name-rev: avoid cutoff timestamp underflow
[git] / builtin / name-rev.c
1 #include "builtin.h"
2 #include "cache.h"
3 #include "repository.h"
4 #include "config.h"
5 #include "commit.h"
6 #include "tag.h"
7 #include "refs.h"
8 #include "parse-options.h"
9 #include "sha1-lookup.h"
10 #include "commit-slab.h"
11
12 /*
13  * One day.  See the 'name a rev shortly after epoch' test in t6120 when
14  * changing this value
15  */
16 #define CUTOFF_DATE_SLOP 86400
17
18 typedef struct rev_name {
19         const char *tip_name;
20         timestamp_t taggerdate;
21         int generation;
22         int distance;
23         int from_tag;
24 } rev_name;
25
26 define_commit_slab(commit_rev_name, struct rev_name *);
27
28 static timestamp_t cutoff = TIME_MAX;
29 static struct commit_rev_name rev_names;
30
31 /* How many generations are maximally preferred over _one_ merge traversal? */
32 #define MERGE_TRAVERSAL_WEIGHT 65535
33
34 static struct rev_name *get_commit_rev_name(struct commit *commit)
35 {
36         struct rev_name **slot = commit_rev_name_peek(&rev_names, commit);
37
38         return slot ? *slot : NULL;
39 }
40
41 static void set_commit_rev_name(struct commit *commit, struct rev_name *name)
42 {
43         *commit_rev_name_at(&rev_names, commit) = name;
44 }
45
46 static int is_better_name(struct rev_name *name,
47                           timestamp_t taggerdate,
48                           int distance,
49                           int from_tag)
50 {
51         /*
52          * When comparing names based on tags, prefer names
53          * based on the older tag, even if it is farther away.
54          */
55         if (from_tag && name->from_tag)
56                 return (name->taggerdate > taggerdate ||
57                         (name->taggerdate == taggerdate &&
58                          name->distance > distance));
59
60         /*
61          * We know that at least one of them is a non-tag at this point.
62          * favor a tag over a non-tag.
63          */
64         if (name->from_tag != from_tag)
65                 return from_tag;
66
67         /*
68          * We are now looking at two non-tags.  Tiebreak to favor
69          * shorter hops.
70          */
71         if (name->distance != distance)
72                 return name->distance > distance;
73
74         /* ... or tiebreak to favor older date */
75         if (name->taggerdate != taggerdate)
76                 return name->taggerdate > taggerdate;
77
78         /* keep the current one if we cannot decide */
79         return 0;
80 }
81
82 static void name_rev(struct commit *commit,
83                 const char *tip_name, timestamp_t taggerdate,
84                 int generation, int distance, int from_tag,
85                 int deref)
86 {
87         struct rev_name *name = get_commit_rev_name(commit);
88         struct commit_list *parents;
89         int parent_number = 1;
90         char *to_free = NULL;
91
92         parse_commit(commit);
93
94         if (commit->date < cutoff)
95                 return;
96
97         if (deref) {
98                 tip_name = to_free = xstrfmt("%s^0", tip_name);
99
100                 if (generation)
101                         die("generation: %d, but deref?", generation);
102         }
103
104         if (name == NULL) {
105                 name = xmalloc(sizeof(rev_name));
106                 set_commit_rev_name(commit, name);
107                 goto copy_data;
108         } else if (is_better_name(name, taggerdate, distance, from_tag)) {
109 copy_data:
110                 name->tip_name = tip_name;
111                 name->taggerdate = taggerdate;
112                 name->generation = generation;
113                 name->distance = distance;
114                 name->from_tag = from_tag;
115         } else {
116                 free(to_free);
117                 return;
118         }
119
120         for (parents = commit->parents;
121                         parents;
122                         parents = parents->next, parent_number++) {
123                 if (parent_number > 1) {
124                         size_t len;
125                         char *new_name;
126
127                         strip_suffix(tip_name, "^0", &len);
128                         if (generation > 0)
129                                 new_name = xstrfmt("%.*s~%d^%d", (int)len, tip_name,
130                                                    generation, parent_number);
131                         else
132                                 new_name = xstrfmt("%.*s^%d", (int)len, tip_name,
133                                                    parent_number);
134
135                         name_rev(parents->item, new_name, taggerdate, 0,
136                                  distance + MERGE_TRAVERSAL_WEIGHT,
137                                  from_tag, 0);
138                 } else {
139                         name_rev(parents->item, tip_name, taggerdate,
140                                  generation + 1, distance + 1,
141                                  from_tag, 0);
142                 }
143         }
144 }
145
146 static int subpath_matches(const char *path, const char *filter)
147 {
148         const char *subpath = path;
149
150         while (subpath) {
151                 if (!wildmatch(filter, subpath, 0))
152                         return subpath - path;
153                 subpath = strchr(subpath, '/');
154                 if (subpath)
155                         subpath++;
156         }
157         return -1;
158 }
159
160 static const char *name_ref_abbrev(const char *refname, int shorten_unambiguous)
161 {
162         if (shorten_unambiguous)
163                 refname = shorten_unambiguous_ref(refname, 0);
164         else if (starts_with(refname, "refs/heads/"))
165                 refname = refname + 11;
166         else if (starts_with(refname, "refs/"))
167                 refname = refname + 5;
168         return refname;
169 }
170
171 struct name_ref_data {
172         int tags_only;
173         int name_only;
174         struct string_list ref_filters;
175         struct string_list exclude_filters;
176 };
177
178 static struct tip_table {
179         struct tip_table_entry {
180                 struct object_id oid;
181                 const char *refname;
182         } *table;
183         int nr;
184         int alloc;
185         int sorted;
186 } tip_table;
187
188 static void add_to_tip_table(const struct object_id *oid, const char *refname,
189                              int shorten_unambiguous)
190 {
191         refname = name_ref_abbrev(refname, shorten_unambiguous);
192
193         ALLOC_GROW(tip_table.table, tip_table.nr + 1, tip_table.alloc);
194         oidcpy(&tip_table.table[tip_table.nr].oid, oid);
195         tip_table.table[tip_table.nr].refname = xstrdup(refname);
196         tip_table.nr++;
197         tip_table.sorted = 0;
198 }
199
200 static int tipcmp(const void *a_, const void *b_)
201 {
202         const struct tip_table_entry *a = a_, *b = b_;
203         return oidcmp(&a->oid, &b->oid);
204 }
205
206 static int name_ref(const char *path, const struct object_id *oid, int flags, void *cb_data)
207 {
208         struct object *o = parse_object(the_repository, oid);
209         struct name_ref_data *data = cb_data;
210         int can_abbreviate_output = data->tags_only && data->name_only;
211         int deref = 0;
212         timestamp_t taggerdate = TIME_MAX;
213
214         if (data->tags_only && !starts_with(path, "refs/tags/"))
215                 return 0;
216
217         if (data->exclude_filters.nr) {
218                 struct string_list_item *item;
219
220                 for_each_string_list_item(item, &data->exclude_filters) {
221                         if (subpath_matches(path, item->string) >= 0)
222                                 return 0;
223                 }
224         }
225
226         if (data->ref_filters.nr) {
227                 struct string_list_item *item;
228                 int matched = 0;
229
230                 /* See if any of the patterns match. */
231                 for_each_string_list_item(item, &data->ref_filters) {
232                         /*
233                          * Check all patterns even after finding a match, so
234                          * that we can see if a match with a subpath exists.
235                          * When a user asked for 'refs/tags/v*' and 'v1.*',
236                          * both of which match, the user is showing her
237                          * willingness to accept a shortened output by having
238                          * the 'v1.*' in the acceptable refnames, so we
239                          * shouldn't stop when seeing 'refs/tags/v1.4' matches
240                          * 'refs/tags/v*'.  We should show it as 'v1.4'.
241                          */
242                         switch (subpath_matches(path, item->string)) {
243                         case -1: /* did not match */
244                                 break;
245                         case 0: /* matched fully */
246                                 matched = 1;
247                                 break;
248                         default: /* matched subpath */
249                                 matched = 1;
250                                 can_abbreviate_output = 1;
251                                 break;
252                         }
253                 }
254
255                 /* If none of the patterns matched, stop now */
256                 if (!matched)
257                         return 0;
258         }
259
260         add_to_tip_table(oid, path, can_abbreviate_output);
261
262         while (o && o->type == OBJ_TAG) {
263                 struct tag *t = (struct tag *) o;
264                 if (!t->tagged)
265                         break; /* broken repository */
266                 o = parse_object(the_repository, &t->tagged->oid);
267                 deref = 1;
268                 taggerdate = t->date;
269         }
270         if (o && o->type == OBJ_COMMIT) {
271                 struct commit *commit = (struct commit *)o;
272                 int from_tag = starts_with(path, "refs/tags/");
273
274                 if (taggerdate == TIME_MAX)
275                         taggerdate = ((struct commit *)o)->date;
276                 path = name_ref_abbrev(path, can_abbreviate_output);
277                 name_rev(commit, xstrdup(path), taggerdate, 0, 0,
278                          from_tag, deref);
279         }
280         return 0;
281 }
282
283 static const unsigned char *nth_tip_table_ent(size_t ix, void *table_)
284 {
285         struct tip_table_entry *table = table_;
286         return table[ix].oid.hash;
287 }
288
289 static const char *get_exact_ref_match(const struct object *o)
290 {
291         int found;
292
293         if (!tip_table.table || !tip_table.nr)
294                 return NULL;
295
296         if (!tip_table.sorted) {
297                 QSORT(tip_table.table, tip_table.nr, tipcmp);
298                 tip_table.sorted = 1;
299         }
300
301         found = sha1_pos(o->oid.hash, tip_table.table, tip_table.nr,
302                          nth_tip_table_ent);
303         if (0 <= found)
304                 return tip_table.table[found].refname;
305         return NULL;
306 }
307
308 /* may return a constant string or use "buf" as scratch space */
309 static const char *get_rev_name(const struct object *o, struct strbuf *buf)
310 {
311         struct rev_name *n;
312         struct commit *c;
313
314         if (o->type != OBJ_COMMIT)
315                 return get_exact_ref_match(o);
316         c = (struct commit *) o;
317         n = get_commit_rev_name(c);
318         if (!n)
319                 return NULL;
320
321         if (!n->generation)
322                 return n->tip_name;
323         else {
324                 int len = strlen(n->tip_name);
325                 if (len > 2 && !strcmp(n->tip_name + len - 2, "^0"))
326                         len -= 2;
327                 strbuf_reset(buf);
328                 strbuf_addf(buf, "%.*s~%d", len, n->tip_name, n->generation);
329                 return buf->buf;
330         }
331 }
332
333 static void show_name(const struct object *obj,
334                       const char *caller_name,
335                       int always, int allow_undefined, int name_only)
336 {
337         const char *name;
338         const struct object_id *oid = &obj->oid;
339         struct strbuf buf = STRBUF_INIT;
340
341         if (!name_only)
342                 printf("%s ", caller_name ? caller_name : oid_to_hex(oid));
343         name = get_rev_name(obj, &buf);
344         if (name)
345                 printf("%s\n", name);
346         else if (allow_undefined)
347                 printf("undefined\n");
348         else if (always)
349                 printf("%s\n", find_unique_abbrev(oid, DEFAULT_ABBREV));
350         else
351                 die("cannot describe '%s'", oid_to_hex(oid));
352         strbuf_release(&buf);
353 }
354
355 static char const * const name_rev_usage[] = {
356         N_("git name-rev [<options>] <commit>..."),
357         N_("git name-rev [<options>] --all"),
358         N_("git name-rev [<options>] --stdin"),
359         NULL
360 };
361
362 static void name_rev_line(char *p, struct name_ref_data *data)
363 {
364         struct strbuf buf = STRBUF_INIT;
365         int counter = 0;
366         char *p_start;
367         const unsigned hexsz = the_hash_algo->hexsz;
368
369         for (p_start = p; *p; p++) {
370 #define ishex(x) (isdigit((x)) || ((x) >= 'a' && (x) <= 'f'))
371                 if (!ishex(*p))
372                         counter = 0;
373                 else if (++counter == hexsz &&
374                          !ishex(*(p+1))) {
375                         struct object_id oid;
376                         const char *name = NULL;
377                         char c = *(p+1);
378                         int p_len = p - p_start + 1;
379
380                         counter = 0;
381
382                         *(p+1) = 0;
383                         if (!get_oid(p - (hexsz - 1), &oid)) {
384                                 struct object *o =
385                                         lookup_object(the_repository, &oid);
386                                 if (o)
387                                         name = get_rev_name(o, &buf);
388                         }
389                         *(p+1) = c;
390
391                         if (!name)
392                                 continue;
393
394                         if (data->name_only)
395                                 printf("%.*s%s", p_len - hexsz, p_start, name);
396                         else
397                                 printf("%.*s (%s)", p_len, p_start, name);
398                         p_start = p + 1;
399                 }
400         }
401
402         /* flush */
403         if (p_start != p)
404                 fwrite(p_start, p - p_start, 1, stdout);
405
406         strbuf_release(&buf);
407 }
408
409 int cmd_name_rev(int argc, const char **argv, const char *prefix)
410 {
411         struct object_array revs = OBJECT_ARRAY_INIT;
412         int all = 0, transform_stdin = 0, allow_undefined = 1, always = 0, peel_tag = 0;
413         struct name_ref_data data = { 0, 0, STRING_LIST_INIT_NODUP, STRING_LIST_INIT_NODUP };
414         struct option opts[] = {
415                 OPT_BOOL(0, "name-only", &data.name_only, N_("print only names (no SHA-1)")),
416                 OPT_BOOL(0, "tags", &data.tags_only, N_("only use tags to name the commits")),
417                 OPT_STRING_LIST(0, "refs", &data.ref_filters, N_("pattern"),
418                                    N_("only use refs matching <pattern>")),
419                 OPT_STRING_LIST(0, "exclude", &data.exclude_filters, N_("pattern"),
420                                    N_("ignore refs matching <pattern>")),
421                 OPT_GROUP(""),
422                 OPT_BOOL(0, "all", &all, N_("list all commits reachable from all refs")),
423                 OPT_BOOL(0, "stdin", &transform_stdin, N_("read from stdin")),
424                 OPT_BOOL(0, "undefined", &allow_undefined, N_("allow to print `undefined` names (default)")),
425                 OPT_BOOL(0, "always",     &always,
426                            N_("show abbreviated commit object as fallback")),
427                 {
428                         /* A Hidden OPT_BOOL */
429                         OPTION_SET_INT, 0, "peel-tag", &peel_tag, NULL,
430                         N_("dereference tags in the input (internal use)"),
431                         PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1,
432                 },
433                 OPT_END(),
434         };
435
436         init_commit_rev_name(&rev_names);
437         git_config(git_default_config, NULL);
438         argc = parse_options(argc, argv, prefix, opts, name_rev_usage, 0);
439         if (all + transform_stdin + !!argc > 1) {
440                 error("Specify either a list, or --all, not both!");
441                 usage_with_options(name_rev_usage, opts);
442         }
443         if (all || transform_stdin)
444                 cutoff = 0;
445
446         for (; argc; argc--, argv++) {
447                 struct object_id oid;
448                 struct object *object;
449                 struct commit *commit;
450
451                 if (get_oid(*argv, &oid)) {
452                         fprintf(stderr, "Could not get sha1 for %s. Skipping.\n",
453                                         *argv);
454                         continue;
455                 }
456
457                 commit = NULL;
458                 object = parse_object(the_repository, &oid);
459                 if (object) {
460                         struct object *peeled = deref_tag(the_repository,
461                                                           object, *argv, 0);
462                         if (peeled && peeled->type == OBJ_COMMIT)
463                                 commit = (struct commit *)peeled;
464                 }
465
466                 if (!object) {
467                         fprintf(stderr, "Could not get object for %s. Skipping.\n",
468                                         *argv);
469                         continue;
470                 }
471
472                 if (commit) {
473                         if (cutoff > commit->date)
474                                 cutoff = commit->date;
475                 }
476
477                 if (peel_tag) {
478                         if (!commit) {
479                                 fprintf(stderr, "Could not get commit for %s. Skipping.\n",
480                                         *argv);
481                                 continue;
482                         }
483                         object = (struct object *)commit;
484                 }
485                 add_object_array(object, *argv, &revs);
486         }
487
488         if (cutoff) {
489                 /* check for undeflow */
490                 if (cutoff > TIME_MIN + CUTOFF_DATE_SLOP)
491                         cutoff = cutoff - CUTOFF_DATE_SLOP;
492                 else
493                         cutoff = TIME_MIN;
494         }
495         for_each_ref(name_ref, &data);
496
497         if (transform_stdin) {
498                 char buffer[2048];
499
500                 while (!feof(stdin)) {
501                         char *p = fgets(buffer, sizeof(buffer), stdin);
502                         if (!p)
503                                 break;
504                         name_rev_line(p, &data);
505                 }
506         } else if (all) {
507                 int i, max;
508
509                 max = get_max_object_index();
510                 for (i = 0; i < max; i++) {
511                         struct object *obj = get_indexed_object(i);
512                         if (!obj || obj->type != OBJ_COMMIT)
513                                 continue;
514                         show_name(obj, NULL,
515                                   always, allow_undefined, data.name_only);
516                 }
517         } else {
518                 int i;
519                 for (i = 0; i < revs.nr; i++)
520                         show_name(revs.objects[i].item, revs.objects[i].name,
521                                   always, allow_undefined, data.name_only);
522         }
523
524         UNLEAK(revs);
525         return 0;
526 }