?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
The seventh batch
[git.git] / line-log.c
blob8ff6ccb77241fd249f58827ecc1add65356648a4
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 if (r->pair)
252 diff_free_filepair(r->pair);
253 }
254
255 static void free_line_log_data(struct line_log_data *r)
256 {
257 while (r) {
258 struct line_log_data *next = r->next;
259 line_log_data_clear(r);
260 free(r);
261 r = next;
262 }
263 }
264
265 static struct line_log_data *
266 search_line_log_data(struct line_log_data *list, const char *path,
267 struct line_log_data **insertion_point)
268 {
269 struct line_log_data *p = list;
270 if (insertion_point)
271 *insertion_point = NULL;
272 while (p) {
273 int cmp = strcmp(p->path, path);
274 if (!cmp)
275 return p;
276 if (insertion_point && cmp < 0)
277 *insertion_point = p;
278 p = p->next;
279 }
280 return NULL;
281 }
282
283 /*
284 * Note: takes ownership of 'path', which happens to be what the only
285 * caller needs.
286 */
287 static void line_log_data_insert(struct line_log_data **list,
288 char *path,
289 long begin, long end)
290 {
291 struct line_log_data *ip;
292 struct line_log_data *p = search_line_log_data(*list, path, &ip);
293
294 if (p) {
295 range_set_append_unsafe(&p->ranges, begin, end);
296 free(path);
297 return;
298 }
299
300 CALLOC_ARRAY(p, 1);
301 p->path = path;
302 range_set_append(&p->ranges, begin, end);
303 if (ip) {
304 p->next = ip->next;
305 ip->next = p;
306 } else {
307 p->next = *list;
308 *list = p;
309 }
310 }
311
312 struct collect_diff_cbdata {
313 struct diff_ranges *diff;
314 };
315
316 static int collect_diff_cb(long start_a, long count_a,
317 long start_b, long count_b,
318 void *data)
319 {
320 struct collect_diff_cbdata *d = data;
321
322 if (count_a >= 0)
323 range_set_append(&d->diff->parent, start_a, start_a + count_a);
324 if (count_b >= 0)
325 range_set_append(&d->diff->target, start_b, start_b + count_b);
326
327 return 0;
328 }
329
330 static int collect_diff(mmfile_t *parent, mmfile_t *target, struct diff_ranges *out)
331 {
332 struct collect_diff_cbdata cbdata = {NULL};
333 xpparam_t xpp;
334 xdemitconf_t xecfg;
335 xdemitcb_t ecb;
336
337 memset(&xpp, 0, sizeof(xpp));
338 memset(&xecfg, 0, sizeof(xecfg));
339 xecfg.ctxlen = xecfg.interhunkctxlen = 0;
340
341 cbdata.diff = out;
342 xecfg.hunk_func = collect_diff_cb;
343 memset(&ecb, 0, sizeof(ecb));
344 ecb.priv = &cbdata;
345 return xdi_diff(parent, target, &xpp, &xecfg, &ecb);
346 }
347
348 /*
349 * These are handy for debugging. Removing them with #if 0 silences
350 * the "unused function" warning.
351 */
352 #if 0
353 static void dump_range_set(struct range_set *rs, const char *desc)
354 {
355 int i;
356 printf("range set %s (%d items):\n", desc, rs->nr);
357 for (i = 0; i < rs->nr; i++)
358 printf("\t[%ld,%ld]\n", rs->ranges[i].start, rs->ranges[i].end);
359 }
360
361 static void dump_line_log_data(struct line_log_data *r)
362 {
363 char buf[4096];
364 while (r) {
365 snprintf(buf, 4096, "file %s\n", r->path);
366 dump_range_set(&r->ranges, buf);
367 r = r->next;
368 }
369 }
370
371 static void dump_diff_ranges(struct diff_ranges *diff, const char *desc)
372 {
373 int i;
374 assert(diff->parent.nr == diff->target.nr);
375 printf("diff ranges %s (%d items):\n", desc, diff->parent.nr);
376 printf("\tparent\ttarget\n");
377 for (i = 0; i < diff->parent.nr; i++) {
378 printf("\t[%ld,%ld]\t[%ld,%ld]\n",
379 diff->parent.ranges[i].start,
380 diff->parent.ranges[i].end,
381 diff->target.ranges[i].start,
382 diff->target.ranges[i].end);
383 }
384 }
385 #endif
386
387
388 static int ranges_overlap(struct range *a, struct range *b)
389 {
390 return !(a->end <= b->start || b->end <= a->start);
391 }
392
393 /*
394 * Given a diff and the set of interesting ranges, determine all hunks
395 * of the diff which touch (overlap) at least one of the interesting
396 * ranges in the target.
397 */
398 static void diff_ranges_filter_touched(struct diff_ranges *out,
399 struct diff_ranges *diff,
400 struct range_set *rs)
401 {
402 unsigned int i, j = 0;
403
404 assert(out->target.nr == 0);
405
406 for (i = 0; i < diff->target.nr; i++) {
407 while (diff->target.ranges[i].start > rs->ranges[j].end) {
408 j++;
409 if (j == rs->nr)
410 return;
411 }
412 if (ranges_overlap(&diff->target.ranges[i], &rs->ranges[j])) {
413 range_set_append(&out->parent,
414 diff->parent.ranges[i].start,
415 diff->parent.ranges[i].end);
416 range_set_append(&out->target,
417 diff->target.ranges[i].start,
418 diff->target.ranges[i].end);
419 }
420 }
421 }
422
423 /*
424 * Adjust the line counts in 'rs' to account for the lines
425 * added/removed in the diff.
426 */
427 static void range_set_shift_diff(struct range_set *out,
428 struct range_set *rs,
429 struct diff_ranges *diff)
430 {
431 unsigned int i, j = 0;
432 long offset = 0;
433 struct range *src = rs->ranges;
434 struct range *target = diff->target.ranges;
435 struct range *parent = diff->parent.ranges;
436
437 for (i = 0; i < rs->nr; i++) {
438 while (j < diff->target.nr && src[i].start >= target[j].start) {
439 offset += (parent[j].end-parent[j].start)
440 - (target[j].end-target[j].start);
441 j++;
442 }
443 range_set_append(out, src[i].start+offset, src[i].end+offset);
444 }
445 }
446
447 /*
448 * Given a diff and the set of interesting ranges, map the ranges
449 * across the diff. That is: observe that the target commit takes
450 * blame for all the + (target-side) ranges. So for every pair of
451 * ranges in the diff that was touched, we remove the latter and add
452 * its parent side.
453 */
454 static void range_set_map_across_diff(struct range_set *out,
455 struct range_set *rs,
456 struct diff_ranges *diff,
457 struct diff_ranges **touched_out)
458 {
459 struct diff_ranges *touched = xmalloc(sizeof(*touched));
460 struct range_set tmp1 = RANGE_SET_INIT;
461 struct range_set tmp2 = RANGE_SET_INIT;
462
463 diff_ranges_init(touched);
464 diff_ranges_filter_touched(touched, diff, rs);
465 range_set_difference(&tmp1, rs, &touched->target);
466 range_set_shift_diff(&tmp2, &tmp1, diff);
467 range_set_union(out, &tmp2, &touched->parent);
468 range_set_release(&tmp1);
469 range_set_release(&tmp2);
470
471 *touched_out = touched;
472 }
473
474 static struct commit *check_single_commit(struct rev_info *revs)
475 {
476 struct object *commit = NULL;
477 int found = -1;
478 int i;
479
480 for (i = 0; i < revs->pending.nr; i++) {
481 struct object *obj = revs->pending.objects[i].item;
482 if (obj->flags & UNINTERESTING)
483 continue;
484 obj = deref_tag(revs->repo, obj, NULL, 0);
485 if (!obj || obj->type != OBJ_COMMIT)
486 die("Non commit %s?", revs->pending.objects[i].name);
487 if (commit)
488 die("More than one commit to dig from: %s and %s?",
489 revs->pending.objects[i].name,
490 revs->pending.objects[found].name);
491 commit = obj;
492 found = i;
493 }
494
495 if (!commit)
496 die("No commit specified?");
497
498 return (struct commit *) commit;
499 }
500
501 static void fill_blob_sha1(struct repository *r, struct commit *commit,
502 struct diff_filespec *spec)
503 {
504 unsigned short mode;
505 struct object_id oid;
506
507 if (get_tree_entry(r, &commit->object.oid, spec->path, &oid, &mode))
508 die("There is no path %s in the commit", spec->path);
509 fill_filespec(spec, &oid, 1, mode);
510
511 return;
512 }
513
514 static void fill_line_ends(struct repository *r,
515 struct diff_filespec *spec,
516 long *lines,
517 unsigned long **line_ends)
518 {
519 int num = 0, size = 50;
520 long cur = 0;
521 unsigned long *ends = NULL;
522 char *data = NULL;
523
524 if (diff_populate_filespec(r, spec, NULL))
525 die("Cannot read blob %s", oid_to_hex(&spec->oid));
526
527 ALLOC_ARRAY(ends, size);
528 ends[cur++] = 0;
529 data = spec->data;
530 while (num < spec->size) {
531 if (data[num] == '\n' || num == spec->size - 1) {
532 ALLOC_GROW(ends, (cur + 1), size);
533 ends[cur++] = num;
534 }
535 num++;
536 }
537
538 /* shrink the array to fit the elements */
539 REALLOC_ARRAY(ends, cur);
540 *lines = cur-1;
541 *line_ends = ends;
542 }
543
544 struct nth_line_cb {
545 struct diff_filespec *spec;
546 long lines;
547 unsigned long *line_ends;
548 };
549
550 static const char *nth_line(void *data, long line)
551 {
552 struct nth_line_cb *d = data;
553 assert(d && line <= d->lines);
554 assert(d->spec && d->spec->data);
555
556 if (line == 0)
557 return (char *)d->spec->data;
558 else
559 return (char *)d->spec->data + d->line_ends[line] + 1;
560 }
561
562 static struct line_log_data *
563 parse_lines(struct repository *r, struct commit *commit,
564 const char *prefix, struct string_list *args)
565 {
566 long lines = 0;
567 unsigned long *ends = NULL;
568 struct nth_line_cb cb_data;
569 struct string_list_item *item;
570 struct line_log_data *ranges = NULL;
571 struct line_log_data *p;
572
573 for_each_string_list_item(item, args) {
574 const char *name_part, *range_part;
575 char *full_name;
576 struct diff_filespec *spec;
577 long begin = 0, end = 0;
578 long anchor;
579
580 name_part = skip_range_arg(item->string, r->index);
581 if (!name_part || *name_part != ':' || !name_part[1])
582 die("-L argument not 'start,end:file' or ':funcname:file': %s",
583 item->string);
584 range_part = xstrndup(item->string, name_part - item->string);
585 name_part++;
586
587 full_name = prefix_path(prefix, prefix ? strlen(prefix) : 0,
588 name_part);
589
590 spec = alloc_filespec(full_name);
591 fill_blob_sha1(r, commit, spec);
592 fill_line_ends(r, spec, &lines, &ends);
593 cb_data.spec = spec;
594 cb_data.lines = lines;
595 cb_data.line_ends = ends;
596
597 p = search_line_log_data(ranges, full_name, NULL);
598 if (p && p->ranges.nr)
599 anchor = p->ranges.ranges[p->ranges.nr - 1].end + 1;
600 else
601 anchor = 1;
602
603 if (parse_range_arg(range_part, nth_line, &cb_data,
604 lines, anchor, &begin, &end,
605 full_name, r->index))
606 die("malformed -L argument '%s'", range_part);
607 if ((!lines && (begin || end)) || lines < begin)
608 die("file %s has only %lu lines", name_part, lines);
609 if (begin < 1)
610 begin = 1;
611 if (end < 1 || lines < end)
612 end = lines;
613 begin--;
614 line_log_data_insert(&ranges, full_name, begin, end);
615
616 free_filespec(spec);
617 FREE_AND_NULL(ends);
618 }
619
620 for (p = ranges; p; p = p->next)
621 sort_and_merge_range_set(&p->ranges);
622
623 return ranges;
624 }
625
626 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
627 {
628 struct line_log_data *ret = xmalloc(sizeof(*ret));
629
630 assert(r);
631 line_log_data_init(ret);
632 range_set_copy(&ret->ranges, &r->ranges);
633
634 ret->path = xstrdup(r->path);
635
636 return ret;
637 }
638
639 static struct line_log_data *
640 line_log_data_copy(struct line_log_data *r)
641 {
642 struct line_log_data *ret = NULL;
643 struct line_log_data *tmp = NULL, *prev = NULL;
644
645 assert(r);
646 ret = tmp = prev = line_log_data_copy_one(r);
647 r = r->next;
648 while (r) {
649 tmp = line_log_data_copy_one(r);
650 prev->next = tmp;
651 prev = tmp;
652 r = r->next;
653 }
654
655 return ret;
656 }
657
658 /* merge two range sets across files */
659 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
660 struct line_log_data *b)
661 {
662 struct line_log_data *head = NULL, **pp = &head;
663
664 while (a || b) {
665 struct line_log_data *src;
666 struct line_log_data *src2 = NULL;
667 struct line_log_data *d;
668 int cmp;
669 if (!a)
670 cmp = 1;
671 else if (!b)
672 cmp = -1;
673 else
674 cmp = strcmp(a->path, b->path);
675 if (cmp < 0) {
676 src = a;
677 a = a->next;
678 } else if (cmp == 0) {
679 src = a;
680 a = a->next;
681 src2 = b;
682 b = b->next;
683 } else {
684 src = b;
685 b = b->next;
686 }
687 d = xmalloc(sizeof(struct line_log_data));
688 line_log_data_init(d);
689 d->path = xstrdup(src->path);
690 *pp = d;
691 pp = &d->next;
692 if (src2)
693 range_set_union(&d->ranges, &src->ranges, &src2->ranges);
694 else
695 range_set_copy(&d->ranges, &src->ranges);
696 }
697
698 return head;
699 }
700
701 static void add_line_range(struct rev_info *revs, struct commit *commit,
702 struct line_log_data *range)
703 {
704 struct line_log_data *old_line = NULL;
705 struct line_log_data *new_line = NULL;
706
707 old_line = lookup_decoration(&revs->line_log_data, &commit->object);
708 if (old_line && range) {
709 new_line = line_log_data_merge(old_line, range);
710 free_line_log_data(old_line);
711 } else if (range)
712 new_line = line_log_data_copy(range);
713
714 if (new_line)
715 add_decoration(&revs->line_log_data, &commit->object, new_line);
716 }
717
718 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
719 {
720 struct line_log_data *r;
721 r = lookup_decoration(&revs->line_log_data, &commit->object);
722 if (!r)
723 return;
724 free_line_log_data(r);
725 add_decoration(&revs->line_log_data, &commit->object, NULL);
726 }
727
728 static struct line_log_data *lookup_line_range(struct rev_info *revs,
729 struct commit *commit)
730 {
731 struct line_log_data *ret = NULL;
732 struct line_log_data *d;
733
734 ret = lookup_decoration(&revs->line_log_data, &commit->object);
735
736 for (d = ret; d; d = d->next)
737 range_set_check_invariants(&d->ranges);
738
739 return ret;
740 }
741
742 static int same_paths_in_pathspec_and_range(struct pathspec *pathspec,
743 struct line_log_data *range)
744 {
745 int i;
746 struct line_log_data *r;
747
748 for (i = 0, r = range; i < pathspec->nr && r; i++, r = r->next)
749 if (strcmp(pathspec->items[i].match, r->path))
750 return 0;
751 if (i < pathspec->nr || r)
752 /* different number of pathspec items and ranges */
753 return 0;
754
755 return 1;
756 }
757
758 static void parse_pathspec_from_ranges(struct pathspec *pathspec,
759 struct line_log_data *range)
760 {
761 struct line_log_data *r;
762 struct strvec array = STRVEC_INIT;
763 const char **paths;
764
765 for (r = range; r; r = r->next)
766 strvec_push(&array, r->path);
767 paths = strvec_detach(&array);
768
769 parse_pathspec(pathspec, 0, PATHSPEC_PREFER_FULL, "", paths);
770 /* strings are now owned by pathspec */
771 free(paths);
772 }
773
774 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
775 {
776 struct commit *commit = NULL;
777 struct line_log_data *range;
778
779 commit = check_single_commit(rev);
780 range = parse_lines(rev->diffopt.repo, commit, prefix, args);
781 add_line_range(rev, commit, range);
782
783 parse_pathspec_from_ranges(&rev->diffopt.pathspec, range);
784 }
785
786 static void move_diff_queue(struct diff_queue_struct *dst,
787 struct diff_queue_struct *src)
788 {
789 assert(src != dst);
790 memcpy(dst, src, sizeof(struct diff_queue_struct));
791 DIFF_QUEUE_CLEAR(src);
792 }
793
794 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
795 {
796 int i;
797 struct diff_queue_struct outq;
798 DIFF_QUEUE_CLEAR(&outq);
799
800 for (i = 0; i < diff_queued_diff.nr; i++) {
801 struct diff_filepair *p = diff_queued_diff.queue[i];
802 struct line_log_data *rg = NULL;
803
804 if (!DIFF_FILE_VALID(p->two)) {
805 if (keep_deletions)
806 diff_q(&outq, p);
807 else
808 diff_free_filepair(p);
809 continue;
810 }
811 for (rg = range; rg; rg = rg->next) {
812 if (!strcmp(rg->path, p->two->path))
813 break;
814 }
815 if (rg)
816 diff_q(&outq, p);
817 else
818 diff_free_filepair(p);
819 }
820 free(diff_queued_diff.queue);
821 diff_queued_diff = outq;
822 }
823
824 static inline int diff_might_be_rename(void)
825 {
826 int i;
827 for (i = 0; i < diff_queued_diff.nr; i++)
828 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
829 /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
830 /* diff_queued_diff.queue[i]->two->path); */
831 return 1;
832 }
833 return 0;
834 }
835
836 static void queue_diffs(struct line_log_data *range,
837 struct diff_options *opt,
838 struct diff_queue_struct *queue,
839 struct commit *commit, struct commit *parent)
840 {
841 struct object_id *tree_oid, *parent_tree_oid;
842
843 assert(commit);
844
845 tree_oid = get_commit_tree_oid(commit);
846 parent_tree_oid = parent ? get_commit_tree_oid(parent) : NULL;
847
848 if (opt->detect_rename &&
849 !same_paths_in_pathspec_and_range(&opt->pathspec, range)) {
850 clear_pathspec(&opt->pathspec);
851 parse_pathspec_from_ranges(&opt->pathspec, range);
852 }
853 DIFF_QUEUE_CLEAR(&diff_queued_diff);
854 diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
855 if (opt->detect_rename && diff_might_be_rename()) {
856 /* must look at the full tree diff to detect renames */
857 clear_pathspec(&opt->pathspec);
858 DIFF_QUEUE_CLEAR(&diff_queued_diff);
859
860 diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
861
862 filter_diffs_for_paths(range, 1);
863 diffcore_std(opt);
864 filter_diffs_for_paths(range, 0);
865 }
866 move_diff_queue(queue, &diff_queued_diff);
867 }
868
869 static char *get_nth_line(long line, unsigned long *ends, void *data)
870 {
871 if (line == 0)
872 return (char *)data;
873 else
874 return (char *)data + ends[line] + 1;
875 }
876
877 static void print_line(const char *prefix, char first,
878 long line, unsigned long *ends, void *data,
879 const char *color, const char *reset, FILE *file)
880 {
881 char *begin = get_nth_line(line, ends, data);
882 char *end = get_nth_line(line+1, ends, data);
883 int had_nl = 0;
884
885 if (end > begin && end[-1] == '\n') {
886 end--;
887 had_nl = 1;
888 }
889
890 fputs(prefix, file);
891 fputs(color, file);
892 putc(first, file);
893 fwrite(begin, 1, end-begin, file);
894 fputs(reset, file);
895 putc('\n', file);
896 if (!had_nl)
897 fputs("\\ No newline at end of file\n", file);
898 }
899
900 static char *output_prefix(struct diff_options *opt)
901 {
902 char *prefix = "";
903
904 if (opt->output_prefix) {
905 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
906 prefix = sb->buf;
907 }
908
909 return prefix;
910 }
911
912 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
913 {
914 unsigned int i, j = 0;
915 long p_lines, t_lines;
916 unsigned long *p_ends = NULL, *t_ends = NULL;
917 struct diff_filepair *pair = range->pair;
918 struct diff_ranges *diff = &range->diff;
919
920 struct diff_options *opt = &rev->diffopt;
921 char *prefix = output_prefix(opt);
922 const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
923 const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
924 const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
925 const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
926 const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
927 const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
928
929 if (!pair || !diff)
930 return;
931
932 if (pair->one->oid_valid)
933 fill_line_ends(rev->diffopt.repo, pair->one, &p_lines, &p_ends);
934 fill_line_ends(rev->diffopt.repo, pair->two, &t_lines, &t_ends);
935
936 fprintf(opt->file, "%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
937 fprintf(opt->file, "%s%s--- %s%s%s\n", prefix, c_meta,
938 pair->one->oid_valid ? "a/" : "",
939 pair->one->oid_valid ? pair->one->path : "/dev/null",
940 c_reset);
941 fprintf(opt->file, "%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
942 for (i = 0; i < range->ranges.nr; i++) {
943 long p_start, p_end;
944 long t_start = range->ranges.ranges[i].start;
945 long t_end = range->ranges.ranges[i].end;
946 long t_cur = t_start;
947 unsigned int j_last;
948
949 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
950 j++;
951 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
952 continue;
953
954 /* Scan ahead to determine the last diff that falls in this range */
955 j_last = j;
956 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
957 j_last++;
958 if (j_last > j)
959 j_last--;
960
961 /*
962 * Compute parent hunk headers: we know that the diff
963 * has the correct line numbers (but not all hunks).
964 * So it suffices to shift the start/end according to
965 * the line numbers of the first/last hunk(s) that
966 * fall in this range.
967 */
968 if (t_start < diff->target.ranges[j].start)
969 p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
970 else
971 p_start = diff->parent.ranges[j].start;
972 if (t_end > diff->target.ranges[j_last].end)
973 p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
974 else
975 p_end = diff->parent.ranges[j_last].end;
976
977 if (!p_start && !p_end) {
978 p_start = -1;
979 p_end = -1;
980 }
981
982 /* Now output a diff hunk for this range */
983 fprintf(opt->file, "%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
984 prefix, c_frag,
985 p_start+1, p_end-p_start, t_start+1, t_end-t_start,
986 c_reset);
987 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
988 int k;
989 for (; t_cur < diff->target.ranges[j].start; t_cur++)
990 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
991 c_context, c_reset, opt->file);
992 for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
993 print_line(prefix, '-', k, p_ends, pair->one->data,
994 c_old, c_reset, opt->file);
995 for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
996 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
997 c_new, c_reset, opt->file);
998 j++;
999 }
1000 for (; t_cur < t_end; t_cur++)
1001 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
1002 c_context, c_reset, opt->file);
1003 }
1004
1005 free(p_ends);
1006 free(t_ends);
1007 }
1008
1009 /*
1010 * NEEDSWORK: manually building a diff here is not the Right
1011 * Thing(tm). log -L should be built into the diff pipeline.
1012 */
1013 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
1014 {
1015 fprintf(rev->diffopt.file, "%s\n", output_prefix(&rev->diffopt));
1016 while (range) {
1017 dump_diff_hacky_one(rev, range);
1018 range = range->next;
1019 }
1020 }
1021
1022 /*
1023 * Unlike most other functions, this destructively operates on
1024 * 'range'.
1025 */
1026 static int process_diff_filepair(struct rev_info *rev,
1027 struct diff_filepair *pair,
1028 struct line_log_data *range,
1029 struct diff_ranges **diff_out)
1030 {
1031 struct line_log_data *rg = range;
1032 struct range_set tmp;
1033 struct diff_ranges diff;
1034 mmfile_t file_parent, file_target;
1035
1036 assert(pair->two->path);
1037 while (rg) {
1038 assert(rg->path);
1039 if (!strcmp(rg->path, pair->two->path))
1040 break;
1041 rg = rg->next;
1042 }
1043
1044 if (!rg)
1045 return 0;
1046 if (rg->ranges.nr == 0)
1047 return 0;
1048
1049 assert(pair->two->oid_valid);
1050 diff_populate_filespec(rev->diffopt.repo, pair->two, NULL);
1051 file_target.ptr = pair->two->data;
1052 file_target.size = pair->two->size;
1053
1054 if (pair->one->oid_valid) {
1055 diff_populate_filespec(rev->diffopt.repo, pair->one, NULL);
1056 file_parent.ptr = pair->one->data;
1057 file_parent.size = pair->one->size;
1058 } else {
1059 file_parent.ptr = "";
1060 file_parent.size = 0;
1061 }
1062
1063 diff_ranges_init(&diff);
1064 if (collect_diff(&file_parent, &file_target, &diff))
1065 die("unable to generate diff for %s", pair->one->path);
1066
1067 /* NEEDSWORK should apply some heuristics to prevent mismatches */
1068 free(rg->path);
1069 rg->path = xstrdup(pair->one->path);
1070
1071 range_set_init(&tmp, 0);
1072 range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1073 range_set_release(&rg->ranges);
1074 range_set_move(&rg->ranges, &tmp);
1075
1076 diff_ranges_release(&diff);
1077
1078 return ((*diff_out)->parent.nr > 0);
1079 }
1080
1081 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1082 {
1083 struct diff_filepair *new_filepair = xmalloc(sizeof(struct diff_filepair));
1084 new_filepair->one = pair->one;
1085 new_filepair->two = pair->two;
1086 new_filepair->one->count++;
1087 new_filepair->two->count++;
1088 return new_filepair;
1089 }
1090
1091 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1092 {
1093 for (int i = 0; i < n; i++)
1094 diff_free_queue(&dq[i]);
1095 free(dq);
1096 }
1097
1098 static int process_all_files(struct line_log_data **range_out,
1099 struct rev_info *rev,
1100 struct diff_queue_struct *queue,
1101 struct line_log_data *range)
1102 {
1103 int i, changed = 0;
1104
1105 *range_out = line_log_data_copy(range);
1106
1107 for (i = 0; i < queue->nr; i++) {
1108 struct diff_ranges *pairdiff = NULL;
1109 struct diff_filepair *pair = queue->queue[i];
1110 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1111 /*
1112 * Store away the diff for later output. We
1113 * tuck it in the ranges we got as _input_,
1114 * since that's the commit that caused the
1115 * diff.
1116 *
1117 * NEEDSWORK not enough when we get around to
1118 * doing something interesting with merges;
1119 * currently each invocation on a merge parent
1120 * trashes the previous one's diff.
1121 *
1122 * NEEDSWORK tramples over data structures not owned here
1123 */
1124 struct line_log_data *rg = range;
1125 changed++;
1126 while (rg && strcmp(rg->path, pair->two->path))
1127 rg = rg->next;
1128 assert(rg);
1129 rg->pair = diff_filepair_dup(queue->queue[i]);
1130 memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1131 }
1132 free(pairdiff);
1133 }
1134
1135 return changed;
1136 }
1137
1138 int line_log_print(struct rev_info *rev, struct commit *commit)
1139 {
1140
1141 show_log(rev);
1142 if (!(rev->diffopt.output_format & DIFF_FORMAT_NO_OUTPUT)) {
1143 struct line_log_data *range = lookup_line_range(rev, commit);
1144 dump_diff_hacky(rev, range);
1145 }
1146 return 1;
1147 }
1148
1149 static int bloom_filter_check(struct rev_info *rev,
1150 struct commit *commit,
1151 struct line_log_data *range)
1152 {
1153 struct bloom_filter *filter;
1154 struct bloom_key key;
1155 int result = 0;
1156
1157 if (!commit->parents)
1158 return 1;
1159
1160 if (!rev->bloom_filter_settings ||
1161 !(filter = get_bloom_filter(rev->repo, commit)))
1162 return 1;
1163
1164 if (!range)
1165 return 0;
1166
1167 while (!result && range) {
1168 fill_bloom_key(range->path, strlen(range->path), &key, rev->bloom_filter_settings);
1169
1170 if (bloom_filter_contains(filter, &key, rev->bloom_filter_settings))
1171 result = 1;
1172
1173 clear_bloom_key(&key);
1174 range = range->next;
1175 }
1176
1177 return result;
1178 }
1179
1180 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1181 struct line_log_data *range)
1182 {
1183 struct commit *parent = NULL;
1184 struct diff_queue_struct queue;
1185 struct line_log_data *parent_range;
1186 int changed;
1187
1188 if (commit->parents)
1189 parent = commit->parents->item;
1190
1191 queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1192 changed = process_all_files(&parent_range, rev, &queue, range);
1193
1194 if (parent)
1195 add_line_range(rev, parent, parent_range);
1196 free_line_log_data(parent_range);
1197 diff_free_queue(&queue);
1198 return changed;
1199 }
1200
1201 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1202 struct line_log_data *range)
1203 {
1204 struct diff_queue_struct *diffqueues;
1205 struct line_log_data **cand;
1206 struct commit **parents;
1207 struct commit_list *p;
1208 int i;
1209 int nparents = commit_list_count(commit->parents);
1210
1211 if (nparents > 1 && rev->first_parent_only)
1212 nparents = 1;
1213
1214 ALLOC_ARRAY(diffqueues, nparents);
1215 ALLOC_ARRAY(cand, nparents);
1216 ALLOC_ARRAY(parents, nparents);
1217
1218 p = commit->parents;
1219 for (i = 0; i < nparents; i++) {
1220 parents[i] = p->item;
1221 p = p->next;
1222 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1223 }
1224
1225 for (i = 0; i < nparents; i++) {
1226 int changed;
1227 cand[i] = NULL;
1228 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1229 if (!changed) {
1230 /*
1231 * This parent can take all the blame, so we
1232 * don't follow any other path in history
1233 */
1234 add_line_range(rev, parents[i], cand[i]);
1235 clear_commit_line_range(rev, commit);
1236 commit_list_append(parents[i], &commit->parents);
1237 free(parents);
1238 free(cand);
1239 free_diffqueues(nparents, diffqueues);
1240 /* NEEDSWORK leaking like a sieve */
1241 return 0;
1242 }
1243 }
1244
1245 /*
1246 * No single parent took the blame. We add the candidates
1247 * from the above loop to the parents.
1248 */
1249 for (i = 0; i < nparents; i++) {
1250 add_line_range(rev, parents[i], cand[i]);
1251 }
1252
1253 clear_commit_line_range(rev, commit);
1254 free(parents);
1255 free(cand);
1256 free_diffqueues(nparents, diffqueues);
1257 return 1;
1258
1259 /* NEEDSWORK evil merge detection stuff */
1260 /* NEEDSWORK leaking like a sieve */
1261 }
1262
1263 int line_log_process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1264 {
1265 struct line_log_data *range = lookup_line_range(rev, commit);
1266 int changed = 0;
1267
1268 if (range) {
1269 if (commit->parents && !bloom_filter_check(rev, commit, range)) {
1270 struct line_log_data *prange = line_log_data_copy(range);
1271 add_line_range(rev, commit->parents->item, prange);
1272 clear_commit_line_range(rev, commit);
1273 } else if (!commit->parents || !commit->parents->next)
1274 changed = process_ranges_ordinary_commit(rev, commit, range);
1275 else
1276 changed = process_ranges_merge_commit(rev, commit, range);
1277 }
1278
1279 if (!changed)
1280 commit->object.flags |= TREESAME;
1281
1282 return changed;
1283 }
1284
1285 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev UNUSED,
1286 struct commit **pp)
1287 {
1288 for (;;) {
1289 struct commit *p = *pp;
1290 if (p->parents && p->parents->next)
1291 return rewrite_one_ok;
1292 if (p->object.flags & UNINTERESTING)
1293 return rewrite_one_ok;
1294 if (!(p->object.flags & TREESAME))
1295 return rewrite_one_ok;
1296 if (!p->parents)
1297 return rewrite_one_noparents;
1298 *pp = p->parents->item;
1299 }
1300 }
1301
1302 int line_log_filter(struct rev_info *rev)
1303 {
1304 struct commit *commit;
1305 struct commit_list *list = rev->commits;
1306 struct commit_list *out = NULL, **pp = &out;
1307
1308 while (list) {
1309 struct commit_list *to_free = NULL;
1310 commit = list->item;
1311 if (line_log_process_ranges_arbitrary_commit(rev, commit)) {
1312 *pp = list;
1313 pp = &list->next;
1314 } else
1315 to_free = list;
1316 list = list->next;
1317 free(to_free);
1318 }
1319 *pp = NULL;
1320
1321 for (list = out; list; list = list->next)
1322 rewrite_parents(rev, list->item, line_log_rewrite_one);
1323
1324 rev->commits = out;
1325
1326 return 0;
1327 }
1328
1329 static void free_void_line_log_data(void *data)
1330 {
1331 free_line_log_data(data);
1332 }
1333
1334 void line_log_free(struct rev_info *rev)
1335 {
1336 clear_decoration(&rev->line_log_data, free_void_line_log_data);
1337 }
The Best VCS Around