Sync with maint
[git] / fsck.c
1 #include "cache.h"
2 #include "object.h"
3 #include "blob.h"
4 #include "tree.h"
5 #include "tree-walk.h"
6 #include "commit.h"
7 #include "tag.h"
8 #include "fsck.h"
9 #include "refs.h"
10 #include "utf8.h"
11 #include "sha1-array.h"
12 #include "decorate.h"
13
14 #define FSCK_FATAL -1
15 #define FSCK_INFO -2
16
17 #define FOREACH_MSG_ID(FUNC) \
18         /* fatal errors */ \
19         FUNC(NUL_IN_HEADER, FATAL) \
20         FUNC(UNTERMINATED_HEADER, FATAL) \
21         /* errors */ \
22         FUNC(BAD_DATE, ERROR) \
23         FUNC(BAD_DATE_OVERFLOW, ERROR) \
24         FUNC(BAD_EMAIL, ERROR) \
25         FUNC(BAD_NAME, ERROR) \
26         FUNC(BAD_OBJECT_SHA1, ERROR) \
27         FUNC(BAD_PARENT_SHA1, ERROR) \
28         FUNC(BAD_TAG_OBJECT, ERROR) \
29         FUNC(BAD_TIMEZONE, ERROR) \
30         FUNC(BAD_TREE, ERROR) \
31         FUNC(BAD_TREE_SHA1, ERROR) \
32         FUNC(BAD_TYPE, ERROR) \
33         FUNC(DUPLICATE_ENTRIES, ERROR) \
34         FUNC(MISSING_AUTHOR, ERROR) \
35         FUNC(MISSING_COMMITTER, ERROR) \
36         FUNC(MISSING_EMAIL, ERROR) \
37         FUNC(MISSING_GRAFT, ERROR) \
38         FUNC(MISSING_NAME_BEFORE_EMAIL, ERROR) \
39         FUNC(MISSING_OBJECT, ERROR) \
40         FUNC(MISSING_PARENT, ERROR) \
41         FUNC(MISSING_SPACE_BEFORE_DATE, ERROR) \
42         FUNC(MISSING_SPACE_BEFORE_EMAIL, ERROR) \
43         FUNC(MISSING_TAG, ERROR) \
44         FUNC(MISSING_TAG_ENTRY, ERROR) \
45         FUNC(MISSING_TAG_OBJECT, ERROR) \
46         FUNC(MISSING_TREE, ERROR) \
47         FUNC(MISSING_TYPE, ERROR) \
48         FUNC(MISSING_TYPE_ENTRY, ERROR) \
49         FUNC(MULTIPLE_AUTHORS, ERROR) \
50         FUNC(TAG_OBJECT_NOT_TAG, ERROR) \
51         FUNC(TREE_NOT_SORTED, ERROR) \
52         FUNC(UNKNOWN_TYPE, ERROR) \
53         FUNC(ZERO_PADDED_DATE, ERROR) \
54         /* warnings */ \
55         FUNC(BAD_FILEMODE, WARN) \
56         FUNC(EMPTY_NAME, WARN) \
57         FUNC(FULL_PATHNAME, WARN) \
58         FUNC(HAS_DOT, WARN) \
59         FUNC(HAS_DOTDOT, WARN) \
60         FUNC(HAS_DOTGIT, WARN) \
61         FUNC(NULL_SHA1, WARN) \
62         FUNC(ZERO_PADDED_FILEMODE, WARN) \
63         FUNC(NUL_IN_COMMIT, WARN) \
64         /* infos (reported as warnings, but ignored by default) */ \
65         FUNC(BAD_TAG_NAME, INFO) \
66         FUNC(MISSING_TAGGER_ENTRY, INFO)
67
68 #define MSG_ID(id, msg_type) FSCK_MSG_##id,
69 enum fsck_msg_id {
70         FOREACH_MSG_ID(MSG_ID)
71         FSCK_MSG_MAX
72 };
73 #undef MSG_ID
74
75 #define STR(x) #x
76 #define MSG_ID(id, msg_type) { STR(id), NULL, FSCK_##msg_type },
77 static struct {
78         const char *id_string;
79         const char *downcased;
80         int msg_type;
81 } msg_id_info[FSCK_MSG_MAX + 1] = {
82         FOREACH_MSG_ID(MSG_ID)
83         { NULL, NULL, -1 }
84 };
85 #undef MSG_ID
86
87 static int parse_msg_id(const char *text)
88 {
89         int i;
90
91         if (!msg_id_info[0].downcased) {
92                 /* convert id_string to lower case, without underscores. */
93                 for (i = 0; i < FSCK_MSG_MAX; i++) {
94                         const char *p = msg_id_info[i].id_string;
95                         int len = strlen(p);
96                         char *q = xmalloc(len);
97
98                         msg_id_info[i].downcased = q;
99                         while (*p)
100                                 if (*p == '_')
101                                         p++;
102                                 else
103                                         *(q)++ = tolower(*(p)++);
104                         *q = '\0';
105                 }
106         }
107
108         for (i = 0; i < FSCK_MSG_MAX; i++)
109                 if (!strcmp(text, msg_id_info[i].downcased))
110                         return i;
111
112         return -1;
113 }
114
115 static int fsck_msg_type(enum fsck_msg_id msg_id,
116         struct fsck_options *options)
117 {
118         int msg_type;
119
120         assert(msg_id >= 0 && msg_id < FSCK_MSG_MAX);
121
122         if (options->msg_type)
123                 msg_type = options->msg_type[msg_id];
124         else {
125                 msg_type = msg_id_info[msg_id].msg_type;
126                 if (options->strict && msg_type == FSCK_WARN)
127                         msg_type = FSCK_ERROR;
128         }
129
130         return msg_type;
131 }
132
133 static void init_skiplist(struct fsck_options *options, const char *path)
134 {
135         static struct sha1_array skiplist = SHA1_ARRAY_INIT;
136         int sorted, fd;
137         char buffer[41];
138         unsigned char sha1[20];
139
140         if (options->skiplist)
141                 sorted = options->skiplist->sorted;
142         else {
143                 sorted = 1;
144                 options->skiplist = &skiplist;
145         }
146
147         fd = open(path, O_RDONLY);
148         if (fd < 0)
149                 die("Could not open skip list: %s", path);
150         for (;;) {
151                 int result = read_in_full(fd, buffer, sizeof(buffer));
152                 if (result < 0)
153                         die_errno("Could not read '%s'", path);
154                 if (!result)
155                         break;
156                 if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n')
157                         die("Invalid SHA-1: %s", buffer);
158                 sha1_array_append(&skiplist, sha1);
159                 if (sorted && skiplist.nr > 1 &&
160                                 hashcmp(skiplist.sha1[skiplist.nr - 2],
161                                         sha1) > 0)
162                         sorted = 0;
163         }
164         close(fd);
165
166         if (sorted)
167                 skiplist.sorted = 1;
168 }
169
170 static int parse_msg_type(const char *str)
171 {
172         if (!strcmp(str, "error"))
173                 return FSCK_ERROR;
174         else if (!strcmp(str, "warn"))
175                 return FSCK_WARN;
176         else if (!strcmp(str, "ignore"))
177                 return FSCK_IGNORE;
178         else
179                 die("Unknown fsck message type: '%s'", str);
180 }
181
182 int is_valid_msg_type(const char *msg_id, const char *msg_type)
183 {
184         if (parse_msg_id(msg_id) < 0)
185                 return 0;
186         parse_msg_type(msg_type);
187         return 1;
188 }
189
190 void fsck_set_msg_type(struct fsck_options *options,
191                 const char *msg_id, const char *msg_type)
192 {
193         int id = parse_msg_id(msg_id), type;
194
195         if (id < 0)
196                 die("Unhandled message id: %s", msg_id);
197         type = parse_msg_type(msg_type);
198
199         if (type != FSCK_ERROR && msg_id_info[id].msg_type == FSCK_FATAL)
200                 die("Cannot demote %s to %s", msg_id, msg_type);
201
202         if (!options->msg_type) {
203                 int i;
204                 int *msg_type;
205                 ALLOC_ARRAY(msg_type, FSCK_MSG_MAX);
206                 for (i = 0; i < FSCK_MSG_MAX; i++)
207                         msg_type[i] = fsck_msg_type(i, options);
208                 options->msg_type = msg_type;
209         }
210
211         options->msg_type[id] = type;
212 }
213
214 void fsck_set_msg_types(struct fsck_options *options, const char *values)
215 {
216         char *buf = xstrdup(values), *to_free = buf;
217         int done = 0;
218
219         while (!done) {
220                 int len = strcspn(buf, " ,|"), equal;
221
222                 done = !buf[len];
223                 if (!len) {
224                         buf++;
225                         continue;
226                 }
227                 buf[len] = '\0';
228
229                 for (equal = 0;
230                      equal < len && buf[equal] != '=' && buf[equal] != ':';
231                      equal++)
232                         buf[equal] = tolower(buf[equal]);
233                 buf[equal] = '\0';
234
235                 if (!strcmp(buf, "skiplist")) {
236                         if (equal == len)
237                                 die("skiplist requires a path");
238                         init_skiplist(options, buf + equal + 1);
239                         buf += len + 1;
240                         continue;
241                 }
242
243                 if (equal == len)
244                         die("Missing '=': '%s'", buf);
245
246                 fsck_set_msg_type(options, buf, buf + equal + 1);
247                 buf += len + 1;
248         }
249         free(to_free);
250 }
251
252 static void append_msg_id(struct strbuf *sb, const char *msg_id)
253 {
254         for (;;) {
255                 char c = *(msg_id)++;
256
257                 if (!c)
258                         break;
259                 if (c != '_')
260                         strbuf_addch(sb, tolower(c));
261                 else {
262                         assert(*msg_id);
263                         strbuf_addch(sb, *(msg_id)++);
264                 }
265         }
266
267         strbuf_addstr(sb, ": ");
268 }
269
270 __attribute__((format (printf, 4, 5)))
271 static int report(struct fsck_options *options, struct object *object,
272         enum fsck_msg_id id, const char *fmt, ...)
273 {
274         va_list ap;
275         struct strbuf sb = STRBUF_INIT;
276         int msg_type = fsck_msg_type(id, options), result;
277
278         if (msg_type == FSCK_IGNORE)
279                 return 0;
280
281         if (options->skiplist && object &&
282                         sha1_array_lookup(options->skiplist, object->oid.hash) >= 0)
283                 return 0;
284
285         if (msg_type == FSCK_FATAL)
286                 msg_type = FSCK_ERROR;
287         else if (msg_type == FSCK_INFO)
288                 msg_type = FSCK_WARN;
289
290         append_msg_id(&sb, msg_id_info[id].id_string);
291
292         va_start(ap, fmt);
293         strbuf_vaddf(&sb, fmt, ap);
294         result = options->error_func(options, object, msg_type, sb.buf);
295         strbuf_release(&sb);
296         va_end(ap);
297
298         return result;
299 }
300
301 static char *get_object_name(struct fsck_options *options, struct object *obj)
302 {
303         if (!options->object_names)
304                 return NULL;
305         return lookup_decoration(options->object_names, obj);
306 }
307
308 static void put_object_name(struct fsck_options *options, struct object *obj,
309         const char *fmt, ...)
310 {
311         va_list ap;
312         struct strbuf buf = STRBUF_INIT;
313         char *existing;
314
315         if (!options->object_names)
316                 return;
317         existing = lookup_decoration(options->object_names, obj);
318         if (existing)
319                 return;
320         va_start(ap, fmt);
321         strbuf_vaddf(&buf, fmt, ap);
322         add_decoration(options->object_names, obj, strbuf_detach(&buf, NULL));
323         va_end(ap);
324 }
325
326 static const char *describe_object(struct fsck_options *o, struct object *obj)
327 {
328         static struct strbuf buf = STRBUF_INIT;
329         char *name;
330
331         strbuf_reset(&buf);
332         strbuf_addstr(&buf, oid_to_hex(&obj->oid));
333         if (o->object_names && (name = lookup_decoration(o->object_names, obj)))
334                 strbuf_addf(&buf, " (%s)", name);
335
336         return buf.buf;
337 }
338
339 static int fsck_walk_tree(struct tree *tree, void *data, struct fsck_options *options)
340 {
341         struct tree_desc desc;
342         struct name_entry entry;
343         int res = 0;
344         const char *name;
345
346         if (parse_tree(tree))
347                 return -1;
348
349         name = get_object_name(options, &tree->object);
350         if (init_tree_desc_gently(&desc, tree->buffer, tree->size))
351                 return -1;
352         while (tree_entry_gently(&desc, &entry)) {
353                 struct object *obj;
354                 int result;
355
356                 if (S_ISGITLINK(entry.mode))
357                         continue;
358
359                 if (S_ISDIR(entry.mode)) {
360                         obj = &lookup_tree(entry.oid->hash)->object;
361                         if (name)
362                                 put_object_name(options, obj, "%s%s/", name,
363                                         entry.path);
364                         result = options->walk(obj, OBJ_TREE, data, options);
365                 }
366                 else if (S_ISREG(entry.mode) || S_ISLNK(entry.mode)) {
367                         obj = &lookup_blob(entry.oid->hash)->object;
368                         if (name)
369                                 put_object_name(options, obj, "%s%s", name,
370                                         entry.path);
371                         result = options->walk(obj, OBJ_BLOB, data, options);
372                 }
373                 else {
374                         result = error("in tree %s: entry %s has bad mode %.6o",
375                                         describe_object(options, &tree->object), entry.path, entry.mode);
376                 }
377                 if (result < 0)
378                         return result;
379                 if (!res)
380                         res = result;
381         }
382         return res;
383 }
384
385 static int fsck_walk_commit(struct commit *commit, void *data, struct fsck_options *options)
386 {
387         int counter = 0, generation = 0, name_prefix_len = 0;
388         struct commit_list *parents;
389         int res;
390         int result;
391         const char *name;
392
393         if (parse_commit(commit))
394                 return -1;
395
396         name = get_object_name(options, &commit->object);
397         if (name)
398                 put_object_name(options, &commit->tree->object, "%s:", name);
399
400         result = options->walk((struct object *)commit->tree, OBJ_TREE, data, options);
401         if (result < 0)
402                 return result;
403         res = result;
404
405         parents = commit->parents;
406         if (name && parents) {
407                 int len = strlen(name), power;
408
409                 if (len && name[len - 1] == '^') {
410                         generation = 1;
411                         name_prefix_len = len - 1;
412                 }
413                 else { /* parse ~<generation> suffix */
414                         for (generation = 0, power = 1;
415                              len && isdigit(name[len - 1]);
416                              power *= 10)
417                                 generation += power * (name[--len] - '0');
418                         if (power > 1 && len && name[len - 1] == '~')
419                                 name_prefix_len = len - 1;
420                 }
421         }
422
423         while (parents) {
424                 if (name) {
425                         struct object *obj = &parents->item->object;
426
427                         if (++counter > 1)
428                                 put_object_name(options, obj, "%s^%d",
429                                         name, counter);
430                         else if (generation > 0)
431                                 put_object_name(options, obj, "%.*s~%d",
432                                         name_prefix_len, name, generation + 1);
433                         else
434                                 put_object_name(options, obj, "%s^", name);
435                 }
436                 result = options->walk((struct object *)parents->item, OBJ_COMMIT, data, options);
437                 if (result < 0)
438                         return result;
439                 if (!res)
440                         res = result;
441                 parents = parents->next;
442         }
443         return res;
444 }
445
446 static int fsck_walk_tag(struct tag *tag, void *data, struct fsck_options *options)
447 {
448         char *name = get_object_name(options, &tag->object);
449
450         if (parse_tag(tag))
451                 return -1;
452         if (name)
453                 put_object_name(options, tag->tagged, "%s", name);
454         return options->walk(tag->tagged, OBJ_ANY, data, options);
455 }
456
457 int fsck_walk(struct object *obj, void *data, struct fsck_options *options)
458 {
459         if (!obj)
460                 return -1;
461
462         if (obj->type == OBJ_NONE)
463                 parse_object(obj->oid.hash);
464
465         switch (obj->type) {
466         case OBJ_BLOB:
467                 return 0;
468         case OBJ_TREE:
469                 return fsck_walk_tree((struct tree *)obj, data, options);
470         case OBJ_COMMIT:
471                 return fsck_walk_commit((struct commit *)obj, data, options);
472         case OBJ_TAG:
473                 return fsck_walk_tag((struct tag *)obj, data, options);
474         default:
475                 error("Unknown object type for %s", describe_object(options, obj));
476                 return -1;
477         }
478 }
479
480 /*
481  * The entries in a tree are ordered in the _path_ order,
482  * which means that a directory entry is ordered by adding
483  * a slash to the end of it.
484  *
485  * So a directory called "a" is ordered _after_ a file
486  * called "a.c", because "a/" sorts after "a.c".
487  */
488 #define TREE_UNORDERED (-1)
489 #define TREE_HAS_DUPS  (-2)
490
491 static int verify_ordered(unsigned mode1, const char *name1, unsigned mode2, const char *name2)
492 {
493         int len1 = strlen(name1);
494         int len2 = strlen(name2);
495         int len = len1 < len2 ? len1 : len2;
496         unsigned char c1, c2;
497         int cmp;
498
499         cmp = memcmp(name1, name2, len);
500         if (cmp < 0)
501                 return 0;
502         if (cmp > 0)
503                 return TREE_UNORDERED;
504
505         /*
506          * Ok, the first <len> characters are the same.
507          * Now we need to order the next one, but turn
508          * a '\0' into a '/' for a directory entry.
509          */
510         c1 = name1[len];
511         c2 = name2[len];
512         if (!c1 && !c2)
513                 /*
514                  * git-write-tree used to write out a nonsense tree that has
515                  * entries with the same name, one blob and one tree.  Make
516                  * sure we do not have duplicate entries.
517                  */
518                 return TREE_HAS_DUPS;
519         if (!c1 && S_ISDIR(mode1))
520                 c1 = '/';
521         if (!c2 && S_ISDIR(mode2))
522                 c2 = '/';
523         return c1 < c2 ? 0 : TREE_UNORDERED;
524 }
525
526 static int fsck_tree(struct tree *item, struct fsck_options *options)
527 {
528         int retval = 0;
529         int has_null_sha1 = 0;
530         int has_full_path = 0;
531         int has_empty_name = 0;
532         int has_dot = 0;
533         int has_dotdot = 0;
534         int has_dotgit = 0;
535         int has_zero_pad = 0;
536         int has_bad_modes = 0;
537         int has_dup_entries = 0;
538         int not_properly_sorted = 0;
539         struct tree_desc desc;
540         unsigned o_mode;
541         const char *o_name;
542
543         if (init_tree_desc_gently(&desc, item->buffer, item->size)) {
544                 retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
545                 return retval;
546         }
547
548         o_mode = 0;
549         o_name = NULL;
550
551         while (desc.size) {
552                 unsigned mode;
553                 const char *name;
554                 const struct object_id *oid;
555
556                 oid = tree_entry_extract(&desc, &name, &mode);
557
558                 has_null_sha1 |= is_null_oid(oid);
559                 has_full_path |= !!strchr(name, '/');
560                 has_empty_name |= !*name;
561                 has_dot |= !strcmp(name, ".");
562                 has_dotdot |= !strcmp(name, "..");
563                 has_dotgit |= (!strcmp(name, ".git") ||
564                                is_hfs_dotgit(name) ||
565                                is_ntfs_dotgit(name));
566                 has_zero_pad |= *(char *)desc.buffer == '0';
567                 if (update_tree_entry_gently(&desc)) {
568                         retval += report(options, &item->object, FSCK_MSG_BAD_TREE, "cannot be parsed as a tree");
569                         break;
570                 }
571
572                 switch (mode) {
573                 /*
574                  * Standard modes..
575                  */
576                 case S_IFREG | 0755:
577                 case S_IFREG | 0644:
578                 case S_IFLNK:
579                 case S_IFDIR:
580                 case S_IFGITLINK:
581                         break;
582                 /*
583                  * This is nonstandard, but we had a few of these
584                  * early on when we honored the full set of mode
585                  * bits..
586                  */
587                 case S_IFREG | 0664:
588                         if (!options->strict)
589                                 break;
590                 default:
591                         has_bad_modes = 1;
592                 }
593
594                 if (o_name) {
595                         switch (verify_ordered(o_mode, o_name, mode, name)) {
596                         case TREE_UNORDERED:
597                                 not_properly_sorted = 1;
598                                 break;
599                         case TREE_HAS_DUPS:
600                                 has_dup_entries = 1;
601                                 break;
602                         default:
603                                 break;
604                         }
605                 }
606
607                 o_mode = mode;
608                 o_name = name;
609         }
610
611         if (has_null_sha1)
612                 retval += report(options, &item->object, FSCK_MSG_NULL_SHA1, "contains entries pointing to null sha1");
613         if (has_full_path)
614                 retval += report(options, &item->object, FSCK_MSG_FULL_PATHNAME, "contains full pathnames");
615         if (has_empty_name)
616                 retval += report(options, &item->object, FSCK_MSG_EMPTY_NAME, "contains empty pathname");
617         if (has_dot)
618                 retval += report(options, &item->object, FSCK_MSG_HAS_DOT, "contains '.'");
619         if (has_dotdot)
620                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTDOT, "contains '..'");
621         if (has_dotgit)
622                 retval += report(options, &item->object, FSCK_MSG_HAS_DOTGIT, "contains '.git'");
623         if (has_zero_pad)
624                 retval += report(options, &item->object, FSCK_MSG_ZERO_PADDED_FILEMODE, "contains zero-padded file modes");
625         if (has_bad_modes)
626                 retval += report(options, &item->object, FSCK_MSG_BAD_FILEMODE, "contains bad file modes");
627         if (has_dup_entries)
628                 retval += report(options, &item->object, FSCK_MSG_DUPLICATE_ENTRIES, "contains duplicate file entries");
629         if (not_properly_sorted)
630                 retval += report(options, &item->object, FSCK_MSG_TREE_NOT_SORTED, "not properly sorted");
631         return retval;
632 }
633
634 static int verify_headers(const void *data, unsigned long size,
635                           struct object *obj, struct fsck_options *options)
636 {
637         const char *buffer = (const char *)data;
638         unsigned long i;
639
640         for (i = 0; i < size; i++) {
641                 switch (buffer[i]) {
642                 case '\0':
643                         return report(options, obj,
644                                 FSCK_MSG_NUL_IN_HEADER,
645                                 "unterminated header: NUL at offset %ld", i);
646                 case '\n':
647                         if (i + 1 < size && buffer[i + 1] == '\n')
648                                 return 0;
649                 }
650         }
651
652         /*
653          * We did not find double-LF that separates the header
654          * and the body.  Not having a body is not a crime but
655          * we do want to see the terminating LF for the last header
656          * line.
657          */
658         if (size && buffer[size - 1] == '\n')
659                 return 0;
660
661         return report(options, obj,
662                 FSCK_MSG_UNTERMINATED_HEADER, "unterminated header");
663 }
664
665 static int fsck_ident(const char **ident, struct object *obj, struct fsck_options *options)
666 {
667         const char *p = *ident;
668         char *end;
669
670         *ident = strchrnul(*ident, '\n');
671         if (**ident == '\n')
672                 (*ident)++;
673
674         if (*p == '<')
675                 return report(options, obj, FSCK_MSG_MISSING_NAME_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
676         p += strcspn(p, "<>\n");
677         if (*p == '>')
678                 return report(options, obj, FSCK_MSG_BAD_NAME, "invalid author/committer line - bad name");
679         if (*p != '<')
680                 return report(options, obj, FSCK_MSG_MISSING_EMAIL, "invalid author/committer line - missing email");
681         if (p[-1] != ' ')
682                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_EMAIL, "invalid author/committer line - missing space before email");
683         p++;
684         p += strcspn(p, "<>\n");
685         if (*p != '>')
686                 return report(options, obj, FSCK_MSG_BAD_EMAIL, "invalid author/committer line - bad email");
687         p++;
688         if (*p != ' ')
689                 return report(options, obj, FSCK_MSG_MISSING_SPACE_BEFORE_DATE, "invalid author/committer line - missing space before date");
690         p++;
691         if (*p == '0' && p[1] != ' ')
692                 return report(options, obj, FSCK_MSG_ZERO_PADDED_DATE, "invalid author/committer line - zero-padded date");
693         if (date_overflows(strtoul(p, &end, 10)))
694                 return report(options, obj, FSCK_MSG_BAD_DATE_OVERFLOW, "invalid author/committer line - date causes integer overflow");
695         if ((end == p || *end != ' '))
696                 return report(options, obj, FSCK_MSG_BAD_DATE, "invalid author/committer line - bad date");
697         p = end + 1;
698         if ((*p != '+' && *p != '-') ||
699             !isdigit(p[1]) ||
700             !isdigit(p[2]) ||
701             !isdigit(p[3]) ||
702             !isdigit(p[4]) ||
703             (p[5] != '\n'))
704                 return report(options, obj, FSCK_MSG_BAD_TIMEZONE, "invalid author/committer line - bad time zone");
705         p += 6;
706         return 0;
707 }
708
709 static int fsck_commit_buffer(struct commit *commit, const char *buffer,
710         unsigned long size, struct fsck_options *options)
711 {
712         unsigned char tree_sha1[20], sha1[20];
713         struct commit_graft *graft;
714         unsigned parent_count, parent_line_count = 0, author_count;
715         int err;
716         const char *buffer_begin = buffer;
717
718         if (verify_headers(buffer, size, &commit->object, options))
719                 return -1;
720
721         if (!skip_prefix(buffer, "tree ", &buffer))
722                 return report(options, &commit->object, FSCK_MSG_MISSING_TREE, "invalid format - expected 'tree' line");
723         if (get_sha1_hex(buffer, tree_sha1) || buffer[40] != '\n') {
724                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE_SHA1, "invalid 'tree' line format - bad sha1");
725                 if (err)
726                         return err;
727         }
728         buffer += 41;
729         while (skip_prefix(buffer, "parent ", &buffer)) {
730                 if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
731                         err = report(options, &commit->object, FSCK_MSG_BAD_PARENT_SHA1, "invalid 'parent' line format - bad sha1");
732                         if (err)
733                                 return err;
734                 }
735                 buffer += 41;
736                 parent_line_count++;
737         }
738         graft = lookup_commit_graft(commit->object.oid.hash);
739         parent_count = commit_list_count(commit->parents);
740         if (graft) {
741                 if (graft->nr_parent == -1 && !parent_count)
742                         ; /* shallow commit */
743                 else if (graft->nr_parent != parent_count) {
744                         err = report(options, &commit->object, FSCK_MSG_MISSING_GRAFT, "graft objects missing");
745                         if (err)
746                                 return err;
747                 }
748         } else {
749                 if (parent_count != parent_line_count) {
750                         err = report(options, &commit->object, FSCK_MSG_MISSING_PARENT, "parent objects missing");
751                         if (err)
752                                 return err;
753                 }
754         }
755         author_count = 0;
756         while (skip_prefix(buffer, "author ", &buffer)) {
757                 author_count++;
758                 err = fsck_ident(&buffer, &commit->object, options);
759                 if (err)
760                         return err;
761         }
762         if (author_count < 1)
763                 err = report(options, &commit->object, FSCK_MSG_MISSING_AUTHOR, "invalid format - expected 'author' line");
764         else if (author_count > 1)
765                 err = report(options, &commit->object, FSCK_MSG_MULTIPLE_AUTHORS, "invalid format - multiple 'author' lines");
766         if (err)
767                 return err;
768         if (!skip_prefix(buffer, "committer ", &buffer))
769                 return report(options, &commit->object, FSCK_MSG_MISSING_COMMITTER, "invalid format - expected 'committer' line");
770         err = fsck_ident(&buffer, &commit->object, options);
771         if (err)
772                 return err;
773         if (!commit->tree) {
774                 err = report(options, &commit->object, FSCK_MSG_BAD_TREE, "could not load commit's tree %s", sha1_to_hex(tree_sha1));
775                 if (err)
776                         return err;
777         }
778         if (memchr(buffer_begin, '\0', size)) {
779                 err = report(options, &commit->object, FSCK_MSG_NUL_IN_COMMIT,
780                              "NUL byte in the commit object body");
781                 if (err)
782                         return err;
783         }
784         return 0;
785 }
786
787 static int fsck_commit(struct commit *commit, const char *data,
788         unsigned long size, struct fsck_options *options)
789 {
790         const char *buffer = data ?  data : get_commit_buffer(commit, &size);
791         int ret = fsck_commit_buffer(commit, buffer, size, options);
792         if (!data)
793                 unuse_commit_buffer(commit, buffer);
794         return ret;
795 }
796
797 static int fsck_tag_buffer(struct tag *tag, const char *data,
798         unsigned long size, struct fsck_options *options)
799 {
800         unsigned char sha1[20];
801         int ret = 0;
802         const char *buffer;
803         char *to_free = NULL, *eol;
804         struct strbuf sb = STRBUF_INIT;
805
806         if (data)
807                 buffer = data;
808         else {
809                 enum object_type type;
810
811                 buffer = to_free =
812                         read_sha1_file(tag->object.oid.hash, &type, &size);
813                 if (!buffer)
814                         return report(options, &tag->object,
815                                 FSCK_MSG_MISSING_TAG_OBJECT,
816                                 "cannot read tag object");
817
818                 if (type != OBJ_TAG) {
819                         ret = report(options, &tag->object,
820                                 FSCK_MSG_TAG_OBJECT_NOT_TAG,
821                                 "expected tag got %s",
822                             typename(type));
823                         goto done;
824                 }
825         }
826
827         ret = verify_headers(buffer, size, &tag->object, options);
828         if (ret)
829                 goto done;
830
831         if (!skip_prefix(buffer, "object ", &buffer)) {
832                 ret = report(options, &tag->object, FSCK_MSG_MISSING_OBJECT, "invalid format - expected 'object' line");
833                 goto done;
834         }
835         if (get_sha1_hex(buffer, sha1) || buffer[40] != '\n') {
836                 ret = report(options, &tag->object, FSCK_MSG_BAD_OBJECT_SHA1, "invalid 'object' line format - bad sha1");
837                 if (ret)
838                         goto done;
839         }
840         buffer += 41;
841
842         if (!skip_prefix(buffer, "type ", &buffer)) {
843                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE_ENTRY, "invalid format - expected 'type' line");
844                 goto done;
845         }
846         eol = strchr(buffer, '\n');
847         if (!eol) {
848                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TYPE, "invalid format - unexpected end after 'type' line");
849                 goto done;
850         }
851         if (type_from_string_gently(buffer, eol - buffer, 1) < 0)
852                 ret = report(options, &tag->object, FSCK_MSG_BAD_TYPE, "invalid 'type' value");
853         if (ret)
854                 goto done;
855         buffer = eol + 1;
856
857         if (!skip_prefix(buffer, "tag ", &buffer)) {
858                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG_ENTRY, "invalid format - expected 'tag' line");
859                 goto done;
860         }
861         eol = strchr(buffer, '\n');
862         if (!eol) {
863                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAG, "invalid format - unexpected end after 'type' line");
864                 goto done;
865         }
866         strbuf_addf(&sb, "refs/tags/%.*s", (int)(eol - buffer), buffer);
867         if (check_refname_format(sb.buf, 0)) {
868                 ret = report(options, &tag->object, FSCK_MSG_BAD_TAG_NAME,
869                            "invalid 'tag' name: %.*s",
870                            (int)(eol - buffer), buffer);
871                 if (ret)
872                         goto done;
873         }
874         buffer = eol + 1;
875
876         if (!skip_prefix(buffer, "tagger ", &buffer)) {
877                 /* early tags do not contain 'tagger' lines; warn only */
878                 ret = report(options, &tag->object, FSCK_MSG_MISSING_TAGGER_ENTRY, "invalid format - expected 'tagger' line");
879                 if (ret)
880                         goto done;
881         }
882         else
883                 ret = fsck_ident(&buffer, &tag->object, options);
884
885 done:
886         strbuf_release(&sb);
887         free(to_free);
888         return ret;
889 }
890
891 static int fsck_tag(struct tag *tag, const char *data,
892         unsigned long size, struct fsck_options *options)
893 {
894         struct object *tagged = tag->tagged;
895
896         if (!tagged)
897                 return report(options, &tag->object, FSCK_MSG_BAD_TAG_OBJECT, "could not load tagged object");
898
899         return fsck_tag_buffer(tag, data, size, options);
900 }
901
902 int fsck_object(struct object *obj, void *data, unsigned long size,
903         struct fsck_options *options)
904 {
905         if (!obj)
906                 return report(options, obj, FSCK_MSG_BAD_OBJECT_SHA1, "no valid object to fsck");
907
908         if (obj->type == OBJ_BLOB)
909                 return 0;
910         if (obj->type == OBJ_TREE)
911                 return fsck_tree((struct tree *) obj, options);
912         if (obj->type == OBJ_COMMIT)
913                 return fsck_commit((struct commit *) obj, (const char *) data,
914                         size, options);
915         if (obj->type == OBJ_TAG)
916                 return fsck_tag((struct tag *) obj, (const char *) data,
917                         size, options);
918
919         return report(options, obj, FSCK_MSG_UNKNOWN_TYPE, "unknown type '%d' (internal fsck error)",
920                           obj->type);
921 }
922
923 int fsck_error_function(struct fsck_options *o,
924         struct object *obj, int msg_type, const char *message)
925 {
926         if (msg_type == FSCK_WARN) {
927                 warning("object %s: %s", describe_object(o, obj), message);
928                 return 0;
929         }
930         error("object %s: %s", describe_object(o, obj), message);
931         return 1;
932 }