2 * LibXDiff by Davide Libenzi ( File Differential Library )
3 * Copyright (C) 2003 Davide Libenzi
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public
7 * License as published by the Free Software Foundation; either
8 * version 2.1 of the License, or (at your option) any later version.
10 * This library is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Lesser General Public License for more details.
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with this library; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19 * Davide Libenzi <davidel@xmailserver.org>
27 #define XDL_MAX_COST_MIN 256
28 #define XDL_HEUR_MIN_COST 256
29 #define XDL_LINE_MAX (long)((1UL << (CHAR_BIT * sizeof(long) - 1)) - 1)
30 #define XDL_SNAKE_CNT 20
35 typedef struct s_xdpsplit {
43 static long xdl_split(unsigned long const *ha1, long off1, long lim1,
44 unsigned long const *ha2, long off2, long lim2,
45 long *kvdf, long *kvdb, int need_min, xdpsplit_t *spl,
47 static xdchange_t *xdl_add_change(xdchange_t *xscr, long i1, long i2, long chg1, long chg2);
54 * See "An O(ND) Difference Algorithm and its Variations", by Eugene Myers.
55 * Basically considers a "box" (off1, off2, lim1, lim2) and scan from both
56 * the forward diagonal starting from (off1, off2) and the backward diagonal
57 * starting from (lim1, lim2). If the K values on the same diagonal crosses
58 * returns the furthest point of reach. We might end up having to expensive
59 * cases using this algorithm is full, so a little bit of heuristic is needed
60 * to cut the search and to return a suboptimal point.
62 static long xdl_split(unsigned long const *ha1, long off1, long lim1,
63 unsigned long const *ha2, long off2, long lim2,
64 long *kvdf, long *kvdb, int need_min, xdpsplit_t *spl,
66 long dmin = off1 - lim2, dmax = lim1 - off2;
67 long fmid = off1 - off2, bmid = lim1 - lim2;
68 long odd = (fmid - bmid) & 1;
69 long fmin = fmid, fmax = fmid;
70 long bmin = bmid, bmax = bmid;
71 long ec, d, i1, i2, prev1, best, dd, v, k;
74 * Set initial diagonal values for both forward and backward path.
83 * We need to extent the diagonal "domain" by one. If the next
84 * values exits the box boundaries we need to change it in the
85 * opposite direction because (max - min) must be a power of two.
86 * Also we initialize the external K value to -1 so that we can
87 * avoid extra conditions check inside the core loop.
90 kvdf[--fmin - 1] = -1;
94 kvdf[++fmax + 1] = -1;
98 for (d = fmax; d >= fmin; d -= 2) {
99 if (kvdf[d - 1] >= kvdf[d + 1])
100 i1 = kvdf[d - 1] + 1;
105 for (; i1 < lim1 && i2 < lim2 && ha1[i1] == ha2[i2]; i1++, i2++);
106 if (i1 - prev1 > xenv->snake_cnt)
109 if (odd && bmin <= d && d <= bmax && kvdb[d] <= i1) {
112 spl->min_lo = spl->min_hi = 1;
118 * We need to extent the diagonal "domain" by one. If the next
119 * values exits the box boundaries we need to change it in the
120 * opposite direction because (max - min) must be a power of two.
121 * Also we initialize the external K value to -1 so that we can
122 * avoid extra conditions check inside the core loop.
125 kvdb[--bmin - 1] = XDL_LINE_MAX;
129 kvdb[++bmax + 1] = XDL_LINE_MAX;
133 for (d = bmax; d >= bmin; d -= 2) {
134 if (kvdb[d - 1] < kvdb[d + 1])
137 i1 = kvdb[d + 1] - 1;
140 for (; i1 > off1 && i2 > off2 && ha1[i1 - 1] == ha2[i2 - 1]; i1--, i2--);
141 if (prev1 - i1 > xenv->snake_cnt)
144 if (!odd && fmin <= d && d <= fmax && i1 <= kvdf[d]) {
147 spl->min_lo = spl->min_hi = 1;
156 * If the edit cost is above the heuristic trigger and if
157 * we got a good snake, we sample current diagonals to see
158 * if some of the, have reached an "interesting" path. Our
159 * measure is a function of the distance from the diagonal
160 * corner (i1 + i2) penalized with the distance from the
161 * mid diagonal itself. If this value is above the current
162 * edit cost times a magic factor (XDL_K_HEUR) we consider
165 if (got_snake && ec > xenv->heur_min) {
166 for (best = 0, d = fmax; d >= fmin; d -= 2) {
167 dd = d > fmid ? d - fmid: fmid - d;
170 v = (i1 - off1) + (i2 - off2) - dd;
172 if (v > XDL_K_HEUR * ec && v > best &&
173 off1 + xenv->snake_cnt <= i1 && i1 < lim1 &&
174 off2 + xenv->snake_cnt <= i2 && i2 < lim2) {
175 for (k = 1; ha1[i1 - k] == ha2[i2 - k]; k++)
176 if (k == xenv->snake_cnt) {
190 for (best = 0, d = bmax; d >= bmin; d -= 2) {
191 dd = d > bmid ? d - bmid: bmid - d;
194 v = (lim1 - i1) + (lim2 - i2) - dd;
196 if (v > XDL_K_HEUR * ec && v > best &&
197 off1 < i1 && i1 <= lim1 - xenv->snake_cnt &&
198 off2 < i2 && i2 <= lim2 - xenv->snake_cnt) {
199 for (k = 0; ha1[i1 + k] == ha2[i2 + k]; k++)
200 if (k == xenv->snake_cnt - 1) {
216 * Enough is enough. We spent too much time here and now we collect
217 * the furthest reaching path using the (i1 + i2) measure.
219 if (ec >= xenv->mxcost) {
220 long fbest, fbest1, bbest, bbest1;
223 for (d = fmax; d >= fmin; d -= 2) {
224 i1 = XDL_MIN(kvdf[d], lim1);
227 i1 = lim2 + d, i2 = lim2;
228 if (fbest < i1 + i2) {
234 bbest = bbest1 = XDL_LINE_MAX;
235 for (d = bmax; d >= bmin; d -= 2) {
236 i1 = XDL_MAX(off1, kvdb[d]);
239 i1 = off2 + d, i2 = off2;
240 if (i1 + i2 < bbest) {
246 if ((lim1 + lim2) - bbest < fbest - (off1 + off2)) {
248 spl->i2 = fbest - fbest1;
253 spl->i2 = bbest - bbest1;
264 * Rule: "Divide et Impera". Recursively split the box in sub-boxes by calling
265 * the box splitting function. Note that the real job (marking changed lines)
266 * is done in the two boundary reaching checks.
268 int xdl_recs_cmp(diffdata_t *dd1, long off1, long lim1,
269 diffdata_t *dd2, long off2, long lim2,
270 long *kvdf, long *kvdb, int need_min, xdalgoenv_t *xenv) {
271 unsigned long const *ha1 = dd1->ha, *ha2 = dd2->ha;
274 * Shrink the box by walking through each diagonal snake (SW and NE).
276 for (; off1 < lim1 && off2 < lim2 && ha1[off1] == ha2[off2]; off1++, off2++);
277 for (; off1 < lim1 && off2 < lim2 && ha1[lim1 - 1] == ha2[lim2 - 1]; lim1--, lim2--);
280 * If one dimension is empty, then all records on the other one must
281 * be obviously changed.
284 char *rchg2 = dd2->rchg;
285 long *rindex2 = dd2->rindex;
287 for (; off2 < lim2; off2++)
288 rchg2[rindex2[off2]] = 1;
289 } else if (off2 == lim2) {
290 char *rchg1 = dd1->rchg;
291 long *rindex1 = dd1->rindex;
293 for (; off1 < lim1; off1++)
294 rchg1[rindex1[off1]] = 1;
302 if (xdl_split(ha1, off1, lim1, ha2, off2, lim2, kvdf, kvdb,
303 need_min, &spl, xenv) < 0) {
311 if (xdl_recs_cmp(dd1, off1, spl.i1, dd2, off2, spl.i2,
312 kvdf, kvdb, spl.min_lo, xenv) < 0 ||
313 xdl_recs_cmp(dd1, spl.i1, lim1, dd2, spl.i2, lim2,
314 kvdf, kvdb, spl.min_hi, xenv) < 0) {
324 int xdl_do_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
327 long *kvd, *kvdf, *kvdb;
331 if (XDF_DIFF_ALG(xpp->flags) == XDF_PATIENCE_DIFF)
332 return xdl_do_patience_diff(mf1, mf2, xpp, xe);
334 if (XDF_DIFF_ALG(xpp->flags) == XDF_HISTOGRAM_DIFF)
335 return xdl_do_histogram_diff(mf1, mf2, xpp, xe);
337 if (xdl_prepare_env(mf1, mf2, xpp, xe) < 0) {
343 * Allocate and setup K vectors to be used by the differential algorithm.
344 * One is to store the forward path and one to store the backward path.
346 ndiags = xe->xdf1.nreff + xe->xdf2.nreff + 3;
347 if (!(kvd = (long *) xdl_malloc((2 * ndiags + 2) * sizeof(long)))) {
353 kvdb = kvdf + ndiags;
354 kvdf += xe->xdf2.nreff + 1;
355 kvdb += xe->xdf2.nreff + 1;
357 xenv.mxcost = xdl_bogosqrt(ndiags);
358 if (xenv.mxcost < XDL_MAX_COST_MIN)
359 xenv.mxcost = XDL_MAX_COST_MIN;
360 xenv.snake_cnt = XDL_SNAKE_CNT;
361 xenv.heur_min = XDL_HEUR_MIN_COST;
363 dd1.nrec = xe->xdf1.nreff;
364 dd1.ha = xe->xdf1.ha;
365 dd1.rchg = xe->xdf1.rchg;
366 dd1.rindex = xe->xdf1.rindex;
367 dd2.nrec = xe->xdf2.nreff;
368 dd2.ha = xe->xdf2.ha;
369 dd2.rchg = xe->xdf2.rchg;
370 dd2.rindex = xe->xdf2.rindex;
372 if (xdl_recs_cmp(&dd1, 0, dd1.nrec, &dd2, 0, dd2.nrec,
373 kvdf, kvdb, (xpp->flags & XDF_NEED_MINIMAL) != 0, &xenv) < 0) {
386 static xdchange_t *xdl_add_change(xdchange_t *xscr, long i1, long i2, long chg1, long chg2) {
389 if (!(xch = (xdchange_t *) xdl_malloc(sizeof(xdchange_t))))
403 static int recs_match(xrecord_t *rec1, xrecord_t *rec2, long flags)
405 return (rec1->ha == rec2->ha &&
406 xdl_recmatch(rec1->ptr, rec1->size,
407 rec2->ptr, rec2->size,
412 * If a line is indented more than this, get_indent() just returns this value.
413 * This avoids having to do absurd amounts of work for data that are not
414 * human-readable text, and also ensures that the output of get_indent fits within
417 #define MAX_INDENT 200
420 * Return the amount of indentation of the specified line, treating TAB as 8
421 * columns. Return -1 if line is empty or contains only whitespace. Clamp the
422 * output value at MAX_INDENT.
424 static int get_indent(xrecord_t *rec)
429 for (i = 0; i < rec->size; i++) {
430 char c = rec->ptr[i];
438 /* ignore other whitespace characters */
440 if (ret >= MAX_INDENT)
444 /* The line contains only whitespace. */
449 * If more than this number of consecutive blank rows are found, just return this
450 * value. This avoids requiring O(N^2) work for pathological cases, and also
451 * ensures that the output of score_split fits in an int.
453 #define MAX_BLANKS 20
455 /* Characteristics measured about a hypothetical split position. */
456 struct split_measurement {
458 * Is the split at the end of the file (aside from any blank lines)?
463 * How much is the line immediately following the split indented (or -1 if
464 * the line is blank):
469 * How many consecutive lines above the split are blank?
474 * How much is the nearest non-blank line above the split indented (or -1
475 * if there is no such line)?
480 * How many lines after the line following the split are blank?
485 * How much is the nearest non-blank line after the line following the
486 * split indented (or -1 if there is no such line)?
492 /* The effective indent of this split (smaller is preferred). */
493 int effective_indent;
495 /* Penalty for this split (smaller is preferred). */
500 * Fill m with information about a hypothetical split of xdf above line split.
502 static void measure_split(const xdfile_t *xdf, long split,
503 struct split_measurement *m)
507 if (split >= xdf->nrec) {
512 m->indent = get_indent(xdf->recs[split]);
517 for (i = split - 1; i >= 0; i--) {
518 m->pre_indent = get_indent(xdf->recs[i]);
519 if (m->pre_indent != -1)
522 if (m->pre_blank == MAX_BLANKS) {
530 for (i = split + 1; i < xdf->nrec; i++) {
531 m->post_indent = get_indent(xdf->recs[i]);
532 if (m->post_indent != -1)
535 if (m->post_blank == MAX_BLANKS) {
543 * The empirically-determined weight factors used by score_split() below.
544 * Larger values means that the position is a less favorable place to split.
546 * Note that scores are only ever compared against each other, so multiplying
547 * all of these weight/penalty values by the same factor wouldn't change the
548 * heuristic's behavior. Still, we need to set that arbitrary scale *somehow*.
549 * In practice, these numbers are chosen to be large enough that they can be
550 * adjusted relative to each other with sufficient precision despite using
554 /* Penalty if there are no non-blank lines before the split */
555 #define START_OF_FILE_PENALTY 1
557 /* Penalty if there are no non-blank lines after the split */
558 #define END_OF_FILE_PENALTY 21
560 /* Multiplier for the number of blank lines around the split */
561 #define TOTAL_BLANK_WEIGHT (-30)
563 /* Multiplier for the number of blank lines after the split */
564 #define POST_BLANK_WEIGHT 6
567 * Penalties applied if the line is indented more than its predecessor
569 #define RELATIVE_INDENT_PENALTY (-4)
570 #define RELATIVE_INDENT_WITH_BLANK_PENALTY 10
573 * Penalties applied if the line is indented less than both its predecessor and
576 #define RELATIVE_OUTDENT_PENALTY 24
577 #define RELATIVE_OUTDENT_WITH_BLANK_PENALTY 17
580 * Penalties applied if the line is indented less than its predecessor but not
581 * less than its successor
583 #define RELATIVE_DEDENT_PENALTY 23
584 #define RELATIVE_DEDENT_WITH_BLANK_PENALTY 17
587 * We only consider whether the sum of the effective indents for splits are
588 * less than (-1), equal to (0), or greater than (+1) each other. The resulting
589 * value is multiplied by the following weight and combined with the penalty to
590 * determine the better of two scores.
592 #define INDENT_WEIGHT 60
595 * Compute a badness score for the hypothetical split whose measurements are
596 * stored in m. The weight factors were determined empirically using the tools and
597 * corpus described in
599 * https://github.com/mhagger/diff-slider-tools
601 * Also see that project if you want to improve the weights based on, for example,
602 * a larger or more diverse corpus.
604 static void score_add_split(const struct split_measurement *m, struct split_score *s)
607 * A place to accumulate penalty factors (positive makes this index more
610 int post_blank, total_blank, indent, any_blanks;
612 if (m->pre_indent == -1 && m->pre_blank == 0)
613 s->penalty += START_OF_FILE_PENALTY;
616 s->penalty += END_OF_FILE_PENALTY;
619 * Set post_blank to the number of blank lines following the split,
620 * including the line immediately after the split:
622 post_blank = (m->indent == -1) ? 1 + m->post_blank : 0;
623 total_blank = m->pre_blank + post_blank;
625 /* Penalties based on nearby blank lines: */
626 s->penalty += TOTAL_BLANK_WEIGHT * total_blank;
627 s->penalty += POST_BLANK_WEIGHT * post_blank;
632 indent = m->post_indent;
634 any_blanks = (total_blank != 0);
636 /* Note that the effective indent is -1 at the end of the file: */
637 s->effective_indent += indent;
640 /* No additional adjustments needed. */
641 } else if (m->pre_indent == -1) {
642 /* No additional adjustments needed. */
643 } else if (indent > m->pre_indent) {
645 * The line is indented more than its predecessor.
647 s->penalty += any_blanks ?
648 RELATIVE_INDENT_WITH_BLANK_PENALTY :
649 RELATIVE_INDENT_PENALTY;
650 } else if (indent == m->pre_indent) {
652 * The line has the same indentation level as its predecessor.
653 * No additional adjustments needed.
657 * The line is indented less than its predecessor. It could be
658 * the block terminator of the previous block, but it could
659 * also be the start of a new block (e.g., an "else" block, or
660 * maybe the previous block didn't have a block terminator).
661 * Try to distinguish those cases based on what comes next:
663 if (m->post_indent != -1 && m->post_indent > indent) {
665 * The following line is indented more. So it is likely
666 * that this line is the start of a block.
668 s->penalty += any_blanks ?
669 RELATIVE_OUTDENT_WITH_BLANK_PENALTY :
670 RELATIVE_OUTDENT_PENALTY;
673 * That was probably the end of a block.
675 s->penalty += any_blanks ?
676 RELATIVE_DEDENT_WITH_BLANK_PENALTY :
677 RELATIVE_DEDENT_PENALTY;
682 static int score_cmp(struct split_score *s1, struct split_score *s2)
684 /* -1 if s1.effective_indent < s2->effective_indent, etc. */
685 int cmp_indents = ((s1->effective_indent > s2->effective_indent) -
686 (s1->effective_indent < s2->effective_indent));
688 return INDENT_WEIGHT * cmp_indents + (s1->penalty - s2->penalty);
692 * Represent a group of changed lines in an xdfile_t (i.e., a contiguous group
693 * of lines that was inserted or deleted from the corresponding version of the
694 * file). We consider there to be such a group at the beginning of the file, at
695 * the end of the file, and between any two unchanged lines, though most such
696 * groups will usually be empty.
698 * If the first line in a group is equal to the line following the group, then
699 * the group can be slid down. Similarly, if the last line in a group is equal
700 * to the line preceding the group, then the group can be slid up. See
701 * group_slide_down() and group_slide_up().
703 * Note that loops that are testing for changed lines in xdf->rchg do not need
704 * index bounding since the array is prepared with a zero at position -1 and N.
708 * The index of the first changed line in the group, or the index of
709 * the unchanged line above which the (empty) group is located.
714 * The index of the first unchanged line after the group. For an empty
715 * group, end is equal to start.
721 * Initialize g to point at the first group in xdf.
723 static void group_init(xdfile_t *xdf, struct xdlgroup *g)
725 g->start = g->end = 0;
726 while (xdf->rchg[g->end])
731 * Move g to describe the next (possibly empty) group in xdf and return 0. If g
732 * is already at the end of the file, do nothing and return -1.
734 static inline int group_next(xdfile_t *xdf, struct xdlgroup *g)
736 if (g->end == xdf->nrec)
739 g->start = g->end + 1;
740 for (g->end = g->start; xdf->rchg[g->end]; g->end++)
747 * Move g to describe the previous (possibly empty) group in xdf and return 0.
748 * If g is already at the beginning of the file, do nothing and return -1.
750 static inline int group_previous(xdfile_t *xdf, struct xdlgroup *g)
755 g->end = g->start - 1;
756 for (g->start = g->end; xdf->rchg[g->start - 1]; g->start--)
763 * If g can be slid toward the end of the file, do so, and if it bumps into a
764 * following group, expand this group to include it. Return 0 on success or -1
765 * if g cannot be slid down.
767 static int group_slide_down(xdfile_t *xdf, struct xdlgroup *g, long flags)
769 if (g->end < xdf->nrec &&
770 recs_match(xdf->recs[g->start], xdf->recs[g->end], flags)) {
771 xdf->rchg[g->start++] = 0;
772 xdf->rchg[g->end++] = 1;
774 while (xdf->rchg[g->end])
784 * If g can be slid toward the beginning of the file, do so, and if it bumps
785 * into a previous group, expand this group to include it. Return 0 on success
786 * or -1 if g cannot be slid up.
788 static int group_slide_up(xdfile_t *xdf, struct xdlgroup *g, long flags)
791 recs_match(xdf->recs[g->start - 1], xdf->recs[g->end - 1], flags)) {
792 xdf->rchg[--g->start] = 1;
793 xdf->rchg[--g->end] = 0;
795 while (xdf->rchg[g->start - 1])
804 static void xdl_bug(const char *msg)
806 fprintf(stderr, "BUG: %s\n", msg);
811 * Move back and forward change groups for a consistent and pretty diff output.
812 * This also helps in finding joinable change groups and reducing the diff
815 int xdl_change_compact(xdfile_t *xdf, xdfile_t *xdfo, long flags) {
816 struct xdlgroup g, go;
817 long earliest_end, end_matching_other;
821 group_init(xdfo, &go);
824 /* If the group is empty in the to-be-compacted file, skip it: */
825 if (g.end == g.start)
829 * Now shift the change up and then down as far as possible in
830 * each direction. If it bumps into any other changes, merge them.
833 groupsize = g.end - g.start;
836 * Keep track of the last "end" index that causes this
837 * group to align with a group of changed lines in the
838 * other file. -1 indicates that we haven't found such
841 end_matching_other = -1;
843 /* Shift the group backward as much as possible: */
844 while (!group_slide_up(xdf, &g, flags))
845 if (group_previous(xdfo, &go))
846 xdl_bug("group sync broken sliding up");
849 * This is this highest that this group can be shifted.
850 * Record its end index:
852 earliest_end = g.end;
854 if (go.end > go.start)
855 end_matching_other = g.end;
857 /* Now shift the group forward as far as possible: */
859 if (group_slide_down(xdf, &g, flags))
861 if (group_next(xdfo, &go))
862 xdl_bug("group sync broken sliding down");
864 if (go.end > go.start)
865 end_matching_other = g.end;
867 } while (groupsize != g.end - g.start);
870 * If the group can be shifted, then we can possibly use this
871 * freedom to produce a more intuitive diff.
873 * The group is currently shifted as far down as possible, so the
874 * heuristics below only have to handle upwards shifts.
877 if (g.end == earliest_end) {
878 /* no shifting was possible */
879 } else if (end_matching_other != -1) {
881 * Move the possibly merged group of changes back to line
882 * up with the last group of changes from the other file
883 * that it can align with.
885 while (go.end == go.start) {
886 if (group_slide_up(xdf, &g, flags))
887 xdl_bug("match disappeared");
888 if (group_previous(xdfo, &go))
889 xdl_bug("group sync broken sliding to match");
891 } else if (flags & XDF_INDENT_HEURISTIC) {
893 * Indent heuristic: a group of pure add/delete lines
894 * implies two splits, one between the end of the "before"
895 * context and the start of the group, and another between
896 * the end of the group and the beginning of the "after"
897 * context. Some splits are aesthetically better and some
898 * are worse. We compute a badness "score" for each split,
899 * and add the scores for the two splits to define a
900 * "score" for each position that the group can be shifted
901 * to. Then we pick the shift with the lowest score.
903 long shift, best_shift = -1;
904 struct split_score best_score;
906 for (shift = earliest_end; shift <= g.end; shift++) {
907 struct split_measurement m;
908 struct split_score score = {0, 0};
910 measure_split(xdf, shift, &m);
911 score_add_split(&m, &score);
912 measure_split(xdf, shift - groupsize, &m);
913 score_add_split(&m, &score);
914 if (best_shift == -1 ||
915 score_cmp(&score, &best_score) <= 0) {
916 best_score.effective_indent = score.effective_indent;
917 best_score.penalty = score.penalty;
922 while (g.end > best_shift) {
923 if (group_slide_up(xdf, &g, flags))
924 xdl_bug("best shift unreached");
925 if (group_previous(xdfo, &go))
926 xdl_bug("group sync broken sliding to blank line");
931 /* Move past the just-processed group: */
932 if (group_next(xdf, &g))
934 if (group_next(xdfo, &go))
935 xdl_bug("group sync broken moving to next group");
938 if (!group_next(xdfo, &go))
939 xdl_bug("group sync broken at end of file");
945 int xdl_build_script(xdfenv_t *xe, xdchange_t **xscr) {
946 xdchange_t *cscr = NULL, *xch;
947 char *rchg1 = xe->xdf1.rchg, *rchg2 = xe->xdf2.rchg;
951 * Trivial. Collects "groups" of changes and creates an edit script.
953 for (i1 = xe->xdf1.nrec, i2 = xe->xdf2.nrec; i1 >= 0 || i2 >= 0; i1--, i2--)
954 if (rchg1[i1 - 1] || rchg2[i2 - 1]) {
955 for (l1 = i1; rchg1[i1 - 1]; i1--);
956 for (l2 = i2; rchg2[i2 - 1]; i2--);
958 if (!(xch = xdl_add_change(cscr, i1, i2, l1 - i1, l2 - i2))) {
959 xdl_free_script(cscr);
971 void xdl_free_script(xdchange_t *xscr) {
974 while ((xch = xscr) != NULL) {
980 static int xdl_call_hunk_func(xdfenv_t *xe, xdchange_t *xscr, xdemitcb_t *ecb,
981 xdemitconf_t const *xecfg)
983 xdchange_t *xch, *xche;
985 for (xch = xscr; xch; xch = xche->next) {
986 xche = xdl_get_hunk(&xch, xecfg);
989 if (xecfg->hunk_func(xch->i1, xche->i1 + xche->chg1 - xch->i1,
990 xch->i2, xche->i2 + xche->chg2 - xch->i2,
997 static void xdl_mark_ignorable(xdchange_t *xscr, xdfenv_t *xe, long flags)
1001 for (xch = xscr; xch; xch = xch->next) {
1006 rec = &xe->xdf1.recs[xch->i1];
1007 for (i = 0; i < xch->chg1 && ignore; i++)
1008 ignore = xdl_blankline(rec[i]->ptr, rec[i]->size, flags);
1010 rec = &xe->xdf2.recs[xch->i2];
1011 for (i = 0; i < xch->chg2 && ignore; i++)
1012 ignore = xdl_blankline(rec[i]->ptr, rec[i]->size, flags);
1014 xch->ignore = ignore;
1018 int xdl_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
1019 xdemitconf_t const *xecfg, xdemitcb_t *ecb) {
1022 emit_func_t ef = xecfg->hunk_func ? xdl_call_hunk_func : xdl_emit_diff;
1024 if (xdl_do_diff(mf1, mf2, xpp, &xe) < 0) {
1028 if (xdl_change_compact(&xe.xdf1, &xe.xdf2, xpp->flags) < 0 ||
1029 xdl_change_compact(&xe.xdf2, &xe.xdf1, xpp->flags) < 0 ||
1030 xdl_build_script(&xe, &xscr) < 0) {
1036 if (xpp->flags & XDF_IGNORE_BLANK_LINES)
1037 xdl_mark_ignorable(xscr, &xe, xpp->flags);
1039 if (ef(&xe, xscr, ecb, xecfg) < 0) {
1041 xdl_free_script(xscr);
1045 xdl_free_script(xscr);