?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'bc/more-git-var'
[git/debian.git] / line-log.c
blob2eff914bf369ee0d794294d8632e42bf58606f1d
1 #include "git-compat-util.h"
2 #include "alloc.h"
3 #include "line-range.h"
4 #include "hex.h"
5 #include "tag.h"
6 #include "blob.h"
7 #include "tree.h"
8 #include "diff.h"
9 #include "commit.h"
10 #include "decorate.h"
11 #include "repository.h"
12 #include "revision.h"
13 #include "xdiff-interface.h"
14 #include "strbuf.h"
15 #include "log-tree.h"
16 #include "graph.h"
17 #include "userdiff.h"
18 #include "line-log.h"
19 #include "setup.h"
20 #include "strvec.h"
21 #include "bloom.h"
22 #include "tree-walk.h"
23
24 static void range_set_grow(struct range_set *rs, size_t extra)
25 {
26 ALLOC_GROW(rs->ranges, rs->nr + extra, rs->alloc);
27 }
28
29 /* Either initialization would be fine */
30 #define RANGE_SET_INIT {0}
31
32 void range_set_init(struct range_set *rs, size_t prealloc)
33 {
34 rs->alloc = rs->nr = 0;
35 rs->ranges = NULL;
36 if (prealloc)
37 range_set_grow(rs, prealloc);
38 }
39
40 void range_set_release(struct range_set *rs)
41 {
42 FREE_AND_NULL(rs->ranges);
43 rs->alloc = rs->nr = 0;
44 }
45
46 /* dst must be uninitialized! */
47 static void range_set_copy(struct range_set *dst, struct range_set *src)
48 {
49 range_set_init(dst, src->nr);
50 COPY_ARRAY(dst->ranges, src->ranges, src->nr);
51 dst->nr = src->nr;
52 }
53
54 static void range_set_move(struct range_set *dst, struct range_set *src)
55 {
56 range_set_release(dst);
57 dst->ranges = src->ranges;
58 dst->nr = src->nr;
59 dst->alloc = src->alloc;
60 src->ranges = NULL;
61 src->alloc = src->nr = 0;
62 }
63
64 /* tack on a _new_ range _at the end_ */
65 void range_set_append_unsafe(struct range_set *rs, long a, long b)
66 {
67 assert(a <= b);
68 range_set_grow(rs, 1);
69 rs->ranges[rs->nr].start = a;
70 rs->ranges[rs->nr].end = b;
71 rs->nr++;
72 }
73
74 void range_set_append(struct range_set *rs, long a, long b)
75 {
76 assert(rs->nr == 0 || rs->ranges[rs->nr-1].end <= a);
77 range_set_append_unsafe(rs, a, b);
78 }
79
80 static int range_cmp(const void *_r, const void *_s)
81 {
82 const struct range *r = _r;
83 const struct range *s = _s;
84
85 /* this could be simply 'return r.start-s.start', but for the types */
86 if (r->start == s->start)
87 return 0;
88 if (r->start < s->start)
89 return -1;
90 return 1;
91 }
92
93 /*
94 * Check that the ranges are non-empty, sorted and non-overlapping
95 */
96 static void range_set_check_invariants(struct range_set *rs)
97 {
98 unsigned int i;
99
100 if (!rs)
101 return;
102
103 if (rs->nr)
104 assert(rs->ranges[0].start < rs->ranges[0].end);
105
106 for (i = 1; i < rs->nr; i++) {
107 assert(rs->ranges[i-1].end < rs->ranges[i].start);
108 assert(rs->ranges[i].start < rs->ranges[i].end);
109 }
110 }
111
112 /*
113 * In-place pass of sorting and merging the ranges in the range set,
114 * to establish the invariants when we get the ranges from the user
115 */
116 void sort_and_merge_range_set(struct range_set *rs)
117 {
118 unsigned int i;
119 unsigned int o = 0; /* output cursor */
120
121 QSORT(rs->ranges, rs->nr, range_cmp);
122
123 for (i = 0; i < rs->nr; i++) {
124 if (rs->ranges[i].start == rs->ranges[i].end)
125 continue;
126 if (o > 0 && rs->ranges[i].start <= rs->ranges[o-1].end) {
127 if (rs->ranges[o-1].end < rs->ranges[i].end)
128 rs->ranges[o-1].end = rs->ranges[i].end;
129 } else {
130 rs->ranges[o].start = rs->ranges[i].start;
131 rs->ranges[o].end = rs->ranges[i].end;
132 o++;
133 }
134 }
135 assert(o <= rs->nr);
136 rs->nr = o;
137
138 range_set_check_invariants(rs);
139 }
140
141 /*
142 * Union of range sets (i.e., sets of line numbers). Used to merge
143 * them when searches meet at a common ancestor.
144 *
145 * This is also where the ranges are consolidated into canonical form:
146 * overlapping and adjacent ranges are merged, and empty ranges are
147 * removed.
148 */
149 static void range_set_union(struct range_set *out,
150 struct range_set *a, struct range_set *b)
151 {
152 unsigned int i = 0, j = 0;
153 struct range *ra = a->ranges;
154 struct range *rb = b->ranges;
155 /* cannot make an alias of out->ranges: it may change during grow */
156
157 assert(out->nr == 0);
158 while (i < a->nr || j < b->nr) {
159 struct range *new_range;
160 if (i < a->nr && j < b->nr) {
161 if (ra[i].start < rb[j].start)
162 new_range = &ra[i++];
163 else if (ra[i].start > rb[j].start)
164 new_range = &rb[j++];
165 else if (ra[i].end < rb[j].end)
166 new_range = &ra[i++];
167 else
168 new_range = &rb[j++];
169 } else if (i < a->nr) /* b exhausted */
170 new_range = &ra[i++];
171 else /* a exhausted */
172 new_range = &rb[j++];
173 if (new_range->start == new_range->end)
174 ; /* empty range */
175 else if (!out->nr || out->ranges[out->nr-1].end < new_range->start) {
176 range_set_grow(out, 1);
177 out->ranges[out->nr].start = new_range->start;
178 out->ranges[out->nr].end = new_range->end;
179 out->nr++;
180 } else if (out->ranges[out->nr-1].end < new_range->end) {
181 out->ranges[out->nr-1].end = new_range->end;
182 }
183 }
184 }
185
186 /*
187 * Difference of range sets (out = a \ b). Pass the "interesting"
188 * ranges as 'a' and the target side of the diff as 'b': it removes
189 * the ranges for which the commit is responsible.
190 */
191 static void range_set_difference(struct range_set *out,
192 struct range_set *a, struct range_set *b)
193 {
194 unsigned int i, j = 0;
195 for (i = 0; i < a->nr; i++) {
196 long start = a->ranges[i].start;
197 long end = a->ranges[i].end;
198 while (start < end) {
199 while (j < b->nr && start >= b->ranges[j].end)
200 /*
201 * a: |-------
202 * b: ------|
203 */
204 j++;
205 if (j >= b->nr || end < b->ranges[j].start) {
206 /*
207 * b exhausted, or
208 * a: ----|
209 * b: |----
210 */
211 range_set_append(out, start, end);
212 break;
213 }
214 if (start >= b->ranges[j].start) {
215 /*
216 * a: |--????
217 * b: |------|
218 */
219 start = b->ranges[j].end;
220 } else if (end > b->ranges[j].start) {
221 /*
222 * a: |-----|
223 * b: |--?????
224 */
225 if (start < b->ranges[j].start)
226 range_set_append(out, start, b->ranges[j].start);
227 start = b->ranges[j].end;
228 }
229 }
230 }
231 }
232
233 static void diff_ranges_init(struct diff_ranges *diff)
234 {
235 range_set_init(&diff->parent, 0);
236 range_set_init(&diff->target, 0);
237 }
238
239 static void diff_ranges_release(struct diff_ranges *diff)
240 {
241 range_set_release(&diff->parent);
242 range_set_release(&diff->target);
243 }
244
245 static void line_log_data_init(struct line_log_data *r)
246 {
247 memset(r, 0, sizeof(struct line_log_data));
248 range_set_init(&r->ranges, 0);
249 }
250
251 static void line_log_data_clear(struct line_log_data *r)
252 {
253 range_set_release(&r->ranges);
254 if (r->pair)
255 diff_free_filepair(r->pair);
256 }
257
258 static void free_line_log_data(struct line_log_data *r)
259 {
260 while (r) {
261 struct line_log_data *next = r->next;
262 line_log_data_clear(r);
263 free(r);
264 r = next;
265 }
266 }
267
268 static struct line_log_data *
269 search_line_log_data(struct line_log_data *list, const char *path,
270 struct line_log_data **insertion_point)
271 {
272 struct line_log_data *p = list;
273 if (insertion_point)
274 *insertion_point = NULL;
275 while (p) {
276 int cmp = strcmp(p->path, path);
277 if (!cmp)
278 return p;
279 if (insertion_point && cmp < 0)
280 *insertion_point = p;
281 p = p->next;
282 }
283 return NULL;
284 }
285
286 /*
287 * Note: takes ownership of 'path', which happens to be what the only
288 * caller needs.
289 */
290 static void line_log_data_insert(struct line_log_data **list,
291 char *path,
292 long begin, long end)
293 {
294 struct line_log_data *ip;
295 struct line_log_data *p = search_line_log_data(*list, path, &ip);
296
297 if (p) {
298 range_set_append_unsafe(&p->ranges, begin, end);
299 free(path);
300 return;
301 }
302
303 CALLOC_ARRAY(p, 1);
304 p->path = path;
305 range_set_append(&p->ranges, begin, end);
306 if (ip) {
307 p->next = ip->next;
308 ip->next = p;
309 } else {
310 p->next = *list;
311 *list = p;
312 }
313 }
314
315 struct collect_diff_cbdata {
316 struct diff_ranges *diff;
317 };
318
319 static int collect_diff_cb(long start_a, long count_a,
320 long start_b, long count_b,
321 void *data)
322 {
323 struct collect_diff_cbdata *d = data;
324
325 if (count_a >= 0)
326 range_set_append(&d->diff->parent, start_a, start_a + count_a);
327 if (count_b >= 0)
328 range_set_append(&d->diff->target, start_b, start_b + count_b);
329
330 return 0;
331 }
332
333 static int collect_diff(mmfile_t *parent, mmfile_t *target, struct diff_ranges *out)
334 {
335 struct collect_diff_cbdata cbdata = {NULL};
336 xpparam_t xpp;
337 xdemitconf_t xecfg;
338 xdemitcb_t ecb;
339
340 memset(&xpp, 0, sizeof(xpp));
341 memset(&xecfg, 0, sizeof(xecfg));
342 xecfg.ctxlen = xecfg.interhunkctxlen = 0;
343
344 cbdata.diff = out;
345 xecfg.hunk_func = collect_diff_cb;
346 memset(&ecb, 0, sizeof(ecb));
347 ecb.priv = &cbdata;
348 return xdi_diff(parent, target, &xpp, &xecfg, &ecb);
349 }
350
351 /*
352 * These are handy for debugging. Removing them with #if 0 silences
353 * the "unused function" warning.
354 */
355 #if 0
356 static void dump_range_set(struct range_set *rs, const char *desc)
357 {
358 int i;
359 printf("range set %s (%d items):\n", desc, rs->nr);
360 for (i = 0; i < rs->nr; i++)
361 printf("\t[%ld,%ld]\n", rs->ranges[i].start, rs->ranges[i].end);
362 }
363
364 static void dump_line_log_data(struct line_log_data *r)
365 {
366 char buf[4096];
367 while (r) {
368 snprintf(buf, 4096, "file %s\n", r->path);
369 dump_range_set(&r->ranges, buf);
370 r = r->next;
371 }
372 }
373
374 static void dump_diff_ranges(struct diff_ranges *diff, const char *desc)
375 {
376 int i;
377 assert(diff->parent.nr == diff->target.nr);
378 printf("diff ranges %s (%d items):\n", desc, diff->parent.nr);
379 printf("\tparent\ttarget\n");
380 for (i = 0; i < diff->parent.nr; i++) {
381 printf("\t[%ld,%ld]\t[%ld,%ld]\n",
382 diff->parent.ranges[i].start,
383 diff->parent.ranges[i].end,
384 diff->target.ranges[i].start,
385 diff->target.ranges[i].end);
386 }
387 }
388 #endif
389
390
391 static int ranges_overlap(struct range *a, struct range *b)
392 {
393 return !(a->end <= b->start || b->end <= a->start);
394 }
395
396 /*
397 * Given a diff and the set of interesting ranges, determine all hunks
398 * of the diff which touch (overlap) at least one of the interesting
399 * ranges in the target.
400 */
401 static void diff_ranges_filter_touched(struct diff_ranges *out,
402 struct diff_ranges *diff,
403 struct range_set *rs)
404 {
405 unsigned int i, j = 0;
406
407 assert(out->target.nr == 0);
408
409 for (i = 0; i < diff->target.nr; i++) {
410 while (diff->target.ranges[i].start > rs->ranges[j].end) {
411 j++;
412 if (j == rs->nr)
413 return;
414 }
415 if (ranges_overlap(&diff->target.ranges[i], &rs->ranges[j])) {
416 range_set_append(&out->parent,
417 diff->parent.ranges[i].start,
418 diff->parent.ranges[i].end);
419 range_set_append(&out->target,
420 diff->target.ranges[i].start,
421 diff->target.ranges[i].end);
422 }
423 }
424 }
425
426 /*
427 * Adjust the line counts in 'rs' to account for the lines
428 * added/removed in the diff.
429 */
430 static void range_set_shift_diff(struct range_set *out,
431 struct range_set *rs,
432 struct diff_ranges *diff)
433 {
434 unsigned int i, j = 0;
435 long offset = 0;
436 struct range *src = rs->ranges;
437 struct range *target = diff->target.ranges;
438 struct range *parent = diff->parent.ranges;
439
440 for (i = 0; i < rs->nr; i++) {
441 while (j < diff->target.nr && src[i].start >= target[j].start) {
442 offset += (parent[j].end-parent[j].start)
443 - (target[j].end-target[j].start);
444 j++;
445 }
446 range_set_append(out, src[i].start+offset, src[i].end+offset);
447 }
448 }
449
450 /*
451 * Given a diff and the set of interesting ranges, map the ranges
452 * across the diff. That is: observe that the target commit takes
453 * blame for all the + (target-side) ranges. So for every pair of
454 * ranges in the diff that was touched, we remove the latter and add
455 * its parent side.
456 */
457 static void range_set_map_across_diff(struct range_set *out,
458 struct range_set *rs,
459 struct diff_ranges *diff,
460 struct diff_ranges **touched_out)
461 {
462 struct diff_ranges *touched = xmalloc(sizeof(*touched));
463 struct range_set tmp1 = RANGE_SET_INIT;
464 struct range_set tmp2 = RANGE_SET_INIT;
465
466 diff_ranges_init(touched);
467 diff_ranges_filter_touched(touched, diff, rs);
468 range_set_difference(&tmp1, rs, &touched->target);
469 range_set_shift_diff(&tmp2, &tmp1, diff);
470 range_set_union(out, &tmp2, &touched->parent);
471 range_set_release(&tmp1);
472 range_set_release(&tmp2);
473
474 *touched_out = touched;
475 }
476
477 static struct commit *check_single_commit(struct rev_info *revs)
478 {
479 struct object *commit = NULL;
480 int found = -1;
481 int i;
482
483 for (i = 0; i < revs->pending.nr; i++) {
484 struct object *obj = revs->pending.objects[i].item;
485 if (obj->flags & UNINTERESTING)
486 continue;
487 obj = deref_tag(revs->repo, obj, NULL, 0);
488 if (!obj || obj->type != OBJ_COMMIT)
489 die("Non commit %s?", revs->pending.objects[i].name);
490 if (commit)
491 die("More than one commit to dig from: %s and %s?",
492 revs->pending.objects[i].name,
493 revs->pending.objects[found].name);
494 commit = obj;
495 found = i;
496 }
497
498 if (!commit)
499 die("No commit specified?");
500
501 return (struct commit *) commit;
502 }
503
504 static void fill_blob_sha1(struct repository *r, struct commit *commit,
505 struct diff_filespec *spec)
506 {
507 unsigned short mode;
508 struct object_id oid;
509
510 if (get_tree_entry(r, &commit->object.oid, spec->path, &oid, &mode))
511 die("There is no path %s in the commit", spec->path);
512 fill_filespec(spec, &oid, 1, mode);
513
514 return;
515 }
516
517 static void fill_line_ends(struct repository *r,
518 struct diff_filespec *spec,
519 long *lines,
520 unsigned long **line_ends)
521 {
522 int num = 0, size = 50;
523 long cur = 0;
524 unsigned long *ends = NULL;
525 char *data = NULL;
526
527 if (diff_populate_filespec(r, spec, NULL))
528 die("Cannot read blob %s", oid_to_hex(&spec->oid));
529
530 ALLOC_ARRAY(ends, size);
531 ends[cur++] = 0;
532 data = spec->data;
533 while (num < spec->size) {
534 if (data[num] == '\n' || num == spec->size - 1) {
535 ALLOC_GROW(ends, (cur + 1), size);
536 ends[cur++] = num;
537 }
538 num++;
539 }
540
541 /* shrink the array to fit the elements */
542 REALLOC_ARRAY(ends, cur);
543 *lines = cur-1;
544 *line_ends = ends;
545 }
546
547 struct nth_line_cb {
548 struct diff_filespec *spec;
549 long lines;
550 unsigned long *line_ends;
551 };
552
553 static const char *nth_line(void *data, long line)
554 {
555 struct nth_line_cb *d = data;
556 assert(d && line <= d->lines);
557 assert(d->spec && d->spec->data);
558
559 if (line == 0)
560 return (char *)d->spec->data;
561 else
562 return (char *)d->spec->data + d->line_ends[line] + 1;
563 }
564
565 static struct line_log_data *
566 parse_lines(struct repository *r, struct commit *commit,
567 const char *prefix, struct string_list *args)
568 {
569 long lines = 0;
570 unsigned long *ends = NULL;
571 struct nth_line_cb cb_data;
572 struct string_list_item *item;
573 struct line_log_data *ranges = NULL;
574 struct line_log_data *p;
575
576 for_each_string_list_item(item, args) {
577 const char *name_part, *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 }
622
623 for (p = ranges; p; p = p->next)
624 sort_and_merge_range_set(&p->ranges);
625
626 return ranges;
627 }
628
629 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
630 {
631 struct line_log_data *ret = xmalloc(sizeof(*ret));
632
633 assert(r);
634 line_log_data_init(ret);
635 range_set_copy(&ret->ranges, &r->ranges);
636
637 ret->path = xstrdup(r->path);
638
639 return ret;
640 }
641
642 static struct line_log_data *
643 line_log_data_copy(struct line_log_data *r)
644 {
645 struct line_log_data *ret = NULL;
646 struct line_log_data *tmp = NULL, *prev = NULL;
647
648 assert(r);
649 ret = tmp = prev = line_log_data_copy_one(r);
650 r = r->next;
651 while (r) {
652 tmp = line_log_data_copy_one(r);
653 prev->next = tmp;
654 prev = tmp;
655 r = r->next;
656 }
657
658 return ret;
659 }
660
661 /* merge two range sets across files */
662 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
663 struct line_log_data *b)
664 {
665 struct line_log_data *head = NULL, **pp = &head;
666
667 while (a || b) {
668 struct line_log_data *src;
669 struct line_log_data *src2 = NULL;
670 struct line_log_data *d;
671 int cmp;
672 if (!a)
673 cmp = 1;
674 else if (!b)
675 cmp = -1;
676 else
677 cmp = strcmp(a->path, b->path);
678 if (cmp < 0) {
679 src = a;
680 a = a->next;
681 } else if (cmp == 0) {
682 src = a;
683 a = a->next;
684 src2 = b;
685 b = b->next;
686 } else {
687 src = b;
688 b = b->next;
689 }
690 d = xmalloc(sizeof(struct line_log_data));
691 line_log_data_init(d);
692 d->path = xstrdup(src->path);
693 *pp = d;
694 pp = &d->next;
695 if (src2)
696 range_set_union(&d->ranges, &src->ranges, &src2->ranges);
697 else
698 range_set_copy(&d->ranges, &src->ranges);
699 }
700
701 return head;
702 }
703
704 static void add_line_range(struct rev_info *revs, struct commit *commit,
705 struct line_log_data *range)
706 {
707 struct line_log_data *old_line = NULL;
708 struct line_log_data *new_line = NULL;
709
710 old_line = lookup_decoration(&revs->line_log_data, &commit->object);
711 if (old_line && range) {
712 new_line = line_log_data_merge(old_line, range);
713 free_line_log_data(old_line);
714 } else if (range)
715 new_line = line_log_data_copy(range);
716
717 if (new_line)
718 add_decoration(&revs->line_log_data, &commit->object, new_line);
719 }
720
721 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
722 {
723 struct line_log_data *r;
724 r = lookup_decoration(&revs->line_log_data, &commit->object);
725 if (!r)
726 return;
727 free_line_log_data(r);
728 add_decoration(&revs->line_log_data, &commit->object, NULL);
729 }
730
731 static struct line_log_data *lookup_line_range(struct rev_info *revs,
732 struct commit *commit)
733 {
734 struct line_log_data *ret = NULL;
735 struct line_log_data *d;
736
737 ret = lookup_decoration(&revs->line_log_data, &commit->object);
738
739 for (d = ret; d; d = d->next)
740 range_set_check_invariants(&d->ranges);
741
742 return ret;
743 }
744
745 static int same_paths_in_pathspec_and_range(struct pathspec *pathspec,
746 struct line_log_data *range)
747 {
748 int i;
749 struct line_log_data *r;
750
751 for (i = 0, r = range; i < pathspec->nr && r; i++, r = r->next)
752 if (strcmp(pathspec->items[i].match, r->path))
753 return 0;
754 if (i < pathspec->nr || r)
755 /* different number of pathspec items and ranges */
756 return 0;
757
758 return 1;
759 }
760
761 static void parse_pathspec_from_ranges(struct pathspec *pathspec,
762 struct line_log_data *range)
763 {
764 struct line_log_data *r;
765 struct strvec array = STRVEC_INIT;
766 const char **paths;
767
768 for (r = range; r; r = r->next)
769 strvec_push(&array, r->path);
770 paths = strvec_detach(&array);
771
772 parse_pathspec(pathspec, 0, PATHSPEC_PREFER_FULL, "", paths);
773 /* strings are now owned by pathspec */
774 free(paths);
775 }
776
777 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
778 {
779 struct commit *commit = NULL;
780 struct line_log_data *range;
781
782 commit = check_single_commit(rev);
783 range = parse_lines(rev->diffopt.repo, commit, prefix, args);
784 add_line_range(rev, commit, range);
785
786 parse_pathspec_from_ranges(&rev->diffopt.pathspec, range);
787 }
788
789 static void move_diff_queue(struct diff_queue_struct *dst,
790 struct diff_queue_struct *src)
791 {
792 assert(src != dst);
793 memcpy(dst, src, sizeof(struct diff_queue_struct));
794 DIFF_QUEUE_CLEAR(src);
795 }
796
797 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
798 {
799 int i;
800 struct diff_queue_struct outq;
801 DIFF_QUEUE_CLEAR(&outq);
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 char *output_prefix(struct diff_options *opt)
904 {
905 char *prefix = "";
906
907 if (opt->output_prefix) {
908 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
909 prefix = sb->buf;
910 }
911
912 return prefix;
913 }
914
915 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
916 {
917 unsigned int i, j = 0;
918 long p_lines, t_lines;
919 unsigned long *p_ends = NULL, *t_ends = NULL;
920 struct diff_filepair *pair = range->pair;
921 struct diff_ranges *diff = &range->diff;
922
923 struct diff_options *opt = &rev->diffopt;
924 char *prefix = output_prefix(opt);
925 const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
926 const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
927 const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
928 const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
929 const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
930 const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
931
932 if (!pair || !diff)
933 return;
934
935 if (pair->one->oid_valid)
936 fill_line_ends(rev->diffopt.repo, pair->one, &p_lines, &p_ends);
937 fill_line_ends(rev->diffopt.repo, pair->two, &t_lines, &t_ends);
938
939 fprintf(opt->file, "%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
940 fprintf(opt->file, "%s%s--- %s%s%s\n", prefix, c_meta,
941 pair->one->oid_valid ? "a/" : "",
942 pair->one->oid_valid ? pair->one->path : "/dev/null",
943 c_reset);
944 fprintf(opt->file, "%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
945 for (i = 0; i < range->ranges.nr; i++) {
946 long p_start, p_end;
947 long t_start = range->ranges.ranges[i].start;
948 long t_end = range->ranges.ranges[i].end;
949 long t_cur = t_start;
950 unsigned int j_last;
951
952 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
953 j++;
954 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
955 continue;
956
957 /* Scan ahead to determine the last diff that falls in this range */
958 j_last = j;
959 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
960 j_last++;
961 if (j_last > j)
962 j_last--;
963
964 /*
965 * Compute parent hunk headers: we know that the diff
966 * has the correct line numbers (but not all hunks).
967 * So it suffices to shift the start/end according to
968 * the line numbers of the first/last hunk(s) that
969 * fall in this range.
970 */
971 if (t_start < diff->target.ranges[j].start)
972 p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
973 else
974 p_start = diff->parent.ranges[j].start;
975 if (t_end > diff->target.ranges[j_last].end)
976 p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
977 else
978 p_end = diff->parent.ranges[j_last].end;
979
980 if (!p_start && !p_end) {
981 p_start = -1;
982 p_end = -1;
983 }
984
985 /* Now output a diff hunk for this range */
986 fprintf(opt->file, "%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
987 prefix, c_frag,
988 p_start+1, p_end-p_start, t_start+1, t_end-t_start,
989 c_reset);
990 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
991 int k;
992 for (; t_cur < diff->target.ranges[j].start; t_cur++)
993 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
994 c_context, c_reset, opt->file);
995 for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
996 print_line(prefix, '-', k, p_ends, pair->one->data,
997 c_old, c_reset, opt->file);
998 for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
999 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
1000 c_new, c_reset, opt->file);
1001 j++;
1002 }
1003 for (; t_cur < t_end; t_cur++)
1004 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
1005 c_context, c_reset, opt->file);
1006 }
1007
1008 free(p_ends);
1009 free(t_ends);
1010 }
1011
1012 /*
1013 * NEEDSWORK: manually building a diff here is not the Right
1014 * Thing(tm). log -L should be built into the diff pipeline.
1015 */
1016 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
1017 {
1018 fprintf(rev->diffopt.file, "%s\n", output_prefix(&rev->diffopt));
1019 while (range) {
1020 dump_diff_hacky_one(rev, range);
1021 range = range->next;
1022 }
1023 }
1024
1025 /*
1026 * Unlike most other functions, this destructively operates on
1027 * 'range'.
1028 */
1029 static int process_diff_filepair(struct rev_info *rev,
1030 struct diff_filepair *pair,
1031 struct line_log_data *range,
1032 struct diff_ranges **diff_out)
1033 {
1034 struct line_log_data *rg = range;
1035 struct range_set tmp;
1036 struct diff_ranges diff;
1037 mmfile_t file_parent, file_target;
1038
1039 assert(pair->two->path);
1040 while (rg) {
1041 assert(rg->path);
1042 if (!strcmp(rg->path, pair->two->path))
1043 break;
1044 rg = rg->next;
1045 }
1046
1047 if (!rg)
1048 return 0;
1049 if (rg->ranges.nr == 0)
1050 return 0;
1051
1052 assert(pair->two->oid_valid);
1053 diff_populate_filespec(rev->diffopt.repo, pair->two, NULL);
1054 file_target.ptr = pair->two->data;
1055 file_target.size = pair->two->size;
1056
1057 if (pair->one->oid_valid) {
1058 diff_populate_filespec(rev->diffopt.repo, pair->one, NULL);
1059 file_parent.ptr = pair->one->data;
1060 file_parent.size = pair->one->size;
1061 } else {
1062 file_parent.ptr = "";
1063 file_parent.size = 0;
1064 }
1065
1066 diff_ranges_init(&diff);
1067 if (collect_diff(&file_parent, &file_target, &diff))
1068 die("unable to generate diff for %s", pair->one->path);
1069
1070 /* NEEDSWORK should apply some heuristics to prevent mismatches */
1071 free(rg->path);
1072 rg->path = xstrdup(pair->one->path);
1073
1074 range_set_init(&tmp, 0);
1075 range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1076 range_set_release(&rg->ranges);
1077 range_set_move(&rg->ranges, &tmp);
1078
1079 diff_ranges_release(&diff);
1080
1081 return ((*diff_out)->parent.nr > 0);
1082 }
1083
1084 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1085 {
1086 struct diff_filepair *new_filepair = xmalloc(sizeof(struct diff_filepair));
1087 new_filepair->one = pair->one;
1088 new_filepair->two = pair->two;
1089 new_filepair->one->count++;
1090 new_filepair->two->count++;
1091 return new_filepair;
1092 }
1093
1094 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1095 {
1096 for (int i = 0; i < n; i++)
1097 diff_free_queue(&dq[i]);
1098 free(dq);
1099 }
1100
1101 static int process_all_files(struct line_log_data **range_out,
1102 struct rev_info *rev,
1103 struct diff_queue_struct *queue,
1104 struct line_log_data *range)
1105 {
1106 int i, changed = 0;
1107
1108 *range_out = line_log_data_copy(range);
1109
1110 for (i = 0; i < queue->nr; i++) {
1111 struct diff_ranges *pairdiff = NULL;
1112 struct diff_filepair *pair = queue->queue[i];
1113 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1114 /*
1115 * Store away the diff for later output. We
1116 * tuck it in the ranges we got as _input_,
1117 * since that's the commit that caused the
1118 * diff.
1119 *
1120 * NEEDSWORK not enough when we get around to
1121 * doing something interesting with merges;
1122 * currently each invocation on a merge parent
1123 * trashes the previous one's diff.
1124 *
1125 * NEEDSWORK tramples over data structures not owned here
1126 */
1127 struct line_log_data *rg = range;
1128 changed++;
1129 while (rg && strcmp(rg->path, pair->two->path))
1130 rg = rg->next;
1131 assert(rg);
1132 rg->pair = diff_filepair_dup(queue->queue[i]);
1133 memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1134 }
1135 free(pairdiff);
1136 }
1137
1138 return changed;
1139 }
1140
1141 int line_log_print(struct rev_info *rev, struct commit *commit)
1142 {
1143
1144 show_log(rev);
1145 if (!(rev->diffopt.output_format & DIFF_FORMAT_NO_OUTPUT)) {
1146 struct line_log_data *range = lookup_line_range(rev, commit);
1147 dump_diff_hacky(rev, range);
1148 }
1149 return 1;
1150 }
1151
1152 static int bloom_filter_check(struct rev_info *rev,
1153 struct commit *commit,
1154 struct line_log_data *range)
1155 {
1156 struct bloom_filter *filter;
1157 struct bloom_key key;
1158 int result = 0;
1159
1160 if (!commit->parents)
1161 return 1;
1162
1163 if (!rev->bloom_filter_settings ||
1164 !(filter = get_bloom_filter(rev->repo, commit)))
1165 return 1;
1166
1167 if (!range)
1168 return 0;
1169
1170 while (!result && range) {
1171 fill_bloom_key(range->path, strlen(range->path), &key, rev->bloom_filter_settings);
1172
1173 if (bloom_filter_contains(filter, &key, rev->bloom_filter_settings))
1174 result = 1;
1175
1176 clear_bloom_key(&key);
1177 range = range->next;
1178 }
1179
1180 return result;
1181 }
1182
1183 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1184 struct line_log_data *range)
1185 {
1186 struct commit *parent = NULL;
1187 struct diff_queue_struct queue;
1188 struct line_log_data *parent_range;
1189 int changed;
1190
1191 if (commit->parents)
1192 parent = commit->parents->item;
1193
1194 queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1195 changed = process_all_files(&parent_range, rev, &queue, range);
1196
1197 if (parent)
1198 add_line_range(rev, parent, parent_range);
1199 free_line_log_data(parent_range);
1200 diff_free_queue(&queue);
1201 return changed;
1202 }
1203
1204 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1205 struct line_log_data *range)
1206 {
1207 struct diff_queue_struct *diffqueues;
1208 struct line_log_data **cand;
1209 struct commit **parents;
1210 struct commit_list *p;
1211 int i;
1212 int nparents = commit_list_count(commit->parents);
1213
1214 if (nparents > 1 && rev->first_parent_only)
1215 nparents = 1;
1216
1217 ALLOC_ARRAY(diffqueues, nparents);
1218 ALLOC_ARRAY(cand, nparents);
1219 ALLOC_ARRAY(parents, nparents);
1220
1221 p = commit->parents;
1222 for (i = 0; i < nparents; i++) {
1223 parents[i] = p->item;
1224 p = p->next;
1225 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1226 }
1227
1228 for (i = 0; i < nparents; i++) {
1229 int changed;
1230 cand[i] = NULL;
1231 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1232 if (!changed) {
1233 /*
1234 * This parent can take all the blame, so we
1235 * don't follow any other path in history
1236 */
1237 add_line_range(rev, parents[i], cand[i]);
1238 clear_commit_line_range(rev, commit);
1239 commit_list_append(parents[i], &commit->parents);
1240 free(parents);
1241 free(cand);
1242 free_diffqueues(nparents, diffqueues);
1243 /* NEEDSWORK leaking like a sieve */
1244 return 0;
1245 }
1246 }
1247
1248 /*
1249 * No single parent took the blame. We add the candidates
1250 * from the above loop to the parents.
1251 */
1252 for (i = 0; i < nparents; i++) {
1253 add_line_range(rev, parents[i], cand[i]);
1254 }
1255
1256 clear_commit_line_range(rev, commit);
1257 free(parents);
1258 free(cand);
1259 free_diffqueues(nparents, diffqueues);
1260 return 1;
1261
1262 /* NEEDSWORK evil merge detection stuff */
1263 /* NEEDSWORK leaking like a sieve */
1264 }
1265
1266 int line_log_process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1267 {
1268 struct line_log_data *range = lookup_line_range(rev, commit);
1269 int changed = 0;
1270
1271 if (range) {
1272 if (commit->parents && !bloom_filter_check(rev, commit, range)) {
1273 struct line_log_data *prange = line_log_data_copy(range);
1274 add_line_range(rev, commit->parents->item, prange);
1275 clear_commit_line_range(rev, commit);
1276 } else if (!commit->parents || !commit->parents->next)
1277 changed = process_ranges_ordinary_commit(rev, commit, range);
1278 else
1279 changed = process_ranges_merge_commit(rev, commit, range);
1280 }
1281
1282 if (!changed)
1283 commit->object.flags |= TREESAME;
1284
1285 return changed;
1286 }
1287
1288 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev UNUSED,
1289 struct commit **pp)
1290 {
1291 for (;;) {
1292 struct commit *p = *pp;
1293 if (p->parents && p->parents->next)
1294 return rewrite_one_ok;
1295 if (p->object.flags & UNINTERESTING)
1296 return rewrite_one_ok;
1297 if (!(p->object.flags & TREESAME))
1298 return rewrite_one_ok;
1299 if (!p->parents)
1300 return rewrite_one_noparents;
1301 *pp = p->parents->item;
1302 }
1303 }
1304
1305 int line_log_filter(struct rev_info *rev)
1306 {
1307 struct commit *commit;
1308 struct commit_list *list = rev->commits;
1309 struct commit_list *out = NULL, **pp = &out;
1310
1311 while (list) {
1312 struct commit_list *to_free = NULL;
1313 commit = list->item;
1314 if (line_log_process_ranges_arbitrary_commit(rev, commit)) {
1315 *pp = list;
1316 pp = &list->next;
1317 } else
1318 to_free = list;
1319 list = list->next;
1320 free(to_free);
1321 }
1322 *pp = NULL;
1323
1324 for (list = out; list; list = list->next)
1325 rewrite_parents(rev, list->item, line_log_rewrite_one);
1326
1327 rev->commits = out;
1328
1329 return 0;
1330 }
git for Debian