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