Merge branch 'ad/commit-have-m-option' into maint
[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(rs->ranges);
38         rs->alloc = rs->nr = 0;
39         rs->ranges = NULL;
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         memcpy(dst->ranges, src->ranges, src->nr*sizeof(struct range_set));
47         dst->nr = src->nr;
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         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         int i;
114         int o = 0; /* output cursor */
115
116         qsort(rs->ranges, rs->nr, sizeof(struct range), 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         int i = 0, j = 0, o = 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;
155                 if (i < a->nr && j < b->nr) {
156                         if (ra[i].start < rb[j].start)
157                                 new = &ra[i++];
158                         else if (ra[i].start > rb[j].start)
159                                 new = &rb[j++];
160                         else if (ra[i].end < rb[j].end)
161                                 new = &ra[i++];
162                         else
163                                 new = &rb[j++];
164                 } else if (i < a->nr)      /* b exhausted */
165                         new = &ra[i++];
166                 else                       /* a exhausted */
167                         new = &rb[j++];
168                 if (new->start == new->end)
169                         ; /* empty range */
170                 else if (!o || out->ranges[o-1].end < new->start) {
171                         range_set_grow(out, 1);
172                         out->ranges[o].start = new->start;
173                         out->ranges[o].end = new->end;
174                         o++;
175                 } else if (out->ranges[o-1].end < new->end) {
176                         out->ranges[o-1].end = new->end;
177                 }
178         }
179         out->nr = o;
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         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         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         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                 while (obj->type == OBJ_TAG)
484                         obj = deref_tag(obj, NULL, 0);
485                 if (obj->type != OBJ_COMMIT)
486                         die("Non commit %s?", revs->pending.objects[i].name);
487                 if (commit)
488                         die("More than one commit to dig from: %s and %s?",
489                             revs->pending.objects[i].name,
490                             revs->pending.objects[found].name);
491                 commit = obj;
492                 found = i;
493         }
494
495         if (!commit)
496                 die("No commit specified?");
497
498         return (struct commit *) commit;
499 }
500
501 static void fill_blob_sha1(struct commit *commit, struct diff_filespec *spec)
502 {
503         unsigned mode;
504         unsigned char sha1[20];
505
506         if (get_tree_entry(commit->object.oid.hash, spec->path,
507                            sha1, &mode))
508                 die("There is no path %s in the commit", spec->path);
509         fill_filespec(spec, sha1, 1, mode);
510
511         return;
512 }
513
514 static void fill_line_ends(struct diff_filespec *spec, long *lines,
515                            unsigned long **line_ends)
516 {
517         int num = 0, size = 50;
518         long cur = 0;
519         unsigned long *ends = NULL;
520         char *data = NULL;
521
522         if (diff_populate_filespec(spec, 0))
523                 die("Cannot read blob %s", sha1_to_hex(spec->sha1));
524
525         ALLOC_ARRAY(ends, size);
526         ends[cur++] = 0;
527         data = spec->data;
528         while (num < spec->size) {
529                 if (data[num] == '\n' || num == spec->size - 1) {
530                         ALLOC_GROW(ends, (cur + 1), size);
531                         ends[cur++] = num;
532                 }
533                 num++;
534         }
535
536         /* shrink the array to fit the elements */
537         REALLOC_ARRAY(ends, cur);
538         *lines = cur-1;
539         *line_ends = ends;
540 }
541
542 struct nth_line_cb {
543         struct diff_filespec *spec;
544         long lines;
545         unsigned long *line_ends;
546 };
547
548 static const char *nth_line(void *data, long line)
549 {
550         struct nth_line_cb *d = data;
551         assert(d && line <= d->lines);
552         assert(d->spec && d->spec->data);
553
554         if (line == 0)
555                 return (char *)d->spec->data;
556         else
557                 return (char *)d->spec->data + d->line_ends[line] + 1;
558 }
559
560 static struct line_log_data *
561 parse_lines(struct commit *commit, 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);
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(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))
603                         die("malformed -L argument '%s'", range_part);
604                 if (lines < end || ((lines || begin) && lines < begin))
605                         die("file %s has only %lu lines", name_part, lines);
606                 if (begin < 1)
607                         begin = 1;
608                 if (end < 1)
609                         end = lines;
610                 begin--;
611                 line_log_data_insert(&ranges, full_name, begin, end);
612
613                 free_filespec(spec);
614                 free(ends);
615                 ends = NULL;
616         }
617
618         for (p = ranges; p; p = p->next)
619                 sort_and_merge_range_set(&p->ranges);
620
621         return ranges;
622 }
623
624 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
625 {
626         struct line_log_data *ret = xmalloc(sizeof(*ret));
627
628         assert(r);
629         line_log_data_init(ret);
630         range_set_copy(&ret->ranges, &r->ranges);
631
632         ret->path = xstrdup(r->path);
633
634         return ret;
635 }
636
637 static struct line_log_data *
638 line_log_data_copy(struct line_log_data *r)
639 {
640         struct line_log_data *ret = NULL;
641         struct line_log_data *tmp = NULL, *prev = NULL;
642
643         assert(r);
644         ret = tmp = prev = line_log_data_copy_one(r);
645         r = r->next;
646         while (r) {
647                 tmp = line_log_data_copy_one(r);
648                 prev->next = tmp;
649                 prev = tmp;
650                 r = r->next;
651         }
652
653         return ret;
654 }
655
656 /* merge two range sets across files */
657 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
658                                                  struct line_log_data *b)
659 {
660         struct line_log_data *head = NULL, **pp = &head;
661
662         while (a || b) {
663                 struct line_log_data *src;
664                 struct line_log_data *src2 = NULL;
665                 struct line_log_data *d;
666                 int cmp;
667                 if (!a)
668                         cmp = 1;
669                 else if (!b)
670                         cmp = -1;
671                 else
672                         cmp = strcmp(a->path, b->path);
673                 if (cmp < 0) {
674                         src = a;
675                         a = a->next;
676                 } else if (cmp == 0) {
677                         src = a;
678                         a = a->next;
679                         src2 = b;
680                         b = b->next;
681                 } else {
682                         src = b;
683                         b = b->next;
684                 }
685                 d = xmalloc(sizeof(struct line_log_data));
686                 line_log_data_init(d);
687                 d->path = xstrdup(src->path);
688                 *pp = d;
689                 pp = &d->next;
690                 if (src2)
691                         range_set_union(&d->ranges, &src->ranges, &src2->ranges);
692                 else
693                         range_set_copy(&d->ranges, &src->ranges);
694         }
695
696         return head;
697 }
698
699 static void add_line_range(struct rev_info *revs, struct commit *commit,
700                            struct line_log_data *range)
701 {
702         struct line_log_data *old = NULL;
703         struct line_log_data *new = NULL;
704
705         old = lookup_decoration(&revs->line_log_data, &commit->object);
706         if (old && range) {
707                 new = line_log_data_merge(old, range);
708                 free_line_log_data(old);
709         } else if (range)
710                 new = line_log_data_copy(range);
711
712         if (new)
713                 add_decoration(&revs->line_log_data, &commit->object, new);
714 }
715
716 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
717 {
718         struct line_log_data *r;
719         r = lookup_decoration(&revs->line_log_data, &commit->object);
720         if (!r)
721                 return;
722         free_line_log_data(r);
723         add_decoration(&revs->line_log_data, &commit->object, NULL);
724 }
725
726 static struct line_log_data *lookup_line_range(struct rev_info *revs,
727                                                struct commit *commit)
728 {
729         struct line_log_data *ret = NULL;
730         struct line_log_data *d;
731
732         ret = lookup_decoration(&revs->line_log_data, &commit->object);
733
734         for (d = ret; d; d = d->next)
735                 range_set_check_invariants(&d->ranges);
736
737         return ret;
738 }
739
740 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
741 {
742         struct commit *commit = NULL;
743         struct line_log_data *range;
744
745         commit = check_single_commit(rev);
746         range = parse_lines(commit, prefix, args);
747         add_line_range(rev, commit, range);
748
749         if (!rev->diffopt.detect_rename) {
750                 struct line_log_data *r;
751                 struct argv_array array = ARGV_ARRAY_INIT;
752                 const char **paths;
753
754                 for (r = range; r; r = r->next)
755                         argv_array_push(&array, r->path);
756                 paths = argv_array_detach(&array);
757
758                 parse_pathspec(&rev->diffopt.pathspec, 0,
759                                PATHSPEC_PREFER_FULL, "", paths);
760                 /* strings are now owned by pathspec */
761                 free(paths);
762         }
763 }
764
765 static void move_diff_queue(struct diff_queue_struct *dst,
766                             struct diff_queue_struct *src)
767 {
768         assert(src != dst);
769         memcpy(dst, src, sizeof(struct diff_queue_struct));
770         DIFF_QUEUE_CLEAR(src);
771 }
772
773 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
774 {
775         int i;
776         struct diff_queue_struct outq;
777         DIFF_QUEUE_CLEAR(&outq);
778
779         for (i = 0; i < diff_queued_diff.nr; i++) {
780                 struct diff_filepair *p = diff_queued_diff.queue[i];
781                 struct line_log_data *rg = NULL;
782
783                 if (!DIFF_FILE_VALID(p->two)) {
784                         if (keep_deletions)
785                                 diff_q(&outq, p);
786                         else
787                                 diff_free_filepair(p);
788                         continue;
789                 }
790                 for (rg = range; rg; rg = rg->next) {
791                         if (!strcmp(rg->path, p->two->path))
792                                 break;
793                 }
794                 if (rg)
795                         diff_q(&outq, p);
796                 else
797                         diff_free_filepair(p);
798         }
799         free(diff_queued_diff.queue);
800         diff_queued_diff = outq;
801 }
802
803 static inline int diff_might_be_rename(void)
804 {
805         int i;
806         for (i = 0; i < diff_queued_diff.nr; i++)
807                 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
808                         /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
809                         /*      diff_queued_diff.queue[i]->two->path); */
810                         return 1;
811                 }
812         return 0;
813 }
814
815 static void queue_diffs(struct line_log_data *range,
816                         struct diff_options *opt,
817                         struct diff_queue_struct *queue,
818                         struct commit *commit, struct commit *parent)
819 {
820         assert(commit);
821
822         DIFF_QUEUE_CLEAR(&diff_queued_diff);
823         diff_tree_sha1(parent ? parent->tree->object.oid.hash : NULL,
824                         commit->tree->object.oid.hash, "", opt);
825         if (opt->detect_rename) {
826                 filter_diffs_for_paths(range, 1);
827                 if (diff_might_be_rename())
828                         diffcore_std(opt);
829                 filter_diffs_for_paths(range, 0);
830         }
831         move_diff_queue(queue, &diff_queued_diff);
832 }
833
834 static char *get_nth_line(long line, unsigned long *ends, void *data)
835 {
836         if (line == 0)
837                 return (char *)data;
838         else
839                 return (char *)data + ends[line] + 1;
840 }
841
842 static void print_line(const char *prefix, char first,
843                        long line, unsigned long *ends, void *data,
844                        const char *color, const char *reset)
845 {
846         char *begin = get_nth_line(line, ends, data);
847         char *end = get_nth_line(line+1, ends, data);
848         int had_nl = 0;
849
850         if (end > begin && end[-1] == '\n') {
851                 end--;
852                 had_nl = 1;
853         }
854
855         fputs(prefix, stdout);
856         fputs(color, stdout);
857         putchar(first);
858         fwrite(begin, 1, end-begin, stdout);
859         fputs(reset, stdout);
860         putchar('\n');
861         if (!had_nl)
862                 fputs("\\ No newline at end of file\n", stdout);
863 }
864
865 static char *output_prefix(struct diff_options *opt)
866 {
867         char *prefix = "";
868
869         if (opt->output_prefix) {
870                 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
871                 prefix = sb->buf;
872         }
873
874         return prefix;
875 }
876
877 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
878 {
879         int i, j = 0;
880         long p_lines, t_lines;
881         unsigned long *p_ends = NULL, *t_ends = NULL;
882         struct diff_filepair *pair = range->pair;
883         struct diff_ranges *diff = &range->diff;
884
885         struct diff_options *opt = &rev->diffopt;
886         char *prefix = output_prefix(opt);
887         const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
888         const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
889         const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
890         const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
891         const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
892         const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
893
894         if (!pair || !diff)
895                 return;
896
897         if (pair->one->sha1_valid)
898                 fill_line_ends(pair->one, &p_lines, &p_ends);
899         fill_line_ends(pair->two, &t_lines, &t_ends);
900
901         printf("%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
902         printf("%s%s--- %s%s%s\n", prefix, c_meta,
903                pair->one->sha1_valid ? "a/" : "",
904                pair->one->sha1_valid ? pair->one->path : "/dev/null",
905                c_reset);
906         printf("%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
907         for (i = 0; i < range->ranges.nr; i++) {
908                 long p_start, p_end;
909                 long t_start = range->ranges.ranges[i].start;
910                 long t_end = range->ranges.ranges[i].end;
911                 long t_cur = t_start;
912                 int j_last;
913
914                 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
915                         j++;
916                 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
917                         continue;
918
919                 /* Scan ahead to determine the last diff that falls in this range */
920                 j_last = j;
921                 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
922                         j_last++;
923                 if (j_last > j)
924                         j_last--;
925
926                 /*
927                  * Compute parent hunk headers: we know that the diff
928                  * has the correct line numbers (but not all hunks).
929                  * So it suffices to shift the start/end according to
930                  * the line numbers of the first/last hunk(s) that
931                  * fall in this range.
932                  */
933                 if (t_start < diff->target.ranges[j].start)
934                         p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
935                 else
936                         p_start = diff->parent.ranges[j].start;
937                 if (t_end > diff->target.ranges[j_last].end)
938                         p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
939                 else
940                         p_end = diff->parent.ranges[j_last].end;
941
942                 if (!p_start && !p_end) {
943                         p_start = -1;
944                         p_end = -1;
945                 }
946
947                 /* Now output a diff hunk for this range */
948                 printf("%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
949                        prefix, c_frag,
950                        p_start+1, p_end-p_start, t_start+1, t_end-t_start,
951                        c_reset);
952                 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
953                         int k;
954                         for (; t_cur < diff->target.ranges[j].start; t_cur++)
955                                 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
956                                            c_context, c_reset);
957                         for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
958                                 print_line(prefix, '-', k, p_ends, pair->one->data,
959                                            c_old, c_reset);
960                         for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
961                                 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
962                                            c_new, c_reset);
963                         j++;
964                 }
965                 for (; t_cur < t_end; t_cur++)
966                         print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
967                                    c_context, c_reset);
968         }
969
970         free(p_ends);
971         free(t_ends);
972 }
973
974 /*
975  * NEEDSWORK: manually building a diff here is not the Right
976  * Thing(tm).  log -L should be built into the diff pipeline.
977  */
978 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
979 {
980         puts(output_prefix(&rev->diffopt));
981         while (range) {
982                 dump_diff_hacky_one(rev, range);
983                 range = range->next;
984         }
985 }
986
987 /*
988  * Unlike most other functions, this destructively operates on
989  * 'range'.
990  */
991 static int process_diff_filepair(struct rev_info *rev,
992                                  struct diff_filepair *pair,
993                                  struct line_log_data *range,
994                                  struct diff_ranges **diff_out)
995 {
996         struct line_log_data *rg = range;
997         struct range_set tmp;
998         struct diff_ranges diff;
999         mmfile_t file_parent, file_target;
1000
1001         assert(pair->two->path);
1002         while (rg) {
1003                 assert(rg->path);
1004                 if (!strcmp(rg->path, pair->two->path))
1005                         break;
1006                 rg = rg->next;
1007         }
1008
1009         if (!rg)
1010                 return 0;
1011         if (rg->ranges.nr == 0)
1012                 return 0;
1013
1014         assert(pair->two->sha1_valid);
1015         diff_populate_filespec(pair->two, 0);
1016         file_target.ptr = pair->two->data;
1017         file_target.size = pair->two->size;
1018
1019         if (pair->one->sha1_valid) {
1020                 diff_populate_filespec(pair->one, 0);
1021                 file_parent.ptr = pair->one->data;
1022                 file_parent.size = pair->one->size;
1023         } else {
1024                 file_parent.ptr = "";
1025                 file_parent.size = 0;
1026         }
1027
1028         diff_ranges_init(&diff);
1029         if (collect_diff(&file_parent, &file_target, &diff))
1030                 die("unable to generate diff for %s", pair->one->path);
1031
1032         /* NEEDSWORK should apply some heuristics to prevent mismatches */
1033         free(rg->path);
1034         rg->path = xstrdup(pair->one->path);
1035
1036         range_set_init(&tmp, 0);
1037         range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1038         range_set_release(&rg->ranges);
1039         range_set_move(&rg->ranges, &tmp);
1040
1041         diff_ranges_release(&diff);
1042
1043         return ((*diff_out)->parent.nr > 0);
1044 }
1045
1046 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1047 {
1048         struct diff_filepair *new = xmalloc(sizeof(struct diff_filepair));
1049         new->one = pair->one;
1050         new->two = pair->two;
1051         new->one->count++;
1052         new->two->count++;
1053         return new;
1054 }
1055
1056 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1057 {
1058         int i, j;
1059         for (i = 0; i < n; i++)
1060                 for (j = 0; j < dq[i].nr; j++)
1061                         diff_free_filepair(dq[i].queue[j]);
1062         free(dq);
1063 }
1064
1065 static int process_all_files(struct line_log_data **range_out,
1066                              struct rev_info *rev,
1067                              struct diff_queue_struct *queue,
1068                              struct line_log_data *range)
1069 {
1070         int i, changed = 0;
1071
1072         *range_out = line_log_data_copy(range);
1073
1074         for (i = 0; i < queue->nr; i++) {
1075                 struct diff_ranges *pairdiff = NULL;
1076                 struct diff_filepair *pair = queue->queue[i];
1077                 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1078                         /*
1079                          * Store away the diff for later output.  We
1080                          * tuck it in the ranges we got as _input_,
1081                          * since that's the commit that caused the
1082                          * diff.
1083                          *
1084                          * NEEDSWORK not enough when we get around to
1085                          * doing something interesting with merges;
1086                          * currently each invocation on a merge parent
1087                          * trashes the previous one's diff.
1088                          *
1089                          * NEEDSWORK tramples over data structures not owned here
1090                          */
1091                         struct line_log_data *rg = range;
1092                         changed++;
1093                         while (rg && strcmp(rg->path, pair->two->path))
1094                                 rg = rg->next;
1095                         assert(rg);
1096                         rg->pair = diff_filepair_dup(queue->queue[i]);
1097                         memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1098                 }
1099                 free(pairdiff);
1100         }
1101
1102         return changed;
1103 }
1104
1105 int line_log_print(struct rev_info *rev, struct commit *commit)
1106 {
1107         struct line_log_data *range = lookup_line_range(rev, commit);
1108
1109         show_log(rev);
1110         dump_diff_hacky(rev, range);
1111         return 1;
1112 }
1113
1114 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1115                                           struct line_log_data *range)
1116 {
1117         struct commit *parent = NULL;
1118         struct diff_queue_struct queue;
1119         struct line_log_data *parent_range;
1120         int changed;
1121
1122         if (commit->parents)
1123                 parent = commit->parents->item;
1124
1125         queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1126         changed = process_all_files(&parent_range, rev, &queue, range);
1127         if (parent)
1128                 add_line_range(rev, parent, 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 }