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