Documentation: fix git-clone manpage not to refer to itself
[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 #include "log-tree.h"
13
14 /* bits #0-15 in revision.h */
15
16 #define COUNTED         (1u<<16)
17
18 static const char rev_list_usage[] =
19 "git-rev-list [OPTION] <commit-id>... [ -- paths... ]\n"
20 "  limiting output:\n"
21 "    --max-count=nr\n"
22 "    --max-age=epoch\n"
23 "    --min-age=epoch\n"
24 "    --sparse\n"
25 "    --no-merges\n"
26 "    --remove-empty\n"
27 "    --all\n"
28 "    --stdin\n"
29 "    --quiet\n"
30 "  ordering output:\n"
31 "    --topo-order\n"
32 "    --date-order\n"
33 "  formatting output:\n"
34 "    --parents\n"
35 "    --objects | --objects-edge\n"
36 "    --unpacked\n"
37 "    --header | --pretty\n"
38 "    --abbrev=nr | --no-abbrev\n"
39 "    --abbrev-commit\n"
40 "    --left-right\n"
41 "  special purpose:\n"
42 "    --bisect\n"
43 "    --bisect-vars\n"
44 "    --bisect-all"
45 ;
46
47 static struct rev_info revs;
48
49 static int bisect_list;
50 static int show_timestamp;
51 static int hdr_termination;
52 static const char *header_prefix;
53
54 static void finish_commit(struct commit *commit);
55 static void show_commit(struct commit *commit)
56 {
57         if (show_timestamp)
58                 printf("%lu ", commit->date);
59         if (header_prefix)
60                 fputs(header_prefix, stdout);
61         if (commit->object.flags & BOUNDARY)
62                 putchar('-');
63         else if (revs.left_right) {
64                 if (commit->object.flags & SYMMETRIC_LEFT)
65                         putchar('<');
66                 else
67                         putchar('>');
68         }
69         if (revs.abbrev_commit && revs.abbrev)
70                 fputs(find_unique_abbrev(commit->object.sha1, revs.abbrev),
71                       stdout);
72         else
73                 fputs(sha1_to_hex(commit->object.sha1), stdout);
74         if (revs.parents) {
75                 struct commit_list *parents = commit->parents;
76                 while (parents) {
77                         printf(" %s", sha1_to_hex(parents->item->object.sha1));
78                         parents = parents->next;
79                 }
80         }
81         show_decorations(commit);
82         if (revs.commit_format == CMIT_FMT_ONELINE)
83                 putchar(' ');
84         else
85                 putchar('\n');
86
87         if (revs.verbose_header) {
88                 struct strbuf buf;
89                 strbuf_init(&buf, 0);
90                 pretty_print_commit(revs.commit_format, commit,
91                                     &buf, revs.abbrev, NULL, NULL,
92                                     revs.date_mode, 0);
93                 if (buf.len)
94                         printf("%s%c", buf.buf, hdr_termination);
95                 strbuf_release(&buf);
96         }
97         maybe_flush_or_die(stdout, "stdout");
98         finish_commit(commit);
99 }
100
101 static void finish_commit(struct commit *commit)
102 {
103         if (commit->parents) {
104                 free_commit_list(commit->parents);
105                 commit->parents = NULL;
106         }
107         free(commit->buffer);
108         commit->buffer = NULL;
109 }
110
111 static void finish_object(struct object_array_entry *p)
112 {
113         if (p->item->type == OBJ_BLOB && !has_sha1_file(p->item->sha1))
114                 die("missing blob object '%s'", sha1_to_hex(p->item->sha1));
115 }
116
117 static void show_object(struct object_array_entry *p)
118 {
119         /* An object with name "foo\n0000000..." can be used to
120          * confuse downstream git-pack-objects very badly.
121          */
122         const char *ep = strchr(p->name, '\n');
123
124         finish_object(p);
125         if (ep) {
126                 printf("%s %.*s\n", sha1_to_hex(p->item->sha1),
127                        (int) (ep - p->name),
128                        p->name);
129         }
130         else
131                 printf("%s %s\n", sha1_to_hex(p->item->sha1), p->name);
132 }
133
134 static void show_edge(struct commit *commit)
135 {
136         printf("-%s\n", sha1_to_hex(commit->object.sha1));
137 }
138
139 /*
140  * This is a truly stupid algorithm, but it's only
141  * used for bisection, and we just don't care enough.
142  *
143  * We care just barely enough to avoid recursing for
144  * non-merge entries.
145  */
146 static int count_distance(struct commit_list *entry)
147 {
148         int nr = 0;
149
150         while (entry) {
151                 struct commit *commit = entry->item;
152                 struct commit_list *p;
153
154                 if (commit->object.flags & (UNINTERESTING | COUNTED))
155                         break;
156                 if (!revs.prune_fn || (commit->object.flags & TREECHANGE))
157                         nr++;
158                 commit->object.flags |= COUNTED;
159                 p = commit->parents;
160                 entry = p;
161                 if (p) {
162                         p = p->next;
163                         while (p) {
164                                 nr += count_distance(p);
165                                 p = p->next;
166                         }
167                 }
168         }
169
170         return nr;
171 }
172
173 static void clear_distance(struct commit_list *list)
174 {
175         while (list) {
176                 struct commit *commit = list->item;
177                 commit->object.flags &= ~COUNTED;
178                 list = list->next;
179         }
180 }
181
182 #define DEBUG_BISECT 0
183
184 static inline int weight(struct commit_list *elem)
185 {
186         return *((int*)(elem->item->util));
187 }
188
189 static inline void weight_set(struct commit_list *elem, int weight)
190 {
191         *((int*)(elem->item->util)) = weight;
192 }
193
194 static int count_interesting_parents(struct commit *commit)
195 {
196         struct commit_list *p;
197         int count;
198
199         for (count = 0, p = commit->parents; p; p = p->next) {
200                 if (p->item->object.flags & UNINTERESTING)
201                         continue;
202                 count++;
203         }
204         return count;
205 }
206
207 static inline int halfway(struct commit_list *p, int nr)
208 {
209         /*
210          * Don't short-cut something we are not going to return!
211          */
212         if (revs.prune_fn && !(p->item->object.flags & TREECHANGE))
213                 return 0;
214         if (DEBUG_BISECT)
215                 return 0;
216         /*
217          * 2 and 3 are halfway of 5.
218          * 3 is halfway of 6 but 2 and 4 are not.
219          */
220         switch (2 * weight(p) - nr) {
221         case -1: case 0: case 1:
222                 return 1;
223         default:
224                 return 0;
225         }
226 }
227
228 #if !DEBUG_BISECT
229 #define show_list(a,b,c,d) do { ; } while (0)
230 #else
231 static void show_list(const char *debug, int counted, int nr,
232                       struct commit_list *list)
233 {
234         struct commit_list *p;
235
236         fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
237
238         for (p = list; p; p = p->next) {
239                 struct commit_list *pp;
240                 struct commit *commit = p->item;
241                 unsigned flags = commit->object.flags;
242                 enum object_type type;
243                 unsigned long size;
244                 char *buf = read_sha1_file(commit->object.sha1, &type, &size);
245                 char *ep, *sp;
246
247                 fprintf(stderr, "%c%c%c ",
248                         (flags & TREECHANGE) ? 'T' : ' ',
249                         (flags & UNINTERESTING) ? 'U' : ' ',
250                         (flags & COUNTED) ? 'C' : ' ');
251                 if (commit->util)
252                         fprintf(stderr, "%3d", weight(p));
253                 else
254                         fprintf(stderr, "---");
255                 fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.sha1));
256                 for (pp = commit->parents; pp; pp = pp->next)
257                         fprintf(stderr, " %.*s", 8,
258                                 sha1_to_hex(pp->item->object.sha1));
259
260                 sp = strstr(buf, "\n\n");
261                 if (sp) {
262                         sp += 2;
263                         for (ep = sp; *ep && *ep != '\n'; ep++)
264                                 ;
265                         fprintf(stderr, " %.*s", (int)(ep - sp), sp);
266                 }
267                 fprintf(stderr, "\n");
268         }
269 }
270 #endif /* DEBUG_BISECT */
271
272 static struct commit_list *best_bisection(struct commit_list *list, int nr)
273 {
274         struct commit_list *p, *best;
275         int best_distance = -1;
276
277         best = list;
278         for (p = list; p; p = p->next) {
279                 int distance;
280                 unsigned flags = p->item->object.flags;
281
282                 if (revs.prune_fn && !(flags & TREECHANGE))
283                         continue;
284                 distance = weight(p);
285                 if (nr - distance < distance)
286                         distance = nr - distance;
287                 if (distance > best_distance) {
288                         best = p;
289                         best_distance = distance;
290                 }
291         }
292
293         return best;
294 }
295
296 struct commit_dist {
297         struct commit *commit;
298         int distance;
299 };
300
301 static int compare_commit_dist(const void *a_, const void *b_)
302 {
303         struct commit_dist *a, *b;
304
305         a = (struct commit_dist *)a_;
306         b = (struct commit_dist *)b_;
307         if (a->distance != b->distance)
308                 return b->distance - a->distance; /* desc sort */
309         return hashcmp(a->commit->object.sha1, b->commit->object.sha1);
310 }
311
312 static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
313 {
314         struct commit_list *p;
315         struct commit_dist *array = xcalloc(nr, sizeof(*array));
316         int cnt, i;
317
318         for (p = list, cnt = 0; p; p = p->next) {
319                 int distance;
320                 unsigned flags = p->item->object.flags;
321
322                 if (revs.prune_fn && !(flags & TREECHANGE))
323                         continue;
324                 distance = weight(p);
325                 if (nr - distance < distance)
326                         distance = nr - distance;
327                 array[cnt].commit = p->item;
328                 array[cnt].distance = distance;
329                 cnt++;
330         }
331         qsort(array, cnt, sizeof(*array), compare_commit_dist);
332         for (p = list, i = 0; i < cnt; i++) {
333                 struct name_decoration *r = xmalloc(sizeof(*r) + 100);
334                 struct object *obj = &(array[i].commit->object);
335
336                 sprintf(r->name, "dist=%d", array[i].distance);
337                 r->next = add_decoration(&name_decoration, obj, r);
338                 p->item = array[i].commit;
339                 p = p->next;
340         }
341         if (p)
342                 p->next = NULL;
343         free(array);
344         return list;
345 }
346
347 /*
348  * zero or positive weight is the number of interesting commits it can
349  * reach, including itself.  Especially, weight = 0 means it does not
350  * reach any tree-changing commits (e.g. just above uninteresting one
351  * but traversal is with pathspec).
352  *
353  * weight = -1 means it has one parent and its distance is yet to
354  * be computed.
355  *
356  * weight = -2 means it has more than one parent and its distance is
357  * unknown.  After running count_distance() first, they will get zero
358  * or positive distance.
359  */
360 static struct commit_list *do_find_bisection(struct commit_list *list,
361                                              int nr, int *weights,
362                                              int find_all)
363 {
364         int n, counted;
365         struct commit_list *p;
366
367         counted = 0;
368
369         for (n = 0, p = list; p; p = p->next) {
370                 struct commit *commit = p->item;
371                 unsigned flags = commit->object.flags;
372
373                 p->item->util = &weights[n++];
374                 switch (count_interesting_parents(commit)) {
375                 case 0:
376                         if (!revs.prune_fn || (flags & TREECHANGE)) {
377                                 weight_set(p, 1);
378                                 counted++;
379                                 show_list("bisection 2 count one",
380                                           counted, nr, list);
381                         }
382                         /*
383                          * otherwise, it is known not to reach any
384                          * tree-changing commit and gets weight 0.
385                          */
386                         break;
387                 case 1:
388                         weight_set(p, -1);
389                         break;
390                 default:
391                         weight_set(p, -2);
392                         break;
393                 }
394         }
395
396         show_list("bisection 2 initialize", counted, nr, list);
397
398         /*
399          * If you have only one parent in the resulting set
400          * then you can reach one commit more than that parent
401          * can reach.  So we do not have to run the expensive
402          * count_distance() for single strand of pearls.
403          *
404          * However, if you have more than one parents, you cannot
405          * just add their distance and one for yourself, since
406          * they usually reach the same ancestor and you would
407          * end up counting them twice that way.
408          *
409          * So we will first count distance of merges the usual
410          * way, and then fill the blanks using cheaper algorithm.
411          */
412         for (p = list; p; p = p->next) {
413                 if (p->item->object.flags & UNINTERESTING)
414                         continue;
415                 if (weight(p) != -2)
416                         continue;
417                 weight_set(p, count_distance(p));
418                 clear_distance(list);
419
420                 /* Does it happen to be at exactly half-way? */
421                 if (!find_all && halfway(p, nr))
422                         return p;
423                 counted++;
424         }
425
426         show_list("bisection 2 count_distance", counted, nr, list);
427
428         while (counted < nr) {
429                 for (p = list; p; p = p->next) {
430                         struct commit_list *q;
431                         unsigned flags = p->item->object.flags;
432
433                         if (0 <= weight(p))
434                                 continue;
435                         for (q = p->item->parents; q; q = q->next) {
436                                 if (q->item->object.flags & UNINTERESTING)
437                                         continue;
438                                 if (0 <= weight(q))
439                                         break;
440                         }
441                         if (!q)
442                                 continue;
443
444                         /*
445                          * weight for p is unknown but q is known.
446                          * add one for p itself if p is to be counted,
447                          * otherwise inherit it from q directly.
448                          */
449                         if (!revs.prune_fn || (flags & TREECHANGE)) {
450                                 weight_set(p, weight(q)+1);
451                                 counted++;
452                                 show_list("bisection 2 count one",
453                                           counted, nr, list);
454                         }
455                         else
456                                 weight_set(p, weight(q));
457
458                         /* Does it happen to be at exactly half-way? */
459                         if (!find_all && halfway(p, nr))
460                                 return p;
461                 }
462         }
463
464         show_list("bisection 2 counted all", counted, nr, list);
465
466         if (!find_all)
467                 return best_bisection(list, nr);
468         else
469                 return best_bisection_sorted(list, nr);
470 }
471
472 static struct commit_list *find_bisection(struct commit_list *list,
473                                           int *reaches, int *all,
474                                           int find_all)
475 {
476         int nr, on_list;
477         struct commit_list *p, *best, *next, *last;
478         int *weights;
479
480         show_list("bisection 2 entry", 0, 0, list);
481
482         /*
483          * Count the number of total and tree-changing items on the
484          * list, while reversing the list.
485          */
486         for (nr = on_list = 0, last = NULL, p = list;
487              p;
488              p = next) {
489                 unsigned flags = p->item->object.flags;
490
491                 next = p->next;
492                 if (flags & UNINTERESTING)
493                         continue;
494                 p->next = last;
495                 last = p;
496                 if (!revs.prune_fn || (flags & TREECHANGE))
497                         nr++;
498                 on_list++;
499         }
500         list = last;
501         show_list("bisection 2 sorted", 0, nr, list);
502
503         *all = nr;
504         weights = xcalloc(on_list, sizeof(*weights));
505
506         /* Do the real work of finding bisection commit. */
507         best = do_find_bisection(list, nr, weights, find_all);
508         if (best) {
509                 if (!find_all)
510                         best->next = NULL;
511                 *reaches = weight(best);
512         }
513         free(weights);
514         return best;
515 }
516
517 static void read_revisions_from_stdin(struct rev_info *revs)
518 {
519         char line[1000];
520
521         while (fgets(line, sizeof(line), stdin) != NULL) {
522                 int len = strlen(line);
523                 if (line[len - 1] == '\n')
524                         line[--len] = 0;
525                 if (!len)
526                         break;
527                 if (line[0] == '-')
528                         die("options not supported in --stdin mode");
529                 if (handle_revision_arg(line, revs, 0, 1))
530                         die("bad revision '%s'", line);
531         }
532 }
533
534 int cmd_rev_list(int argc, const char **argv, const char *prefix)
535 {
536         struct commit_list *list;
537         int i;
538         int read_from_stdin = 0;
539         int bisect_show_vars = 0;
540         int bisect_find_all = 0;
541         int quiet = 0;
542
543         git_config(git_default_config);
544         init_revisions(&revs, prefix);
545         revs.abbrev = 0;
546         revs.commit_format = CMIT_FMT_UNSPECIFIED;
547         argc = setup_revisions(argc, argv, &revs, NULL);
548
549         for (i = 1 ; i < argc; i++) {
550                 const char *arg = argv[i];
551
552                 if (!strcmp(arg, "--header")) {
553                         revs.verbose_header = 1;
554                         continue;
555                 }
556                 if (!strcmp(arg, "--timestamp")) {
557                         show_timestamp = 1;
558                         continue;
559                 }
560                 if (!strcmp(arg, "--bisect")) {
561                         bisect_list = 1;
562                         continue;
563                 }
564                 if (!strcmp(arg, "--bisect-all")) {
565                         bisect_list = 1;
566                         bisect_find_all = 1;
567                         continue;
568                 }
569                 if (!strcmp(arg, "--bisect-vars")) {
570                         bisect_list = 1;
571                         bisect_show_vars = 1;
572                         continue;
573                 }
574                 if (!strcmp(arg, "--stdin")) {
575                         if (read_from_stdin++)
576                                 die("--stdin given twice?");
577                         read_revisions_from_stdin(&revs);
578                         continue;
579                 }
580                 if (!strcmp(arg, "--quiet")) {
581                         quiet = 1;
582                         continue;
583                 }
584                 usage(rev_list_usage);
585
586         }
587         if (revs.commit_format != CMIT_FMT_UNSPECIFIED) {
588                 /* The command line has a --pretty  */
589                 hdr_termination = '\n';
590                 if (revs.commit_format == CMIT_FMT_ONELINE)
591                         header_prefix = "";
592                 else
593                         header_prefix = "commit ";
594         }
595         else if (revs.verbose_header)
596                 /* Only --header was specified */
597                 revs.commit_format = CMIT_FMT_RAW;
598
599         list = revs.commits;
600
601         if ((!list &&
602              (!(revs.tag_objects||revs.tree_objects||revs.blob_objects) &&
603               !revs.pending.nr)) ||
604             revs.diff)
605                 usage(rev_list_usage);
606
607         save_commit_buffer = revs.verbose_header || revs.grep_filter;
608         track_object_refs = 0;
609         if (bisect_list)
610                 revs.limited = 1;
611
612         prepare_revision_walk(&revs);
613         if (revs.tree_objects)
614                 mark_edges_uninteresting(revs.commits, &revs, show_edge);
615
616         if (bisect_list) {
617                 int reaches = reaches, all = all;
618
619                 revs.commits = find_bisection(revs.commits, &reaches, &all,
620                                               bisect_find_all);
621                 if (bisect_show_vars) {
622                         int cnt;
623                         char hex[41];
624                         if (!revs.commits)
625                                 return 1;
626                         /*
627                          * revs.commits can reach "reaches" commits among
628                          * "all" commits.  If it is good, then there are
629                          * (all-reaches) commits left to be bisected.
630                          * On the other hand, if it is bad, then the set
631                          * to bisect is "reaches".
632                          * A bisect set of size N has (N-1) commits further
633                          * to test, as we already know one bad one.
634                          */
635                         cnt = all - reaches;
636                         if (cnt < reaches)
637                                 cnt = reaches;
638                         strcpy(hex, sha1_to_hex(revs.commits->item->object.sha1));
639
640                         if (bisect_find_all) {
641                                 traverse_commit_list(&revs, show_commit, show_object);
642                                 printf("------\n");
643                         }
644
645                         printf("bisect_rev=%s\n"
646                                "bisect_nr=%d\n"
647                                "bisect_good=%d\n"
648                                "bisect_bad=%d\n"
649                                "bisect_all=%d\n",
650                                hex,
651                                cnt - 1,
652                                all - reaches - 1,
653                                reaches - 1,
654                                all);
655                         return 0;
656                 }
657         }
658
659         traverse_commit_list(&revs,
660                 quiet ? finish_commit : show_commit,
661                 quiet ? finish_object : show_object);
662
663         return 0;
664 }