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