?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
The eleventh batch
[git/gitster.git] / line-log.c
blobbca9bd804073df9b77b2859063ac5ad4d3b24e0f
1 #include "git-compat-util.h"
2 #include "diffcore.h"
3 #include "line-range.h"
4 #include "hex.h"
5 #include "tag.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 "line-log.h"
16 #include "setup.h"
17 #include "strvec.h"
18 #include "bloom.h"
19 #include "tree-walk.h"
20
21 static void range_set_grow(struct range_set *rs, size_t extra)
22 {
23 ALLOC_GROW(rs->ranges, rs->nr + extra, rs->alloc);
24 }
25
26 /* Either initialization would be fine */
27 #define RANGE_SET_INIT {0}
28
29 void range_set_init(struct range_set *rs, size_t prealloc)
30 {
31 rs->alloc = rs->nr = 0;
32 rs->ranges = NULL;
33 if (prealloc)
34 range_set_grow(rs, prealloc);
35 }
36
37 void range_set_release(struct range_set *rs)
38 {
39 FREE_AND_NULL(rs->ranges);
40 rs->alloc = rs->nr = 0;
41 }
42
43 /* dst must be uninitialized! */
44 static void range_set_copy(struct range_set *dst, struct range_set *src)
45 {
46 range_set_init(dst, src->nr);
47 COPY_ARRAY(dst->ranges, src->ranges, src->nr);
48 dst->nr = src->nr;
49 }
50
51 static void range_set_move(struct range_set *dst, struct range_set *src)
52 {
53 range_set_release(dst);
54 dst->ranges = src->ranges;
55 dst->nr = src->nr;
56 dst->alloc = src->alloc;
57 src->ranges = NULL;
58 src->alloc = src->nr = 0;
59 }
60
61 /* tack on a _new_ range _at the end_ */
62 void range_set_append_unsafe(struct range_set *rs, long a, long b)
63 {
64 assert(a <= b);
65 range_set_grow(rs, 1);
66 rs->ranges[rs->nr].start = a;
67 rs->ranges[rs->nr].end = b;
68 rs->nr++;
69 }
70
71 void range_set_append(struct range_set *rs, long a, long b)
72 {
73 assert(rs->nr == 0 || rs->ranges[rs->nr-1].end <= a);
74 range_set_append_unsafe(rs, a, b);
75 }
76
77 static int range_cmp(const void *_r, const void *_s)
78 {
79 const struct range *r = _r;
80 const struct range *s = _s;
81
82 /* this could be simply 'return r.start-s.start', but for the types */
83 if (r->start == s->start)
84 return 0;
85 if (r->start < s->start)
86 return -1;
87 return 1;
88 }
89
90 /*
91 * Check that the ranges are non-empty, sorted and non-overlapping
92 */
93 static void range_set_check_invariants(struct range_set *rs)
94 {
95 unsigned int i;
96
97 if (!rs)
98 return;
99
100 if (rs->nr)
101 assert(rs->ranges[0].start < rs->ranges[0].end);
102
103 for (i = 1; i < rs->nr; i++) {
104 assert(rs->ranges[i-1].end < rs->ranges[i].start);
105 assert(rs->ranges[i].start < rs->ranges[i].end);
106 }
107 }
108
109 /*
110 * In-place pass of sorting and merging the ranges in the range set,
111 * to establish the invariants when we get the ranges from the user
112 */
113 void sort_and_merge_range_set(struct range_set *rs)
114 {
115 unsigned int i;
116 unsigned int o = 0; /* output cursor */
117
118 QSORT(rs->ranges, rs->nr, range_cmp);
119
120 for (i = 0; i < rs->nr; i++) {
121 if (rs->ranges[i].start == rs->ranges[i].end)
122 continue;
123 if (o > 0 && rs->ranges[i].start <= rs->ranges[o-1].end) {
124 if (rs->ranges[o-1].end < rs->ranges[i].end)
125 rs->ranges[o-1].end = rs->ranges[i].end;
126 } else {
127 rs->ranges[o].start = rs->ranges[i].start;
128 rs->ranges[o].end = rs->ranges[i].end;
129 o++;
130 }
131 }
132 assert(o <= rs->nr);
133 rs->nr = o;
134
135 range_set_check_invariants(rs);
136 }
137
138 /*
139 * Union of range sets (i.e., sets of line numbers). Used to merge
140 * them when searches meet at a common ancestor.
141 *
142 * This is also where the ranges are consolidated into canonical form:
143 * overlapping and adjacent ranges are merged, and empty ranges are
144 * removed.
145 */
146 static void range_set_union(struct range_set *out,
147 struct range_set *a, struct range_set *b)
148 {
149 unsigned int i = 0, j = 0;
150 struct range *ra = a->ranges;
151 struct range *rb = b->ranges;
152 /* cannot make an alias of out->ranges: it may change during grow */
153
154 assert(out->nr == 0);
155 while (i < a->nr || j < b->nr) {
156 struct range *new_range;
157 if (i < a->nr && j < b->nr) {
158 if (ra[i].start < rb[j].start)
159 new_range = &ra[i++];
160 else if (ra[i].start > rb[j].start)
161 new_range = &rb[j++];
162 else if (ra[i].end < rb[j].end)
163 new_range = &ra[i++];
164 else
165 new_range = &rb[j++];
166 } else if (i < a->nr) /* b exhausted */
167 new_range = &ra[i++];
168 else /* a exhausted */
169 new_range = &rb[j++];
170 if (new_range->start == new_range->end)
171 ; /* empty range */
172 else if (!out->nr || out->ranges[out->nr-1].end < new_range->start) {
173 range_set_grow(out, 1);
174 out->ranges[out->nr].start = new_range->start;
175 out->ranges[out->nr].end = new_range->end;
176 out->nr++;
177 } else if (out->ranges[out->nr-1].end < new_range->end) {
178 out->ranges[out->nr-1].end = new_range->end;
179 }
180 }
181 }
182
183 /*
184 * Difference of range sets (out = a \ b). Pass the "interesting"
185 * ranges as 'a' and the target side of the diff as 'b': it removes
186 * the ranges for which the commit is responsible.
187 */
188 static void range_set_difference(struct range_set *out,
189 struct range_set *a, struct range_set *b)
190 {
191 unsigned int i, j = 0;
192 for (i = 0; i < a->nr; i++) {
193 long start = a->ranges[i].start;
194 long end = a->ranges[i].end;
195 while (start < end) {
196 while (j < b->nr && start >= b->ranges[j].end)
197 /*
198 * a: |-------
199 * b: ------|
200 */
201 j++;
202 if (j >= b->nr || end < b->ranges[j].start) {
203 /*
204 * b exhausted, or
205 * a: ----|
206 * b: |----
207 */
208 range_set_append(out, start, end);
209 break;
210 }
211 if (start >= b->ranges[j].start) {
212 /*
213 * a: |--????
214 * b: |------|
215 */
216 start = b->ranges[j].end;
217 } else if (end > b->ranges[j].start) {
218 /*
219 * a: |-----|
220 * b: |--?????
221 */
222 if (start < b->ranges[j].start)
223 range_set_append(out, start, b->ranges[j].start);
224 start = b->ranges[j].end;
225 }
226 }
227 }
228 }
229
230 static void diff_ranges_init(struct diff_ranges *diff)
231 {
232 range_set_init(&diff->parent, 0);
233 range_set_init(&diff->target, 0);
234 }
235
236 static void diff_ranges_release(struct diff_ranges *diff)
237 {
238 range_set_release(&diff->parent);
239 range_set_release(&diff->target);
240 }
241
242 static void line_log_data_init(struct line_log_data *r)
243 {
244 memset(r, 0, sizeof(struct line_log_data));
245 range_set_init(&r->ranges, 0);
246 }
247
248 static void line_log_data_clear(struct line_log_data *r)
249 {
250 range_set_release(&r->ranges);
251 free(r->path);
252 if (r->pair)
253 diff_free_filepair(r->pair);
254 diff_ranges_release(&r->diff);
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;
577 char *range_part;
578 char *full_name;
579 struct diff_filespec *spec;
580 long begin = 0, end = 0;
581 long anchor;
582
583 name_part = skip_range_arg(item->string, r->index);
584 if (!name_part || *name_part != ':' || !name_part[1])
585 die("-L argument not 'start,end:file' or ':funcname:file': %s",
586 item->string);
587 range_part = xstrndup(item->string, name_part - item->string);
588 name_part++;
589
590 full_name = prefix_path(prefix, prefix ? strlen(prefix) : 0,
591 name_part);
592
593 spec = alloc_filespec(full_name);
594 fill_blob_sha1(r, commit, spec);
595 fill_line_ends(r, spec, &lines, &ends);
596 cb_data.spec = spec;
597 cb_data.lines = lines;
598 cb_data.line_ends = ends;
599
600 p = search_line_log_data(ranges, full_name, NULL);
601 if (p && p->ranges.nr)
602 anchor = p->ranges.ranges[p->ranges.nr - 1].end + 1;
603 else
604 anchor = 1;
605
606 if (parse_range_arg(range_part, nth_line, &cb_data,
607 lines, anchor, &begin, &end,
608 full_name, r->index))
609 die("malformed -L argument '%s'", range_part);
610 if ((!lines && (begin || end)) || lines < begin)
611 die("file %s has only %lu lines", name_part, lines);
612 if (begin < 1)
613 begin = 1;
614 if (end < 1 || lines < end)
615 end = lines;
616 begin--;
617 line_log_data_insert(&ranges, full_name, begin, end);
618
619 free_filespec(spec);
620 FREE_AND_NULL(ends);
621 free(range_part);
622 }
623
624 for (p = ranges; p; p = p->next)
625 sort_and_merge_range_set(&p->ranges);
626
627 return ranges;
628 }
629
630 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
631 {
632 struct line_log_data *ret = xmalloc(sizeof(*ret));
633
634 assert(r);
635 line_log_data_init(ret);
636 range_set_copy(&ret->ranges, &r->ranges);
637
638 ret->path = xstrdup(r->path);
639
640 return ret;
641 }
642
643 static struct line_log_data *
644 line_log_data_copy(struct line_log_data *r)
645 {
646 struct line_log_data *ret = NULL;
647 struct line_log_data *tmp = NULL, *prev = NULL;
648
649 assert(r);
650 ret = tmp = prev = line_log_data_copy_one(r);
651 r = r->next;
652 while (r) {
653 tmp = line_log_data_copy_one(r);
654 prev->next = tmp;
655 prev = tmp;
656 r = r->next;
657 }
658
659 return ret;
660 }
661
662 /* merge two range sets across files */
663 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
664 struct line_log_data *b)
665 {
666 struct line_log_data *head = NULL, **pp = &head;
667
668 while (a || b) {
669 struct line_log_data *src;
670 struct line_log_data *src2 = NULL;
671 struct line_log_data *d;
672 int cmp;
673 if (!a)
674 cmp = 1;
675 else if (!b)
676 cmp = -1;
677 else
678 cmp = strcmp(a->path, b->path);
679 if (cmp < 0) {
680 src = a;
681 a = a->next;
682 } else if (cmp == 0) {
683 src = a;
684 a = a->next;
685 src2 = b;
686 b = b->next;
687 } else {
688 src = b;
689 b = b->next;
690 }
691 d = xmalloc(sizeof(struct line_log_data));
692 line_log_data_init(d);
693 d->path = xstrdup(src->path);
694 *pp = d;
695 pp = &d->next;
696 if (src2)
697 range_set_union(&d->ranges, &src->ranges, &src2->ranges);
698 else
699 range_set_copy(&d->ranges, &src->ranges);
700 }
701
702 return head;
703 }
704
705 static void add_line_range(struct rev_info *revs, struct commit *commit,
706 struct line_log_data *range)
707 {
708 struct line_log_data *old_line = NULL;
709 struct line_log_data *new_line = NULL;
710
711 old_line = lookup_decoration(&revs->line_log_data, &commit->object);
712 if (old_line && range) {
713 new_line = line_log_data_merge(old_line, range);
714 free_line_log_data(old_line);
715 } else if (range)
716 new_line = line_log_data_copy(range);
717
718 if (new_line)
719 add_decoration(&revs->line_log_data, &commit->object, new_line);
720 }
721
722 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
723 {
724 struct line_log_data *r;
725 r = lookup_decoration(&revs->line_log_data, &commit->object);
726 if (!r)
727 return;
728 free_line_log_data(r);
729 add_decoration(&revs->line_log_data, &commit->object, NULL);
730 }
731
732 static struct line_log_data *lookup_line_range(struct rev_info *revs,
733 struct commit *commit)
734 {
735 struct line_log_data *ret = NULL;
736 struct line_log_data *d;
737
738 ret = lookup_decoration(&revs->line_log_data, &commit->object);
739
740 for (d = ret; d; d = d->next)
741 range_set_check_invariants(&d->ranges);
742
743 return ret;
744 }
745
746 static int same_paths_in_pathspec_and_range(struct pathspec *pathspec,
747 struct line_log_data *range)
748 {
749 int i;
750 struct line_log_data *r;
751
752 for (i = 0, r = range; i < pathspec->nr && r; i++, r = r->next)
753 if (strcmp(pathspec->items[i].match, r->path))
754 return 0;
755 if (i < pathspec->nr || r)
756 /* different number of pathspec items and ranges */
757 return 0;
758
759 return 1;
760 }
761
762 static void parse_pathspec_from_ranges(struct pathspec *pathspec,
763 struct line_log_data *range)
764 {
765 struct line_log_data *r;
766 struct strvec array = STRVEC_INIT;
767
768 for (r = range; r; r = r->next)
769 strvec_push(&array, r->path);
770
771 parse_pathspec(pathspec, 0, PATHSPEC_PREFER_FULL, "", array.v);
772
773 strvec_clear(&array);
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 free_line_log_data(range);
788 }
789
790 static void move_diff_queue(struct diff_queue_struct *dst,
791 struct diff_queue_struct *src)
792 {
793 assert(src != dst);
794 memcpy(dst, src, sizeof(*dst));
795 diff_queue_init(src);
796 }
797
798 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
799 {
800 int i;
801 struct diff_queue_struct outq = DIFF_QUEUE_INIT;
802
803 for (i = 0; i < diff_queued_diff.nr; i++) {
804 struct diff_filepair *p = diff_queued_diff.queue[i];
805 struct line_log_data *rg = NULL;
806
807 if (!DIFF_FILE_VALID(p->two)) {
808 if (keep_deletions)
809 diff_q(&outq, p);
810 else
811 diff_free_filepair(p);
812 continue;
813 }
814 for (rg = range; rg; rg = rg->next) {
815 if (!strcmp(rg->path, p->two->path))
816 break;
817 }
818 if (rg)
819 diff_q(&outq, p);
820 else
821 diff_free_filepair(p);
822 }
823 free(diff_queued_diff.queue);
824 diff_queued_diff = outq;
825 }
826
827 static inline int diff_might_be_rename(void)
828 {
829 int i;
830 for (i = 0; i < diff_queued_diff.nr; i++)
831 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
832 /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
833 /* diff_queued_diff.queue[i]->two->path); */
834 return 1;
835 }
836 return 0;
837 }
838
839 static void queue_diffs(struct line_log_data *range,
840 struct diff_options *opt,
841 struct diff_queue_struct *queue,
842 struct commit *commit, struct commit *parent)
843 {
844 struct object_id *tree_oid, *parent_tree_oid;
845
846 assert(commit);
847
848 tree_oid = get_commit_tree_oid(commit);
849 parent_tree_oid = parent ? get_commit_tree_oid(parent) : NULL;
850
851 if (opt->detect_rename &&
852 !same_paths_in_pathspec_and_range(&opt->pathspec, range)) {
853 clear_pathspec(&opt->pathspec);
854 parse_pathspec_from_ranges(&opt->pathspec, range);
855 }
856 diff_queue_clear(&diff_queued_diff);
857 diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
858 if (opt->detect_rename && diff_might_be_rename()) {
859 /* must look at the full tree diff to detect renames */
860 clear_pathspec(&opt->pathspec);
861 diff_queue_clear(&diff_queued_diff);
862
863 diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
864
865 filter_diffs_for_paths(range, 1);
866 diffcore_std(opt);
867 filter_diffs_for_paths(range, 0);
868 }
869 move_diff_queue(queue, &diff_queued_diff);
870 }
871
872 static char *get_nth_line(long line, unsigned long *ends, void *data)
873 {
874 if (line == 0)
875 return (char *)data;
876 else
877 return (char *)data + ends[line] + 1;
878 }
879
880 static void print_line(const char *prefix, char first,
881 long line, unsigned long *ends, void *data,
882 const char *color, const char *reset, FILE *file)
883 {
884 char *begin = get_nth_line(line, ends, data);
885 char *end = get_nth_line(line+1, ends, data);
886 int had_nl = 0;
887
888 if (end > begin && end[-1] == '\n') {
889 end--;
890 had_nl = 1;
891 }
892
893 fputs(prefix, file);
894 fputs(color, file);
895 putc(first, file);
896 fwrite(begin, 1, end-begin, file);
897 fputs(reset, file);
898 putc('\n', file);
899 if (!had_nl)
900 fputs("\\ No newline at end of file\n", file);
901 }
902
903 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
904 {
905 unsigned int i, j = 0;
906 long p_lines, t_lines;
907 unsigned long *p_ends = NULL, *t_ends = NULL;
908 struct diff_filepair *pair = range->pair;
909 struct diff_ranges *diff = &range->diff;
910
911 struct diff_options *opt = &rev->diffopt;
912 const char *prefix = diff_line_prefix(opt);
913 const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
914 const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
915 const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
916 const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
917 const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
918 const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
919
920 if (!pair || !diff)
921 goto out;
922
923 if (pair->one->oid_valid)
924 fill_line_ends(rev->diffopt.repo, pair->one, &p_lines, &p_ends);
925 fill_line_ends(rev->diffopt.repo, pair->two, &t_lines, &t_ends);
926
927 fprintf(opt->file, "%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
928 fprintf(opt->file, "%s%s--- %s%s%s\n", prefix, c_meta,
929 pair->one->oid_valid ? "a/" : "",
930 pair->one->oid_valid ? pair->one->path : "/dev/null",
931 c_reset);
932 fprintf(opt->file, "%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
933 for (i = 0; i < range->ranges.nr; i++) {
934 long p_start, p_end;
935 long t_start = range->ranges.ranges[i].start;
936 long t_end = range->ranges.ranges[i].end;
937 long t_cur = t_start;
938 unsigned int j_last;
939
940 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
941 j++;
942 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
943 continue;
944
945 /* Scan ahead to determine the last diff that falls in this range */
946 j_last = j;
947 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
948 j_last++;
949 if (j_last > j)
950 j_last--;
951
952 /*
953 * Compute parent hunk headers: we know that the diff
954 * has the correct line numbers (but not all hunks).
955 * So it suffices to shift the start/end according to
956 * the line numbers of the first/last hunk(s) that
957 * fall in this range.
958 */
959 if (t_start < diff->target.ranges[j].start)
960 p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
961 else
962 p_start = diff->parent.ranges[j].start;
963 if (t_end > diff->target.ranges[j_last].end)
964 p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
965 else
966 p_end = diff->parent.ranges[j_last].end;
967
968 if (!p_start && !p_end) {
969 p_start = -1;
970 p_end = -1;
971 }
972
973 /* Now output a diff hunk for this range */
974 fprintf(opt->file, "%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
975 prefix, c_frag,
976 p_start+1, p_end-p_start, t_start+1, t_end-t_start,
977 c_reset);
978 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
979 int k;
980 for (; t_cur < diff->target.ranges[j].start; t_cur++)
981 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
982 c_context, c_reset, opt->file);
983 for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
984 print_line(prefix, '-', k, p_ends, pair->one->data,
985 c_old, c_reset, opt->file);
986 for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
987 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
988 c_new, c_reset, opt->file);
989 j++;
990 }
991 for (; t_cur < t_end; t_cur++)
992 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
993 c_context, c_reset, opt->file);
994 }
995
996 out:
997 free(p_ends);
998 free(t_ends);
999 }
1000
1001 /*
1002 * NEEDSWORK: manually building a diff here is not the Right
1003 * Thing(tm). log -L should be built into the diff pipeline.
1004 */
1005 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
1006 {
1007 const char *prefix = diff_line_prefix(&rev->diffopt);
1008
1009 fprintf(rev->diffopt.file, "%s\n", prefix);
1010
1011 while (range) {
1012 dump_diff_hacky_one(rev, range);
1013 range = range->next;
1014 }
1015 }
1016
1017 /*
1018 * Unlike most other functions, this destructively operates on
1019 * 'range'.
1020 */
1021 static int process_diff_filepair(struct rev_info *rev,
1022 struct diff_filepair *pair,
1023 struct line_log_data *range,
1024 struct diff_ranges **diff_out)
1025 {
1026 struct line_log_data *rg = range;
1027 struct range_set tmp;
1028 struct diff_ranges diff;
1029 mmfile_t file_parent, file_target;
1030 char *parent_data_to_free = NULL;
1031
1032 assert(pair->two->path);
1033 while (rg) {
1034 assert(rg->path);
1035 if (!strcmp(rg->path, pair->two->path))
1036 break;
1037 rg = rg->next;
1038 }
1039
1040 if (!rg)
1041 return 0;
1042 if (rg->ranges.nr == 0)
1043 return 0;
1044
1045 assert(pair->two->oid_valid);
1046 diff_populate_filespec(rev->diffopt.repo, pair->two, NULL);
1047 file_target.ptr = pair->two->data;
1048 file_target.size = pair->two->size;
1049
1050 if (pair->one->oid_valid) {
1051 diff_populate_filespec(rev->diffopt.repo, pair->one, NULL);
1052 file_parent.ptr = pair->one->data;
1053 file_parent.size = pair->one->size;
1054 } else {
1055 file_parent.ptr = parent_data_to_free = xstrdup("");
1056 file_parent.size = 0;
1057 }
1058
1059 diff_ranges_init(&diff);
1060 if (collect_diff(&file_parent, &file_target, &diff))
1061 die("unable to generate diff for %s", pair->one->path);
1062
1063 /* NEEDSWORK should apply some heuristics to prevent mismatches */
1064 free(rg->path);
1065 rg->path = xstrdup(pair->one->path);
1066
1067 range_set_init(&tmp, 0);
1068 range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1069 range_set_release(&rg->ranges);
1070 range_set_move(&rg->ranges, &tmp);
1071
1072 diff_ranges_release(&diff);
1073
1074 free(parent_data_to_free);
1075 return ((*diff_out)->parent.nr > 0);
1076 }
1077
1078 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1079 {
1080 struct diff_filepair *new_filepair = xmalloc(sizeof(struct diff_filepair));
1081 new_filepair->one = pair->one;
1082 new_filepair->two = pair->two;
1083 new_filepair->one->count++;
1084 new_filepair->two->count++;
1085 return new_filepair;
1086 }
1087
1088 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1089 {
1090 for (int i = 0; i < n; i++)
1091 diff_queue_clear(&dq[i]);
1092 free(dq);
1093 }
1094
1095 static int process_all_files(struct line_log_data **range_out,
1096 struct rev_info *rev,
1097 struct diff_queue_struct *queue,
1098 struct line_log_data *range)
1099 {
1100 int i, changed = 0;
1101
1102 *range_out = line_log_data_copy(range);
1103
1104 for (i = 0; i < queue->nr; i++) {
1105 struct diff_ranges *pairdiff = NULL;
1106 struct diff_filepair *pair = queue->queue[i];
1107 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1108 /*
1109 * Store away the diff for later output. We
1110 * tuck it in the ranges we got as _input_,
1111 * since that's the commit that caused the
1112 * diff.
1113 *
1114 * NEEDSWORK not enough when we get around to
1115 * doing something interesting with merges;
1116 * currently each invocation on a merge parent
1117 * trashes the previous one's diff.
1118 *
1119 * NEEDSWORK tramples over data structures not owned here
1120 */
1121 struct line_log_data *rg = range;
1122 changed++;
1123 while (rg && strcmp(rg->path, pair->two->path))
1124 rg = rg->next;
1125 assert(rg);
1126 if (rg->pair)
1127 diff_free_filepair(rg->pair);
1128 rg->pair = diff_filepair_dup(queue->queue[i]);
1129 diff_ranges_release(&rg->diff);
1130 memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1131 FREE_AND_NULL(pairdiff);
1132 }
1133
1134 if (pairdiff) {
1135 diff_ranges_release(pairdiff);
1136 free(pairdiff);
1137 }
1138 }
1139
1140 return changed;
1141 }
1142
1143 int line_log_print(struct rev_info *rev, struct commit *commit)
1144 {
1145
1146 show_log(rev);
1147 if (!(rev->diffopt.output_format & DIFF_FORMAT_NO_OUTPUT)) {
1148 struct line_log_data *range = lookup_line_range(rev, commit);
1149 dump_diff_hacky(rev, range);
1150 }
1151 return 1;
1152 }
1153
1154 static int bloom_filter_check(struct rev_info *rev,
1155 struct commit *commit,
1156 struct line_log_data *range)
1157 {
1158 struct bloom_filter *filter;
1159 struct bloom_key key;
1160 int result = 0;
1161
1162 if (!commit->parents)
1163 return 1;
1164
1165 if (!rev->bloom_filter_settings ||
1166 !(filter = get_bloom_filter(rev->repo, commit)))
1167 return 1;
1168
1169 if (!range)
1170 return 0;
1171
1172 while (!result && range) {
1173 fill_bloom_key(range->path, strlen(range->path), &key, rev->bloom_filter_settings);
1174
1175 if (bloom_filter_contains(filter, &key, rev->bloom_filter_settings))
1176 result = 1;
1177
1178 clear_bloom_key(&key);
1179 range = range->next;
1180 }
1181
1182 return result;
1183 }
1184
1185 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1186 struct line_log_data *range)
1187 {
1188 struct commit *parent = NULL;
1189 struct diff_queue_struct queue;
1190 struct line_log_data *parent_range;
1191 int changed;
1192
1193 if (commit->parents)
1194 parent = commit->parents->item;
1195
1196 queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1197 changed = process_all_files(&parent_range, rev, &queue, range);
1198
1199 if (parent)
1200 add_line_range(rev, parent, parent_range);
1201 free_line_log_data(parent_range);
1202 diff_queue_clear(&queue);
1203 return changed;
1204 }
1205
1206 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1207 struct line_log_data *range)
1208 {
1209 struct diff_queue_struct *diffqueues;
1210 struct line_log_data **cand;
1211 struct commit **parents;
1212 struct commit_list *p;
1213 int i;
1214 int nparents = commit_list_count(commit->parents);
1215 int ret;
1216
1217 if (nparents > 1 && rev->first_parent_only)
1218 nparents = 1;
1219
1220 ALLOC_ARRAY(diffqueues, nparents);
1221 CALLOC_ARRAY(cand, nparents);
1222 ALLOC_ARRAY(parents, nparents);
1223
1224 p = commit->parents;
1225 for (i = 0; i < nparents; i++) {
1226 parents[i] = p->item;
1227 p = p->next;
1228 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1229 }
1230
1231 for (i = 0; i < nparents; i++) {
1232 int changed;
1233 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1234 if (!changed) {
1235 /*
1236 * This parent can take all the blame, so we
1237 * don't follow any other path in history
1238 */
1239 add_line_range(rev, parents[i], cand[i]);
1240 commit_list_append(parents[i], &commit->parents);
1241
1242 ret = 0;
1243 goto out;
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 ret = 1;
1255
1256 out:
1257 clear_commit_line_range(rev, commit);
1258 free(parents);
1259 for (i = 0; i < nparents; i++) {
1260 if (!cand[i])
1261 continue;
1262 line_log_data_clear(cand[i]);
1263 free(cand[i]);
1264 }
1265 free(cand);
1266 free_diffqueues(nparents, diffqueues);
1267 return ret;
1268
1269 /* NEEDSWORK evil merge detection stuff */
1270 }
1271
1272 int line_log_process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1273 {
1274 struct line_log_data *range = lookup_line_range(rev, commit);
1275 int changed = 0;
1276
1277 if (range) {
1278 if (commit->parents && !bloom_filter_check(rev, commit, range)) {
1279 struct line_log_data *prange = line_log_data_copy(range);
1280 add_line_range(rev, commit->parents->item, prange);
1281 clear_commit_line_range(rev, commit);
1282 } else if (!commit->parents || !commit->parents->next)
1283 changed = process_ranges_ordinary_commit(rev, commit, range);
1284 else
1285 changed = process_ranges_merge_commit(rev, commit, range);
1286 }
1287
1288 if (!changed)
1289 commit->object.flags |= TREESAME;
1290
1291 return changed;
1292 }
1293
1294 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev UNUSED,
1295 struct commit **pp)
1296 {
1297 for (;;) {
1298 struct commit *p = *pp;
1299 if (p->parents && p->parents->next)
1300 return rewrite_one_ok;
1301 if (p->object.flags & UNINTERESTING)
1302 return rewrite_one_ok;
1303 if (!(p->object.flags & TREESAME))
1304 return rewrite_one_ok;
1305 if (!p->parents)
1306 return rewrite_one_noparents;
1307 *pp = p->parents->item;
1308 }
1309 }
1310
1311 int line_log_filter(struct rev_info *rev)
1312 {
1313 struct commit *commit;
1314 struct commit_list *list = rev->commits;
1315 struct commit_list *out = NULL, **pp = &out;
1316
1317 while (list) {
1318 struct commit_list *to_free = NULL;
1319 commit = list->item;
1320 if (line_log_process_ranges_arbitrary_commit(rev, commit)) {
1321 *pp = list;
1322 pp = &list->next;
1323 } else
1324 to_free = list;
1325 list = list->next;
1326 free(to_free);
1327 }
1328 *pp = NULL;
1329
1330 for (list = out; list; list = list->next)
1331 rewrite_parents(rev, list->item, line_log_rewrite_one);
1332
1333 rev->commits = out;
1334
1335 return 0;
1336 }
1337
1338 static void free_void_line_log_data(void *data)
1339 {
1340 free_line_log_data(data);
1341 }
1342
1343 void line_log_free(struct rev_info *rev)
1344 {
1345 clear_decoration(&rev->line_log_data, free_void_line_log_data);
1346 }
Mirror of Junio's Git repository with all the branches