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