line-range.c: remove implicit dependency on the_index
[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(the_repository, 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 commit *commit, struct diff_filespec *spec)
500 {
501         unsigned mode;
502         struct object_id oid;
503
504         if (get_tree_entry(&commit->object.oid, spec->path, &oid, &mode))
505                 die("There is no path %s in the commit", spec->path);
506         fill_filespec(spec, &oid, 1, mode);
507
508         return;
509 }
510
511 static void fill_line_ends(struct repository *r,
512                            struct diff_filespec *spec,
513                            long *lines,
514                            unsigned long **line_ends)
515 {
516         int num = 0, size = 50;
517         long cur = 0;
518         unsigned long *ends = NULL;
519         char *data = NULL;
520
521         if (diff_populate_filespec(r, spec, 0))
522                 die("Cannot read blob %s", oid_to_hex(&spec->oid));
523
524         ALLOC_ARRAY(ends, size);
525         ends[cur++] = 0;
526         data = spec->data;
527         while (num < spec->size) {
528                 if (data[num] == '\n' || num == spec->size - 1) {
529                         ALLOC_GROW(ends, (cur + 1), size);
530                         ends[cur++] = num;
531                 }
532                 num++;
533         }
534
535         /* shrink the array to fit the elements */
536         REALLOC_ARRAY(ends, cur);
537         *lines = cur-1;
538         *line_ends = ends;
539 }
540
541 struct nth_line_cb {
542         struct diff_filespec *spec;
543         long lines;
544         unsigned long *line_ends;
545 };
546
547 static const char *nth_line(void *data, long line)
548 {
549         struct nth_line_cb *d = data;
550         assert(d && line <= d->lines);
551         assert(d->spec && d->spec->data);
552
553         if (line == 0)
554                 return (char *)d->spec->data;
555         else
556                 return (char *)d->spec->data + d->line_ends[line] + 1;
557 }
558
559 static struct line_log_data *
560 parse_lines(struct repository *r, struct commit *commit,
561             const char *prefix, struct string_list *args)
562 {
563         long lines = 0;
564         unsigned long *ends = NULL;
565         struct nth_line_cb cb_data;
566         struct string_list_item *item;
567         struct line_log_data *ranges = NULL;
568         struct line_log_data *p;
569
570         for_each_string_list_item(item, args) {
571                 const char *name_part, *range_part;
572                 char *full_name;
573                 struct diff_filespec *spec;
574                 long begin = 0, end = 0;
575                 long anchor;
576
577                 name_part = skip_range_arg(item->string, r->index);
578                 if (!name_part || *name_part != ':' || !name_part[1])
579                         die("-L argument not 'start,end:file' or ':funcname:file': %s",
580                             item->string);
581                 range_part = xstrndup(item->string, name_part - item->string);
582                 name_part++;
583
584                 full_name = prefix_path(prefix, prefix ? strlen(prefix) : 0,
585                                         name_part);
586
587                 spec = alloc_filespec(full_name);
588                 fill_blob_sha1(commit, spec);
589                 fill_line_ends(r, spec, &lines, &ends);
590                 cb_data.spec = spec;
591                 cb_data.lines = lines;
592                 cb_data.line_ends = ends;
593
594                 p = search_line_log_data(ranges, full_name, NULL);
595                 if (p && p->ranges.nr)
596                         anchor = p->ranges.ranges[p->ranges.nr - 1].end + 1;
597                 else
598                         anchor = 1;
599
600                 if (parse_range_arg(range_part, nth_line, &cb_data,
601                                     lines, anchor, &begin, &end,
602                                     full_name, r->index))
603                         die("malformed -L argument '%s'", range_part);
604                 if ((!lines && (begin || end)) || lines < begin)
605                         die("file %s has only %lu lines", name_part, lines);
606                 if (begin < 1)
607                         begin = 1;
608                 if (end < 1 || lines < end)
609                         end = lines;
610                 begin--;
611                 line_log_data_insert(&ranges, full_name, begin, end);
612
613                 free_filespec(spec);
614                 FREE_AND_NULL(ends);
615         }
616
617         for (p = ranges; p; p = p->next)
618                 sort_and_merge_range_set(&p->ranges);
619
620         return ranges;
621 }
622
623 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
624 {
625         struct line_log_data *ret = xmalloc(sizeof(*ret));
626
627         assert(r);
628         line_log_data_init(ret);
629         range_set_copy(&ret->ranges, &r->ranges);
630
631         ret->path = xstrdup(r->path);
632
633         return ret;
634 }
635
636 static struct line_log_data *
637 line_log_data_copy(struct line_log_data *r)
638 {
639         struct line_log_data *ret = NULL;
640         struct line_log_data *tmp = NULL, *prev = NULL;
641
642         assert(r);
643         ret = tmp = prev = line_log_data_copy_one(r);
644         r = r->next;
645         while (r) {
646                 tmp = line_log_data_copy_one(r);
647                 prev->next = tmp;
648                 prev = tmp;
649                 r = r->next;
650         }
651
652         return ret;
653 }
654
655 /* merge two range sets across files */
656 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
657                                                  struct line_log_data *b)
658 {
659         struct line_log_data *head = NULL, **pp = &head;
660
661         while (a || b) {
662                 struct line_log_data *src;
663                 struct line_log_data *src2 = NULL;
664                 struct line_log_data *d;
665                 int cmp;
666                 if (!a)
667                         cmp = 1;
668                 else if (!b)
669                         cmp = -1;
670                 else
671                         cmp = strcmp(a->path, b->path);
672                 if (cmp < 0) {
673                         src = a;
674                         a = a->next;
675                 } else if (cmp == 0) {
676                         src = a;
677                         a = a->next;
678                         src2 = b;
679                         b = b->next;
680                 } else {
681                         src = b;
682                         b = b->next;
683                 }
684                 d = xmalloc(sizeof(struct line_log_data));
685                 line_log_data_init(d);
686                 d->path = xstrdup(src->path);
687                 *pp = d;
688                 pp = &d->next;
689                 if (src2)
690                         range_set_union(&d->ranges, &src->ranges, &src2->ranges);
691                 else
692                         range_set_copy(&d->ranges, &src->ranges);
693         }
694
695         return head;
696 }
697
698 static void add_line_range(struct rev_info *revs, struct commit *commit,
699                            struct line_log_data *range)
700 {
701         struct line_log_data *old_line = NULL;
702         struct line_log_data *new_line = NULL;
703
704         old_line = lookup_decoration(&revs->line_log_data, &commit->object);
705         if (old_line && range) {
706                 new_line = line_log_data_merge(old_line, range);
707                 free_line_log_data(old_line);
708         } else if (range)
709                 new_line = line_log_data_copy(range);
710
711         if (new_line)
712                 add_decoration(&revs->line_log_data, &commit->object, new_line);
713 }
714
715 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
716 {
717         struct line_log_data *r;
718         r = lookup_decoration(&revs->line_log_data, &commit->object);
719         if (!r)
720                 return;
721         free_line_log_data(r);
722         add_decoration(&revs->line_log_data, &commit->object, NULL);
723 }
724
725 static struct line_log_data *lookup_line_range(struct rev_info *revs,
726                                                struct commit *commit)
727 {
728         struct line_log_data *ret = NULL;
729         struct line_log_data *d;
730
731         ret = lookup_decoration(&revs->line_log_data, &commit->object);
732
733         for (d = ret; d; d = d->next)
734                 range_set_check_invariants(&d->ranges);
735
736         return ret;
737 }
738
739 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
740 {
741         struct commit *commit = NULL;
742         struct line_log_data *range;
743
744         commit = check_single_commit(rev);
745         range = parse_lines(rev->diffopt.repo, commit, prefix, args);
746         add_line_range(rev, commit, range);
747
748         if (!rev->diffopt.detect_rename) {
749                 struct line_log_data *r;
750                 struct argv_array array = ARGV_ARRAY_INIT;
751                 const char **paths;
752
753                 for (r = range; r; r = r->next)
754                         argv_array_push(&array, r->path);
755                 paths = argv_array_detach(&array);
756
757                 parse_pathspec(&rev->diffopt.pathspec, 0,
758                                PATHSPEC_PREFER_FULL, "", paths);
759                 /* strings are now owned by pathspec */
760                 free(paths);
761         }
762 }
763
764 static void move_diff_queue(struct diff_queue_struct *dst,
765                             struct diff_queue_struct *src)
766 {
767         assert(src != dst);
768         memcpy(dst, src, sizeof(struct diff_queue_struct));
769         DIFF_QUEUE_CLEAR(src);
770 }
771
772 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
773 {
774         int i;
775         struct diff_queue_struct outq;
776         DIFF_QUEUE_CLEAR(&outq);
777
778         for (i = 0; i < diff_queued_diff.nr; i++) {
779                 struct diff_filepair *p = diff_queued_diff.queue[i];
780                 struct line_log_data *rg = NULL;
781
782                 if (!DIFF_FILE_VALID(p->two)) {
783                         if (keep_deletions)
784                                 diff_q(&outq, p);
785                         else
786                                 diff_free_filepair(p);
787                         continue;
788                 }
789                 for (rg = range; rg; rg = rg->next) {
790                         if (!strcmp(rg->path, p->two->path))
791                                 break;
792                 }
793                 if (rg)
794                         diff_q(&outq, p);
795                 else
796                         diff_free_filepair(p);
797         }
798         free(diff_queued_diff.queue);
799         diff_queued_diff = outq;
800 }
801
802 static inline int diff_might_be_rename(void)
803 {
804         int i;
805         for (i = 0; i < diff_queued_diff.nr; i++)
806                 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
807                         /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
808                         /*      diff_queued_diff.queue[i]->two->path); */
809                         return 1;
810                 }
811         return 0;
812 }
813
814 static void queue_diffs(struct line_log_data *range,
815                         struct diff_options *opt,
816                         struct diff_queue_struct *queue,
817                         struct commit *commit, struct commit *parent)
818 {
819         assert(commit);
820
821         DIFF_QUEUE_CLEAR(&diff_queued_diff);
822         diff_tree_oid(parent ? get_commit_tree_oid(parent) : NULL,
823                       get_commit_tree_oid(commit), "", opt);
824         if (opt->detect_rename) {
825                 filter_diffs_for_paths(range, 1);
826                 if (diff_might_be_rename())
827                         diffcore_std(opt);
828                 filter_diffs_for_paths(range, 0);
829         }
830         move_diff_queue(queue, &diff_queued_diff);
831 }
832
833 static char *get_nth_line(long line, unsigned long *ends, void *data)
834 {
835         if (line == 0)
836                 return (char *)data;
837         else
838                 return (char *)data + ends[line] + 1;
839 }
840
841 static void print_line(const char *prefix, char first,
842                        long line, unsigned long *ends, void *data,
843                        const char *color, const char *reset, FILE *file)
844 {
845         char *begin = get_nth_line(line, ends, data);
846         char *end = get_nth_line(line+1, ends, data);
847         int had_nl = 0;
848
849         if (end > begin && end[-1] == '\n') {
850                 end--;
851                 had_nl = 1;
852         }
853
854         fputs(prefix, file);
855         fputs(color, file);
856         putc(first, file);
857         fwrite(begin, 1, end-begin, file);
858         fputs(reset, file);
859         putc('\n', file);
860         if (!had_nl)
861                 fputs("\\ No newline at end of file\n", file);
862 }
863
864 static char *output_prefix(struct diff_options *opt)
865 {
866         char *prefix = "";
867
868         if (opt->output_prefix) {
869                 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
870                 prefix = sb->buf;
871         }
872
873         return prefix;
874 }
875
876 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
877 {
878         unsigned int i, j = 0;
879         long p_lines, t_lines;
880         unsigned long *p_ends = NULL, *t_ends = NULL;
881         struct diff_filepair *pair = range->pair;
882         struct diff_ranges *diff = &range->diff;
883
884         struct diff_options *opt = &rev->diffopt;
885         char *prefix = output_prefix(opt);
886         const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
887         const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
888         const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
889         const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
890         const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
891         const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
892
893         if (!pair || !diff)
894                 return;
895
896         if (pair->one->oid_valid)
897                 fill_line_ends(rev->diffopt.repo, pair->one, &p_lines, &p_ends);
898         fill_line_ends(rev->diffopt.repo, pair->two, &t_lines, &t_ends);
899
900         fprintf(opt->file, "%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
901         fprintf(opt->file, "%s%s--- %s%s%s\n", prefix, c_meta,
902                pair->one->oid_valid ? "a/" : "",
903                pair->one->oid_valid ? pair->one->path : "/dev/null",
904                c_reset);
905         fprintf(opt->file, "%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
906         for (i = 0; i < range->ranges.nr; i++) {
907                 long p_start, p_end;
908                 long t_start = range->ranges.ranges[i].start;
909                 long t_end = range->ranges.ranges[i].end;
910                 long t_cur = t_start;
911                 unsigned int j_last;
912
913                 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
914                         j++;
915                 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
916                         continue;
917
918                 /* Scan ahead to determine the last diff that falls in this range */
919                 j_last = j;
920                 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
921                         j_last++;
922                 if (j_last > j)
923                         j_last--;
924
925                 /*
926                  * Compute parent hunk headers: we know that the diff
927                  * has the correct line numbers (but not all hunks).
928                  * So it suffices to shift the start/end according to
929                  * the line numbers of the first/last hunk(s) that
930                  * fall in this range.
931                  */
932                 if (t_start < diff->target.ranges[j].start)
933                         p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
934                 else
935                         p_start = diff->parent.ranges[j].start;
936                 if (t_end > diff->target.ranges[j_last].end)
937                         p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
938                 else
939                         p_end = diff->parent.ranges[j_last].end;
940
941                 if (!p_start && !p_end) {
942                         p_start = -1;
943                         p_end = -1;
944                 }
945
946                 /* Now output a diff hunk for this range */
947                 fprintf(opt->file, "%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
948                        prefix, c_frag,
949                        p_start+1, p_end-p_start, t_start+1, t_end-t_start,
950                        c_reset);
951                 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
952                         int k;
953                         for (; t_cur < diff->target.ranges[j].start; t_cur++)
954                                 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
955                                            c_context, c_reset, opt->file);
956                         for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
957                                 print_line(prefix, '-', k, p_ends, pair->one->data,
958                                            c_old, c_reset, opt->file);
959                         for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
960                                 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
961                                            c_new, c_reset, opt->file);
962                         j++;
963                 }
964                 for (; t_cur < t_end; t_cur++)
965                         print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
966                                    c_context, c_reset, opt->file);
967         }
968
969         free(p_ends);
970         free(t_ends);
971 }
972
973 /*
974  * NEEDSWORK: manually building a diff here is not the Right
975  * Thing(tm).  log -L should be built into the diff pipeline.
976  */
977 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
978 {
979         fprintf(rev->diffopt.file, "%s\n", output_prefix(&rev->diffopt));
980         while (range) {
981                 dump_diff_hacky_one(rev, range);
982                 range = range->next;
983         }
984 }
985
986 /*
987  * Unlike most other functions, this destructively operates on
988  * 'range'.
989  */
990 static int process_diff_filepair(struct rev_info *rev,
991                                  struct diff_filepair *pair,
992                                  struct line_log_data *range,
993                                  struct diff_ranges **diff_out)
994 {
995         struct line_log_data *rg = range;
996         struct range_set tmp;
997         struct diff_ranges diff;
998         mmfile_t file_parent, file_target;
999
1000         assert(pair->two->path);
1001         while (rg) {
1002                 assert(rg->path);
1003                 if (!strcmp(rg->path, pair->two->path))
1004                         break;
1005                 rg = rg->next;
1006         }
1007
1008         if (!rg)
1009                 return 0;
1010         if (rg->ranges.nr == 0)
1011                 return 0;
1012
1013         assert(pair->two->oid_valid);
1014         diff_populate_filespec(rev->diffopt.repo, pair->two, 0);
1015         file_target.ptr = pair->two->data;
1016         file_target.size = pair->two->size;
1017
1018         if (pair->one->oid_valid) {
1019                 diff_populate_filespec(rev->diffopt.repo, pair->one, 0);
1020                 file_parent.ptr = pair->one->data;
1021                 file_parent.size = pair->one->size;
1022         } else {
1023                 file_parent.ptr = "";
1024                 file_parent.size = 0;
1025         }
1026
1027         diff_ranges_init(&diff);
1028         if (collect_diff(&file_parent, &file_target, &diff))
1029                 die("unable to generate diff for %s", pair->one->path);
1030
1031         /* NEEDSWORK should apply some heuristics to prevent mismatches */
1032         free(rg->path);
1033         rg->path = xstrdup(pair->one->path);
1034
1035         range_set_init(&tmp, 0);
1036         range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1037         range_set_release(&rg->ranges);
1038         range_set_move(&rg->ranges, &tmp);
1039
1040         diff_ranges_release(&diff);
1041
1042         return ((*diff_out)->parent.nr > 0);
1043 }
1044
1045 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1046 {
1047         struct diff_filepair *new_filepair = xmalloc(sizeof(struct diff_filepair));
1048         new_filepair->one = pair->one;
1049         new_filepair->two = pair->two;
1050         new_filepair->one->count++;
1051         new_filepair->two->count++;
1052         return new_filepair;
1053 }
1054
1055 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1056 {
1057         int i, j;
1058         for (i = 0; i < n; i++)
1059                 for (j = 0; j < dq[i].nr; j++)
1060                         diff_free_filepair(dq[i].queue[j]);
1061         free(dq);
1062 }
1063
1064 static int process_all_files(struct line_log_data **range_out,
1065                              struct rev_info *rev,
1066                              struct diff_queue_struct *queue,
1067                              struct line_log_data *range)
1068 {
1069         int i, changed = 0;
1070
1071         *range_out = line_log_data_copy(range);
1072
1073         for (i = 0; i < queue->nr; i++) {
1074                 struct diff_ranges *pairdiff = NULL;
1075                 struct diff_filepair *pair = queue->queue[i];
1076                 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1077                         /*
1078                          * Store away the diff for later output.  We
1079                          * tuck it in the ranges we got as _input_,
1080                          * since that's the commit that caused the
1081                          * diff.
1082                          *
1083                          * NEEDSWORK not enough when we get around to
1084                          * doing something interesting with merges;
1085                          * currently each invocation on a merge parent
1086                          * trashes the previous one's diff.
1087                          *
1088                          * NEEDSWORK tramples over data structures not owned here
1089                          */
1090                         struct line_log_data *rg = range;
1091                         changed++;
1092                         while (rg && strcmp(rg->path, pair->two->path))
1093                                 rg = rg->next;
1094                         assert(rg);
1095                         rg->pair = diff_filepair_dup(queue->queue[i]);
1096                         memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1097                 }
1098                 free(pairdiff);
1099         }
1100
1101         return changed;
1102 }
1103
1104 int line_log_print(struct rev_info *rev, struct commit *commit)
1105 {
1106         struct line_log_data *range = lookup_line_range(rev, commit);
1107
1108         show_log(rev);
1109         dump_diff_hacky(rev, range);
1110         return 1;
1111 }
1112
1113 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1114                                           struct line_log_data *range)
1115 {
1116         struct commit *parent = NULL;
1117         struct diff_queue_struct queue;
1118         struct line_log_data *parent_range;
1119         int changed;
1120
1121         if (commit->parents)
1122                 parent = commit->parents->item;
1123
1124         queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1125         changed = process_all_files(&parent_range, rev, &queue, range);
1126         if (parent)
1127                 add_line_range(rev, parent, parent_range);
1128         free_line_log_data(parent_range);
1129         return changed;
1130 }
1131
1132 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1133                                        struct line_log_data *range)
1134 {
1135         struct diff_queue_struct *diffqueues;
1136         struct line_log_data **cand;
1137         struct commit **parents;
1138         struct commit_list *p;
1139         int i;
1140         int nparents = commit_list_count(commit->parents);
1141
1142         if (nparents > 1 && rev->first_parent_only)
1143                 nparents = 1;
1144
1145         ALLOC_ARRAY(diffqueues, nparents);
1146         ALLOC_ARRAY(cand, nparents);
1147         ALLOC_ARRAY(parents, nparents);
1148
1149         p = commit->parents;
1150         for (i = 0; i < nparents; i++) {
1151                 parents[i] = p->item;
1152                 p = p->next;
1153                 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1154         }
1155
1156         for (i = 0; i < nparents; i++) {
1157                 int changed;
1158                 cand[i] = NULL;
1159                 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1160                 if (!changed) {
1161                         /*
1162                          * This parent can take all the blame, so we
1163                          * don't follow any other path in history
1164                          */
1165                         add_line_range(rev, parents[i], cand[i]);
1166                         clear_commit_line_range(rev, commit);
1167                         commit_list_append(parents[i], &commit->parents);
1168                         free(parents);
1169                         free(cand);
1170                         free_diffqueues(nparents, diffqueues);
1171                         /* NEEDSWORK leaking like a sieve */
1172                         return 0;
1173                 }
1174         }
1175
1176         /*
1177          * No single parent took the blame.  We add the candidates
1178          * from the above loop to the parents.
1179          */
1180         for (i = 0; i < nparents; i++) {
1181                 add_line_range(rev, parents[i], cand[i]);
1182         }
1183
1184         clear_commit_line_range(rev, commit);
1185         free(parents);
1186         free(cand);
1187         free_diffqueues(nparents, diffqueues);
1188         return 1;
1189
1190         /* NEEDSWORK evil merge detection stuff */
1191         /* NEEDSWORK leaking like a sieve */
1192 }
1193
1194 static int process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1195 {
1196         struct line_log_data *range = lookup_line_range(rev, commit);
1197         int changed = 0;
1198
1199         if (range) {
1200                 if (!commit->parents || !commit->parents->next)
1201                         changed = process_ranges_ordinary_commit(rev, commit, range);
1202                 else
1203                         changed = process_ranges_merge_commit(rev, commit, range);
1204         }
1205
1206         if (!changed)
1207                 commit->object.flags |= TREESAME;
1208
1209         return changed;
1210 }
1211
1212 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev, struct commit **pp)
1213 {
1214         for (;;) {
1215                 struct commit *p = *pp;
1216                 if (p->parents && p->parents->next)
1217                         return rewrite_one_ok;
1218                 if (p->object.flags & UNINTERESTING)
1219                         return rewrite_one_ok;
1220                 if (!(p->object.flags & TREESAME))
1221                         return rewrite_one_ok;
1222                 if (!p->parents)
1223                         return rewrite_one_noparents;
1224                 *pp = p->parents->item;
1225         }
1226 }
1227
1228 int line_log_filter(struct rev_info *rev)
1229 {
1230         struct commit *commit;
1231         struct commit_list *list = rev->commits;
1232         struct commit_list *out = NULL, **pp = &out;
1233
1234         while (list) {
1235                 struct commit_list *to_free = NULL;
1236                 commit = list->item;
1237                 if (process_ranges_arbitrary_commit(rev, commit)) {
1238                         *pp = list;
1239                         pp = &list->next;
1240                 } else
1241                         to_free = list;
1242                 list = list->next;
1243                 free(to_free);
1244         }
1245         *pp = NULL;
1246
1247         for (list = out; list; list = list->next)
1248                 rewrite_parents(rev, list->item, line_log_rewrite_one);
1249
1250         rev->commits = out;
1251
1252         return 0;
1253 }