Merge branch 'jk/submodule-name-verify-fix' into jk/submodule-name-verify-fsck
[git] / submodule-config.c
1 #include "cache.h"
2 #include "repository.h"
3 #include "config.h"
4 #include "submodule-config.h"
5 #include "submodule.h"
6 #include "strbuf.h"
7 #include "parse-options.h"
8
9 /*
10  * submodule cache lookup structure
11  * There is one shared set of 'struct submodule' entries which can be
12  * looked up by their sha1 blob id of the .gitmodules file and either
13  * using path or name as key.
14  * for_path stores submodule entries with path as key
15  * for_name stores submodule entries with name as key
16  */
17 struct submodule_cache {
18         struct hashmap for_path;
19         struct hashmap for_name;
20         unsigned initialized:1;
21         unsigned gitmodules_read:1;
22 };
23
24 /*
25  * thin wrapper struct needed to insert 'struct submodule' entries to
26  * the hashmap
27  */
28 struct submodule_entry {
29         struct hashmap_entry ent;
30         struct submodule *config;
31 };
32
33 enum lookup_type {
34         lookup_name,
35         lookup_path
36 };
37
38 static int config_path_cmp(const void *unused_cmp_data,
39                            const void *entry,
40                            const void *entry_or_key,
41                            const void *unused_keydata)
42 {
43         const struct submodule_entry *a = entry;
44         const struct submodule_entry *b = entry_or_key;
45
46         return strcmp(a->config->path, b->config->path) ||
47                hashcmp(a->config->gitmodules_sha1, b->config->gitmodules_sha1);
48 }
49
50 static int config_name_cmp(const void *unused_cmp_data,
51                            const void *entry,
52                            const void *entry_or_key,
53                            const void *unused_keydata)
54 {
55         const struct submodule_entry *a = entry;
56         const struct submodule_entry *b = entry_or_key;
57
58         return strcmp(a->config->name, b->config->name) ||
59                hashcmp(a->config->gitmodules_sha1, b->config->gitmodules_sha1);
60 }
61
62 static struct submodule_cache *submodule_cache_alloc(void)
63 {
64         return xcalloc(1, sizeof(struct submodule_cache));
65 }
66
67 static void submodule_cache_init(struct submodule_cache *cache)
68 {
69         hashmap_init(&cache->for_path, config_path_cmp, NULL, 0);
70         hashmap_init(&cache->for_name, config_name_cmp, NULL, 0);
71         cache->initialized = 1;
72 }
73
74 static void free_one_config(struct submodule_entry *entry)
75 {
76         free((void *) entry->config->path);
77         free((void *) entry->config->name);
78         free((void *) entry->config->branch);
79         free((void *) entry->config->update_strategy.command);
80         free(entry->config);
81 }
82
83 static void submodule_cache_clear(struct submodule_cache *cache)
84 {
85         struct hashmap_iter iter;
86         struct submodule_entry *entry;
87
88         if (!cache->initialized)
89                 return;
90
91         /*
92          * We iterate over the name hash here to be symmetric with the
93          * allocation of struct submodule entries. Each is allocated by
94          * their .gitmodules blob sha1 and submodule name.
95          */
96         hashmap_iter_init(&cache->for_name, &iter);
97         while ((entry = hashmap_iter_next(&iter)))
98                 free_one_config(entry);
99
100         hashmap_free(&cache->for_path, 1);
101         hashmap_free(&cache->for_name, 1);
102         cache->initialized = 0;
103         cache->gitmodules_read = 0;
104 }
105
106 void submodule_cache_free(struct submodule_cache *cache)
107 {
108         submodule_cache_clear(cache);
109         free(cache);
110 }
111
112 static unsigned int hash_sha1_string(const unsigned char *sha1,
113                                      const char *string)
114 {
115         return memhash(sha1, 20) + strhash(string);
116 }
117
118 static void cache_put_path(struct submodule_cache *cache,
119                            struct submodule *submodule)
120 {
121         unsigned int hash = hash_sha1_string(submodule->gitmodules_sha1,
122                                              submodule->path);
123         struct submodule_entry *e = xmalloc(sizeof(*e));
124         hashmap_entry_init(e, hash);
125         e->config = submodule;
126         hashmap_put(&cache->for_path, e);
127 }
128
129 static void cache_remove_path(struct submodule_cache *cache,
130                               struct submodule *submodule)
131 {
132         unsigned int hash = hash_sha1_string(submodule->gitmodules_sha1,
133                                              submodule->path);
134         struct submodule_entry e;
135         struct submodule_entry *removed;
136         hashmap_entry_init(&e, hash);
137         e.config = submodule;
138         removed = hashmap_remove(&cache->for_path, &e, NULL);
139         free(removed);
140 }
141
142 static void cache_add(struct submodule_cache *cache,
143                       struct submodule *submodule)
144 {
145         unsigned int hash = hash_sha1_string(submodule->gitmodules_sha1,
146                                              submodule->name);
147         struct submodule_entry *e = xmalloc(sizeof(*e));
148         hashmap_entry_init(e, hash);
149         e->config = submodule;
150         hashmap_add(&cache->for_name, e);
151 }
152
153 static const struct submodule *cache_lookup_path(struct submodule_cache *cache,
154                 const unsigned char *gitmodules_sha1, const char *path)
155 {
156         struct submodule_entry *entry;
157         unsigned int hash = hash_sha1_string(gitmodules_sha1, path);
158         struct submodule_entry key;
159         struct submodule key_config;
160
161         hashcpy(key_config.gitmodules_sha1, gitmodules_sha1);
162         key_config.path = path;
163
164         hashmap_entry_init(&key, hash);
165         key.config = &key_config;
166
167         entry = hashmap_get(&cache->for_path, &key, NULL);
168         if (entry)
169                 return entry->config;
170         return NULL;
171 }
172
173 static struct submodule *cache_lookup_name(struct submodule_cache *cache,
174                 const unsigned char *gitmodules_sha1, const char *name)
175 {
176         struct submodule_entry *entry;
177         unsigned int hash = hash_sha1_string(gitmodules_sha1, name);
178         struct submodule_entry key;
179         struct submodule key_config;
180
181         hashcpy(key_config.gitmodules_sha1, gitmodules_sha1);
182         key_config.name = name;
183
184         hashmap_entry_init(&key, hash);
185         key.config = &key_config;
186
187         entry = hashmap_get(&cache->for_name, &key, NULL);
188         if (entry)
189                 return entry->config;
190         return NULL;
191 }
192
193 int check_submodule_name(const char *name)
194 {
195         /* Disallow empty names */
196         if (!*name)
197                 return -1;
198
199         /*
200          * Look for '..' as a path component. Check both '/' and '\\' as
201          * separators rather than is_dir_sep(), because we want the name rules
202          * to be consistent across platforms.
203          */
204         goto in_component; /* always start inside component */
205         while (*name) {
206                 char c = *name++;
207                 if (c == '/' || c == '\\') {
208 in_component:
209                         if (name[0] == '.' && name[1] == '.' &&
210                             (!name[2] || name[2] == '/' || name[2] == '\\'))
211                                 return -1;
212                 }
213         }
214
215         return 0;
216 }
217
218 static int name_and_item_from_var(const char *var, struct strbuf *name,
219                                   struct strbuf *item)
220 {
221         const char *subsection, *key;
222         int subsection_len, parse;
223         parse = parse_config_key(var, "submodule", &subsection,
224                         &subsection_len, &key);
225         if (parse < 0 || !subsection)
226                 return 0;
227
228         strbuf_add(name, subsection, subsection_len);
229         if (check_submodule_name(name->buf) < 0) {
230                 warning(_("ignoring suspicious submodule name: %s"), name->buf);
231                 strbuf_release(name);
232                 return 0;
233         }
234
235         strbuf_addstr(item, key);
236
237         return 1;
238 }
239
240 static struct submodule *lookup_or_create_by_name(struct submodule_cache *cache,
241                 const unsigned char *gitmodules_sha1, const char *name)
242 {
243         struct submodule *submodule;
244         struct strbuf name_buf = STRBUF_INIT;
245
246         submodule = cache_lookup_name(cache, gitmodules_sha1, name);
247         if (submodule)
248                 return submodule;
249
250         submodule = xmalloc(sizeof(*submodule));
251
252         strbuf_addstr(&name_buf, name);
253         submodule->name = strbuf_detach(&name_buf, NULL);
254
255         submodule->path = NULL;
256         submodule->url = NULL;
257         submodule->update_strategy.type = SM_UPDATE_UNSPECIFIED;
258         submodule->update_strategy.command = NULL;
259         submodule->fetch_recurse = RECURSE_SUBMODULES_NONE;
260         submodule->ignore = NULL;
261         submodule->branch = NULL;
262         submodule->recommend_shallow = -1;
263
264         hashcpy(submodule->gitmodules_sha1, gitmodules_sha1);
265
266         cache_add(cache, submodule);
267
268         return submodule;
269 }
270
271 static int parse_fetch_recurse(const char *opt, const char *arg,
272                                int die_on_error)
273 {
274         switch (git_parse_maybe_bool(arg)) {
275         case 1:
276                 return RECURSE_SUBMODULES_ON;
277         case 0:
278                 return RECURSE_SUBMODULES_OFF;
279         default:
280                 if (!strcmp(arg, "on-demand"))
281                         return RECURSE_SUBMODULES_ON_DEMAND;
282
283                 if (die_on_error)
284                         die("bad %s argument: %s", opt, arg);
285                 else
286                         return RECURSE_SUBMODULES_ERROR;
287         }
288 }
289
290 int parse_submodule_fetchjobs(const char *var, const char *value)
291 {
292         int fetchjobs = git_config_int(var, value);
293         if (fetchjobs < 0)
294                 die(_("negative values not allowed for submodule.fetchjobs"));
295         return fetchjobs;
296 }
297
298 int parse_fetch_recurse_submodules_arg(const char *opt, const char *arg)
299 {
300         return parse_fetch_recurse(opt, arg, 1);
301 }
302
303 int option_fetch_parse_recurse_submodules(const struct option *opt,
304                                           const char *arg, int unset)
305 {
306         int *v;
307
308         if (!opt->value)
309                 return -1;
310
311         v = opt->value;
312
313         if (unset) {
314                 *v = RECURSE_SUBMODULES_OFF;
315         } else {
316                 if (arg)
317                         *v = parse_fetch_recurse_submodules_arg(opt->long_name, arg);
318                 else
319                         *v = RECURSE_SUBMODULES_ON;
320         }
321         return 0;
322 }
323
324 static int parse_update_recurse(const char *opt, const char *arg,
325                                 int die_on_error)
326 {
327         switch (git_parse_maybe_bool(arg)) {
328         case 1:
329                 return RECURSE_SUBMODULES_ON;
330         case 0:
331                 return RECURSE_SUBMODULES_OFF;
332         default:
333                 if (die_on_error)
334                         die("bad %s argument: %s", opt, arg);
335                 return RECURSE_SUBMODULES_ERROR;
336         }
337 }
338
339 int parse_update_recurse_submodules_arg(const char *opt, const char *arg)
340 {
341         return parse_update_recurse(opt, arg, 1);
342 }
343
344 static int parse_push_recurse(const char *opt, const char *arg,
345                                int die_on_error)
346 {
347         switch (git_parse_maybe_bool(arg)) {
348         case 1:
349                 /* There's no simple "on" value when pushing */
350                 if (die_on_error)
351                         die("bad %s argument: %s", opt, arg);
352                 else
353                         return RECURSE_SUBMODULES_ERROR;
354         case 0:
355                 return RECURSE_SUBMODULES_OFF;
356         default:
357                 if (!strcmp(arg, "on-demand"))
358                         return RECURSE_SUBMODULES_ON_DEMAND;
359                 else if (!strcmp(arg, "check"))
360                         return RECURSE_SUBMODULES_CHECK;
361                 else if (!strcmp(arg, "only"))
362                         return RECURSE_SUBMODULES_ONLY;
363                 else if (die_on_error)
364                         die("bad %s argument: %s", opt, arg);
365                 else
366                         return RECURSE_SUBMODULES_ERROR;
367         }
368 }
369
370 int parse_push_recurse_submodules_arg(const char *opt, const char *arg)
371 {
372         return parse_push_recurse(opt, arg, 1);
373 }
374
375 static void warn_multiple_config(const unsigned char *treeish_name,
376                                  const char *name, const char *option)
377 {
378         const char *commit_string = "WORKTREE";
379         if (treeish_name)
380                 commit_string = sha1_to_hex(treeish_name);
381         warning("%s:.gitmodules, multiple configurations found for "
382                         "'submodule.%s.%s'. Skipping second one!",
383                         commit_string, name, option);
384 }
385
386 struct parse_config_parameter {
387         struct submodule_cache *cache;
388         const unsigned char *treeish_name;
389         const unsigned char *gitmodules_sha1;
390         int overwrite;
391 };
392
393 static int parse_config(const char *var, const char *value, void *data)
394 {
395         struct parse_config_parameter *me = data;
396         struct submodule *submodule;
397         struct strbuf name = STRBUF_INIT, item = STRBUF_INIT;
398         int ret = 0;
399
400         /* this also ensures that we only parse submodule entries */
401         if (!name_and_item_from_var(var, &name, &item))
402                 return 0;
403
404         submodule = lookup_or_create_by_name(me->cache,
405                                              me->gitmodules_sha1,
406                                              name.buf);
407
408         if (!strcmp(item.buf, "path")) {
409                 if (!value)
410                         ret = config_error_nonbool(var);
411                 else if (!me->overwrite && submodule->path)
412                         warn_multiple_config(me->treeish_name, submodule->name,
413                                         "path");
414                 else {
415                         if (submodule->path)
416                                 cache_remove_path(me->cache, submodule);
417                         free((void *) submodule->path);
418                         submodule->path = xstrdup(value);
419                         cache_put_path(me->cache, submodule);
420                 }
421         } else if (!strcmp(item.buf, "fetchrecursesubmodules")) {
422                 /* when parsing worktree configurations we can die early */
423                 int die_on_error = is_null_sha1(me->gitmodules_sha1);
424                 if (!me->overwrite &&
425                     submodule->fetch_recurse != RECURSE_SUBMODULES_NONE)
426                         warn_multiple_config(me->treeish_name, submodule->name,
427                                         "fetchrecursesubmodules");
428                 else
429                         submodule->fetch_recurse = parse_fetch_recurse(
430                                                                 var, value,
431                                                                 die_on_error);
432         } else if (!strcmp(item.buf, "ignore")) {
433                 if (!value)
434                         ret = config_error_nonbool(var);
435                 else if (!me->overwrite && submodule->ignore)
436                         warn_multiple_config(me->treeish_name, submodule->name,
437                                         "ignore");
438                 else if (strcmp(value, "untracked") &&
439                          strcmp(value, "dirty") &&
440                          strcmp(value, "all") &&
441                          strcmp(value, "none"))
442                         warning("Invalid parameter '%s' for config option "
443                                         "'submodule.%s.ignore'", value, name.buf);
444                 else {
445                         free((void *) submodule->ignore);
446                         submodule->ignore = xstrdup(value);
447                 }
448         } else if (!strcmp(item.buf, "url")) {
449                 if (!value) {
450                         ret = config_error_nonbool(var);
451                 } else if (!me->overwrite && submodule->url) {
452                         warn_multiple_config(me->treeish_name, submodule->name,
453                                         "url");
454                 } else {
455                         free((void *) submodule->url);
456                         submodule->url = xstrdup(value);
457                 }
458         } else if (!strcmp(item.buf, "update")) {
459                 if (!value)
460                         ret = config_error_nonbool(var);
461                 else if (!me->overwrite &&
462                          submodule->update_strategy.type != SM_UPDATE_UNSPECIFIED)
463                         warn_multiple_config(me->treeish_name, submodule->name,
464                                              "update");
465                 else if (parse_submodule_update_strategy(value,
466                          &submodule->update_strategy) < 0)
467                                 die(_("invalid value for %s"), var);
468         } else if (!strcmp(item.buf, "shallow")) {
469                 if (!me->overwrite && submodule->recommend_shallow != -1)
470                         warn_multiple_config(me->treeish_name, submodule->name,
471                                              "shallow");
472                 else
473                         submodule->recommend_shallow =
474                                 git_config_bool(var, value);
475         } else if (!strcmp(item.buf, "branch")) {
476                 if (!me->overwrite && submodule->branch)
477                         warn_multiple_config(me->treeish_name, submodule->name,
478                                              "branch");
479                 else {
480                         free((void *)submodule->branch);
481                         submodule->branch = xstrdup(value);
482                 }
483         }
484
485         strbuf_release(&name);
486         strbuf_release(&item);
487
488         return ret;
489 }
490
491 static int gitmodule_oid_from_commit(const struct object_id *treeish_name,
492                                      struct object_id *gitmodules_oid,
493                                      struct strbuf *rev)
494 {
495         int ret = 0;
496
497         if (is_null_oid(treeish_name)) {
498                 oidclr(gitmodules_oid);
499                 return 1;
500         }
501
502         strbuf_addf(rev, "%s:.gitmodules", oid_to_hex(treeish_name));
503         if (get_oid(rev->buf, gitmodules_oid) >= 0)
504                 ret = 1;
505
506         return ret;
507 }
508
509 /* This does a lookup of a submodule configuration by name or by path
510  * (key) with on-demand reading of the appropriate .gitmodules from
511  * revisions.
512  */
513 static const struct submodule *config_from(struct submodule_cache *cache,
514                 const struct object_id *treeish_name, const char *key,
515                 enum lookup_type lookup_type)
516 {
517         struct strbuf rev = STRBUF_INIT;
518         unsigned long config_size;
519         char *config = NULL;
520         struct object_id oid;
521         enum object_type type;
522         const struct submodule *submodule = NULL;
523         struct parse_config_parameter parameter;
524
525         /*
526          * If any parameter except the cache is a NULL pointer just
527          * return the first submodule. Can be used to check whether
528          * there are any submodules parsed.
529          */
530         if (!treeish_name || !key) {
531                 struct hashmap_iter iter;
532                 struct submodule_entry *entry;
533
534                 entry = hashmap_iter_first(&cache->for_name, &iter);
535                 if (!entry)
536                         return NULL;
537                 return entry->config;
538         }
539
540         if (!gitmodule_oid_from_commit(treeish_name, &oid, &rev))
541                 goto out;
542
543         switch (lookup_type) {
544         case lookup_name:
545                 submodule = cache_lookup_name(cache, oid.hash, key);
546                 break;
547         case lookup_path:
548                 submodule = cache_lookup_path(cache, oid.hash, key);
549                 break;
550         }
551         if (submodule)
552                 goto out;
553
554         config = read_sha1_file(oid.hash, &type, &config_size);
555         if (!config || type != OBJ_BLOB)
556                 goto out;
557
558         /* fill the submodule config into the cache */
559         parameter.cache = cache;
560         parameter.treeish_name = treeish_name->hash;
561         parameter.gitmodules_sha1 = oid.hash;
562         parameter.overwrite = 0;
563         git_config_from_mem(parse_config, CONFIG_ORIGIN_SUBMODULE_BLOB, rev.buf,
564                         config, config_size, &parameter);
565         strbuf_release(&rev);
566         free(config);
567
568         switch (lookup_type) {
569         case lookup_name:
570                 return cache_lookup_name(cache, oid.hash, key);
571         case lookup_path:
572                 return cache_lookup_path(cache, oid.hash, key);
573         default:
574                 return NULL;
575         }
576
577 out:
578         strbuf_release(&rev);
579         free(config);
580         return submodule;
581 }
582
583 static void submodule_cache_check_init(struct repository *repo)
584 {
585         if (repo->submodule_cache && repo->submodule_cache->initialized)
586                 return;
587
588         if (!repo->submodule_cache)
589                 repo->submodule_cache = submodule_cache_alloc();
590
591         submodule_cache_init(repo->submodule_cache);
592 }
593
594 static int gitmodules_cb(const char *var, const char *value, void *data)
595 {
596         struct repository *repo = data;
597         struct parse_config_parameter parameter;
598
599         parameter.cache = repo->submodule_cache;
600         parameter.treeish_name = NULL;
601         parameter.gitmodules_sha1 = null_sha1;
602         parameter.overwrite = 1;
603
604         return parse_config(var, value, &parameter);
605 }
606
607 void repo_read_gitmodules(struct repository *repo)
608 {
609         submodule_cache_check_init(repo);
610
611         if (repo->worktree) {
612                 char *gitmodules;
613
614                 if (repo_read_index(repo) < 0)
615                         return;
616
617                 gitmodules = repo_worktree_path(repo, GITMODULES_FILE);
618
619                 if (!is_gitmodules_unmerged(repo->index))
620                         git_config_from_file(gitmodules_cb, gitmodules, repo);
621
622                 free(gitmodules);
623         }
624
625         repo->submodule_cache->gitmodules_read = 1;
626 }
627
628 void gitmodules_config_oid(const struct object_id *commit_oid)
629 {
630         struct strbuf rev = STRBUF_INIT;
631         struct object_id oid;
632
633         submodule_cache_check_init(the_repository);
634
635         if (gitmodule_oid_from_commit(commit_oid, &oid, &rev)) {
636                 git_config_from_blob_oid(gitmodules_cb, rev.buf,
637                                          &oid, the_repository);
638         }
639         strbuf_release(&rev);
640
641         the_repository->submodule_cache->gitmodules_read = 1;
642 }
643
644 static void gitmodules_read_check(struct repository *repo)
645 {
646         submodule_cache_check_init(repo);
647
648         /* read the repo's .gitmodules file if it hasn't been already */
649         if (!repo->submodule_cache->gitmodules_read)
650                 repo_read_gitmodules(repo);
651 }
652
653 const struct submodule *submodule_from_name(const struct object_id *treeish_name,
654                 const char *name)
655 {
656         gitmodules_read_check(the_repository);
657         return config_from(the_repository->submodule_cache, treeish_name, name, lookup_name);
658 }
659
660 const struct submodule *submodule_from_path(const struct object_id *treeish_name,
661                 const char *path)
662 {
663         gitmodules_read_check(the_repository);
664         return config_from(the_repository->submodule_cache, treeish_name, path, lookup_path);
665 }
666
667 const struct submodule *submodule_from_cache(struct repository *repo,
668                                              const struct object_id *treeish_name,
669                                              const char *key)
670 {
671         gitmodules_read_check(repo);
672         return config_from(repo->submodule_cache, treeish_name,
673                            key, lookup_path);
674 }
675
676 void submodule_free(void)
677 {
678         if (the_repository->submodule_cache)
679                 submodule_cache_clear(the_repository->submodule_cache);
680 }