2 * GIT - The information manager from hell
4 * Copyright (C) Linus Torvalds, 2005
6 #define USE_THE_INDEX_COMPATIBILITY_MACROS
11 #include "parse-options.h"
13 #include "streaming.h"
14 #include "tree-walk.h"
15 #include "sha1-array.h"
17 #include "object-store.h"
18 #include "promisor-remote.h"
20 struct batch_options
{
27 int cmdmode
; /* may be 'w' or 'c' for --filters or --textconv */
31 static const char *force_path
;
33 static int filter_object(const char *path
, unsigned mode
,
34 const struct object_id
*oid
,
35 char **buf
, unsigned long *size
)
37 enum object_type type
;
39 *buf
= read_object_file(oid
, &type
, size
);
41 return error(_("cannot read object %s '%s'"),
42 oid_to_hex(oid
), path
);
43 if ((type
== OBJ_BLOB
) && S_ISREG(mode
)) {
44 struct strbuf strbuf
= STRBUF_INIT
;
45 if (convert_to_working_tree(&the_index
, path
, *buf
, *size
, &strbuf
)) {
48 *buf
= strbuf_detach(&strbuf
, NULL
);
55 static int stream_blob(const struct object_id
*oid
)
57 if (stream_blob_to_fd(1, oid
, NULL
, 0))
58 die("unable to stream %s to stdout", oid_to_hex(oid
));
62 static int cat_one_file(int opt
, const char *exp_type
, const char *obj_name
,
66 enum object_type type
;
69 struct object_context obj_context
;
70 struct object_info oi
= OBJECT_INFO_INIT
;
71 struct strbuf sb
= STRBUF_INIT
;
72 unsigned flags
= OBJECT_INFO_LOOKUP_REPLACE
;
73 const char *path
= force_path
;
76 flags
|= OBJECT_INFO_ALLOW_UNKNOWN_TYPE
;
78 if (get_oid_with_context(the_repository
, obj_name
,
81 die("Not a valid object name %s", obj_name
);
84 path
= obj_context
.path
;
85 if (obj_context
.mode
== S_IFINVALID
)
86 obj_context
.mode
= 0100644;
92 if (oid_object_info_extended(the_repository
, &oid
, &oi
, flags
) < 0)
93 die("git cat-file: could not get object info");
95 printf("%s\n", sb
.buf
);
103 if (oid_object_info_extended(the_repository
, &oid
, &oi
, flags
) < 0)
104 die("git cat-file: could not get object info");
105 printf("%"PRIuMAX
"\n", (uintmax_t)size
);
109 return !has_object_file(&oid
);
113 die("git cat-file --filters %s: <object> must be "
114 "<sha1:path>", obj_name
);
116 if (filter_object(path
, obj_context
.mode
,
123 die("git cat-file --textconv %s: <object> must be <sha1:path>",
126 if (textconv_object(the_repository
, path
, obj_context
.mode
,
127 &oid
, 1, &buf
, &size
))
129 /* else fallthrough */
132 type
= oid_object_info(the_repository
, &oid
, NULL
);
134 die("Not a valid object name %s", obj_name
);
136 /* custom pretty-print here */
137 if (type
== OBJ_TREE
) {
138 const char *ls_args
[3] = { NULL
};
139 ls_args
[0] = "ls-tree";
140 ls_args
[1] = obj_name
;
141 return cmd_ls_tree(2, ls_args
, NULL
);
144 if (type
== OBJ_BLOB
)
145 return stream_blob(&oid
);
146 buf
= read_object_file(&oid
, &type
, &size
);
148 die("Cannot read object %s", obj_name
);
150 /* otherwise just spit out the data */
154 if (type_from_string(exp_type
) == OBJ_BLOB
) {
155 struct object_id blob_oid
;
156 if (oid_object_info(the_repository
, &oid
, NULL
) == OBJ_TAG
) {
157 char *buffer
= read_object_file(&oid
, &type
,
160 if (!skip_prefix(buffer
, "object ", &target
) ||
161 get_oid_hex(target
, &blob_oid
))
162 die("%s not a valid tag", oid_to_hex(&oid
));
165 oidcpy(&blob_oid
, &oid
);
167 if (oid_object_info(the_repository
, &blob_oid
, NULL
) == OBJ_BLOB
)
168 return stream_blob(&blob_oid
);
170 * we attempted to dereference a tag to a blob
171 * and failed; there may be new dereference
172 * mechanisms this code is not aware of.
173 * fall-back to the usual case.
176 buf
= read_object_with_reference(the_repository
,
177 &oid
, exp_type
, &size
, NULL
);
181 die("git cat-file: unknown option: %s", exp_type
);
185 die("git cat-file %s: bad file", obj_name
);
187 write_or_die(1, buf
, size
);
189 free(obj_context
.path
);
194 struct object_id oid
;
195 enum object_type type
;
199 struct object_id delta_base_oid
;
202 * If mark_query is true, we do not expand anything, but rather
203 * just mark the object_info with items we wish to query.
208 * Whether to split the input on whitespace before feeding it to
209 * get_sha1; this is decided during the mark_query phase based on
210 * whether we have a %(rest) token in our format.
212 int split_on_whitespace
;
215 * After a mark_query run, this object_info is set up to be
216 * passed to oid_object_info_extended. It will point to the data
217 * elements above, so you can retrieve the response from there.
219 struct object_info info
;
222 * This flag will be true if the requested batch format and options
223 * don't require us to call oid_object_info, which can then be
226 unsigned skip_object_info
: 1;
229 static int is_atom(const char *atom
, const char *s
, int slen
)
231 int alen
= strlen(atom
);
232 return alen
== slen
&& !memcmp(atom
, s
, alen
);
235 static void expand_atom(struct strbuf
*sb
, const char *atom
, int len
,
238 struct expand_data
*data
= vdata
;
240 if (is_atom("objectname", atom
, len
)) {
241 if (!data
->mark_query
)
242 strbuf_addstr(sb
, oid_to_hex(&data
->oid
));
243 } else if (is_atom("objecttype", atom
, len
)) {
244 if (data
->mark_query
)
245 data
->info
.typep
= &data
->type
;
247 strbuf_addstr(sb
, type_name(data
->type
));
248 } else if (is_atom("objectsize", atom
, len
)) {
249 if (data
->mark_query
)
250 data
->info
.sizep
= &data
->size
;
252 strbuf_addf(sb
, "%"PRIuMAX
, (uintmax_t)data
->size
);
253 } else if (is_atom("objectsize:disk", atom
, len
)) {
254 if (data
->mark_query
)
255 data
->info
.disk_sizep
= &data
->disk_size
;
257 strbuf_addf(sb
, "%"PRIuMAX
, (uintmax_t)data
->disk_size
);
258 } else if (is_atom("rest", atom
, len
)) {
259 if (data
->mark_query
)
260 data
->split_on_whitespace
= 1;
262 strbuf_addstr(sb
, data
->rest
);
263 } else if (is_atom("deltabase", atom
, len
)) {
264 if (data
->mark_query
)
265 data
->info
.delta_base_sha1
= data
->delta_base_oid
.hash
;
268 oid_to_hex(&data
->delta_base_oid
));
270 die("unknown format element: %.*s", len
, atom
);
273 static size_t expand_format(struct strbuf
*sb
, const char *start
, void *data
)
279 end
= strchr(start
+ 1, ')');
281 die("format element '%s' does not end in ')'", start
);
283 expand_atom(sb
, start
+ 1, end
- start
- 1, data
);
285 return end
- start
+ 1;
288 static void batch_write(struct batch_options
*opt
, const void *data
, int len
)
290 if (opt
->buffer_output
) {
291 if (fwrite(data
, 1, len
, stdout
) != len
)
292 die_errno("unable to write to stdout");
294 write_or_die(1, data
, len
);
297 static void print_object_or_die(struct batch_options
*opt
, struct expand_data
*data
)
299 const struct object_id
*oid
= &data
->oid
;
301 assert(data
->info
.typep
);
303 if (data
->type
== OBJ_BLOB
) {
304 if (opt
->buffer_output
)
311 die("missing path for '%s'", oid_to_hex(oid
));
313 if (opt
->cmdmode
== 'w') {
314 if (filter_object(data
->rest
, 0100644, oid
,
316 die("could not convert '%s' %s",
317 oid_to_hex(oid
), data
->rest
);
318 } else if (opt
->cmdmode
== 'c') {
319 enum object_type type
;
320 if (!textconv_object(the_repository
,
321 data
->rest
, 0100644, oid
,
322 1, &contents
, &size
))
323 contents
= read_object_file(oid
,
327 die("could not convert '%s' %s",
328 oid_to_hex(oid
), data
->rest
);
330 BUG("invalid cmdmode: %c", opt
->cmdmode
);
331 batch_write(opt
, contents
, size
);
338 enum object_type type
;
342 contents
= read_object_file(oid
, &type
, &size
);
344 die("object %s disappeared", oid_to_hex(oid
));
345 if (type
!= data
->type
)
346 die("object %s changed type!?", oid_to_hex(oid
));
347 if (data
->info
.sizep
&& size
!= data
->size
)
348 die("object %s changed size!?", oid_to_hex(oid
));
350 batch_write(opt
, contents
, size
);
355 static void batch_object_write(const char *obj_name
,
356 struct strbuf
*scratch
,
357 struct batch_options
*opt
,
358 struct expand_data
*data
)
360 if (!data
->skip_object_info
&&
361 oid_object_info_extended(the_repository
, &data
->oid
, &data
->info
,
362 OBJECT_INFO_LOOKUP_REPLACE
) < 0) {
363 printf("%s missing\n",
364 obj_name
? obj_name
: oid_to_hex(&data
->oid
));
369 strbuf_reset(scratch
);
370 strbuf_expand(scratch
, opt
->format
, expand_format
, data
);
371 strbuf_addch(scratch
, '\n');
372 batch_write(opt
, scratch
->buf
, scratch
->len
);
374 if (opt
->print_contents
) {
375 print_object_or_die(opt
, data
);
376 batch_write(opt
, "\n", 1);
380 static void batch_one_object(const char *obj_name
,
381 struct strbuf
*scratch
,
382 struct batch_options
*opt
,
383 struct expand_data
*data
)
385 struct object_context ctx
;
386 int flags
= opt
->follow_symlinks
? GET_OID_FOLLOW_SYMLINKS
: 0;
387 enum get_oid_result result
;
389 result
= get_oid_with_context(the_repository
, obj_name
,
390 flags
, &data
->oid
, &ctx
);
391 if (result
!= FOUND
) {
394 printf("%s missing\n", obj_name
);
396 case SHORT_NAME_AMBIGUOUS
:
397 printf("%s ambiguous\n", obj_name
);
399 case DANGLING_SYMLINK
:
400 printf("dangling %"PRIuMAX
"\n%s\n",
401 (uintmax_t)strlen(obj_name
), obj_name
);
404 printf("loop %"PRIuMAX
"\n%s\n",
405 (uintmax_t)strlen(obj_name
), obj_name
);
408 printf("notdir %"PRIuMAX
"\n%s\n",
409 (uintmax_t)strlen(obj_name
), obj_name
);
412 BUG("unknown get_sha1_with_context result %d\n",
421 printf("symlink %"PRIuMAX
"\n%s\n",
422 (uintmax_t)ctx
.symlink_path
.len
,
423 ctx
.symlink_path
.buf
);
428 batch_object_write(obj_name
, scratch
, opt
, data
);
431 struct object_cb_data
{
432 struct batch_options
*opt
;
433 struct expand_data
*expand
;
435 struct strbuf
*scratch
;
438 static int batch_object_cb(const struct object_id
*oid
, void *vdata
)
440 struct object_cb_data
*data
= vdata
;
441 oidcpy(&data
->expand
->oid
, oid
);
442 batch_object_write(NULL
, data
->scratch
, data
->opt
, data
->expand
);
446 static int collect_loose_object(const struct object_id
*oid
,
450 oid_array_append(data
, oid
);
454 static int collect_packed_object(const struct object_id
*oid
,
455 struct packed_git
*pack
,
459 oid_array_append(data
, oid
);
463 static int batch_unordered_object(const struct object_id
*oid
, void *vdata
)
465 struct object_cb_data
*data
= vdata
;
467 if (oidset_insert(data
->seen
, oid
))
470 return batch_object_cb(oid
, data
);
473 static int batch_unordered_loose(const struct object_id
*oid
,
477 return batch_unordered_object(oid
, data
);
480 static int batch_unordered_packed(const struct object_id
*oid
,
481 struct packed_git
*pack
,
485 return batch_unordered_object(oid
, data
);
488 static int batch_objects(struct batch_options
*opt
)
490 struct strbuf input
= STRBUF_INIT
;
491 struct strbuf output
= STRBUF_INIT
;
492 struct expand_data data
;
497 opt
->format
= "%(objectname) %(objecttype) %(objectsize)";
500 * Expand once with our special mark_query flag, which will prime the
501 * object_info to be handed to oid_object_info_extended for each
504 memset(&data
, 0, sizeof(data
));
506 strbuf_expand(&output
, opt
->format
, expand_format
, &data
);
508 strbuf_release(&output
);
510 data
.split_on_whitespace
= 1;
512 if (opt
->all_objects
) {
513 struct object_info empty
= OBJECT_INFO_INIT
;
514 if (!memcmp(&data
.info
, &empty
, sizeof(empty
)))
515 data
.skip_object_info
= 1;
519 * If we are printing out the object, then always fill in the type,
520 * since we will want to decide whether or not to stream.
522 if (opt
->print_contents
)
523 data
.info
.typep
= &data
.type
;
525 if (opt
->all_objects
) {
526 struct object_cb_data cb
;
528 if (has_promisor_remote())
529 warning("This repository uses promisor remotes. Some objects may not be loaded.");
533 cb
.scratch
= &output
;
535 if (opt
->unordered
) {
536 struct oidset seen
= OIDSET_INIT
;
540 for_each_loose_object(batch_unordered_loose
, &cb
, 0);
541 for_each_packed_object(batch_unordered_packed
, &cb
,
542 FOR_EACH_OBJECT_PACK_ORDER
);
546 struct oid_array sa
= OID_ARRAY_INIT
;
548 for_each_loose_object(collect_loose_object
, &sa
, 0);
549 for_each_packed_object(collect_packed_object
, &sa
, 0);
551 oid_array_for_each_unique(&sa
, batch_object_cb
, &cb
);
553 oid_array_clear(&sa
);
556 strbuf_release(&output
);
561 * We are going to call get_sha1 on a potentially very large number of
562 * objects. In most large cases, these will be actual object sha1s. The
563 * cost to double-check that each one is not also a ref (just so we can
564 * warn) ends up dwarfing the actual cost of the object lookups
565 * themselves. We can work around it by just turning off the warning.
567 save_warning
= warn_on_object_refname_ambiguity
;
568 warn_on_object_refname_ambiguity
= 0;
570 while (strbuf_getline(&input
, stdin
) != EOF
) {
571 if (data
.split_on_whitespace
) {
573 * Split at first whitespace, tying off the beginning
574 * of the string and saving the remainder (or NULL) in
577 char *p
= strpbrk(input
.buf
, " \t");
579 while (*p
&& strchr(" \t", *p
))
585 batch_one_object(input
.buf
, &output
, opt
, &data
);
588 strbuf_release(&input
);
589 strbuf_release(&output
);
590 warn_on_object_refname_ambiguity
= save_warning
;
594 static const char * const cat_file_usage
[] = {
595 N_("git cat-file (-t [--allow-unknown-type] | -s [--allow-unknown-type] | -e | -p | <type> | --textconv | --filters) [--path=<path>] <object>"),
596 N_("git cat-file (--batch | --batch-check) [--follow-symlinks] [--textconv | --filters]"),
600 static int git_cat_file_config(const char *var
, const char *value
, void *cb
)
602 if (userdiff_config(var
, value
) < 0)
605 return git_default_config(var
, value
, cb
);
608 static int batch_option_callback(const struct option
*opt
,
612 struct batch_options
*bo
= opt
->value
;
614 BUG_ON_OPT_NEG(unset
);
617 return error(_("only one batch option may be specified"));
621 bo
->print_contents
= !strcmp(opt
->long_name
, "batch");
627 int cmd_cat_file(int argc
, const char **argv
, const char *prefix
)
630 const char *exp_type
= NULL
, *obj_name
= NULL
;
631 struct batch_options batch
= {0};
632 int unknown_type
= 0;
634 const struct option options
[] = {
635 OPT_GROUP(N_("<type> can be one of: blob, tree, commit, tag")),
636 OPT_CMDMODE('t', NULL
, &opt
, N_("show object type"), 't'),
637 OPT_CMDMODE('s', NULL
, &opt
, N_("show object size"), 's'),
638 OPT_CMDMODE('e', NULL
, &opt
,
639 N_("exit with zero when there's no error"), 'e'),
640 OPT_CMDMODE('p', NULL
, &opt
, N_("pretty-print object's content"), 'p'),
641 OPT_CMDMODE(0, "textconv", &opt
,
642 N_("for blob objects, run textconv on object's content"), 'c'),
643 OPT_CMDMODE(0, "filters", &opt
,
644 N_("for blob objects, run filters on object's content"), 'w'),
645 OPT_STRING(0, "path", &force_path
, N_("blob"),
646 N_("use a specific path for --textconv/--filters")),
647 OPT_BOOL(0, "allow-unknown-type", &unknown_type
,
648 N_("allow -s and -t to work with broken/corrupt objects")),
649 OPT_BOOL(0, "buffer", &batch
.buffer_output
, N_("buffer --batch output")),
650 { OPTION_CALLBACK
, 0, "batch", &batch
, "format",
651 N_("show info and content of objects fed from the standard input"),
652 PARSE_OPT_OPTARG
| PARSE_OPT_NONEG
,
653 batch_option_callback
},
654 { OPTION_CALLBACK
, 0, "batch-check", &batch
, "format",
655 N_("show info about objects fed from the standard input"),
656 PARSE_OPT_OPTARG
| PARSE_OPT_NONEG
,
657 batch_option_callback
},
658 OPT_BOOL(0, "follow-symlinks", &batch
.follow_symlinks
,
659 N_("follow in-tree symlinks (used with --batch or --batch-check)")),
660 OPT_BOOL(0, "batch-all-objects", &batch
.all_objects
,
661 N_("show all objects with --batch or --batch-check")),
662 OPT_BOOL(0, "unordered", &batch
.unordered
,
663 N_("do not order --batch-all-objects output")),
667 git_config(git_cat_file_config
, NULL
);
669 batch
.buffer_output
= -1;
670 argc
= parse_options(argc
, argv
, prefix
, options
, cat_file_usage
, 0);
673 if (batch
.enabled
&& (opt
== 'c' || opt
== 'w'))
678 usage_with_options(cat_file_usage
, options
);
680 if (!opt
&& !batch
.enabled
) {
685 usage_with_options(cat_file_usage
, options
);
688 if (batch
.cmdmode
!= opt
|| argc
)
689 usage_with_options(cat_file_usage
, options
);
690 if (batch
.cmdmode
&& batch
.all_objects
)
691 die("--batch-all-objects cannot be combined with "
692 "--textconv nor with --filters");
695 if ((batch
.follow_symlinks
|| batch
.all_objects
) && !batch
.enabled
) {
696 usage_with_options(cat_file_usage
, options
);
699 if (force_path
&& opt
!= 'c' && opt
!= 'w') {
700 error("--path=<path> needs --textconv or --filters");
701 usage_with_options(cat_file_usage
, options
);
704 if (force_path
&& batch
.enabled
) {
705 error("--path=<path> incompatible with --batch");
706 usage_with_options(cat_file_usage
, options
);
709 if (batch
.buffer_output
< 0)
710 batch
.buffer_output
= batch
.all_objects
;
713 return batch_objects(&batch
);
715 if (unknown_type
&& opt
!= 't' && opt
!= 's')
716 die("git cat-file --allow-unknown-type: use with -s or -t");
717 return cat_one_file(opt
, exp_type
, obj_name
, unknown_type
);