Merge branch 'gb/idx'
[git] / builtin-rev-list.c
1 #include "cache.h"
2 #include "refs.h"
3 #include "tag.h"
4 #include "commit.h"
5 #include "tree.h"
6 #include "blob.h"
7 #include "tree-walk.h"
8 #include "diff.h"
9 #include "revision.h"
10 #include "list-objects.h"
11 #include "builtin.h"
12
13 /* bits #0-15 in revision.h */
14
15 #define COUNTED         (1u<<16)
16
17 static const char rev_list_usage[] =
18 "git-rev-list [OPTION] <commit-id>... [ -- paths... ]\n"
19 "  limiting output:\n"
20 "    --max-count=nr\n"
21 "    --max-age=epoch\n"
22 "    --min-age=epoch\n"
23 "    --sparse\n"
24 "    --no-merges\n"
25 "    --remove-empty\n"
26 "    --all\n"
27 "    --stdin\n"
28 "  ordering output:\n"
29 "    --topo-order\n"
30 "    --date-order\n"
31 "  formatting output:\n"
32 "    --parents\n"
33 "    --objects | --objects-edge\n"
34 "    --unpacked\n"
35 "    --header | --pretty\n"
36 "    --abbrev=nr | --no-abbrev\n"
37 "    --abbrev-commit\n"
38 "    --left-right\n"
39 "  special purpose:\n"
40 "    --bisect\n"
41 "    --bisect-vars"
42 ;
43
44 static struct rev_info revs;
45
46 static int bisect_list;
47 static int show_timestamp;
48 static int hdr_termination;
49 static const char *header_prefix;
50
51 static void show_commit(struct commit *commit)
52 {
53         if (show_timestamp)
54                 printf("%lu ", commit->date);
55         if (header_prefix)
56                 fputs(header_prefix, stdout);
57         if (commit->object.flags & BOUNDARY)
58                 putchar('-');
59         else if (revs.left_right) {
60                 if (commit->object.flags & SYMMETRIC_LEFT)
61                         putchar('<');
62                 else
63                         putchar('>');
64         }
65         if (revs.abbrev_commit && revs.abbrev)
66                 fputs(find_unique_abbrev(commit->object.sha1, revs.abbrev),
67                       stdout);
68         else
69                 fputs(sha1_to_hex(commit->object.sha1), stdout);
70         if (revs.parents) {
71                 struct commit_list *parents = commit->parents;
72                 while (parents) {
73                         struct object *o = &(parents->item->object);
74                         parents = parents->next;
75                         if (o->flags & TMP_MARK)
76                                 continue;
77                         printf(" %s", sha1_to_hex(o->sha1));
78                         o->flags |= TMP_MARK;
79                 }
80                 /* TMP_MARK is a general purpose flag that can
81                  * be used locally, but the user should clean
82                  * things up after it is done with them.
83                  */
84                 for (parents = commit->parents;
85                      parents;
86                      parents = parents->next)
87                         parents->item->object.flags &= ~TMP_MARK;
88         }
89         if (revs.commit_format == CMIT_FMT_ONELINE)
90                 putchar(' ');
91         else
92                 putchar('\n');
93
94         if (revs.verbose_header) {
95                 static char pretty_header[16384];
96                 pretty_print_commit(revs.commit_format, commit, ~0,
97                                     pretty_header, sizeof(pretty_header),
98                                     revs.abbrev, NULL, NULL, revs.date_mode);
99                 printf("%s%c", pretty_header, hdr_termination);
100         }
101         fflush(stdout);
102         if (commit->parents) {
103                 free_commit_list(commit->parents);
104                 commit->parents = NULL;
105         }
106         free(commit->buffer);
107         commit->buffer = NULL;
108 }
109
110 static void show_object(struct object_array_entry *p)
111 {
112         /* An object with name "foo\n0000000..." can be used to
113          * confuse downstream git-pack-objects very badly.
114          */
115         const char *ep = strchr(p->name, '\n');
116
117         if (p->item->type == OBJ_BLOB && !has_sha1_file(p->item->sha1))
118                 die("missing blob object '%s'", sha1_to_hex(p->item->sha1));
119
120         if (ep) {
121                 printf("%s %.*s\n", sha1_to_hex(p->item->sha1),
122                        (int) (ep - p->name),
123                        p->name);
124         }
125         else
126                 printf("%s %s\n", sha1_to_hex(p->item->sha1), p->name);
127 }
128
129 static void show_edge(struct commit *commit)
130 {
131         printf("-%s\n", sha1_to_hex(commit->object.sha1));
132 }
133
134 /*
135  * This is a truly stupid algorithm, but it's only
136  * used for bisection, and we just don't care enough.
137  *
138  * We care just barely enough to avoid recursing for
139  * non-merge entries.
140  */
141 static int count_distance(struct commit_list *entry)
142 {
143         int nr = 0;
144
145         while (entry) {
146                 struct commit *commit = entry->item;
147                 struct commit_list *p;
148
149                 if (commit->object.flags & (UNINTERESTING | COUNTED))
150                         break;
151                 if (!revs.prune_fn || (commit->object.flags & TREECHANGE))
152                         nr++;
153                 commit->object.flags |= COUNTED;
154                 p = commit->parents;
155                 entry = p;
156                 if (p) {
157                         p = p->next;
158                         while (p) {
159                                 nr += count_distance(p);
160                                 p = p->next;
161                         }
162                 }
163         }
164
165         return nr;
166 }
167
168 static void clear_distance(struct commit_list *list)
169 {
170         while (list) {
171                 struct commit *commit = list->item;
172                 commit->object.flags &= ~COUNTED;
173                 list = list->next;
174         }
175 }
176
177 #define DEBUG_BISECT 0
178
179 static inline int weight(struct commit_list *elem)
180 {
181         return *((int*)(elem->item->util));
182 }
183
184 static inline void weight_set(struct commit_list *elem, int weight)
185 {
186         *((int*)(elem->item->util)) = weight;
187 }
188
189 static int count_interesting_parents(struct commit *commit)
190 {
191         struct commit_list *p;
192         int count;
193
194         for (count = 0, p = commit->parents; p; p = p->next) {
195                 if (p->item->object.flags & UNINTERESTING)
196                         continue;
197                 count++;
198         }
199         return count;
200 }
201
202 static inline int halfway(struct commit_list *p, int distance, int nr)
203 {
204         /*
205          * Don't short-cut something we are not going to return!
206          */
207         if (revs.prune_fn && !(p->item->object.flags & TREECHANGE))
208                 return 0;
209         if (DEBUG_BISECT)
210                 return 0;
211         /*
212          * 2 and 3 are halfway of 5.
213          * 3 is halfway of 6 but 2 and 4 are not.
214          */
215         distance *= 2;
216         switch (distance - nr) {
217         case -1: case 0: case 1:
218                 return 1;
219         default:
220                 return 0;
221         }
222 }
223
224 #if !DEBUG_BISECT
225 #define show_list(a,b,c,d) do { ; } while (0)
226 #else
227 static void show_list(const char *debug, int counted, int nr,
228                       struct commit_list *list)
229 {
230         struct commit_list *p;
231
232         fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
233
234         for (p = list; p; p = p->next) {
235                 struct commit_list *pp;
236                 struct commit *commit = p->item;
237                 unsigned flags = commit->object.flags;
238                 enum object_type type;
239                 unsigned long size;
240                 char *buf = read_sha1_file(commit->object.sha1, &type, &size);
241                 char *ep, *sp;
242
243                 fprintf(stderr, "%c%c%c ",
244                         (flags & TREECHANGE) ? 'T' : ' ',
245                         (flags & UNINTERESTING) ? 'U' : ' ',
246                         (flags & COUNTED) ? 'C' : ' ');
247                 if (commit->util)
248                         fprintf(stderr, "%3d", weight(p));
249                 else
250                         fprintf(stderr, "---");
251                 fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.sha1));
252                 for (pp = commit->parents; pp; pp = pp->next)
253                         fprintf(stderr, " %.*s", 8,
254                                 sha1_to_hex(pp->item->object.sha1));
255
256                 sp = strstr(buf, "\n\n");
257                 if (sp) {
258                         sp += 2;
259                         for (ep = sp; *ep && *ep != '\n'; ep++)
260                                 ;
261                         fprintf(stderr, " %.*s", (int)(ep - sp), sp);
262                 }
263                 fprintf(stderr, "\n");
264         }
265 }
266 #endif /* DEBUG_BISECT */
267
268 /*
269  * zero or positive weight is the number of interesting commits it can
270  * reach, including itself.  Especially, weight = 0 means it does not
271  * reach any tree-changing commits (e.g. just above uninteresting one
272  * but traversal is with pathspec).
273  *
274  * weight = -1 means it has one parent and its distance is yet to
275  * be computed.
276  *
277  * weight = -2 means it has more than one parent and its distance is
278  * unknown.  After running count_distance() first, they will get zero
279  * or positive distance.
280  */
281
282 static struct commit_list *find_bisection(struct commit_list *list,
283                                           int *reaches, int *all)
284 {
285         int n, nr, on_list, counted, distance;
286         struct commit_list *p, *best, *next, *last;
287         int *weights;
288
289         show_list("bisection 2 entry", 0, 0, list);
290
291         /*
292          * Count the number of total and tree-changing items on the
293          * list, while reversing the list.
294          */
295         for (nr = on_list = 0, last = NULL, p = list;
296              p;
297              p = next) {
298                 unsigned flags = p->item->object.flags;
299
300                 next = p->next;
301                 if (flags & UNINTERESTING)
302                         continue;
303                 p->next = last;
304                 last = p;
305                 if (!revs.prune_fn || (flags & TREECHANGE))
306                         nr++;
307                 on_list++;
308         }
309         list = last;
310         show_list("bisection 2 sorted", 0, nr, list);
311
312         *all = nr;
313         weights = xcalloc(on_list, sizeof(int*));
314         counted = 0;
315
316         for (n = 0, p = list; p; p = p->next) {
317                 struct commit *commit = p->item;
318                 unsigned flags = commit->object.flags;
319
320                 p->item->util = &weights[n++];
321                 switch (count_interesting_parents(commit)) {
322                 case 0:
323                         if (!revs.prune_fn || (flags & TREECHANGE)) {
324                                 weight_set(p, 1);
325                                 counted++;
326                                 show_list("bisection 2 count one",
327                                           counted, nr, list);
328                         }
329                         /*
330                          * otherwise, it is known not to reach any
331                          * tree-changing commit and gets weight 0.
332                          */
333                         break;
334                 case 1:
335                         weight_set(p, -1);
336                         break;
337                 default:
338                         weight_set(p, -2);
339                         break;
340                 }
341         }
342
343         show_list("bisection 2 initialize", counted, nr, list);
344
345         /*
346          * If you have only one parent in the resulting set
347          * then you can reach one commit more than that parent
348          * can reach.  So we do not have to run the expensive
349          * count_distance() for single strand of pearls.
350          *
351          * However, if you have more than one parents, you cannot
352          * just add their distance and one for yourself, since
353          * they usually reach the same ancestor and you would
354          * end up counting them twice that way.
355          *
356          * So we will first count distance of merges the usual
357          * way, and then fill the blanks using cheaper algorithm.
358          */
359         for (p = list; p; p = p->next) {
360                 if (p->item->object.flags & UNINTERESTING)
361                         continue;
362                 n = weight(p);
363                 if (n != -2)
364                         continue;
365                 distance = count_distance(p);
366                 clear_distance(list);
367                 weight_set(p, distance);
368
369                 /* Does it happen to be at exactly half-way? */
370                 if (halfway(p, distance, nr)) {
371                         p->next = NULL;
372                         *reaches = distance;
373                         free(weights);
374                         return p;
375                 }
376                 counted++;
377         }
378
379         show_list("bisection 2 count_distance", counted, nr, list);
380
381         while (counted < nr) {
382                 for (p = list; p; p = p->next) {
383                         struct commit_list *q;
384                         unsigned flags = p->item->object.flags;
385
386                         if (0 <= weight(p))
387                                 continue;
388                         for (q = p->item->parents; q; q = q->next) {
389                                 if (q->item->object.flags & UNINTERESTING)
390                                         continue;
391                                 if (0 <= weight(q))
392                                         break;
393                         }
394                         if (!q)
395                                 continue;
396
397                         /*
398                          * weight for p is unknown but q is known.
399                          * add one for p itself if p is to be counted,
400                          * otherwise inherit it from q directly.
401                          */
402                         if (!revs.prune_fn || (flags & TREECHANGE)) {
403                                 weight_set(p, weight(q)+1);
404                                 counted++;
405                                 show_list("bisection 2 count one",
406                                           counted, nr, list);
407                         }
408                         else
409                                 weight_set(p, weight(q));
410
411                         /* Does it happen to be at exactly half-way? */
412                         distance = weight(p);
413                         if (halfway(p, distance, nr)) {
414                                 p->next = NULL;
415                                 *reaches = distance;
416                                 free(weights);
417                                 return p;
418                         }
419                 }
420         }
421
422         show_list("bisection 2 counted all", counted, nr, list);
423
424         /* Then find the best one */
425         counted = -1;
426         best = list;
427         for (p = list; p; p = p->next) {
428                 unsigned flags = p->item->object.flags;
429
430                 if (revs.prune_fn && !(flags & TREECHANGE))
431                         continue;
432                 distance = weight(p);
433                 if (nr - distance < distance)
434                         distance = nr - distance;
435                 if (distance > counted) {
436                         best = p;
437                         counted = distance;
438                         *reaches = weight(p);
439                 }
440         }
441         if (best)
442                 best->next = NULL;
443         free(weights);
444         return best;
445 }
446
447 static void read_revisions_from_stdin(struct rev_info *revs)
448 {
449         char line[1000];
450
451         while (fgets(line, sizeof(line), stdin) != NULL) {
452                 int len = strlen(line);
453                 if (line[len - 1] == '\n')
454                         line[--len] = 0;
455                 if (!len)
456                         break;
457                 if (line[0] == '-')
458                         die("options not supported in --stdin mode");
459                 if (handle_revision_arg(line, revs, 0, 1))
460                         die("bad revision '%s'", line);
461         }
462 }
463
464 int cmd_rev_list(int argc, const char **argv, const char *prefix)
465 {
466         struct commit_list *list;
467         int i;
468         int read_from_stdin = 0;
469         int bisect_show_vars = 0;
470
471         git_config(git_default_config);
472         init_revisions(&revs, prefix);
473         revs.abbrev = 0;
474         revs.commit_format = CMIT_FMT_UNSPECIFIED;
475         argc = setup_revisions(argc, argv, &revs, NULL);
476
477         for (i = 1 ; i < argc; i++) {
478                 const char *arg = argv[i];
479
480                 if (!strcmp(arg, "--header")) {
481                         revs.verbose_header = 1;
482                         continue;
483                 }
484                 if (!strcmp(arg, "--timestamp")) {
485                         show_timestamp = 1;
486                         continue;
487                 }
488                 if (!strcmp(arg, "--bisect")) {
489                         bisect_list = 1;
490                         continue;
491                 }
492                 if (!strcmp(arg, "--bisect-vars")) {
493                         bisect_list = 1;
494                         bisect_show_vars = 1;
495                         continue;
496                 }
497                 if (!strcmp(arg, "--stdin")) {
498                         if (read_from_stdin++)
499                                 die("--stdin given twice?");
500                         read_revisions_from_stdin(&revs);
501                         continue;
502                 }
503                 usage(rev_list_usage);
504
505         }
506         if (revs.commit_format != CMIT_FMT_UNSPECIFIED) {
507                 /* The command line has a --pretty  */
508                 hdr_termination = '\n';
509                 if (revs.commit_format == CMIT_FMT_ONELINE)
510                         header_prefix = "";
511                 else
512                         header_prefix = "commit ";
513         }
514         else if (revs.verbose_header)
515                 /* Only --header was specified */
516                 revs.commit_format = CMIT_FMT_RAW;
517
518         list = revs.commits;
519
520         if ((!list &&
521              (!(revs.tag_objects||revs.tree_objects||revs.blob_objects) &&
522               !revs.pending.nr)) ||
523             revs.diff)
524                 usage(rev_list_usage);
525
526         save_commit_buffer = revs.verbose_header || revs.grep_filter;
527         track_object_refs = 0;
528         if (bisect_list)
529                 revs.limited = 1;
530
531         prepare_revision_walk(&revs);
532         if (revs.tree_objects)
533                 mark_edges_uninteresting(revs.commits, &revs, show_edge);
534
535         if (bisect_list) {
536                 int reaches = reaches, all = all;
537
538                 revs.commits = find_bisection(revs.commits, &reaches, &all);
539                 if (bisect_show_vars) {
540                         int cnt;
541                         if (!revs.commits)
542                                 return 1;
543                         /*
544                          * revs.commits can reach "reaches" commits among
545                          * "all" commits.  If it is good, then there are
546                          * (all-reaches) commits left to be bisected.
547                          * On the other hand, if it is bad, then the set
548                          * to bisect is "reaches".
549                          * A bisect set of size N has (N-1) commits further
550                          * to test, as we already know one bad one.
551                          */
552                         cnt = all-reaches;
553                         if (cnt < reaches)
554                                 cnt = reaches;
555                         printf("bisect_rev=%s\n"
556                                "bisect_nr=%d\n"
557                                "bisect_good=%d\n"
558                                "bisect_bad=%d\n"
559                                "bisect_all=%d\n",
560                                sha1_to_hex(revs.commits->item->object.sha1),
561                                cnt - 1,
562                                all - reaches - 1,
563                                reaches - 1,
564                                all);
565                         return 0;
566                 }
567         }
568
569         traverse_commit_list(&revs, show_commit, show_object);
570
571         return 0;
572 }