?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
debian: new upstream release
[git/debian.git] / line-log.c
blob24a1ecb67799c23570ef485bc382b4bbdf2779d2
1 #include "git-compat-util.h"
2 #include "line-range.h"
3 #include "hex.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 "repository.h"
11 #include "revision.h"
12 #include "xdiff-interface.h"
13 #include "strbuf.h"
14 #include "log-tree.h"
15 #include "graph.h"
16 #include "userdiff.h"
17 #include "line-log.h"
18 #include "setup.h"
19 #include "strvec.h"
20 #include "bloom.h"
21 #include "tree-walk.h"
22
23 static void range_set_grow(struct range_set *rs, size_t extra)
24 {
25 ALLOC_GROW(rs->ranges, rs->nr + extra, rs->alloc);
26 }
27
28 /* Either initialization would be fine */
29 #define RANGE_SET_INIT {0}
30
31 void range_set_init(struct range_set *rs, size_t prealloc)
32 {
33 rs->alloc = rs->nr = 0;
34 rs->ranges = NULL;
35 if (prealloc)
36 range_set_grow(rs, prealloc);
37 }
38
39 void range_set_release(struct range_set *rs)
40 {
41 FREE_AND_NULL(rs->ranges);
42 rs->alloc = rs->nr = 0;
43 }
44
45 /* dst must be uninitialized! */
46 static void range_set_copy(struct range_set *dst, struct range_set *src)
47 {
48 range_set_init(dst, src->nr);
49 COPY_ARRAY(dst->ranges, src->ranges, src->nr);
50 dst->nr = src->nr;
51 }
52
53 static void range_set_move(struct range_set *dst, struct range_set *src)
54 {
55 range_set_release(dst);
56 dst->ranges = src->ranges;
57 dst->nr = src->nr;
58 dst->alloc = src->alloc;
59 src->ranges = NULL;
60 src->alloc = src->nr = 0;
61 }
62
63 /* tack on a _new_ range _at the end_ */
64 void range_set_append_unsafe(struct range_set *rs, long a, long b)
65 {
66 assert(a <= b);
67 range_set_grow(rs, 1);
68 rs->ranges[rs->nr].start = a;
69 rs->ranges[rs->nr].end = b;
70 rs->nr++;
71 }
72
73 void range_set_append(struct range_set *rs, long a, long b)
74 {
75 assert(rs->nr == 0 || rs->ranges[rs->nr-1].end <= a);
76 range_set_append_unsafe(rs, a, b);
77 }
78
79 static int range_cmp(const void *_r, const void *_s)
80 {
81 const struct range *r = _r;
82 const struct range *s = _s;
83
84 /* this could be simply 'return r.start-s.start', but for the types */
85 if (r->start == s->start)
86 return 0;
87 if (r->start < s->start)
88 return -1;
89 return 1;
90 }
91
92 /*
93 * Check that the ranges are non-empty, sorted and non-overlapping
94 */
95 static void range_set_check_invariants(struct range_set *rs)
96 {
97 unsigned int i;
98
99 if (!rs)
100 return;
101
102 if (rs->nr)
103 assert(rs->ranges[0].start < rs->ranges[0].end);
104
105 for (i = 1; i < rs->nr; i++) {
106 assert(rs->ranges[i-1].end < rs->ranges[i].start);
107 assert(rs->ranges[i].start < rs->ranges[i].end);
108 }
109 }
110
111 /*
112 * In-place pass of sorting and merging the ranges in the range set,
113 * to establish the invariants when we get the ranges from the user
114 */
115 void sort_and_merge_range_set(struct range_set *rs)
116 {
117 unsigned int i;
118 unsigned int o = 0; /* output cursor */
119
120 QSORT(rs->ranges, rs->nr, range_cmp);
121
122 for (i = 0; i < rs->nr; i++) {
123 if (rs->ranges[i].start == rs->ranges[i].end)
124 continue;
125 if (o > 0 && rs->ranges[i].start <= rs->ranges[o-1].end) {
126 if (rs->ranges[o-1].end < rs->ranges[i].end)
127 rs->ranges[o-1].end = rs->ranges[i].end;
128 } else {
129 rs->ranges[o].start = rs->ranges[i].start;
130 rs->ranges[o].end = rs->ranges[i].end;
131 o++;
132 }
133 }
134 assert(o <= rs->nr);
135 rs->nr = o;
136
137 range_set_check_invariants(rs);
138 }
139
140 /*
141 * Union of range sets (i.e., sets of line numbers). Used to merge
142 * them when searches meet at a common ancestor.
143 *
144 * This is also where the ranges are consolidated into canonical form:
145 * overlapping and adjacent ranges are merged, and empty ranges are
146 * removed.
147 */
148 static void range_set_union(struct range_set *out,
149 struct range_set *a, struct range_set *b)
150 {
151 unsigned int i = 0, j = 0;
152 struct range *ra = a->ranges;
153 struct range *rb = b->ranges;
154 /* cannot make an alias of out->ranges: it may change during grow */
155
156 assert(out->nr == 0);
157 while (i < a->nr || j < b->nr) {
158 struct range *new_range;
159 if (i < a->nr && j < b->nr) {
160 if (ra[i].start < rb[j].start)
161 new_range = &ra[i++];
162 else if (ra[i].start > rb[j].start)
163 new_range = &rb[j++];
164 else if (ra[i].end < rb[j].end)
165 new_range = &ra[i++];
166 else
167 new_range = &rb[j++];
168 } else if (i < a->nr) /* b exhausted */
169 new_range = &ra[i++];
170 else /* a exhausted */
171 new_range = &rb[j++];
172 if (new_range->start == new_range->end)
173 ; /* empty range */
174 else if (!out->nr || out->ranges[out->nr-1].end < new_range->start) {
175 range_set_grow(out, 1);
176 out->ranges[out->nr].start = new_range->start;
177 out->ranges[out->nr].end = new_range->end;
178 out->nr++;
179 } else if (out->ranges[out->nr-1].end < new_range->end) {
180 out->ranges[out->nr-1].end = new_range->end;
181 }
182 }
183 }
184
185 /*
186 * Difference of range sets (out = a \ b). Pass the "interesting"
187 * ranges as 'a' and the target side of the diff as 'b': it removes
188 * the ranges for which the commit is responsible.
189 */
190 static void range_set_difference(struct range_set *out,
191 struct range_set *a, struct range_set *b)
192 {
193 unsigned int i, j = 0;
194 for (i = 0; i < a->nr; i++) {
195 long start = a->ranges[i].start;
196 long end = a->ranges[i].end;
197 while (start < end) {
198 while (j < b->nr && start >= b->ranges[j].end)
199 /*
200 * a: |-------
201 * b: ------|
202 */
203 j++;
204 if (j >= b->nr || end < b->ranges[j].start) {
205 /*
206 * b exhausted, or
207 * a: ----|
208 * b: |----
209 */
210 range_set_append(out, start, end);
211 break;
212 }
213 if (start >= b->ranges[j].start) {
214 /*
215 * a: |--????
216 * b: |------|
217 */
218 start = b->ranges[j].end;
219 } else if (end > b->ranges[j].start) {
220 /*
221 * a: |-----|
222 * b: |--?????
223 */
224 if (start < b->ranges[j].start)
225 range_set_append(out, start, b->ranges[j].start);
226 start = b->ranges[j].end;
227 }
228 }
229 }
230 }
231
232 static void diff_ranges_init(struct diff_ranges *diff)
233 {
234 range_set_init(&diff->parent, 0);
235 range_set_init(&diff->target, 0);
236 }
237
238 static void diff_ranges_release(struct diff_ranges *diff)
239 {
240 range_set_release(&diff->parent);
241 range_set_release(&diff->target);
242 }
243
244 static void line_log_data_init(struct line_log_data *r)
245 {
246 memset(r, 0, sizeof(struct line_log_data));
247 range_set_init(&r->ranges, 0);
248 }
249
250 static void line_log_data_clear(struct line_log_data *r)
251 {
252 range_set_release(&r->ranges);
253 if (r->pair)
254 diff_free_filepair(r->pair);
255 }
256
257 static void free_line_log_data(struct line_log_data *r)
258 {
259 while (r) {
260 struct line_log_data *next = r->next;
261 line_log_data_clear(r);
262 free(r);
263 r = next;
264 }
265 }
266
267 static struct line_log_data *
268 search_line_log_data(struct line_log_data *list, const char *path,
269 struct line_log_data **insertion_point)
270 {
271 struct line_log_data *p = list;
272 if (insertion_point)
273 *insertion_point = NULL;
274 while (p) {
275 int cmp = strcmp(p->path, path);
276 if (!cmp)
277 return p;
278 if (insertion_point && cmp < 0)
279 *insertion_point = p;
280 p = p->next;
281 }
282 return NULL;
283 }
284
285 /*
286 * Note: takes ownership of 'path', which happens to be what the only
287 * caller needs.
288 */
289 static void line_log_data_insert(struct line_log_data **list,
290 char *path,
291 long begin, long end)
292 {
293 struct line_log_data *ip;
294 struct line_log_data *p = search_line_log_data(*list, path, &ip);
295
296 if (p) {
297 range_set_append_unsafe(&p->ranges, begin, end);
298 free(path);
299 return;
300 }
301
302 CALLOC_ARRAY(p, 1);
303 p->path = path;
304 range_set_append(&p->ranges, begin, end);
305 if (ip) {
306 p->next = ip->next;
307 ip->next = p;
308 } else {
309 p->next = *list;
310 *list = p;
311 }
312 }
313
314 struct collect_diff_cbdata {
315 struct diff_ranges *diff;
316 };
317
318 static int collect_diff_cb(long start_a, long count_a,
319 long start_b, long count_b,
320 void *data)
321 {
322 struct collect_diff_cbdata *d = data;
323
324 if (count_a >= 0)
325 range_set_append(&d->diff->parent, start_a, start_a + count_a);
326 if (count_b >= 0)
327 range_set_append(&d->diff->target, start_b, start_b + count_b);
328
329 return 0;
330 }
331
332 static int collect_diff(mmfile_t *parent, mmfile_t *target, struct diff_ranges *out)
333 {
334 struct collect_diff_cbdata cbdata = {NULL};
335 xpparam_t xpp;
336 xdemitconf_t xecfg;
337 xdemitcb_t ecb;
338
339 memset(&xpp, 0, sizeof(xpp));
340 memset(&xecfg, 0, sizeof(xecfg));
341 xecfg.ctxlen = xecfg.interhunkctxlen = 0;
342
343 cbdata.diff = out;
344 xecfg.hunk_func = collect_diff_cb;
345 memset(&ecb, 0, sizeof(ecb));
346 ecb.priv = &cbdata;
347 return xdi_diff(parent, target, &xpp, &xecfg, &ecb);
348 }
349
350 /*
351 * These are handy for debugging. Removing them with #if 0 silences
352 * the "unused function" warning.
353 */
354 #if 0
355 static void dump_range_set(struct range_set *rs, const char *desc)
356 {
357 int i;
358 printf("range set %s (%d items):\n", desc, rs->nr);
359 for (i = 0; i < rs->nr; i++)
360 printf("\t[%ld,%ld]\n", rs->ranges[i].start, rs->ranges[i].end);
361 }
362
363 static void dump_line_log_data(struct line_log_data *r)
364 {
365 char buf[4096];
366 while (r) {
367 snprintf(buf, 4096, "file %s\n", r->path);
368 dump_range_set(&r->ranges, buf);
369 r = r->next;
370 }
371 }
372
373 static void dump_diff_ranges(struct diff_ranges *diff, const char *desc)
374 {
375 int i;
376 assert(diff->parent.nr == diff->target.nr);
377 printf("diff ranges %s (%d items):\n", desc, diff->parent.nr);
378 printf("\tparent\ttarget\n");
379 for (i = 0; i < diff->parent.nr; i++) {
380 printf("\t[%ld,%ld]\t[%ld,%ld]\n",
381 diff->parent.ranges[i].start,
382 diff->parent.ranges[i].end,
383 diff->target.ranges[i].start,
384 diff->target.ranges[i].end);
385 }
386 }
387 #endif
388
389
390 static int ranges_overlap(struct range *a, struct range *b)
391 {
392 return !(a->end <= b->start || b->end <= a->start);
393 }
394
395 /*
396 * Given a diff and the set of interesting ranges, determine all hunks
397 * of the diff which touch (overlap) at least one of the interesting
398 * ranges in the target.
399 */
400 static void diff_ranges_filter_touched(struct diff_ranges *out,
401 struct diff_ranges *diff,
402 struct range_set *rs)
403 {
404 unsigned int i, j = 0;
405
406 assert(out->target.nr == 0);
407
408 for (i = 0; i < diff->target.nr; i++) {
409 while (diff->target.ranges[i].start > rs->ranges[j].end) {
410 j++;
411 if (j == rs->nr)
412 return;
413 }
414 if (ranges_overlap(&diff->target.ranges[i], &rs->ranges[j])) {
415 range_set_append(&out->parent,
416 diff->parent.ranges[i].start,
417 diff->parent.ranges[i].end);
418 range_set_append(&out->target,
419 diff->target.ranges[i].start,
420 diff->target.ranges[i].end);
421 }
422 }
423 }
424
425 /*
426 * Adjust the line counts in 'rs' to account for the lines
427 * added/removed in the diff.
428 */
429 static void range_set_shift_diff(struct range_set *out,
430 struct range_set *rs,
431 struct diff_ranges *diff)
432 {
433 unsigned int i, j = 0;
434 long offset = 0;
435 struct range *src = rs->ranges;
436 struct range *target = diff->target.ranges;
437 struct range *parent = diff->parent.ranges;
438
439 for (i = 0; i < rs->nr; i++) {
440 while (j < diff->target.nr && src[i].start >= target[j].start) {
441 offset += (parent[j].end-parent[j].start)
442 - (target[j].end-target[j].start);
443 j++;
444 }
445 range_set_append(out, src[i].start+offset, src[i].end+offset);
446 }
447 }
448
449 /*
450 * Given a diff and the set of interesting ranges, map the ranges
451 * across the diff. That is: observe that the target commit takes
452 * blame for all the + (target-side) ranges. So for every pair of
453 * ranges in the diff that was touched, we remove the latter and add
454 * its parent side.
455 */
456 static void range_set_map_across_diff(struct range_set *out,
457 struct range_set *rs,
458 struct diff_ranges *diff,
459 struct diff_ranges **touched_out)
460 {
461 struct diff_ranges *touched = xmalloc(sizeof(*touched));
462 struct range_set tmp1 = RANGE_SET_INIT;
463 struct range_set tmp2 = RANGE_SET_INIT;
464
465 diff_ranges_init(touched);
466 diff_ranges_filter_touched(touched, diff, rs);
467 range_set_difference(&tmp1, rs, &touched->target);
468 range_set_shift_diff(&tmp2, &tmp1, diff);
469 range_set_union(out, &tmp2, &touched->parent);
470 range_set_release(&tmp1);
471 range_set_release(&tmp2);
472
473 *touched_out = touched;
474 }
475
476 static struct commit *check_single_commit(struct rev_info *revs)
477 {
478 struct object *commit = NULL;
479 int found = -1;
480 int i;
481
482 for (i = 0; i < revs->pending.nr; i++) {
483 struct object *obj = revs->pending.objects[i].item;
484 if (obj->flags & UNINTERESTING)
485 continue;
486 obj = deref_tag(revs->repo, obj, NULL, 0);
487 if (!obj || obj->type != OBJ_COMMIT)
488 die("Non commit %s?", revs->pending.objects[i].name);
489 if (commit)
490 die("More than one commit to dig from: %s and %s?",
491 revs->pending.objects[i].name,
492 revs->pending.objects[found].name);
493 commit = obj;
494 found = i;
495 }
496
497 if (!commit)
498 die("No commit specified?");
499
500 return (struct commit *) commit;
501 }
502
503 static void fill_blob_sha1(struct repository *r, struct commit *commit,
504 struct diff_filespec *spec)
505 {
506 unsigned short mode;
507 struct object_id oid;
508
509 if (get_tree_entry(r, &commit->object.oid, spec->path, &oid, &mode))
510 die("There is no path %s in the commit", spec->path);
511 fill_filespec(spec, &oid, 1, mode);
512
513 return;
514 }
515
516 static void fill_line_ends(struct repository *r,
517 struct diff_filespec *spec,
518 long *lines,
519 unsigned long **line_ends)
520 {
521 int num = 0, size = 50;
522 long cur = 0;
523 unsigned long *ends = NULL;
524 char *data = NULL;
525
526 if (diff_populate_filespec(r, spec, NULL))
527 die("Cannot read blob %s", oid_to_hex(&spec->oid));
528
529 ALLOC_ARRAY(ends, size);
530 ends[cur++] = 0;
531 data = spec->data;
532 while (num < spec->size) {
533 if (data[num] == '\n' || num == spec->size - 1) {
534 ALLOC_GROW(ends, (cur + 1), size);
535 ends[cur++] = num;
536 }
537 num++;
538 }
539
540 /* shrink the array to fit the elements */
541 REALLOC_ARRAY(ends, cur);
542 *lines = cur-1;
543 *line_ends = ends;
544 }
545
546 struct nth_line_cb {
547 struct diff_filespec *spec;
548 long lines;
549 unsigned long *line_ends;
550 };
551
552 static const char *nth_line(void *data, long line)
553 {
554 struct nth_line_cb *d = data;
555 assert(d && line <= d->lines);
556 assert(d->spec && d->spec->data);
557
558 if (line == 0)
559 return (char *)d->spec->data;
560 else
561 return (char *)d->spec->data + d->line_ends[line] + 1;
562 }
563
564 static struct line_log_data *
565 parse_lines(struct repository *r, struct commit *commit,
566 const char *prefix, struct string_list *args)
567 {
568 long lines = 0;
569 unsigned long *ends = NULL;
570 struct nth_line_cb cb_data;
571 struct string_list_item *item;
572 struct line_log_data *ranges = NULL;
573 struct line_log_data *p;
574
575 for_each_string_list_item(item, args) {
576 const char *name_part, *range_part;
577 char *full_name;
578 struct diff_filespec *spec;
579 long begin = 0, end = 0;
580 long anchor;
581
582 name_part = skip_range_arg(item->string, r->index);
583 if (!name_part || *name_part != ':' || !name_part[1])
584 die("-L argument not 'start,end:file' or ':funcname:file': %s",
585 item->string);
586 range_part = xstrndup(item->string, name_part - item->string);
587 name_part++;
588
589 full_name = prefix_path(prefix, prefix ? strlen(prefix) : 0,
590 name_part);
591
592 spec = alloc_filespec(full_name);
593 fill_blob_sha1(r, commit, spec);
594 fill_line_ends(r, spec, &lines, &ends);
595 cb_data.spec = spec;
596 cb_data.lines = lines;
597 cb_data.line_ends = ends;
598
599 p = search_line_log_data(ranges, full_name, NULL);
600 if (p && p->ranges.nr)
601 anchor = p->ranges.ranges[p->ranges.nr - 1].end + 1;
602 else
603 anchor = 1;
604
605 if (parse_range_arg(range_part, nth_line, &cb_data,
606 lines, anchor, &begin, &end,
607 full_name, r->index))
608 die("malformed -L argument '%s'", range_part);
609 if ((!lines && (begin || end)) || lines < begin)
610 die("file %s has only %lu lines", name_part, lines);
611 if (begin < 1)
612 begin = 1;
613 if (end < 1 || lines < end)
614 end = lines;
615 begin--;
616 line_log_data_insert(&ranges, full_name, begin, end);
617
618 free_filespec(spec);
619 FREE_AND_NULL(ends);
620 }
621
622 for (p = ranges; p; p = p->next)
623 sort_and_merge_range_set(&p->ranges);
624
625 return ranges;
626 }
627
628 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
629 {
630 struct line_log_data *ret = xmalloc(sizeof(*ret));
631
632 assert(r);
633 line_log_data_init(ret);
634 range_set_copy(&ret->ranges, &r->ranges);
635
636 ret->path = xstrdup(r->path);
637
638 return ret;
639 }
640
641 static struct line_log_data *
642 line_log_data_copy(struct line_log_data *r)
643 {
644 struct line_log_data *ret = NULL;
645 struct line_log_data *tmp = NULL, *prev = NULL;
646
647 assert(r);
648 ret = tmp = prev = line_log_data_copy_one(r);
649 r = r->next;
650 while (r) {
651 tmp = line_log_data_copy_one(r);
652 prev->next = tmp;
653 prev = tmp;
654 r = r->next;
655 }
656
657 return ret;
658 }
659
660 /* merge two range sets across files */
661 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
662 struct line_log_data *b)
663 {
664 struct line_log_data *head = NULL, **pp = &head;
665
666 while (a || b) {
667 struct line_log_data *src;
668 struct line_log_data *src2 = NULL;
669 struct line_log_data *d;
670 int cmp;
671 if (!a)
672 cmp = 1;
673 else if (!b)
674 cmp = -1;
675 else
676 cmp = strcmp(a->path, b->path);
677 if (cmp < 0) {
678 src = a;
679 a = a->next;
680 } else if (cmp == 0) {
681 src = a;
682 a = a->next;
683 src2 = b;
684 b = b->next;
685 } else {
686 src = b;
687 b = b->next;
688 }
689 d = xmalloc(sizeof(struct line_log_data));
690 line_log_data_init(d);
691 d->path = xstrdup(src->path);
692 *pp = d;
693 pp = &d->next;
694 if (src2)
695 range_set_union(&d->ranges, &src->ranges, &src2->ranges);
696 else
697 range_set_copy(&d->ranges, &src->ranges);
698 }
699
700 return head;
701 }
702
703 static void add_line_range(struct rev_info *revs, struct commit *commit,
704 struct line_log_data *range)
705 {
706 struct line_log_data *old_line = NULL;
707 struct line_log_data *new_line = NULL;
708
709 old_line = lookup_decoration(&revs->line_log_data, &commit->object);
710 if (old_line && range) {
711 new_line = line_log_data_merge(old_line, range);
712 free_line_log_data(old_line);
713 } else if (range)
714 new_line = line_log_data_copy(range);
715
716 if (new_line)
717 add_decoration(&revs->line_log_data, &commit->object, new_line);
718 }
719
720 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
721 {
722 struct line_log_data *r;
723 r = lookup_decoration(&revs->line_log_data, &commit->object);
724 if (!r)
725 return;
726 free_line_log_data(r);
727 add_decoration(&revs->line_log_data, &commit->object, NULL);
728 }
729
730 static struct line_log_data *lookup_line_range(struct rev_info *revs,
731 struct commit *commit)
732 {
733 struct line_log_data *ret = NULL;
734 struct line_log_data *d;
735
736 ret = lookup_decoration(&revs->line_log_data, &commit->object);
737
738 for (d = ret; d; d = d->next)
739 range_set_check_invariants(&d->ranges);
740
741 return ret;
742 }
743
744 static int same_paths_in_pathspec_and_range(struct pathspec *pathspec,
745 struct line_log_data *range)
746 {
747 int i;
748 struct line_log_data *r;
749
750 for (i = 0, r = range; i < pathspec->nr && r; i++, r = r->next)
751 if (strcmp(pathspec->items[i].match, r->path))
752 return 0;
753 if (i < pathspec->nr || r)
754 /* different number of pathspec items and ranges */
755 return 0;
756
757 return 1;
758 }
759
760 static void parse_pathspec_from_ranges(struct pathspec *pathspec,
761 struct line_log_data *range)
762 {
763 struct line_log_data *r;
764 struct strvec array = STRVEC_INIT;
765 const char **paths;
766
767 for (r = range; r; r = r->next)
768 strvec_push(&array, r->path);
769 paths = strvec_detach(&array);
770
771 parse_pathspec(pathspec, 0, PATHSPEC_PREFER_FULL, "", paths);
772 /* strings are now owned by pathspec */
773 free(paths);
774 }
775
776 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
777 {
778 struct commit *commit = NULL;
779 struct line_log_data *range;
780
781 commit = check_single_commit(rev);
782 range = parse_lines(rev->diffopt.repo, commit, prefix, args);
783 add_line_range(rev, commit, range);
784
785 parse_pathspec_from_ranges(&rev->diffopt.pathspec, range);
786 }
787
788 static void move_diff_queue(struct diff_queue_struct *dst,
789 struct diff_queue_struct *src)
790 {
791 assert(src != dst);
792 memcpy(dst, src, sizeof(struct diff_queue_struct));
793 DIFF_QUEUE_CLEAR(src);
794 }
795
796 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
797 {
798 int i;
799 struct diff_queue_struct outq;
800 DIFF_QUEUE_CLEAR(&outq);
801
802 for (i = 0; i < diff_queued_diff.nr; i++) {
803 struct diff_filepair *p = diff_queued_diff.queue[i];
804 struct line_log_data *rg = NULL;
805
806 if (!DIFF_FILE_VALID(p->two)) {
807 if (keep_deletions)
808 diff_q(&outq, p);
809 else
810 diff_free_filepair(p);
811 continue;
812 }
813 for (rg = range; rg; rg = rg->next) {
814 if (!strcmp(rg->path, p->two->path))
815 break;
816 }
817 if (rg)
818 diff_q(&outq, p);
819 else
820 diff_free_filepair(p);
821 }
822 free(diff_queued_diff.queue);
823 diff_queued_diff = outq;
824 }
825
826 static inline int diff_might_be_rename(void)
827 {
828 int i;
829 for (i = 0; i < diff_queued_diff.nr; i++)
830 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
831 /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
832 /* diff_queued_diff.queue[i]->two->path); */
833 return 1;
834 }
835 return 0;
836 }
837
838 static void queue_diffs(struct line_log_data *range,
839 struct diff_options *opt,
840 struct diff_queue_struct *queue,
841 struct commit *commit, struct commit *parent)
842 {
843 struct object_id *tree_oid, *parent_tree_oid;
844
845 assert(commit);
846
847 tree_oid = get_commit_tree_oid(commit);
848 parent_tree_oid = parent ? get_commit_tree_oid(parent) : NULL;
849
850 if (opt->detect_rename &&
851 !same_paths_in_pathspec_and_range(&opt->pathspec, range)) {
852 clear_pathspec(&opt->pathspec);
853 parse_pathspec_from_ranges(&opt->pathspec, range);
854 }
855 DIFF_QUEUE_CLEAR(&diff_queued_diff);
856 diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
857 if (opt->detect_rename && diff_might_be_rename()) {
858 /* must look at the full tree diff to detect renames */
859 clear_pathspec(&opt->pathspec);
860 DIFF_QUEUE_CLEAR(&diff_queued_diff);
861
862 diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
863
864 filter_diffs_for_paths(range, 1);
865 diffcore_std(opt);
866 filter_diffs_for_paths(range, 0);
867 }
868 move_diff_queue(queue, &diff_queued_diff);
869 }
870
871 static char *get_nth_line(long line, unsigned long *ends, void *data)
872 {
873 if (line == 0)
874 return (char *)data;
875 else
876 return (char *)data + ends[line] + 1;
877 }
878
879 static void print_line(const char *prefix, char first,
880 long line, unsigned long *ends, void *data,
881 const char *color, const char *reset, FILE *file)
882 {
883 char *begin = get_nth_line(line, ends, data);
884 char *end = get_nth_line(line+1, ends, data);
885 int had_nl = 0;
886
887 if (end > begin && end[-1] == '\n') {
888 end--;
889 had_nl = 1;
890 }
891
892 fputs(prefix, file);
893 fputs(color, file);
894 putc(first, file);
895 fwrite(begin, 1, end-begin, file);
896 fputs(reset, file);
897 putc('\n', file);
898 if (!had_nl)
899 fputs("\\ No newline at end of file\n", file);
900 }
901
902 static char *output_prefix(struct diff_options *opt)
903 {
904 char *prefix = "";
905
906 if (opt->output_prefix) {
907 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
908 prefix = sb->buf;
909 }
910
911 return prefix;
912 }
913
914 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
915 {
916 unsigned int i, j = 0;
917 long p_lines, t_lines;
918 unsigned long *p_ends = NULL, *t_ends = NULL;
919 struct diff_filepair *pair = range->pair;
920 struct diff_ranges *diff = &range->diff;
921
922 struct diff_options *opt = &rev->diffopt;
923 char *prefix = output_prefix(opt);
924 const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
925 const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
926 const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
927 const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
928 const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
929 const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
930
931 if (!pair || !diff)
932 return;
933
934 if (pair->one->oid_valid)
935 fill_line_ends(rev->diffopt.repo, pair->one, &p_lines, &p_ends);
936 fill_line_ends(rev->diffopt.repo, pair->two, &t_lines, &t_ends);
937
938 fprintf(opt->file, "%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
939 fprintf(opt->file, "%s%s--- %s%s%s\n", prefix, c_meta,
940 pair->one->oid_valid ? "a/" : "",
941 pair->one->oid_valid ? pair->one->path : "/dev/null",
942 c_reset);
943 fprintf(opt->file, "%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
944 for (i = 0; i < range->ranges.nr; i++) {
945 long p_start, p_end;
946 long t_start = range->ranges.ranges[i].start;
947 long t_end = range->ranges.ranges[i].end;
948 long t_cur = t_start;
949 unsigned int j_last;
950
951 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
952 j++;
953 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
954 continue;
955
956 /* Scan ahead to determine the last diff that falls in this range */
957 j_last = j;
958 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
959 j_last++;
960 if (j_last > j)
961 j_last--;
962
963 /*
964 * Compute parent hunk headers: we know that the diff
965 * has the correct line numbers (but not all hunks).
966 * So it suffices to shift the start/end according to
967 * the line numbers of the first/last hunk(s) that
968 * fall in this range.
969 */
970 if (t_start < diff->target.ranges[j].start)
971 p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
972 else
973 p_start = diff->parent.ranges[j].start;
974 if (t_end > diff->target.ranges[j_last].end)
975 p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
976 else
977 p_end = diff->parent.ranges[j_last].end;
978
979 if (!p_start && !p_end) {
980 p_start = -1;
981 p_end = -1;
982 }
983
984 /* Now output a diff hunk for this range */
985 fprintf(opt->file, "%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
986 prefix, c_frag,
987 p_start+1, p_end-p_start, t_start+1, t_end-t_start,
988 c_reset);
989 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
990 int k;
991 for (; t_cur < diff->target.ranges[j].start; t_cur++)
992 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
993 c_context, c_reset, opt->file);
994 for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
995 print_line(prefix, '-', k, p_ends, pair->one->data,
996 c_old, c_reset, opt->file);
997 for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
998 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
999 c_new, c_reset, opt->file);
1000 j++;
1001 }
1002 for (; t_cur < t_end; t_cur++)
1003 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
1004 c_context, c_reset, opt->file);
1005 }
1006
1007 free(p_ends);
1008 free(t_ends);
1009 }
1010
1011 /*
1012 * NEEDSWORK: manually building a diff here is not the Right
1013 * Thing(tm). log -L should be built into the diff pipeline.
1014 */
1015 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
1016 {
1017 fprintf(rev->diffopt.file, "%s\n", output_prefix(&rev->diffopt));
1018 while (range) {
1019 dump_diff_hacky_one(rev, range);
1020 range = range->next;
1021 }
1022 }
1023
1024 /*
1025 * Unlike most other functions, this destructively operates on
1026 * 'range'.
1027 */
1028 static int process_diff_filepair(struct rev_info *rev,
1029 struct diff_filepair *pair,
1030 struct line_log_data *range,
1031 struct diff_ranges **diff_out)
1032 {
1033 struct line_log_data *rg = range;
1034 struct range_set tmp;
1035 struct diff_ranges diff;
1036 mmfile_t file_parent, file_target;
1037
1038 assert(pair->two->path);
1039 while (rg) {
1040 assert(rg->path);
1041 if (!strcmp(rg->path, pair->two->path))
1042 break;
1043 rg = rg->next;
1044 }
1045
1046 if (!rg)
1047 return 0;
1048 if (rg->ranges.nr == 0)
1049 return 0;
1050
1051 assert(pair->two->oid_valid);
1052 diff_populate_filespec(rev->diffopt.repo, pair->two, NULL);
1053 file_target.ptr = pair->two->data;
1054 file_target.size = pair->two->size;
1055
1056 if (pair->one->oid_valid) {
1057 diff_populate_filespec(rev->diffopt.repo, pair->one, NULL);
1058 file_parent.ptr = pair->one->data;
1059 file_parent.size = pair->one->size;
1060 } else {
1061 file_parent.ptr = "";
1062 file_parent.size = 0;
1063 }
1064
1065 diff_ranges_init(&diff);
1066 if (collect_diff(&file_parent, &file_target, &diff))
1067 die("unable to generate diff for %s", pair->one->path);
1068
1069 /* NEEDSWORK should apply some heuristics to prevent mismatches */
1070 free(rg->path);
1071 rg->path = xstrdup(pair->one->path);
1072
1073 range_set_init(&tmp, 0);
1074 range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1075 range_set_release(&rg->ranges);
1076 range_set_move(&rg->ranges, &tmp);
1077
1078 diff_ranges_release(&diff);
1079
1080 return ((*diff_out)->parent.nr > 0);
1081 }
1082
1083 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1084 {
1085 struct diff_filepair *new_filepair = xmalloc(sizeof(struct diff_filepair));
1086 new_filepair->one = pair->one;
1087 new_filepair->two = pair->two;
1088 new_filepair->one->count++;
1089 new_filepair->two->count++;
1090 return new_filepair;
1091 }
1092
1093 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1094 {
1095 for (int i = 0; i < n; i++)
1096 diff_free_queue(&dq[i]);
1097 free(dq);
1098 }
1099
1100 static int process_all_files(struct line_log_data **range_out,
1101 struct rev_info *rev,
1102 struct diff_queue_struct *queue,
1103 struct line_log_data *range)
1104 {
1105 int i, changed = 0;
1106
1107 *range_out = line_log_data_copy(range);
1108
1109 for (i = 0; i < queue->nr; i++) {
1110 struct diff_ranges *pairdiff = NULL;
1111 struct diff_filepair *pair = queue->queue[i];
1112 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1113 /*
1114 * Store away the diff for later output. We
1115 * tuck it in the ranges we got as _input_,
1116 * since that's the commit that caused the
1117 * diff.
1118 *
1119 * NEEDSWORK not enough when we get around to
1120 * doing something interesting with merges;
1121 * currently each invocation on a merge parent
1122 * trashes the previous one's diff.
1123 *
1124 * NEEDSWORK tramples over data structures not owned here
1125 */
1126 struct line_log_data *rg = range;
1127 changed++;
1128 while (rg && strcmp(rg->path, pair->two->path))
1129 rg = rg->next;
1130 assert(rg);
1131 rg->pair = diff_filepair_dup(queue->queue[i]);
1132 memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1133 }
1134 free(pairdiff);
1135 }
1136
1137 return changed;
1138 }
1139
1140 int line_log_print(struct rev_info *rev, struct commit *commit)
1141 {
1142
1143 show_log(rev);
1144 if (!(rev->diffopt.output_format & DIFF_FORMAT_NO_OUTPUT)) {
1145 struct line_log_data *range = lookup_line_range(rev, commit);
1146 dump_diff_hacky(rev, range);
1147 }
1148 return 1;
1149 }
1150
1151 static int bloom_filter_check(struct rev_info *rev,
1152 struct commit *commit,
1153 struct line_log_data *range)
1154 {
1155 struct bloom_filter *filter;
1156 struct bloom_key key;
1157 int result = 0;
1158
1159 if (!commit->parents)
1160 return 1;
1161
1162 if (!rev->bloom_filter_settings ||
1163 !(filter = get_bloom_filter(rev->repo, commit)))
1164 return 1;
1165
1166 if (!range)
1167 return 0;
1168
1169 while (!result && range) {
1170 fill_bloom_key(range->path, strlen(range->path), &key, rev->bloom_filter_settings);
1171
1172 if (bloom_filter_contains(filter, &key, rev->bloom_filter_settings))
1173 result = 1;
1174
1175 clear_bloom_key(&key);
1176 range = range->next;
1177 }
1178
1179 return result;
1180 }
1181
1182 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1183 struct line_log_data *range)
1184 {
1185 struct commit *parent = NULL;
1186 struct diff_queue_struct queue;
1187 struct line_log_data *parent_range;
1188 int changed;
1189
1190 if (commit->parents)
1191 parent = commit->parents->item;
1192
1193 queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1194 changed = process_all_files(&parent_range, rev, &queue, range);
1195
1196 if (parent)
1197 add_line_range(rev, parent, parent_range);
1198 free_line_log_data(parent_range);
1199 diff_free_queue(&queue);
1200 return changed;
1201 }
1202
1203 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1204 struct line_log_data *range)
1205 {
1206 struct diff_queue_struct *diffqueues;
1207 struct line_log_data **cand;
1208 struct commit **parents;
1209 struct commit_list *p;
1210 int i;
1211 int nparents = commit_list_count(commit->parents);
1212
1213 if (nparents > 1 && rev->first_parent_only)
1214 nparents = 1;
1215
1216 ALLOC_ARRAY(diffqueues, nparents);
1217 ALLOC_ARRAY(cand, nparents);
1218 ALLOC_ARRAY(parents, nparents);
1219
1220 p = commit->parents;
1221 for (i = 0; i < nparents; i++) {
1222 parents[i] = p->item;
1223 p = p->next;
1224 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1225 }
1226
1227 for (i = 0; i < nparents; i++) {
1228 int changed;
1229 cand[i] = NULL;
1230 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1231 if (!changed) {
1232 /*
1233 * This parent can take all the blame, so we
1234 * don't follow any other path in history
1235 */
1236 add_line_range(rev, parents[i], cand[i]);
1237 clear_commit_line_range(rev, commit);
1238 commit_list_append(parents[i], &commit->parents);
1239 free(parents);
1240 free(cand);
1241 free_diffqueues(nparents, diffqueues);
1242 /* NEEDSWORK leaking like a sieve */
1243 return 0;
1244 }
1245 }
1246
1247 /*
1248 * No single parent took the blame. We add the candidates
1249 * from the above loop to the parents.
1250 */
1251 for (i = 0; i < nparents; i++) {
1252 add_line_range(rev, parents[i], cand[i]);
1253 }
1254
1255 clear_commit_line_range(rev, commit);
1256 free(parents);
1257 free(cand);
1258 free_diffqueues(nparents, diffqueues);
1259 return 1;
1260
1261 /* NEEDSWORK evil merge detection stuff */
1262 /* NEEDSWORK leaking like a sieve */
1263 }
1264
1265 int line_log_process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1266 {
1267 struct line_log_data *range = lookup_line_range(rev, commit);
1268 int changed = 0;
1269
1270 if (range) {
1271 if (commit->parents && !bloom_filter_check(rev, commit, range)) {
1272 struct line_log_data *prange = line_log_data_copy(range);
1273 add_line_range(rev, commit->parents->item, prange);
1274 clear_commit_line_range(rev, commit);
1275 } else if (!commit->parents || !commit->parents->next)
1276 changed = process_ranges_ordinary_commit(rev, commit, range);
1277 else
1278 changed = process_ranges_merge_commit(rev, commit, range);
1279 }
1280
1281 if (!changed)
1282 commit->object.flags |= TREESAME;
1283
1284 return changed;
1285 }
1286
1287 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev UNUSED,
1288 struct commit **pp)
1289 {
1290 for (;;) {
1291 struct commit *p = *pp;
1292 if (p->parents && p->parents->next)
1293 return rewrite_one_ok;
1294 if (p->object.flags & UNINTERESTING)
1295 return rewrite_one_ok;
1296 if (!(p->object.flags & TREESAME))
1297 return rewrite_one_ok;
1298 if (!p->parents)
1299 return rewrite_one_noparents;
1300 *pp = p->parents->item;
1301 }
1302 }
1303
1304 int line_log_filter(struct rev_info *rev)
1305 {
1306 struct commit *commit;
1307 struct commit_list *list = rev->commits;
1308 struct commit_list *out = NULL, **pp = &out;
1309
1310 while (list) {
1311 struct commit_list *to_free = NULL;
1312 commit = list->item;
1313 if (line_log_process_ranges_arbitrary_commit(rev, commit)) {
1314 *pp = list;
1315 pp = &list->next;
1316 } else
1317 to_free = list;
1318 list = list->next;
1319 free(to_free);
1320 }
1321 *pp = NULL;
1322
1323 for (list = out; list; list = list->next)
1324 rewrite_parents(rev, list->item, line_log_rewrite_one);
1325
1326 rev->commits = out;
1327
1328 return 0;
1329 }
1330
1331 static void free_void_line_log_data(void *data)
1332 {
1333 free_line_log_data(data);
1334 }
1335
1336 void line_log_free(struct rev_info *rev)
1337 {
1338 clear_decoration(&rev->line_log_data, free_void_line_log_data);
1339 }
git for Debian