xvec[xlim - 1] == yvec[ylim - 1])
X --xlim, --ylim;
X
X /* Handle simple cases. */
X if (xoff == xlim)
X while (yoff < ylim)
X files[1].changed_flag[files[1].realindexes[yoff++]] = 1;
X else if (yoff == ylim)
X while (xoff < xlim)
X files[0].changed_flag[files[0].realindexes[xoff++]] = 1;
X else
X {
X int c, d, f, b;
X
X /* Find a point of correspondence in the middle of the files. */
X
X d = diag (xoff, xlim, yoff, ylim, &c);
X f = fdiag[d];
X b = bdiag[d];
X
X if (c == 1)
X {
X /* This should be impossible, because it implies that
X one of the two subsequences is empty,
X and that case was handled above without calling `diag'.
X Let's verify that this is true. */
X abort ();
X#if 0
X /* The two subsequences differ by a single insert or delete;
X record it and we are done. */
X if (d < xoff - yoff)
X files[1].changed_flag[files[1].realindexes[b - d - 1]] = 1;
X else
X files[0].changed_flag[files[0].realindexes[b]] = 1;
X#endif
X }
X else
X {
X /* Use that point to split this problem into two subproblems. */
X compareseq (xoff, b, yoff, b - d);
X /* This used to use f instead of b,
X but that is incorrect!
X It is not necessarily the case that diagonal d
X has a snake from b to f. */
X compareseq (b, xlim, b - d, ylim);
X }
X }
X}
X
X/* Discard lines from one file that have no matches in the other file.
X
X A line which is discarded will not be considered by the actual
X comparison algorithm; it will be as if that line were not in the file.
X The file's `realindexes' table maps virtual line numbers
X (which don't count the discarded lines) into real line numbers;
X this is how the actual comparison algorithm produces results
X that are comprehensible when the discarded lines are counted.
X
X When we discard a line, we also mark it as a deletion or insertion
X so that it will be printed in the output. */
X
Xvoid
Xdiscard_confusing_lines (filevec)
X struct file_data filevec[];
X{
X unsigned int f, i;
X char *discarded[2];
X int *equiv_count[2];
X
X /* Allocate our results. */
X for (f = 0; f < 2; f++)
X {
X filevec[f].undiscarded
X = (int *) xmalloc (filevec[f].buffered_lines * sizeof (int));
X filevec[f].realindexes
X = (int *) xmalloc (filevec[f].buffered_lines * sizeof (int));
X }
X
X /* Set up equiv_count[F][I] as the number of lines in file F
X that fall in equivalence class I. */
X
X equiv_count[0] = (int *) xmalloc (filevec[0].equiv_max * sizeof (int));
X bzero (equiv_count[0], filevec[0].equiv_max * sizeof (int));
X equiv_count[1] = (int *) xmalloc (filevec[1].equiv_max * sizeof (int));
X bzero (equiv_count[1], filevec[1].equiv_max * sizeof (int));
X
X for (i = 0; i < filevec[0].buffered_lines; ++i)
X ++equiv_count[0][filevec[0].equivs[i]];
X for (i = 0; i < filevec[1].buffered_lines; ++i)
X ++equiv_count[1][filevec[1].equivs[i]];
X
X /* Set up tables of which lines are going to be discarded. */
X
X discarded[0] = (char *) xmalloc (filevec[0].buffered_lines);
X discarded[1] = (char *) xmalloc (filevec[1].buffered_lines);
X bzero (discarded[0], filevec[0].buffered_lines);
X bzero (discarded[1], filevec[1].buffered_lines);
X
X /* Mark to be discarded each line that matches no line of the other file.
X If a line matches many lines, mark it as provisionally discardable. */
X
X for (f = 0; f < 2; f++)
X {
X unsigned int end = filevec[f].buffered_lines;
X char *discards = discarded[f];
X int *counts = equiv_count[1 - f];
X int *equivs = filevec[f].equivs;
X unsigned int many = 5;
X unsigned int tem = end / 64;
X
X /* Multiply MANY by approximate square root of number of lines.
X That is the threshold for provisionally discardable lines. */
X while ((tem = tem >> 2) > 0)
X many *= 2;
X
X for (i = 0; i < end; i++)
X {
X int nmatch;
X if (equivs[i] == 0)
X continue;
X nmatch = counts[equivs[i]];
X if (nmatch == 0)
X discards[i] = 1;
X else if (nmatch > many)
X discards[i] = 2;
X }
X }
X
X /* Don't really discard the provisional lines except when they occur
X in a run of discardables, with nonprovisionals at the beginning
X and end. */
X
X for (f = 0; f < 2; f++)
X {
X unsigned int end = filevec[f].buffered_lines;
X register char *discards = discarded[f];
X
X for (i = 0; i < end; i++)
X {
X /* Cancel provisional discards not in middle of run of discards. */
X if (discards[i] == 2)
X discards[i] = 0;
X else if (discards[i] != 0)
X {
X /* We have found a nonprovisional discard. */
X register int j;
X unsigned int length;
X unsigned int provisional = 0;
X
X /* Find end of this run of discardable lines.
X Count how many are provisionally discardable. */
X for (j = i; j < end; j++)
X {
X if (discards[j] == 0)
X break;
X if (discards[j] == 2)
X ++provisional;
X }
X
X /* Cancel provisional discards at end, and shrink the run. */
X while (j > i && discards[j - 1] == 2)
X discards[--j] = 0, --provisional;
X
X /* Now we have the length of a run of discardable lines
X whose first and last are not provisional. */
X length = j - i;
X
X /* If 1/4 of the lines in the run are provisional,
X cancel discarding of all provisional lines in the run. */
X if (provisional * 4 > length)
X {
X while (j > i)
X if (discards[--j] == 2)
X discards[j] = 0;
X }
X else
X {
X register unsigned int consec;
X unsigned int minimum = 1;
X unsigned int tem = length / 4;
X
X /* MINIMUM is approximate square root of LENGTH/4.
X A subrun of two or more provisionals can stand
X when LENGTH is at least 16.
X A subrun of 4 or more can stand when LENGTH >= 64. */
X while ((tem = tem >> 2) > 0)
X minimum *= 2;
X minimum++;
X
X /* Cancel any subrun of MINIMUM or more provisionals
X within the larger run. */
X for (j = 0, consec = 0; j < length; j++)
X if (discards[i + j] != 2)
X consec = 0;
X else if (minimum == ++consec)
X /* Back up to start of subrun, to cancel it all. */
X j -= consec;
X else if (minimum < consec)
X discards[i + j] = 0;
X
X /* Scan from beginning of run
X until we find 3 or more nonprovisionals in a row
X or until the first nonprovisional at least 8 lines in.
X Until that point, cancel any provisionals. */
X for (j = 0, consec = 0; j < length; j++)
X {
X if (j >= 8 && discards[i + j] == 1)
X break;
X if (discards[i + j] == 2)
X consec = 0, discards[i + j] = 0;
X else if (discards[i + j] == 0)
X consec = 0;
X else
X consec++;
X if (consec == 3)
X break;
X }
X
X /* I advances to the last line of the run. */
X i += length - 1;
X
X /* Same thing, from end. */
X for (j = 0, consec = 0; j < length; j++)
X {
X if (j >= 8 && discards[i - j] == 1)
X break;
X if (discards[i - j] == 2)
X consec = 0, discards[i - j] = 0;
X else if (discards[i - j] == 0)
X consec = 0;
X else
X consec++;
X if (consec == 3)
X break;
X }
X }
X }
X }
X }
X
X /* Actually discard the lines. */
X for (f = 0; f < 2; f++)
X {
X char *discards = discarded[f];
X unsigned int end = filevec[f].buffered_lines;
X unsigned int j = 0;
X for (i = 0; i < end; ++i)
X if (no_discards || discards[i] == 0)
X {
X filevec[f].undiscarded[j] = filevec[f].equivs[i];
X filevec[f].realindexes[j++] = i;
X }
X else
X filevec[f].changed_flag[i] = 1;
X filevec[f].nondiscarded_lines = j;
X }
X
X free (discarded[1]);
X free (discarded[0]);
X free (equiv_count[1]);
X free (equiv_count[0]);
X}
X
X/* Adjust inserts/deletes of blank lines to join changes
X as much as possible.
X
X We do something when a run of changed lines include a blank
X line at one end and have an excluded blank line at the other.
X We are free to choose which blank line is included.
X `compareseq' always chooses the one at the beginning,
X but usually it is cleaner to consider the following blank line
X to be the "change". The only exception is if the preceding blank line
X would join this change to other changes. */
X
Xint inhibit;
X
Xstatic void
Xshift_boundaries (filevec)
X struct file_data filevec[];
X{
X int f;
X
X if (inhibit)
X return;
X
X for (f = 0; f < 2; f++)
X {
X char *changed = filevec[f].changed_flag;
X char *other_changed = filevec[1-f].changed_flag;
X int i = 0;
X int j = 0;
X int i_end = filevec[f].buffered_lines;
X int preceding = -1;
X int other_preceding = -1;
X
X while (1)
X {
X int start, end, other_start;
X
X /* Scan forwards to find beginning of another run of changes.
X Also keep track of the corresponding point in the other file. */
X
X while (i < i_end && changed[i] == 0)
X {
X while (other_changed[j++])
X /* Non-corresponding lines in the other file
X will count as the preceding batch of changes. */
X other_preceding = j;
X i++;
X }
X
X if (i == i_end)
X break;
X
X start = i;
X other_start = j;
X
X while (1)
X {
X /* Now find the end of this run of changes. */
X
X while (i < i_end && changed[i] != 0) i++;
X end = i;
X
X /* If the first changed line matches the following unchanged one,
X and this run does not follow right after a previous run,
X and there are no lines deleted from the other file here,
X then classify the first changed line as unchanged
X and the following line as changed in its place. */
X
X /* You might ask, how could this run follow right after another?
X Only because the previous run was shifted here. */
X
X if (end != i_end
X && files[f].equivs[start] == files[f].equivs[end]
X && !other_changed[j]
X && end != i_end
X && !((preceding >= 0 && start == preceding)
X || (other_preceding >= 0
X && other_start == other_preceding)))
X {
X changed[end++] = 1;
X changed[start++] = 0;
X ++i;
X /* Since one line-that-matches is now before this run
X instead of after, we must advance in the other file
X to keep in synch. */
X ++j;
X }
X else
X break;
X }
X
X preceding = i;
X other_preceding = j;
X }
X }
X}
X
X/* Cons an additional entry onto the front of an edit script OLD.
X LINE0 and LINE1 are the first affected lines in the two files (origin 0).
X DELETED is the number of lines deleted here from file 0.
X INSERTED is the number of lines inserted here in file 1.
X
X If DELETED is 0 then LINE0 is the number of the line before
X which the insertion was done; vice versa for INSERTED and LINE1. */
X
Xstatic struct change *
Xadd_change (line0, line1, deleted, inserted, old)
X int line0, line1, deleted, inserted;
X struct change *old;
X{
X struct change *new = (struct change *) xmalloc (sizeof (struct change));
X
X new->line0 = line0;
X new->line1 = line1;
X new->inserted = inserted;
X new->deleted = deleted;
X new->link = old;
X return new;
X}
X
X/* Scan the tables of which lines are inserted and deleted,
X producing an edit script in reverse order. */
X
Xstatic struct change *
Xbuild_reverse_script (filevec)
X struct file_data filevec[];
X{
X struct change *script = 0;
X char *changed0 = filevec[0].changed_flag;
X char *changed1 = filevec[1].changed_flag;
X int len0 = filevec[0].buffered_lines;
X int len1 = filevec[1].buffered_lines;
X
X /* Note that changedN[len0] does exist, and contains 0. */
X
X int i0 = 0, i1 = 0;
X
X while (i0 < len0 || i1 < len1)
X {
X if (changed0[i0] || changed1[i1])
X {
X int line0 = i0, line1 = i1;
X
X /* Find # lines changed here in each file. */
X while (changed0[i0]) ++i0;
X while (changed1[i1]) ++i1;
X
X /* Record this change. */
X script = add_change (line0, line1, i0 - line0, i1 - line1, script);
X }
X
X /* We have reached lines in the two files that match each other. */
X i0++, i1++;
X }
X
X return script;
X}
X
X/* Scan the tables of which lines are inserted and deleted,
X producing an edit script in forward order. */
X
Xstatic struct change *
Xbuild_script (filevec)
X struct file_data filevec[];
X{
X struct change *script = 0;
X char *changed0 = filevec[0].changed_flag;
X char *changed1 = filevec[1].changed_flag;
X int len0 = filevec[0].buffered_lines;
X int len1 = filevec[1].buffered_lines;
X int i0 = len0, i1 = len1;
X
X /* Note that changedN[-1] does exist, and contains 0. */
X
X#if 0 /* Unnecessary since a line includes its trailing newline. */
X /* In RCS comparisons, making the existence or nonexistence of trailing
X newlines really matter. */
X if (output_style == OUTPUT_RCS
X && filevec[0].missing_newline != filevec[1].missing_newline)
X changed0[len0 - 1] = changed1[len1 - 1] = 1;
X#endif
X
X while (i0 >= 0 || i1 >= 0)
X {
X if (changed0[i0 - 1] || changed1[i1 - 1])
X {
X int line0 = i0, line1 = i1;
X
X /* Find # lines changed here in each file. */
X while (changed0[i0 - 1]) --i0;
X while (changed1[i1 - 1]) --i1;
X
X /* Record this change. */
X script = add_change (i0, i1, line0 - i0, line1 - i1, script);
X }
X
X /* We have reached lines in the two files that match each other. */
X i0--, i1--;
X }
X
X return script;
X}
X
X/* Report the differences of two files. DEPTH is the current directory
X depth. */
Xint
Xdiff_2_files (filevec, depth)
X struct file_data filevec[];
X int depth;
X{
X int diags;
X int i;
X struct change *e, *p;
X struct change *script;
X int binary;
X int changes;
X
X /* See if the two named files are actually the same physical file.
X If so, we know they are identical without actually reading them. */
X
X if (output_style != OUTPUT_IFDEF
X && filevec[0].stat.st_ino == filevec[1].stat.st_ino
X && filevec[0].stat.st_dev == filevec[1].stat.st_dev)
X return 0;
X
X binary = read_files (filevec);
X
X /* If we have detected that file 0 is a binary file,
X compare the two files as binary. This can happen
X only when the first chunk is read.
X Also, -q means treat all files as binary. */
X
X if (binary || no_details_flag)
X {
X int differs = (filevec[0].buffered_chars != filevec[1].buffered_chars
X || bcmp (filevec[0].buffer, filevec[1].buffer,
X filevec[1].buffered_chars));
X if (differs)
X message (binary ? "Binary files %s and %s differ\n"
X : "Files %s and %s differ\n",
X filevec[0].name, filevec[1].name);
X
X for (i = 0; i < 2; ++i)
X if (filevec[i].buffer)
X free (filevec[i].buffer);
X return differs;
X }
X
X /* Allocate vectors for the results of comparison:
X a flag for each line of each file, saying whether that line
X is an insertion or deletion.
X Allocate an extra element, always zero, at each end of each vector. */
X
X filevec[0].changed_flag = (char *) xmalloc (filevec[0].buffered_lines + 2);
X filevec[1].changed_flag = (char *) xmalloc (filevec[1].buffered_lines + 2);
X bzero (filevec[0].changed_flag, filevec[0].buffered_lines + 2);
X bzero (filevec[1].changed_flag, filevec[1].buffered_lines + 2);
X filevec[0].changed_flag++;
X filevec[1].changed_flag++;
X
X /* Some lines are obviously insertions or deletions
X because they don't match anything. Detect them now,
X and avoid even thinking about them in the main comparison algorithm. */
X
X discard_confusing_lines (filevec);
X
X /* Now do the main comparison algorithm, considering just the
X undiscarded lines. */
X
X xvec = filevec[0].undiscarded;
X yvec = filevec[1].undiscarded;
X diags = filevec[0].nondiscarded_lines + filevec[1].nondiscarded_lines + 3;
X fdiag = (int *) xmalloc (diags * sizeof (int));
X fdiag += filevec[1].nondiscarded_lines + 1;
X bdiag = (int *) xmalloc (diags * sizeof (int));
X bdiag += filevec[1].nondiscarded_lines + 1;
X
X files[0] = filevec[0];
X files[1] = filevec[1];
X
X compareseq (0, filevec[0].nondiscarded_lines,
X 0, filevec[1].nondiscarded_lines);
X
X bdiag -= filevec[1].nondiscarded_lines + 1;
X free (bdiag);
X fdiag -= filevec[1].nondiscarded_lines + 1;
X free (fdiag);
X
X /* Modify the results slightly to make them prettier
X in cases where that can validly be done. */
X
X shift_boundaries (filevec);
X
X /* Get the results of comparison in the form of a chain
X of `struct change's -- an edit script. */
X
X if (output_style == OUTPUT_ED)
X script = build_reverse_script (filevec);
X else
X script = build_script (filevec);
X
X if (script || output_style == OUTPUT_IFDEF)
X {
X setup_output (files[0].name, files[1].name, depth);
X
X switch (output_style)
X {
X case OUTPUT_CONTEXT:
X print_context_header (files, 0);
X print_context_script (script, 0);
X break;
X
X case OUTPUT_UNIFIED:
X print_context_header (files, 1);
X print_context_script (script, 1);
X break;
X
X case OUTPUT_ED:
X print_ed_script (script);
X break;
X
X case OUTPUT_FORWARD_ED:
X pr_forward_ed_script (script);
X break;
X
X case OUTPUT_RCS:
X print_rcs_script (script);
X break;
X
X case OUTPUT_NORMAL:
X print_normal_script (script);
X break;
X
X case OUTPUT_IFDEF:
X print_ifdef_script (script);
X break;
X }
X
X finish_output ();
X }
X
X /* Set CHANGES if we had any diffs that were printed.
X If some changes are being ignored, we must scan the script to decide. */
X if (ignore_blank_lines_flag || ignore_regexp)
X {
X struct change *next = script;
X changes = 0;
X
X while (next && changes == 0)
X {
X struct change *this, *end;
X int first0, last0, first1, last1, deletes, inserts;
X
X /* Find a set of changes that belong together. */
X this = next;
X end = find_change (next);
X
X /* Disconnect them from the rest of the changes,
X making them a hunk, and remember the rest for next iteration. */
X next = end->link;
X end->link = NULL;
X
X /* Determine whether this hunk was printed. */
X analyze_hunk (this, &first0, &last0, &first1, &last1,
X &deletes, &inserts);
X
X /* Reconnect the script so it will all be freed properly. */
X end->link = next;
X
X if (deletes || inserts)
X changes = 1;
X }
X }
X else
X changes = (script != 0);
X
X for (i = 1; i >= 0; --i)
X {
X free (filevec[i].realindexes);
X free (filevec[i].undiscarded);
X }
X
X for (i = 1; i >= 0; --i)
X free (--filevec[i].changed_flag);
X
X for (i = 1; i >= 0; --i)
X free (filevec[i].equivs);
X
X for (i = 0; i < 2; ++i)
X {
X if (filevec[i].buffer != 0)
X free (filevec[i].buffer);
X free (filevec[i].linbuf);
X }
X
X for (e = script; e; e = p)
X {
X p = e->link;
X free (e);
X }
X
X if (! ROBUST_OUTPUT_STYLE (output_style)
X /* For -D, invent newlines silently. That's ok in C code. */
X && output_style != OUTPUT_IFDEF)
X for (i = 0; i < 2; ++i)
X if (filevec[i].missing_newline)
X {
X error ("No newline at end of file %s", filevec[i].name, "");
X changes = 2;
X }
X
X return changes;
X}
END_OF_FILE
if test 26295 -ne `wc -c <'analyze.c'`; then
echo shar: \"'analyze.c'\" unpacked with wrong size!
fi
# end of 'analyze.c'
fi
echo shar: End of archive 5 \(of 8\).
cp /dev/null ark5isdone
MISSING=""
for I in 1 2 3 4 5 6 7 8 ; do
if test ! -f ark${I}isdone ; then
MISSING="${MISSING} ${I}"
fi
done
if test "${MISSING}" = "" ; then
echo You have unpacked all 8 archives.
rm -f ark[1-9]isdone
else
echo You still need to unpack the following archives:
echo " " ${MISSING}
fi
## End of shell archive.
exit 0
exit 0 # Just in case...
--
Please send comp.sources.unix-related mail to rsalz@uunet.uu.net.
Use a domain-based address or give alternate paths, or you may lose out.