?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
trace2/tr2_tls.h: remove unnecessary include
[git.git] / line-log.c
blobc276ccec54959199b0a6bf4cb1f37757a359a428
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 "userdiff.h"
16 #include "line-log.h"
17 #include "setup.h"
18 #include "strvec.h"
19 #include "bloom.h"
20 #include "tree-walk.h"
21
22 static void range_set_grow(struct range_set *rs, size_t extra)
23 {
24 ALLOC_GROW(rs->ranges, rs->nr + extra, rs->alloc);
25 }
26
27 /* Either initialization would be fine */
28 #define RANGE_SET_INIT {0}
29
30 void range_set_init(struct range_set *rs, size_t prealloc)
31 {
32 rs->alloc = rs->nr = 0;
33 rs->ranges = NULL;
34 if (prealloc)
35 range_set_grow(rs, prealloc);
36 }
37
38 void range_set_release(struct range_set *rs)
39 {
40 FREE_AND_NULL(rs->ranges);
41 rs->alloc = rs->nr = 0;
42 }
43
44 /* dst must be uninitialized! */
45 static void range_set_copy(struct range_set *dst, struct range_set *src)
46 {
47 range_set_init(dst, src->nr);
48 COPY_ARRAY(dst->ranges, src->ranges, src->nr);
49 dst->nr = src->nr;
50 }
51
52 static void range_set_move(struct range_set *dst, struct range_set *src)
53 {
54 range_set_release(dst);
55 dst->ranges = src->ranges;
56 dst->nr = src->nr;
57 dst->alloc = src->alloc;
58 src->ranges = NULL;
59 src->alloc = src->nr = 0;
60 }
61
62 /* tack on a _new_ range _at the end_ */
63 void range_set_append_unsafe(struct range_set *rs, long a, long b)
64 {
65 assert(a <= b);
66 range_set_grow(rs, 1);
67 rs->ranges[rs->nr].start = a;
68 rs->ranges[rs->nr].end = b;
69 rs->nr++;
70 }
71
72 void range_set_append(struct range_set *rs, long a, long b)
73 {
74 assert(rs->nr == 0 || rs->ranges[rs->nr-1].end <= a);
75 range_set_append_unsafe(rs, a, b);
76 }
77
78 static int range_cmp(const void *_r, const void *_s)
79 {
80 const struct range *r = _r;
81 const struct range *s = _s;
82
83 /* this could be simply 'return r.start-s.start', but for the types */
84 if (r->start == s->start)
85 return 0;
86 if (r->start < s->start)
87 return -1;
88 return 1;
89 }
90
91 /*
92 * Check that the ranges are non-empty, sorted and non-overlapping
93 */
94 static void range_set_check_invariants(struct range_set *rs)
95 {
96 unsigned int i;
97
98 if (!rs)
99 return;
100
101 if (rs->nr)
102 assert(rs->ranges[0].start < rs->ranges[0].end);
103
104 for (i = 1; i < rs->nr; i++) {
105 assert(rs->ranges[i-1].end < rs->ranges[i].start);
106 assert(rs->ranges[i].start < rs->ranges[i].end);
107 }
108 }
109
110 /*
111 * In-place pass of sorting and merging the ranges in the range set,
112 * to establish the invariants when we get the ranges from the user
113 */
114 void sort_and_merge_range_set(struct range_set *rs)
115 {
116 unsigned int i;
117 unsigned int o = 0; /* output cursor */
118
119 QSORT(rs->ranges, rs->nr, range_cmp);
120
121 for (i = 0; i < rs->nr; i++) {
122 if (rs->ranges[i].start == rs->ranges[i].end)
123 continue;
124 if (o > 0 && rs->ranges[i].start <= rs->ranges[o-1].end) {
125 if (rs->ranges[o-1].end < rs->ranges[i].end)
126 rs->ranges[o-1].end = rs->ranges[i].end;
127 } else {
128 rs->ranges[o].start = rs->ranges[i].start;
129 rs->ranges[o].end = rs->ranges[i].end;
130 o++;
131 }
132 }
133 assert(o <= rs->nr);
134 rs->nr = o;
135
136 range_set_check_invariants(rs);
137 }
138
139 /*
140 * Union of range sets (i.e., sets of line numbers). Used to merge
141 * them when searches meet at a common ancestor.
142 *
143 * This is also where the ranges are consolidated into canonical form:
144 * overlapping and adjacent ranges are merged, and empty ranges are
145 * removed.
146 */
147 static void range_set_union(struct range_set *out,
148 struct range_set *a, struct range_set *b)
149 {
150 unsigned int i = 0, j = 0;
151 struct range *ra = a->ranges;
152 struct range *rb = b->ranges;
153 /* cannot make an alias of out->ranges: it may change during grow */
154
155 assert(out->nr == 0);
156 while (i < a->nr || j < b->nr) {
157 struct range *new_range;
158 if (i < a->nr && j < b->nr) {
159 if (ra[i].start < rb[j].start)
160 new_range = &ra[i++];
161 else if (ra[i].start > rb[j].start)
162 new_range = &rb[j++];
163 else if (ra[i].end < rb[j].end)
164 new_range = &ra[i++];
165 else
166 new_range = &rb[j++];
167 } else if (i < a->nr) /* b exhausted */
168 new_range = &ra[i++];
169 else /* a exhausted */
170 new_range = &rb[j++];
171 if (new_range->start == new_range->end)
172 ; /* empty range */
173 else if (!out->nr || out->ranges[out->nr-1].end < new_range->start) {
174 range_set_grow(out, 1);
175 out->ranges[out->nr].start = new_range->start;
176 out->ranges[out->nr].end = new_range->end;
177 out->nr++;
178 } else if (out->ranges[out->nr-1].end < new_range->end) {
179 out->ranges[out->nr-1].end = new_range->end;
180 }
181 }
182 }
183
184 /*
185 * Difference of range sets (out = a \ b). Pass the "interesting"
186 * ranges as 'a' and the target side of the diff as 'b': it removes
187 * the ranges for which the commit is responsible.
188 */
189 static void range_set_difference(struct range_set *out,
190 struct range_set *a, struct range_set *b)
191 {
192 unsigned int i, j = 0;
193 for (i = 0; i < a->nr; i++) {
194 long start = a->ranges[i].start;
195 long end = a->ranges[i].end;
196 while (start < end) {
197 while (j < b->nr && start >= b->ranges[j].end)
198 /*
199 * a: |-------
200 * b: ------|
201 */
202 j++;
203 if (j >= b->nr || end < b->ranges[j].start) {
204 /*
205 * b exhausted, or
206 * a: ----|
207 * b: |----
208 */
209 range_set_append(out, start, end);
210 break;
211 }
212 if (start >= b->ranges[j].start) {
213 /*
214 * a: |--????
215 * b: |------|
216 */
217 start = b->ranges[j].end;
218 } else if (end > b->ranges[j].start) {
219 /*
220 * a: |-----|
221 * b: |--?????
222 */
223 if (start < b->ranges[j].start)
224 range_set_append(out, start, b->ranges[j].start);
225 start = b->ranges[j].end;
226 }
227 }
228 }
229 }
230
231 static void diff_ranges_init(struct diff_ranges *diff)
232 {
233 range_set_init(&diff->parent, 0);
234 range_set_init(&diff->target, 0);
235 }
236
237 static void diff_ranges_release(struct diff_ranges *diff)
238 {
239 range_set_release(&diff->parent);
240 range_set_release(&diff->target);
241 }
242
243 static void line_log_data_init(struct line_log_data *r)
244 {
245 memset(r, 0, sizeof(struct line_log_data));
246 range_set_init(&r->ranges, 0);
247 }
248
249 static void line_log_data_clear(struct line_log_data *r)
250 {
251 range_set_release(&r->ranges);
252 if (r->pair)
253 diff_free_filepair(r->pair);
254 }
255
256 static void free_line_log_data(struct line_log_data *r)
257 {
258 while (r) {
259 struct line_log_data *next = r->next;
260 line_log_data_clear(r);
261 free(r);
262 r = next;
263 }
264 }
265
266 static struct line_log_data *
267 search_line_log_data(struct line_log_data *list, const char *path,
268 struct line_log_data **insertion_point)
269 {
270 struct line_log_data *p = list;
271 if (insertion_point)
272 *insertion_point = NULL;
273 while (p) {
274 int cmp = strcmp(p->path, path);
275 if (!cmp)
276 return p;
277 if (insertion_point && cmp < 0)
278 *insertion_point = p;
279 p = p->next;
280 }
281 return NULL;
282 }
283
284 /*
285 * Note: takes ownership of 'path', which happens to be what the only
286 * caller needs.
287 */
288 static void line_log_data_insert(struct line_log_data **list,
289 char *path,
290 long begin, long end)
291 {
292 struct line_log_data *ip;
293 struct line_log_data *p = search_line_log_data(*list, path, &ip);
294
295 if (p) {
296 range_set_append_unsafe(&p->ranges, begin, end);
297 free(path);
298 return;
299 }
300
301 CALLOC_ARRAY(p, 1);
302 p->path = path;
303 range_set_append(&p->ranges, begin, end);
304 if (ip) {
305 p->next = ip->next;
306 ip->next = p;
307 } else {
308 p->next = *list;
309 *list = p;
310 }
311 }
312
313 struct collect_diff_cbdata {
314 struct diff_ranges *diff;
315 };
316
317 static int collect_diff_cb(long start_a, long count_a,
318 long start_b, long count_b,
319 void *data)
320 {
321 struct collect_diff_cbdata *d = data;
322
323 if (count_a >= 0)
324 range_set_append(&d->diff->parent, start_a, start_a + count_a);
325 if (count_b >= 0)
326 range_set_append(&d->diff->target, start_b, start_b + count_b);
327
328 return 0;
329 }
330
331 static int collect_diff(mmfile_t *parent, mmfile_t *target, struct diff_ranges *out)
332 {
333 struct collect_diff_cbdata cbdata = {NULL};
334 xpparam_t xpp;
335 xdemitconf_t xecfg;
336 xdemitcb_t ecb;
337
338 memset(&xpp, 0, sizeof(xpp));
339 memset(&xecfg, 0, sizeof(xecfg));
340 xecfg.ctxlen = xecfg.interhunkctxlen = 0;
341
342 cbdata.diff = out;
343 xecfg.hunk_func = collect_diff_cb;
344 memset(&ecb, 0, sizeof(ecb));
345 ecb.priv = &cbdata;
346 return xdi_diff(parent, target, &xpp, &xecfg, &ecb);
347 }
348
349 /*
350 * These are handy for debugging. Removing them with #if 0 silences
351 * the "unused function" warning.
352 */
353 #if 0
354 static void dump_range_set(struct range_set *rs, const char *desc)
355 {
356 int i;
357 printf("range set %s (%d items):\n", desc, rs->nr);
358 for (i = 0; i < rs->nr; i++)
359 printf("\t[%ld,%ld]\n", rs->ranges[i].start, rs->ranges[i].end);
360 }
361
362 static void dump_line_log_data(struct line_log_data *r)
363 {
364 char buf[4096];
365 while (r) {
366 snprintf(buf, 4096, "file %s\n", r->path);
367 dump_range_set(&r->ranges, buf);
368 r = r->next;
369 }
370 }
371
372 static void dump_diff_ranges(struct diff_ranges *diff, const char *desc)
373 {
374 int i;
375 assert(diff->parent.nr == diff->target.nr);
376 printf("diff ranges %s (%d items):\n", desc, diff->parent.nr);
377 printf("\tparent\ttarget\n");
378 for (i = 0; i < diff->parent.nr; i++) {
379 printf("\t[%ld,%ld]\t[%ld,%ld]\n",
380 diff->parent.ranges[i].start,
381 diff->parent.ranges[i].end,
382 diff->target.ranges[i].start,
383 diff->target.ranges[i].end);
384 }
385 }
386 #endif
387
388
389 static int ranges_overlap(struct range *a, struct range *b)
390 {
391 return !(a->end <= b->start || b->end <= a->start);
392 }
393
394 /*
395 * Given a diff and the set of interesting ranges, determine all hunks
396 * of the diff which touch (overlap) at least one of the interesting
397 * ranges in the target.
398 */
399 static void diff_ranges_filter_touched(struct diff_ranges *out,
400 struct diff_ranges *diff,
401 struct range_set *rs)
402 {
403 unsigned int i, j = 0;
404
405 assert(out->target.nr == 0);
406
407 for (i = 0; i < diff->target.nr; i++) {
408 while (diff->target.ranges[i].start > rs->ranges[j].end) {
409 j++;
410 if (j == rs->nr)
411 return;
412 }
413 if (ranges_overlap(&diff->target.ranges[i], &rs->ranges[j])) {
414 range_set_append(&out->parent,
415 diff->parent.ranges[i].start,
416 diff->parent.ranges[i].end);
417 range_set_append(&out->target,
418 diff->target.ranges[i].start,
419 diff->target.ranges[i].end);
420 }
421 }
422 }
423
424 /*
425 * Adjust the line counts in 'rs' to account for the lines
426 * added/removed in the diff.
427 */
428 static void range_set_shift_diff(struct range_set *out,
429 struct range_set *rs,
430 struct diff_ranges *diff)
431 {
432 unsigned int i, j = 0;
433 long offset = 0;
434 struct range *src = rs->ranges;
435 struct range *target = diff->target.ranges;
436 struct range *parent = diff->parent.ranges;
437
438 for (i = 0; i < rs->nr; i++) {
439 while (j < diff->target.nr && src[i].start >= target[j].start) {
440 offset += (parent[j].end-parent[j].start)
441 - (target[j].end-target[j].start);
442 j++;
443 }
444 range_set_append(out, src[i].start+offset, src[i].end+offset);
445 }
446 }
447
448 /*
449 * Given a diff and the set of interesting ranges, map the ranges
450 * across the diff. That is: observe that the target commit takes
451 * blame for all the + (target-side) ranges. So for every pair of
452 * ranges in the diff that was touched, we remove the latter and add
453 * its parent side.
454 */
455 static void range_set_map_across_diff(struct range_set *out,
456 struct range_set *rs,
457 struct diff_ranges *diff,
458 struct diff_ranges **touched_out)
459 {
460 struct diff_ranges *touched = xmalloc(sizeof(*touched));
461 struct range_set tmp1 = RANGE_SET_INIT;
462 struct range_set tmp2 = RANGE_SET_INIT;
463
464 diff_ranges_init(touched);
465 diff_ranges_filter_touched(touched, diff, rs);
466 range_set_difference(&tmp1, rs, &touched->target);
467 range_set_shift_diff(&tmp2, &tmp1, diff);
468 range_set_union(out, &tmp2, &touched->parent);
469 range_set_release(&tmp1);
470 range_set_release(&tmp2);
471
472 *touched_out = touched;
473 }
474
475 static struct commit *check_single_commit(struct rev_info *revs)
476 {
477 struct object *commit = NULL;
478 int found = -1;
479 int i;
480
481 for (i = 0; i < revs->pending.nr; i++) {
482 struct object *obj = revs->pending.objects[i].item;
483 if (obj->flags & UNINTERESTING)
484 continue;
485 obj = deref_tag(revs->repo, obj, NULL, 0);
486 if (!obj || obj->type != OBJ_COMMIT)
487 die("Non commit %s?", revs->pending.objects[i].name);
488 if (commit)
489 die("More than one commit to dig from: %s and %s?",
490 revs->pending.objects[i].name,
491 revs->pending.objects[found].name);
492 commit = obj;
493 found = i;
494 }
495
496 if (!commit)
497 die("No commit specified?");
498
499 return (struct commit *) commit;
500 }
501
502 static void fill_blob_sha1(struct repository *r, struct commit *commit,
503 struct diff_filespec *spec)
504 {
505 unsigned short mode;
506 struct object_id oid;
507
508 if (get_tree_entry(r, &commit->object.oid, spec->path, &oid, &mode))
509 die("There is no path %s in the commit", spec->path);
510 fill_filespec(spec, &oid, 1, mode);
511
512 return;
513 }
514
515 static void fill_line_ends(struct repository *r,
516 struct diff_filespec *spec,
517 long *lines,
518 unsigned long **line_ends)
519 {
520 int num = 0, size = 50;
521 long cur = 0;
522 unsigned long *ends = NULL;
523 char *data = NULL;
524
525 if (diff_populate_filespec(r, spec, NULL))
526 die("Cannot read blob %s", oid_to_hex(&spec->oid));
527
528 ALLOC_ARRAY(ends, size);
529 ends[cur++] = 0;
530 data = spec->data;
531 while (num < spec->size) {
532 if (data[num] == '\n' || num == spec->size - 1) {
533 ALLOC_GROW(ends, (cur + 1), size);
534 ends[cur++] = num;
535 }
536 num++;
537 }
538
539 /* shrink the array to fit the elements */
540 REALLOC_ARRAY(ends, cur);
541 *lines = cur-1;
542 *line_ends = ends;
543 }
544
545 struct nth_line_cb {
546 struct diff_filespec *spec;
547 long lines;
548 unsigned long *line_ends;
549 };
550
551 static const char *nth_line(void *data, long line)
552 {
553 struct nth_line_cb *d = data;
554 assert(d && line <= d->lines);
555 assert(d->spec && d->spec->data);
556
557 if (line == 0)
558 return (char *)d->spec->data;
559 else
560 return (char *)d->spec->data + d->line_ends[line] + 1;
561 }
562
563 static struct line_log_data *
564 parse_lines(struct repository *r, struct commit *commit,
565 const char *prefix, struct string_list *args)
566 {
567 long lines = 0;
568 unsigned long *ends = NULL;
569 struct nth_line_cb cb_data;
570 struct string_list_item *item;
571 struct line_log_data *ranges = NULL;
572 struct line_log_data *p;
573
574 for_each_string_list_item(item, args) {
575 const char *name_part, *range_part;
576 char *full_name;
577 struct diff_filespec *spec;
578 long begin = 0, end = 0;
579 long anchor;
580
581 name_part = skip_range_arg(item->string, r->index);
582 if (!name_part || *name_part != ':' || !name_part[1])
583 die("-L argument not 'start,end:file' or ':funcname:file': %s",
584 item->string);
585 range_part = xstrndup(item->string, name_part - item->string);
586 name_part++;
587
588 full_name = prefix_path(prefix, prefix ? strlen(prefix) : 0,
589 name_part);
590
591 spec = alloc_filespec(full_name);
592 fill_blob_sha1(r, commit, spec);
593 fill_line_ends(r, spec, &lines, &ends);
594 cb_data.spec = spec;
595 cb_data.lines = lines;
596 cb_data.line_ends = ends;
597
598 p = search_line_log_data(ranges, full_name, NULL);
599 if (p && p->ranges.nr)
600 anchor = p->ranges.ranges[p->ranges.nr - 1].end + 1;
601 else
602 anchor = 1;
603
604 if (parse_range_arg(range_part, nth_line, &cb_data,
605 lines, anchor, &begin, &end,
606 full_name, r->index))
607 die("malformed -L argument '%s'", range_part);
608 if ((!lines && (begin || end)) || lines < begin)
609 die("file %s has only %lu lines", name_part, lines);
610 if (begin < 1)
611 begin = 1;
612 if (end < 1 || lines < end)
613 end = lines;
614 begin--;
615 line_log_data_insert(&ranges, full_name, begin, end);
616
617 free_filespec(spec);
618 FREE_AND_NULL(ends);
619 }
620
621 for (p = ranges; p; p = p->next)
622 sort_and_merge_range_set(&p->ranges);
623
624 return ranges;
625 }
626
627 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
628 {
629 struct line_log_data *ret = xmalloc(sizeof(*ret));
630
631 assert(r);
632 line_log_data_init(ret);
633 range_set_copy(&ret->ranges, &r->ranges);
634
635 ret->path = xstrdup(r->path);
636
637 return ret;
638 }
639
640 static struct line_log_data *
641 line_log_data_copy(struct line_log_data *r)
642 {
643 struct line_log_data *ret = NULL;
644 struct line_log_data *tmp = NULL, *prev = NULL;
645
646 assert(r);
647 ret = tmp = prev = line_log_data_copy_one(r);
648 r = r->next;
649 while (r) {
650 tmp = line_log_data_copy_one(r);
651 prev->next = tmp;
652 prev = tmp;
653 r = r->next;
654 }
655
656 return ret;
657 }
658
659 /* merge two range sets across files */
660 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
661 struct line_log_data *b)
662 {
663 struct line_log_data *head = NULL, **pp = &head;
664
665 while (a || b) {
666 struct line_log_data *src;
667 struct line_log_data *src2 = NULL;
668 struct line_log_data *d;
669 int cmp;
670 if (!a)
671 cmp = 1;
672 else if (!b)
673 cmp = -1;
674 else
675 cmp = strcmp(a->path, b->path);
676 if (cmp < 0) {
677 src = a;
678 a = a->next;
679 } else if (cmp == 0) {
680 src = a;
681 a = a->next;
682 src2 = b;
683 b = b->next;
684 } else {
685 src = b;
686 b = b->next;
687 }
688 d = xmalloc(sizeof(struct line_log_data));
689 line_log_data_init(d);
690 d->path = xstrdup(src->path);
691 *pp = d;
692 pp = &d->next;
693 if (src2)
694 range_set_union(&d->ranges, &src->ranges, &src2->ranges);
695 else
696 range_set_copy(&d->ranges, &src->ranges);
697 }
698
699 return head;
700 }
701
702 static void add_line_range(struct rev_info *revs, struct commit *commit,
703 struct line_log_data *range)
704 {
705 struct line_log_data *old_line = NULL;
706 struct line_log_data *new_line = NULL;
707
708 old_line = lookup_decoration(&revs->line_log_data, &commit->object);
709 if (old_line && range) {
710 new_line = line_log_data_merge(old_line, range);
711 free_line_log_data(old_line);
712 } else if (range)
713 new_line = line_log_data_copy(range);
714
715 if (new_line)
716 add_decoration(&revs->line_log_data, &commit->object, new_line);
717 }
718
719 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
720 {
721 struct line_log_data *r;
722 r = lookup_decoration(&revs->line_log_data, &commit->object);
723 if (!r)
724 return;
725 free_line_log_data(r);
726 add_decoration(&revs->line_log_data, &commit->object, NULL);
727 }
728
729 static struct line_log_data *lookup_line_range(struct rev_info *revs,
730 struct commit *commit)
731 {
732 struct line_log_data *ret = NULL;
733 struct line_log_data *d;
734
735 ret = lookup_decoration(&revs->line_log_data, &commit->object);
736
737 for (d = ret; d; d = d->next)
738 range_set_check_invariants(&d->ranges);
739
740 return ret;
741 }
742
743 static int same_paths_in_pathspec_and_range(struct pathspec *pathspec,
744 struct line_log_data *range)
745 {
746 int i;
747 struct line_log_data *r;
748
749 for (i = 0, r = range; i < pathspec->nr && r; i++, r = r->next)
750 if (strcmp(pathspec->items[i].match, r->path))
751 return 0;
752 if (i < pathspec->nr || r)
753 /* different number of pathspec items and ranges */
754 return 0;
755
756 return 1;
757 }
758
759 static void parse_pathspec_from_ranges(struct pathspec *pathspec,
760 struct line_log_data *range)
761 {
762 struct line_log_data *r;
763 struct strvec array = STRVEC_INIT;
764 const char **paths;
765
766 for (r = range; r; r = r->next)
767 strvec_push(&array, r->path);
768 paths = strvec_detach(&array);
769
770 parse_pathspec(pathspec, 0, PATHSPEC_PREFER_FULL, "", paths);
771 /* strings are now owned by pathspec */
772 free(paths);
773 }
774
775 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
776 {
777 struct commit *commit = NULL;
778 struct line_log_data *range;
779
780 commit = check_single_commit(rev);
781 range = parse_lines(rev->diffopt.repo, commit, prefix, args);
782 add_line_range(rev, commit, range);
783
784 parse_pathspec_from_ranges(&rev->diffopt.pathspec, range);
785 }
786
787 static void move_diff_queue(struct diff_queue_struct *dst,
788 struct diff_queue_struct *src)
789 {
790 assert(src != dst);
791 memcpy(dst, src, sizeof(struct diff_queue_struct));
792 DIFF_QUEUE_CLEAR(src);
793 }
794
795 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
796 {
797 int i;
798 struct diff_queue_struct outq;
799 DIFF_QUEUE_CLEAR(&outq);
800
801 for (i = 0; i < diff_queued_diff.nr; i++) {
802 struct diff_filepair *p = diff_queued_diff.queue[i];
803 struct line_log_data *rg = NULL;
804
805 if (!DIFF_FILE_VALID(p->two)) {
806 if (keep_deletions)
807 diff_q(&outq, p);
808 else
809 diff_free_filepair(p);
810 continue;
811 }
812 for (rg = range; rg; rg = rg->next) {
813 if (!strcmp(rg->path, p->two->path))
814 break;
815 }
816 if (rg)
817 diff_q(&outq, p);
818 else
819 diff_free_filepair(p);
820 }
821 free(diff_queued_diff.queue);
822 diff_queued_diff = outq;
823 }
824
825 static inline int diff_might_be_rename(void)
826 {
827 int i;
828 for (i = 0; i < diff_queued_diff.nr; i++)
829 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
830 /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
831 /* diff_queued_diff.queue[i]->two->path); */
832 return 1;
833 }
834 return 0;
835 }
836
837 static void queue_diffs(struct line_log_data *range,
838 struct diff_options *opt,
839 struct diff_queue_struct *queue,
840 struct commit *commit, struct commit *parent)
841 {
842 struct object_id *tree_oid, *parent_tree_oid;
843
844 assert(commit);
845
846 tree_oid = get_commit_tree_oid(commit);
847 parent_tree_oid = parent ? get_commit_tree_oid(parent) : NULL;
848
849 if (opt->detect_rename &&
850 !same_paths_in_pathspec_and_range(&opt->pathspec, range)) {
851 clear_pathspec(&opt->pathspec);
852 parse_pathspec_from_ranges(&opt->pathspec, range);
853 }
854 DIFF_QUEUE_CLEAR(&diff_queued_diff);
855 diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
856 if (opt->detect_rename && diff_might_be_rename()) {
857 /* must look at the full tree diff to detect renames */
858 clear_pathspec(&opt->pathspec);
859 DIFF_QUEUE_CLEAR(&diff_queued_diff);
860
861 diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
862
863 filter_diffs_for_paths(range, 1);
864 diffcore_std(opt);
865 filter_diffs_for_paths(range, 0);
866 }
867 move_diff_queue(queue, &diff_queued_diff);
868 }
869
870 static char *get_nth_line(long line, unsigned long *ends, void *data)
871 {
872 if (line == 0)
873 return (char *)data;
874 else
875 return (char *)data + ends[line] + 1;
876 }
877
878 static void print_line(const char *prefix, char first,
879 long line, unsigned long *ends, void *data,
880 const char *color, const char *reset, FILE *file)
881 {
882 char *begin = get_nth_line(line, ends, data);
883 char *end = get_nth_line(line+1, ends, data);
884 int had_nl = 0;
885
886 if (end > begin && end[-1] == '\n') {
887 end--;
888 had_nl = 1;
889 }
890
891 fputs(prefix, file);
892 fputs(color, file);
893 putc(first, file);
894 fwrite(begin, 1, end-begin, file);
895 fputs(reset, file);
896 putc('\n', file);
897 if (!had_nl)
898 fputs("\\ No newline at end of file\n", file);
899 }
900
901 static char *output_prefix(struct diff_options *opt)
902 {
903 char *prefix = "";
904
905 if (opt->output_prefix) {
906 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
907 prefix = sb->buf;
908 }
909
910 return prefix;
911 }
912
913 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
914 {
915 unsigned int i, j = 0;
916 long p_lines, t_lines;
917 unsigned long *p_ends = NULL, *t_ends = NULL;
918 struct diff_filepair *pair = range->pair;
919 struct diff_ranges *diff = &range->diff;
920
921 struct diff_options *opt = &rev->diffopt;
922 char *prefix = output_prefix(opt);
923 const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
924 const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
925 const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
926 const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
927 const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
928 const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
929
930 if (!pair || !diff)
931 return;
932
933 if (pair->one->oid_valid)
934 fill_line_ends(rev->diffopt.repo, pair->one, &p_lines, &p_ends);
935 fill_line_ends(rev->diffopt.repo, pair->two, &t_lines, &t_ends);
936
937 fprintf(opt->file, "%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
938 fprintf(opt->file, "%s%s--- %s%s%s\n", prefix, c_meta,
939 pair->one->oid_valid ? "a/" : "",
940 pair->one->oid_valid ? pair->one->path : "/dev/null",
941 c_reset);
942 fprintf(opt->file, "%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
943 for (i = 0; i < range->ranges.nr; i++) {
944 long p_start, p_end;
945 long t_start = range->ranges.ranges[i].start;
946 long t_end = range->ranges.ranges[i].end;
947 long t_cur = t_start;
948 unsigned int j_last;
949
950 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
951 j++;
952 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
953 continue;
954
955 /* Scan ahead to determine the last diff that falls in this range */
956 j_last = j;
957 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
958 j_last++;
959 if (j_last > j)
960 j_last--;
961
962 /*
963 * Compute parent hunk headers: we know that the diff
964 * has the correct line numbers (but not all hunks).
965 * So it suffices to shift the start/end according to
966 * the line numbers of the first/last hunk(s) that
967 * fall in this range.
968 */
969 if (t_start < diff->target.ranges[j].start)
970 p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
971 else
972 p_start = diff->parent.ranges[j].start;
973 if (t_end > diff->target.ranges[j_last].end)
974 p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
975 else
976 p_end = diff->parent.ranges[j_last].end;
977
978 if (!p_start && !p_end) {
979 p_start = -1;
980 p_end = -1;
981 }
982
983 /* Now output a diff hunk for this range */
984 fprintf(opt->file, "%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
985 prefix, c_frag,
986 p_start+1, p_end-p_start, t_start+1, t_end-t_start,
987 c_reset);
988 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
989 int k;
990 for (; t_cur < diff->target.ranges[j].start; t_cur++)
991 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
992 c_context, c_reset, opt->file);
993 for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
994 print_line(prefix, '-', k, p_ends, pair->one->data,
995 c_old, c_reset, opt->file);
996 for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
997 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
998 c_new, c_reset, opt->file);
999 j++;
1000 }
1001 for (; t_cur < t_end; t_cur++)
1002 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
1003 c_context, c_reset, opt->file);
1004 }
1005
1006 free(p_ends);
1007 free(t_ends);
1008 }
1009
1010 /*
1011 * NEEDSWORK: manually building a diff here is not the Right
1012 * Thing(tm). log -L should be built into the diff pipeline.
1013 */
1014 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
1015 {
1016 fprintf(rev->diffopt.file, "%s\n", output_prefix(&rev->diffopt));
1017 while (range) {
1018 dump_diff_hacky_one(rev, range);
1019 range = range->next;
1020 }
1021 }
1022
1023 /*
1024 * Unlike most other functions, this destructively operates on
1025 * 'range'.
1026 */
1027 static int process_diff_filepair(struct rev_info *rev,
1028 struct diff_filepair *pair,
1029 struct line_log_data *range,
1030 struct diff_ranges **diff_out)
1031 {
1032 struct line_log_data *rg = range;
1033 struct range_set tmp;
1034 struct diff_ranges diff;
1035 mmfile_t file_parent, file_target;
1036
1037 assert(pair->two->path);
1038 while (rg) {
1039 assert(rg->path);
1040 if (!strcmp(rg->path, pair->two->path))
1041 break;
1042 rg = rg->next;
1043 }
1044
1045 if (!rg)
1046 return 0;
1047 if (rg->ranges.nr == 0)
1048 return 0;
1049
1050 assert(pair->two->oid_valid);
1051 diff_populate_filespec(rev->diffopt.repo, pair->two, NULL);
1052 file_target.ptr = pair->two->data;
1053 file_target.size = pair->two->size;
1054
1055 if (pair->one->oid_valid) {
1056 diff_populate_filespec(rev->diffopt.repo, pair->one, NULL);
1057 file_parent.ptr = pair->one->data;
1058 file_parent.size = pair->one->size;
1059 } else {
1060 file_parent.ptr = "";
1061 file_parent.size = 0;
1062 }
1063
1064 diff_ranges_init(&diff);
1065 if (collect_diff(&file_parent, &file_target, &diff))
1066 die("unable to generate diff for %s", pair->one->path);
1067
1068 /* NEEDSWORK should apply some heuristics to prevent mismatches */
1069 free(rg->path);
1070 rg->path = xstrdup(pair->one->path);
1071
1072 range_set_init(&tmp, 0);
1073 range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1074 range_set_release(&rg->ranges);
1075 range_set_move(&rg->ranges, &tmp);
1076
1077 diff_ranges_release(&diff);
1078
1079 return ((*diff_out)->parent.nr > 0);
1080 }
1081
1082 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1083 {
1084 struct diff_filepair *new_filepair = xmalloc(sizeof(struct diff_filepair));
1085 new_filepair->one = pair->one;
1086 new_filepair->two = pair->two;
1087 new_filepair->one->count++;
1088 new_filepair->two->count++;
1089 return new_filepair;
1090 }
1091
1092 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1093 {
1094 for (int i = 0; i < n; i++)
1095 diff_free_queue(&dq[i]);
1096 free(dq);
1097 }
1098
1099 static int process_all_files(struct line_log_data **range_out,
1100 struct rev_info *rev,
1101 struct diff_queue_struct *queue,
1102 struct line_log_data *range)
1103 {
1104 int i, changed = 0;
1105
1106 *range_out = line_log_data_copy(range);
1107
1108 for (i = 0; i < queue->nr; i++) {
1109 struct diff_ranges *pairdiff = NULL;
1110 struct diff_filepair *pair = queue->queue[i];
1111 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1112 /*
1113 * Store away the diff for later output. We
1114 * tuck it in the ranges we got as _input_,
1115 * since that's the commit that caused the
1116 * diff.
1117 *
1118 * NEEDSWORK not enough when we get around to
1119 * doing something interesting with merges;
1120 * currently each invocation on a merge parent
1121 * trashes the previous one's diff.
1122 *
1123 * NEEDSWORK tramples over data structures not owned here
1124 */
1125 struct line_log_data *rg = range;
1126 changed++;
1127 while (rg && strcmp(rg->path, pair->two->path))
1128 rg = rg->next;
1129 assert(rg);
1130 rg->pair = diff_filepair_dup(queue->queue[i]);
1131 memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1132 }
1133 free(pairdiff);
1134 }
1135
1136 return changed;
1137 }
1138
1139 int line_log_print(struct rev_info *rev, struct commit *commit)
1140 {
1141
1142 show_log(rev);
1143 if (!(rev->diffopt.output_format & DIFF_FORMAT_NO_OUTPUT)) {
1144 struct line_log_data *range = lookup_line_range(rev, commit);
1145 dump_diff_hacky(rev, range);
1146 }
1147 return 1;
1148 }
1149
1150 static int bloom_filter_check(struct rev_info *rev,
1151 struct commit *commit,
1152 struct line_log_data *range)
1153 {
1154 struct bloom_filter *filter;
1155 struct bloom_key key;
1156 int result = 0;
1157
1158 if (!commit->parents)
1159 return 1;
1160
1161 if (!rev->bloom_filter_settings ||
1162 !(filter = get_bloom_filter(rev->repo, commit)))
1163 return 1;
1164
1165 if (!range)
1166 return 0;
1167
1168 while (!result && range) {
1169 fill_bloom_key(range->path, strlen(range->path), &key, rev->bloom_filter_settings);
1170
1171 if (bloom_filter_contains(filter, &key, rev->bloom_filter_settings))
1172 result = 1;
1173
1174 clear_bloom_key(&key);
1175 range = range->next;
1176 }
1177
1178 return result;
1179 }
1180
1181 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1182 struct line_log_data *range)
1183 {
1184 struct commit *parent = NULL;
1185 struct diff_queue_struct queue;
1186 struct line_log_data *parent_range;
1187 int changed;
1188
1189 if (commit->parents)
1190 parent = commit->parents->item;
1191
1192 queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1193 changed = process_all_files(&parent_range, rev, &queue, range);
1194
1195 if (parent)
1196 add_line_range(rev, parent, parent_range);
1197 free_line_log_data(parent_range);
1198 diff_free_queue(&queue);
1199 return changed;
1200 }
1201
1202 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1203 struct line_log_data *range)
1204 {
1205 struct diff_queue_struct *diffqueues;
1206 struct line_log_data **cand;
1207 struct commit **parents;
1208 struct commit_list *p;
1209 int i;
1210 int nparents = commit_list_count(commit->parents);
1211
1212 if (nparents > 1 && rev->first_parent_only)
1213 nparents = 1;
1214
1215 ALLOC_ARRAY(diffqueues, nparents);
1216 ALLOC_ARRAY(cand, nparents);
1217 ALLOC_ARRAY(parents, nparents);
1218
1219 p = commit->parents;
1220 for (i = 0; i < nparents; i++) {
1221 parents[i] = p->item;
1222 p = p->next;
1223 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1224 }
1225
1226 for (i = 0; i < nparents; i++) {
1227 int changed;
1228 cand[i] = NULL;
1229 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1230 if (!changed) {
1231 /*
1232 * This parent can take all the blame, so we
1233 * don't follow any other path in history
1234 */
1235 add_line_range(rev, parents[i], cand[i]);
1236 clear_commit_line_range(rev, commit);
1237 commit_list_append(parents[i], &commit->parents);
1238 free(parents);
1239 free(cand);
1240 free_diffqueues(nparents, diffqueues);
1241 /* NEEDSWORK leaking like a sieve */
1242 return 0;
1243 }
1244 }
1245
1246 /*
1247 * No single parent took the blame. We add the candidates
1248 * from the above loop to the parents.
1249 */
1250 for (i = 0; i < nparents; i++) {
1251 add_line_range(rev, parents[i], cand[i]);
1252 }
1253
1254 clear_commit_line_range(rev, commit);
1255 free(parents);
1256 free(cand);
1257 free_diffqueues(nparents, diffqueues);
1258 return 1;
1259
1260 /* NEEDSWORK evil merge detection stuff */
1261 /* NEEDSWORK leaking like a sieve */
1262 }
1263
1264 int line_log_process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1265 {
1266 struct line_log_data *range = lookup_line_range(rev, commit);
1267 int changed = 0;
1268
1269 if (range) {
1270 if (commit->parents && !bloom_filter_check(rev, commit, range)) {
1271 struct line_log_data *prange = line_log_data_copy(range);
1272 add_line_range(rev, commit->parents->item, prange);
1273 clear_commit_line_range(rev, commit);
1274 } else if (!commit->parents || !commit->parents->next)
1275 changed = process_ranges_ordinary_commit(rev, commit, range);
1276 else
1277 changed = process_ranges_merge_commit(rev, commit, range);
1278 }
1279
1280 if (!changed)
1281 commit->object.flags |= TREESAME;
1282
1283 return changed;
1284 }
1285
1286 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev UNUSED,
1287 struct commit **pp)
1288 {
1289 for (;;) {
1290 struct commit *p = *pp;
1291 if (p->parents && p->parents->next)
1292 return rewrite_one_ok;
1293 if (p->object.flags & UNINTERESTING)
1294 return rewrite_one_ok;
1295 if (!(p->object.flags & TREESAME))
1296 return rewrite_one_ok;
1297 if (!p->parents)
1298 return rewrite_one_noparents;
1299 *pp = p->parents->item;
1300 }
1301 }
1302
1303 int line_log_filter(struct rev_info *rev)
1304 {
1305 struct commit *commit;
1306 struct commit_list *list = rev->commits;
1307 struct commit_list *out = NULL, **pp = &out;
1308
1309 while (list) {
1310 struct commit_list *to_free = NULL;
1311 commit = list->item;
1312 if (line_log_process_ranges_arbitrary_commit(rev, commit)) {
1313 *pp = list;
1314 pp = &list->next;
1315 } else
1316 to_free = list;
1317 list = list->next;
1318 free(to_free);
1319 }
1320 *pp = NULL;
1321
1322 for (list = out; list; list = list->next)
1323 rewrite_parents(rev, list->item, line_log_rewrite_one);
1324
1325 rev->commits = out;
1326
1327 return 0;
1328 }
1329
1330 static void free_void_line_log_data(void *data)
1331 {
1332 free_line_log_data(data);
1333 }
1334
1335 void line_log_free(struct rev_info *rev)
1336 {
1337 clear_decoration(&rev->line_log_data, free_void_line_log_data);
1338 }
The Best VCS Around