?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
log -L: check range set invariants when we look it up
[git/gitweb.git] / line-log.c
blob5a12ceb012ee656d9495dc43d924ec4d983d61b6
1 #include "git-compat-util.h"
2 #include "line-range.h"
3 #include "cache.h"
4 #include "tag.h"
5 #include "blob.h"
6 #include "tree.h"
7 #include "diff.h"
8 #include "commit.h"
9 #include "decorate.h"
10 #include "revision.h"
11 #include "xdiff-interface.h"
12 #include "strbuf.h"
13 #include "log-tree.h"
14 #include "graph.h"
15 #include "userdiff.h"
16 #include "line-log.h"
17
18 static void range_set_grow(struct range_set *rs, size_t extra)
19 {
20 ALLOC_GROW(rs->ranges, rs->nr + extra, rs->alloc);
21 }
22
23 /* Either initialization would be fine */
24 #define RANGE_SET_INIT {0}
25
26 static void range_set_init(struct range_set *rs, size_t prealloc)
27 {
28 rs->alloc = rs->nr = 0;
29 rs->ranges = NULL;
30 if (prealloc)
31 range_set_grow(rs, prealloc);
32 }
33
34 static void range_set_release(struct range_set *rs)
35 {
36 free(rs->ranges);
37 rs->alloc = rs->nr = 0;
38 rs->ranges = NULL;
39 }
40
41 /* dst must be uninitialized! */
42 static void range_set_copy(struct range_set *dst, struct range_set *src)
43 {
44 range_set_init(dst, src->nr);
45 memcpy(dst->ranges, src->ranges, src->nr*sizeof(struct range_set));
46 dst->nr = src->nr;
47 }
48 static void range_set_move(struct range_set *dst, struct range_set *src)
49 {
50 range_set_release(dst);
51 dst->ranges = src->ranges;
52 dst->nr = src->nr;
53 dst->alloc = src->alloc;
54 src->ranges = NULL;
55 src->alloc = src->nr = 0;
56 }
57
58 /* tack on a _new_ range _at the end_ */
59 static void range_set_append(struct range_set *rs, long a, long b)
60 {
61 assert(a <= b);
62 assert(rs->nr == 0 || rs->ranges[rs->nr-1].end <= a);
63 range_set_grow(rs, 1);
64 rs->ranges[rs->nr].start = a;
65 rs->ranges[rs->nr].end = b;
66 rs->nr++;
67 }
68
69 static int range_cmp(const void *_r, const void *_s)
70 {
71 const struct range *r = _r;
72 const struct range *s = _s;
73
74 /* this could be simply 'return r.start-s.start', but for the types */
75 if (r->start == s->start)
76 return 0;
77 if (r->start < s->start)
78 return -1;
79 return 1;
80 }
81
82 /*
83 * Check that the ranges are non-empty, sorted and non-overlapping
84 */
85 static void range_set_check_invariants(struct range_set *rs)
86 {
87 int i;
88
89 if (!rs)
90 return;
91
92 if (rs->nr)
93 assert(rs->ranges[0].start < rs->ranges[0].end);
94
95 for (i = 1; i < rs->nr; i++) {
96 assert(rs->ranges[i-1].end < rs->ranges[i].start);
97 assert(rs->ranges[i].start < rs->ranges[i].end);
98 }
99 }
100
101 /*
102 * Helper: In-place pass of sorting and merging the ranges in the
103 * range set, to re-establish the invariants after another operation
104 *
105 * NEEDSWORK currently not needed
106 */
107 static void sort_and_merge_range_set(struct range_set *rs)
108 {
109 int i;
110 int o = 1; /* output cursor */
111
112 qsort(rs->ranges, rs->nr, sizeof(struct range), range_cmp);
113
114 for (i = 1; i < rs->nr; i++) {
115 if (rs->ranges[i].start <= rs->ranges[o-1].end) {
116 rs->ranges[o-1].end = rs->ranges[i].end;
117 } else {
118 rs->ranges[o].start = rs->ranges[i].start;
119 rs->ranges[o].end = rs->ranges[i].end;
120 o++;
121 }
122 }
123 assert(o <= rs->nr);
124 rs->nr = o;
125
126 range_set_check_invariants(rs);
127 }
128
129 /*
130 * Union of range sets (i.e., sets of line numbers). Used to merge
131 * them when searches meet at a common ancestor.
132 *
133 * This is also where the ranges are consolidated into canonical form:
134 * overlapping and adjacent ranges are merged, and empty ranges are
135 * removed.
136 */
137 static void range_set_union(struct range_set *out,
138 struct range_set *a, struct range_set *b)
139 {
140 int i = 0, j = 0, o = 0;
141 struct range *ra = a->ranges;
142 struct range *rb = b->ranges;
143 /* cannot make an alias of out->ranges: it may change during grow */
144
145 assert(out->nr == 0);
146 while (i < a->nr || j < b->nr) {
147 struct range *new;
148 if (i < a->nr && j < b->nr) {
149 if (ra[i].start < rb[j].start)
150 new = &ra[i++];
151 else if (ra[i].start > rb[j].start)
152 new = &rb[j++];
153 else if (ra[i].end < rb[j].end)
154 new = &ra[i++];
155 else
156 new = &rb[j++];
157 } else if (i < a->nr) /* b exhausted */
158 new = &ra[i++];
159 else /* a exhausted */
160 new = &rb[j++];
161 if (new->start == new->end)
162 ; /* empty range */
163 else if (!o || out->ranges[o-1].end < new->start) {
164 range_set_grow(out, 1);
165 out->ranges[o].start = new->start;
166 out->ranges[o].end = new->end;
167 o++;
168 } else if (out->ranges[o-1].end < new->end) {
169 out->ranges[o-1].end = new->end;
170 }
171 }
172 out->nr = o;
173 }
174
175 /*
176 * Difference of range sets (out = a \ b). Pass the "interesting"
177 * ranges as 'a' and the target side of the diff as 'b': it removes
178 * the ranges for which the commit is responsible.
179 */
180 static void range_set_difference(struct range_set *out,
181 struct range_set *a, struct range_set *b)
182 {
183 int i, j = 0;
184 for (i = 0; i < a->nr; i++) {
185 long start = a->ranges[i].start;
186 long end = a->ranges[i].end;
187 while (start < end) {
188 while (j < b->nr && start >= b->ranges[j].end)
189 /*
190 * a: |-------
191 * b: ------|
192 */
193 j++;
194 if (j >= b->nr || end < b->ranges[j].start) {
195 /*
196 * b exhausted, or
197 * a: ----|
198 * b: |----
199 */
200 range_set_append(out, start, end);
201 break;
202 }
203 if (start >= b->ranges[j].start) {
204 /*
205 * a: |--????
206 * b: |------|
207 */
208 start = b->ranges[j].end;
209 } else if (end > b->ranges[j].start) {
210 /*
211 * a: |-----|
212 * b: |--?????
213 */
214 if (start < b->ranges[j].start)
215 range_set_append(out, start, b->ranges[j].start);
216 start = b->ranges[j].end;
217 }
218 }
219 }
220 }
221
222 static void diff_ranges_init(struct diff_ranges *diff)
223 {
224 range_set_init(&diff->parent, 0);
225 range_set_init(&diff->target, 0);
226 }
227
228 static void diff_ranges_release(struct diff_ranges *diff)
229 {
230 range_set_release(&diff->parent);
231 range_set_release(&diff->target);
232 }
233
234 void line_log_data_init(struct line_log_data *r)
235 {
236 memset(r, 0, sizeof(struct line_log_data));
237 range_set_init(&r->ranges, 0);
238 }
239
240 static void line_log_data_clear(struct line_log_data *r)
241 {
242 range_set_release(&r->ranges);
243 if (r->pair)
244 diff_free_filepair(r->pair);
245 }
246
247 static void free_line_log_data(struct line_log_data *r)
248 {
249 while (r) {
250 struct line_log_data *next = r->next;
251 line_log_data_clear(r);
252 free(r);
253 r = next;
254 }
255 }
256
257 static struct line_log_data *
258 search_line_log_data(struct line_log_data *list, const char *path,
259 struct line_log_data **insertion_point)
260 {
261 struct line_log_data *p = list;
262 if (insertion_point)
263 *insertion_point = NULL;
264 while (p) {
265 int cmp = strcmp(p->spec->path, path);
266 if (!cmp)
267 return p;
268 if (insertion_point && cmp < 0)
269 *insertion_point = p;
270 p = p->next;
271 }
272 return NULL;
273 }
274
275 static void line_log_data_insert(struct line_log_data **list,
276 struct diff_filespec *spec,
277 long begin, long end)
278 {
279 struct line_log_data *ip;
280 struct line_log_data *p = search_line_log_data(*list, spec->path, &ip);
281
282 if (p) {
283 range_set_append(&p->ranges, begin, end);
284 sort_and_merge_range_set(&p->ranges);
285 free_filespec(spec);
286 return;
287 }
288
289 p = xcalloc(1, sizeof(struct line_log_data));
290 p->spec = spec;
291 range_set_append(&p->ranges, begin, end);
292 if (ip) {
293 p->next = ip->next;
294 ip->next = p;
295 } else {
296 p->next = *list;
297 *list = p;
298 }
299 }
300
301 struct collect_diff_cbdata {
302 struct diff_ranges *diff;
303 };
304
305 static int collect_diff_cb(long start_a, long count_a,
306 long start_b, long count_b,
307 void *data)
308 {
309 struct collect_diff_cbdata *d = data;
310
311 if (count_a >= 0)
312 range_set_append(&d->diff->parent, start_a, start_a + count_a);
313 if (count_b >= 0)
314 range_set_append(&d->diff->target, start_b, start_b + count_b);
315
316 return 0;
317 }
318
319 static void collect_diff(mmfile_t *parent, mmfile_t *target, struct diff_ranges *out)
320 {
321 struct collect_diff_cbdata cbdata = {NULL};
322 xpparam_t xpp;
323 xdemitconf_t xecfg;
324 xdemitcb_t ecb;
325
326 memset(&xpp, 0, sizeof(xpp));
327 memset(&xecfg, 0, sizeof(xecfg));
328 xecfg.ctxlen = xecfg.interhunkctxlen = 0;
329
330 cbdata.diff = out;
331 xecfg.hunk_func = collect_diff_cb;
332 memset(&ecb, 0, sizeof(ecb));
333 ecb.priv = &cbdata;
334 xdi_diff(parent, target, &xpp, &xecfg, &ecb);
335 }
336
337 /*
338 * These are handy for debugging. Removing them with #if 0 silences
339 * the "unused function" warning.
340 */
341 #if 0
342 static void dump_range_set(struct range_set *rs, const char *desc)
343 {
344 int i;
345 printf("range set %s (%d items):\n", desc, rs->nr);
346 for (i = 0; i < rs->nr; i++)
347 printf("\t[%ld,%ld]\n", rs->ranges[i].start, rs->ranges[i].end);
348 }
349
350 static void dump_line_log_data(struct line_log_data *r)
351 {
352 char buf[4096];
353 while (r) {
354 snprintf(buf, 4096, "file %s\n", r->spec->path);
355 dump_range_set(&r->ranges, buf);
356 r = r->next;
357 }
358 }
359
360 static void dump_diff_ranges(struct diff_ranges *diff, const char *desc)
361 {
362 int i;
363 assert(diff->parent.nr == diff->target.nr);
364 printf("diff ranges %s (%d items):\n", desc, diff->parent.nr);
365 printf("\tparent\ttarget\n");
366 for (i = 0; i < diff->parent.nr; i++) {
367 printf("\t[%ld,%ld]\t[%ld,%ld]\n",
368 diff->parent.ranges[i].start,
369 diff->parent.ranges[i].end,
370 diff->target.ranges[i].start,
371 diff->target.ranges[i].end);
372 }
373 }
374 #endif
375
376
377 static int ranges_overlap(struct range *a, struct range *b)
378 {
379 return !(a->end <= b->start || b->end <= a->start);
380 }
381
382 /*
383 * Given a diff and the set of interesting ranges, determine all hunks
384 * of the diff which touch (overlap) at least one of the interesting
385 * ranges in the target.
386 */
387 static void diff_ranges_filter_touched(struct diff_ranges *out,
388 struct diff_ranges *diff,
389 struct range_set *rs)
390 {
391 int i, j = 0;
392
393 assert(out->target.nr == 0);
394
395 for (i = 0; i < diff->target.nr; i++) {
396 while (diff->target.ranges[i].start > rs->ranges[j].end) {
397 j++;
398 if (j == rs->nr)
399 return;
400 }
401 if (ranges_overlap(&diff->target.ranges[i], &rs->ranges[j])) {
402 range_set_append(&out->parent,
403 diff->parent.ranges[i].start,
404 diff->parent.ranges[i].end);
405 range_set_append(&out->target,
406 diff->target.ranges[i].start,
407 diff->target.ranges[i].end);
408 }
409 }
410 }
411
412 /*
413 * Adjust the line counts in 'rs' to account for the lines
414 * added/removed in the diff.
415 */
416 static void range_set_shift_diff(struct range_set *out,
417 struct range_set *rs,
418 struct diff_ranges *diff)
419 {
420 int i, j = 0;
421 long offset = 0;
422 struct range *src = rs->ranges;
423 struct range *target = diff->target.ranges;
424 struct range *parent = diff->parent.ranges;
425
426 for (i = 0; i < rs->nr; i++) {
427 while (j < diff->target.nr && src[i].start >= target[j].start) {
428 offset += (parent[j].end-parent[j].start)
429 - (target[j].end-target[j].start);
430 j++;
431 }
432 range_set_append(out, src[i].start+offset, src[i].end+offset);
433 }
434 }
435
436 /*
437 * Given a diff and the set of interesting ranges, map the ranges
438 * across the diff. That is: observe that the target commit takes
439 * blame for all the + (target-side) ranges. So for every pair of
440 * ranges in the diff that was touched, we remove the latter and add
441 * its parent side.
442 */
443 static void range_set_map_across_diff(struct range_set *out,
444 struct range_set *rs,
445 struct diff_ranges *diff,
446 struct diff_ranges **touched_out)
447 {
448 struct diff_ranges *touched = xmalloc(sizeof(*touched));
449 struct range_set tmp1 = RANGE_SET_INIT;
450 struct range_set tmp2 = RANGE_SET_INIT;
451
452 diff_ranges_init(touched);
453 diff_ranges_filter_touched(touched, diff, rs);
454 range_set_difference(&tmp1, rs, &touched->target);
455 range_set_shift_diff(&tmp2, &tmp1, diff);
456 range_set_union(out, &tmp2, &touched->parent);
457 range_set_release(&tmp1);
458 range_set_release(&tmp2);
459
460 *touched_out = touched;
461 }
462
463 static struct commit *check_single_commit(struct rev_info *revs)
464 {
465 struct object *commit = NULL;
466 int found = -1;
467 int i;
468
469 for (i = 0; i < revs->pending.nr; i++) {
470 struct object *obj = revs->pending.objects[i].item;
471 if (obj->flags & UNINTERESTING)
472 continue;
473 while (obj->type == OBJ_TAG)
474 obj = deref_tag(obj, NULL, 0);
475 if (obj->type != OBJ_COMMIT)
476 die("Non commit %s?", revs->pending.objects[i].name);
477 if (commit)
478 die("More than one commit to dig from: %s and %s?",
479 revs->pending.objects[i].name,
480 revs->pending.objects[found].name);
481 commit = obj;
482 found = i;
483 }
484
485 if (!commit)
486 die("No commit specified?");
487
488 return (struct commit *) commit;
489 }
490
491 static void fill_blob_sha1(struct commit *commit, struct diff_filespec *spec)
492 {
493 unsigned mode;
494 unsigned char sha1[20];
495
496 if (get_tree_entry(commit->object.sha1, spec->path,
497 sha1, &mode))
498 die("There is no path %s in the commit", spec->path);
499 fill_filespec(spec, sha1, 1, mode);
500
501 return;
502 }
503
504 static void fill_line_ends(struct diff_filespec *spec, long *lines,
505 unsigned long **line_ends)
506 {
507 int num = 0, size = 50;
508 long cur = 0;
509 unsigned long *ends = NULL;
510 char *data = NULL;
511
512 if (diff_populate_filespec(spec, 0))
513 die("Cannot read blob %s", sha1_to_hex(spec->sha1));
514
515 ends = xmalloc(size * sizeof(*ends));
516 ends[cur++] = 0;
517 data = spec->data;
518 while (num < spec->size) {
519 if (data[num] == '\n' || num == spec->size - 1) {
520 ALLOC_GROW(ends, (cur + 1), size);
521 ends[cur++] = num;
522 }
523 num++;
524 }
525
526 /* shrink the array to fit the elements */
527 ends = xrealloc(ends, cur * sizeof(*ends));
528 *lines = cur-1;
529 *line_ends = ends;
530 }
531
532 struct nth_line_cb {
533 struct diff_filespec *spec;
534 long lines;
535 unsigned long *line_ends;
536 };
537
538 static const char *nth_line(void *data, long line)
539 {
540 struct nth_line_cb *d = data;
541 assert(d && line <= d->lines);
542 assert(d->spec && d->spec->data);
543
544 if (line == 0)
545 return (char *)d->spec->data;
546 else
547 return (char *)d->spec->data + d->line_ends[line] + 1;
548 }
549
550 static struct line_log_data *
551 parse_lines(struct commit *commit, const char *prefix, struct string_list *args)
552 {
553 long lines = 0;
554 unsigned long *ends = NULL;
555 struct nth_line_cb cb_data;
556 struct string_list_item *item;
557 struct line_log_data *ranges = NULL;
558
559 for_each_string_list_item(item, args) {
560 const char *name_part, *range_part;
561 const char *full_name;
562 struct diff_filespec *spec;
563 long begin = 0, end = 0;
564
565 name_part = skip_range_arg(item->string);
566 if (!name_part || *name_part != ':' || !name_part[1])
567 die("-L argument '%s' not of the form start,end:file",
568 item->string);
569 range_part = xstrndup(item->string, name_part - item->string);
570 name_part++;
571
572 full_name = prefix_path(prefix, prefix ? strlen(prefix) : 0,
573 name_part);
574
575 spec = alloc_filespec(full_name);
576 fill_blob_sha1(commit, spec);
577 fill_line_ends(spec, &lines, &ends);
578 cb_data.spec = spec;
579 cb_data.lines = lines;
580 cb_data.line_ends = ends;
581
582 if (parse_range_arg(range_part, nth_line, &cb_data,
583 lines, &begin, &end,
584 spec->path))
585 die("malformed -L argument '%s'", range_part);
586 if (begin < 1)
587 begin = 1;
588 if (end < 1)
589 end = lines;
590 begin--;
591 if (lines < end || lines < begin)
592 die("file %s has only %ld lines", name_part, lines);
593 line_log_data_insert(&ranges, spec, begin, end);
594
595 free(ends);
596 ends = NULL;
597 }
598
599 return ranges;
600 }
601
602 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
603 {
604 struct line_log_data *ret = xmalloc(sizeof(*ret));
605
606 assert(r);
607 line_log_data_init(ret);
608 range_set_copy(&ret->ranges, &r->ranges);
609
610 ret->spec = r->spec;
611 assert(ret->spec);
612 ret->spec->count++;
613
614 return ret;
615 }
616
617 static struct line_log_data *
618 line_log_data_copy(struct line_log_data *r)
619 {
620 struct line_log_data *ret = NULL;
621 struct line_log_data *tmp = NULL, *prev = NULL;
622
623 assert(r);
624 ret = tmp = prev = line_log_data_copy_one(r);
625 r = r->next;
626 while (r) {
627 tmp = line_log_data_copy_one(r);
628 prev->next = tmp;
629 prev = tmp;
630 r = r->next;
631 }
632
633 return ret;
634 }
635
636 /* merge two range sets across files */
637 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
638 struct line_log_data *b)
639 {
640 struct line_log_data *head = NULL, **pp = &head;
641
642 while (a || b) {
643 struct line_log_data *src;
644 struct line_log_data *src2 = NULL;
645 struct line_log_data *d;
646 int cmp;
647 if (!a)
648 cmp = 1;
649 else if (!b)
650 cmp = -1;
651 else
652 cmp = strcmp(a->spec->path, b->spec->path);
653 if (cmp < 0) {
654 src = a;
655 a = a->next;
656 } else if (cmp == 0) {
657 src = a;
658 a = a->next;
659 src2 = b;
660 b = b->next;
661 } else {
662 src = b;
663 b = b->next;
664 }
665 d = xmalloc(sizeof(struct line_log_data));
666 line_log_data_init(d);
667 d->spec = src->spec;
668 d->spec->count++;
669 *pp = d;
670 pp = &d->next;
671 if (src2)
672 range_set_union(&d->ranges, &src->ranges, &src2->ranges);
673 else
674 range_set_copy(&d->ranges, &src->ranges);
675 }
676
677 return head;
678 }
679
680 static void add_line_range(struct rev_info *revs, struct commit *commit,
681 struct line_log_data *range)
682 {
683 struct line_log_data *old = NULL;
684 struct line_log_data *new = NULL;
685
686 old = lookup_decoration(&revs->line_log_data, &commit->object);
687 if (old && range) {
688 new = line_log_data_merge(old, range);
689 free_line_log_data(old);
690 } else if (range)
691 new = line_log_data_copy(range);
692
693 if (new)
694 add_decoration(&revs->line_log_data, &commit->object, new);
695 }
696
697 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
698 {
699 struct line_log_data *r;
700 r = lookup_decoration(&revs->line_log_data, &commit->object);
701 if (!r)
702 return;
703 free_line_log_data(r);
704 add_decoration(&revs->line_log_data, &commit->object, NULL);
705 }
706
707 static struct line_log_data *lookup_line_range(struct rev_info *revs,
708 struct commit *commit)
709 {
710 struct line_log_data *ret = NULL;
711 struct line_log_data *d;
712
713 ret = lookup_decoration(&revs->line_log_data, &commit->object);
714
715 for (d = ret; d; d = d->next)
716 range_set_check_invariants(&d->ranges);
717
718 return ret;
719 }
720
721 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
722 {
723 struct commit *commit = NULL;
724 struct line_log_data *range;
725
726 commit = check_single_commit(rev);
727 range = parse_lines(commit, prefix, args);
728 add_line_range(rev, commit, range);
729
730 if (!rev->diffopt.detect_rename) {
731 int i, count = 0;
732 struct line_log_data *r = range;
733 const char **paths;
734 while (r) {
735 count++;
736 r = r->next;
737 }
738 paths = xmalloc((count+1)*sizeof(char *));
739 r = range;
740 for (i = 0; i < count; i++) {
741 paths[i] = xstrdup(r->spec->path);
742 r = r->next;
743 }
744 paths[count] = NULL;
745 init_pathspec(&rev->diffopt.pathspec, paths);
746 free(paths);
747 }
748 }
749
750 static void load_tree_desc(struct tree_desc *desc, void **tree,
751 const unsigned char *sha1)
752 {
753 unsigned long size;
754 *tree = read_object_with_reference(sha1, tree_type, &size, NULL);
755 if (!*tree)
756 die("Unable to read tree (%s)", sha1_to_hex(sha1));
757 init_tree_desc(desc, *tree, size);
758 }
759
760 static int count_parents(struct commit *commit)
761 {
762 struct commit_list *parents = commit->parents;
763 int count = 0;
764 while (parents) {
765 count++;
766 parents = parents->next;
767 }
768 return count;
769 }
770
771 static void move_diff_queue(struct diff_queue_struct *dst,
772 struct diff_queue_struct *src)
773 {
774 assert(src != dst);
775 memcpy(dst, src, sizeof(struct diff_queue_struct));
776 DIFF_QUEUE_CLEAR(src);
777 }
778
779 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
780 {
781 int i;
782 struct diff_queue_struct outq;
783 DIFF_QUEUE_CLEAR(&outq);
784
785 for (i = 0; i < diff_queued_diff.nr; i++) {
786 struct diff_filepair *p = diff_queued_diff.queue[i];
787 struct line_log_data *rg = NULL;
788
789 if (!DIFF_FILE_VALID(p->two)) {
790 if (keep_deletions)
791 diff_q(&outq, p);
792 else
793 diff_free_filepair(p);
794 continue;
795 }
796 for (rg = range; rg; rg = rg->next) {
797 if (!strcmp(rg->spec->path, p->two->path))
798 break;
799 }
800 if (rg)
801 diff_q(&outq, p);
802 else
803 diff_free_filepair(p);
804 }
805 free(diff_queued_diff.queue);
806 diff_queued_diff = outq;
807 }
808
809 static inline int diff_might_be_rename(void)
810 {
811 int i;
812 for (i = 0; i < diff_queued_diff.nr; i++)
813 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
814 /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
815 /* diff_queued_diff.queue[i]->two->path); */
816 return 1;
817 }
818 return 0;
819 }
820
821 static void queue_diffs(struct line_log_data *range,
822 struct diff_options *opt,
823 struct diff_queue_struct *queue,
824 struct commit *commit, struct commit *parent)
825 {
826 void *tree1 = NULL, *tree2 = NULL;
827 struct tree_desc desc1, desc2;
828
829 assert(commit);
830 load_tree_desc(&desc2, &tree2, commit->tree->object.sha1);
831 if (parent)
832 load_tree_desc(&desc1, &tree1, parent->tree->object.sha1);
833 else
834 init_tree_desc(&desc1, "", 0);
835
836 DIFF_QUEUE_CLEAR(&diff_queued_diff);
837 diff_tree(&desc1, &desc2, "", opt);
838 if (opt->detect_rename) {
839 filter_diffs_for_paths(range, 1);
840 if (diff_might_be_rename())
841 diffcore_std(opt);
842 filter_diffs_for_paths(range, 0);
843 }
844 move_diff_queue(queue, &diff_queued_diff);
845
846 if (tree1)
847 free(tree1);
848 if (tree2)
849 free(tree2);
850 }
851
852 static char *get_nth_line(long line, unsigned long *ends, void *data)
853 {
854 if (line == 0)
855 return (char *)data;
856 else
857 return (char *)data + ends[line] + 1;
858 }
859
860 static void print_line(const char *prefix, char first,
861 long line, unsigned long *ends, void *data,
862 const char *color, const char *reset)
863 {
864 char *begin = get_nth_line(line, ends, data);
865 char *end = get_nth_line(line+1, ends, data);
866 int had_nl = 0;
867
868 if (end > begin && end[-1] == '\n') {
869 end--;
870 had_nl = 1;
871 }
872
873 fputs(prefix, stdout);
874 fputs(color, stdout);
875 putchar(first);
876 fwrite(begin, 1, end-begin, stdout);
877 fputs(reset, stdout);
878 putchar('\n');
879 if (!had_nl)
880 fputs("\\ No newline at end of file\n", stdout);
881 }
882
883 static char *output_prefix(struct diff_options *opt)
884 {
885 char *prefix = "";
886
887 if (opt->output_prefix) {
888 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
889 prefix = sb->buf;
890 }
891
892 return prefix;
893 }
894
895 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
896 {
897 int i, j = 0;
898 long p_lines, t_lines;
899 unsigned long *p_ends = NULL, *t_ends = NULL;
900 struct diff_filepair *pair = range->pair;
901 struct diff_ranges *diff = &range->diff;
902
903 struct diff_options *opt = &rev->diffopt;
904 char *prefix = output_prefix(opt);
905 const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
906 const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
907 const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
908 const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
909 const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
910 const char *c_plain = diff_get_color(opt->use_color, DIFF_PLAIN);
911
912 if (!pair || !diff)
913 return;
914
915 if (pair->one->sha1_valid)
916 fill_line_ends(pair->one, &p_lines, &p_ends);
917 fill_line_ends(pair->two, &t_lines, &t_ends);
918
919 printf("%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
920 printf("%s%s--- %s%s%s\n", prefix, c_meta,
921 pair->one->sha1_valid ? "a/" : "",
922 pair->one->sha1_valid ? pair->one->path : "/dev/null",
923 c_reset);
924 printf("%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
925 for (i = 0; i < range->ranges.nr; i++) {
926 long p_start, p_end;
927 long t_start = range->ranges.ranges[i].start;
928 long t_end = range->ranges.ranges[i].end;
929 long t_cur = t_start;
930 int j_last;
931
932 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
933 j++;
934 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
935 continue;
936
937 /* Scan ahead to determine the last diff that falls in this range */
938 j_last = j;
939 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
940 j_last++;
941 if (j_last > j)
942 j_last--;
943
944 /*
945 * Compute parent hunk headers: we know that the diff
946 * has the correct line numbers (but not all hunks).
947 * So it suffices to shift the start/end according to
948 * the line numbers of the first/last hunk(s) that
949 * fall in this range.
950 */
951 if (t_start < diff->target.ranges[j].start)
952 p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
953 else
954 p_start = diff->parent.ranges[j].start;
955 if (t_end > diff->target.ranges[j_last].end)
956 p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
957 else
958 p_end = diff->parent.ranges[j_last].end;
959
960 if (!p_start && !p_end) {
961 p_start = -1;
962 p_end = -1;
963 }
964
965 /* Now output a diff hunk for this range */
966 printf("%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
967 prefix, c_frag,
968 p_start+1, p_end-p_start, t_start+1, t_end-t_start,
969 c_reset);
970 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
971 int k;
972 for (; t_cur < diff->target.ranges[j].start; t_cur++)
973 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
974 c_plain, c_reset);
975 for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
976 print_line(prefix, '-', k, p_ends, pair->one->data,
977 c_old, c_reset);
978 for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
979 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
980 c_new, c_reset);
981 j++;
982 }
983 for (; t_cur < t_end; t_cur++)
984 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
985 c_plain, c_reset);
986 }
987
988 free(p_ends);
989 free(t_ends);
990 }
991
992 /*
993 * NEEDSWORK: manually building a diff here is not the Right
994 * Thing(tm). log -L should be built into the diff pipeline.
995 */
996 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
997 {
998 puts(output_prefix(&rev->diffopt));
999 while (range) {
1000 dump_diff_hacky_one(rev, range);
1001 range = range->next;
1002 }
1003 }
1004
1005 /*
1006 * Unlike most other functions, this destructively operates on
1007 * 'range'.
1008 */
1009 static int process_diff_filepair(struct rev_info *rev,
1010 struct diff_filepair *pair,
1011 struct line_log_data *range,
1012 struct diff_ranges **diff_out)
1013 {
1014 struct line_log_data *rg = range;
1015 struct range_set tmp;
1016 struct diff_ranges diff;
1017 mmfile_t file_parent, file_target;
1018
1019 assert(pair->two->path);
1020 while (rg) {
1021 assert(rg->spec->path);
1022 if (!strcmp(rg->spec->path, pair->two->path))
1023 break;
1024 rg = rg->next;
1025 }
1026
1027 if (!rg)
1028 return 0;
1029 if (rg->ranges.nr == 0)
1030 return 0;
1031
1032 assert(pair->two->sha1_valid);
1033 diff_populate_filespec(pair->two, 0);
1034 file_target.ptr = pair->two->data;
1035 file_target.size = pair->two->size;
1036
1037 if (pair->one->sha1_valid) {
1038 diff_populate_filespec(pair->one, 0);
1039 file_parent.ptr = pair->one->data;
1040 file_parent.size = pair->one->size;
1041 } else {
1042 file_parent.ptr = "";
1043 file_parent.size = 0;
1044 }
1045
1046 diff_ranges_init(&diff);
1047 collect_diff(&file_parent, &file_target, &diff);
1048
1049 /* NEEDSWORK should apply some heuristics to prevent mismatches */
1050 rg->spec->path = xstrdup(pair->one->path);
1051
1052 range_set_init(&tmp, 0);
1053 range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1054 range_set_release(&rg->ranges);
1055 range_set_move(&rg->ranges, &tmp);
1056
1057 diff_ranges_release(&diff);
1058
1059 return ((*diff_out)->parent.nr > 0);
1060 }
1061
1062 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1063 {
1064 struct diff_filepair *new = xmalloc(sizeof(struct diff_filepair));
1065 new->one = pair->one;
1066 new->two = pair->two;
1067 new->one->count++;
1068 new->two->count++;
1069 return new;
1070 }
1071
1072 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1073 {
1074 int i, j;
1075 for (i = 0; i < n; i++)
1076 for (j = 0; j < dq[i].nr; j++)
1077 diff_free_filepair(dq[i].queue[j]);
1078 free(dq);
1079 }
1080
1081 static int process_all_files(struct line_log_data **range_out,
1082 struct rev_info *rev,
1083 struct diff_queue_struct *queue,
1084 struct line_log_data *range)
1085 {
1086 int i, changed = 0;
1087
1088 *range_out = line_log_data_copy(range);
1089
1090 for (i = 0; i < queue->nr; i++) {
1091 struct diff_ranges *pairdiff = NULL;
1092 if (process_diff_filepair(rev, queue->queue[i], *range_out, &pairdiff)) {
1093 struct line_log_data *rg = range;
1094 changed++;
1095 /* NEEDSWORK tramples over data structures not owned here */
1096 while (rg && strcmp(rg->spec->path, queue->queue[i]->two->path))
1097 rg = rg->next;
1098 assert(rg);
1099 rg->pair = diff_filepair_dup(queue->queue[i]);
1100 memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1101 }
1102 }
1103
1104 return changed;
1105 }
1106
1107 int line_log_print(struct rev_info *rev, struct commit *commit)
1108 {
1109 struct line_log_data *range = lookup_line_range(rev, commit);
1110
1111 show_log(rev);
1112 dump_diff_hacky(rev, range);
1113 return 1;
1114 }
1115
1116 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1117 struct line_log_data *range)
1118 {
1119 struct commit *parent = NULL;
1120 struct diff_queue_struct queue;
1121 struct line_log_data *parent_range;
1122 int changed;
1123
1124 if (commit->parents)
1125 parent = commit->parents->item;
1126
1127 queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1128 changed = process_all_files(&parent_range, rev, &queue, range);
1129 if (parent)
1130 add_line_range(rev, parent, parent_range);
1131 return changed;
1132 }
1133
1134 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1135 struct line_log_data *range)
1136 {
1137 struct diff_queue_struct *diffqueues;
1138 struct line_log_data **cand;
1139 struct commit **parents;
1140 struct commit_list *p;
1141 int i;
1142 int nparents = count_parents(commit);
1143
1144 diffqueues = xmalloc(nparents * sizeof(*diffqueues));
1145 cand = xmalloc(nparents * sizeof(*cand));
1146 parents = xmalloc(nparents * sizeof(*parents));
1147
1148 p = commit->parents;
1149 for (i = 0; i < nparents; i++) {
1150 parents[i] = p->item;
1151 p = p->next;
1152 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1153 }
1154
1155 for (i = 0; i < nparents; i++) {
1156 int changed;
1157 cand[i] = NULL;
1158 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1159 if (!changed) {
1160 /*
1161 * This parent can take all the blame, so we
1162 * don't follow any other path in history
1163 */
1164 add_line_range(rev, parents[i], cand[i]);
1165 clear_commit_line_range(rev, commit);
1166 commit->parents = xmalloc(sizeof(struct commit_list));
1167 commit->parents->item = parents[i];
1168 commit->parents->next = NULL;
1169 free(parents);
1170 free(cand);
1171 free_diffqueues(nparents, diffqueues);
1172 /* NEEDSWORK leaking like a sieve */
1173 return 0;
1174 }
1175 }
1176
1177 /*
1178 * No single parent took the blame. We add the candidates
1179 * from the above loop to the parents.
1180 */
1181 for (i = 0; i < nparents; i++) {
1182 add_line_range(rev, parents[i], cand[i]);
1183 }
1184
1185 clear_commit_line_range(rev, commit);
1186 free(parents);
1187 free(cand);
1188 free_diffqueues(nparents, diffqueues);
1189 return 1;
1190
1191 /* NEEDSWORK evil merge detection stuff */
1192 /* NEEDSWORK leaking like a sieve */
1193 }
1194
1195 static int process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1196 {
1197 struct line_log_data *range = lookup_line_range(rev, commit);
1198 int changed = 0;
1199
1200 if (range) {
1201 if (!commit->parents || !commit->parents->next)
1202 changed = process_ranges_ordinary_commit(rev, commit, range);
1203 else
1204 changed = process_ranges_merge_commit(rev, commit, range);
1205 }
1206
1207 if (!changed)
1208 commit->object.flags |= TREESAME;
1209
1210 return changed;
1211 }
1212
1213 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev, struct commit **pp)
1214 {
1215 for (;;) {
1216 struct commit *p = *pp;
1217 if (p->parents && p->parents->next)
1218 return rewrite_one_ok;
1219 if (p->object.flags & UNINTERESTING)
1220 return rewrite_one_ok;
1221 if (!(p->object.flags & TREESAME))
1222 return rewrite_one_ok;
1223 if (!p->parents)
1224 return rewrite_one_noparents;
1225 *pp = p->parents->item;
1226 }
1227 }
1228
1229 int line_log_filter(struct rev_info *rev)
1230 {
1231 struct commit *commit;
1232 struct commit_list *list = rev->commits;
1233 struct commit_list *out = NULL, **pp = &out;
1234
1235 while (list) {
1236 struct commit_list *to_free = NULL;
1237 commit = list->item;
1238 if (process_ranges_arbitrary_commit(rev, commit)) {
1239 *pp = list;
1240 pp = &list->next;
1241 } else
1242 to_free = list;
1243 list = list->next;
1244 free(to_free);
1245 }
1246 *pp = NULL;
1247
1248 for (list = out; list; list = list->next)
1249 rewrite_parents(rev, list->item, line_log_rewrite_one);
1250
1251 rev->commits = out;
1252
1253 return 0;
1254 }
Gitweb development nest