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"
19 struct batch_options
{
26 int cmdmode
; /* may be 'w' or 'c' for --filters or --textconv */
30 static const char *force_path
;
32 static int filter_object(const char *path
, unsigned mode
,
33 const struct object_id
*oid
,
34 char **buf
, unsigned long *size
)
36 enum object_type type
;
38 *buf
= read_object_file(oid
, &type
, size
);
40 return error(_("cannot read object %s '%s'"),
41 oid_to_hex(oid
), path
);
42 if ((type
== OBJ_BLOB
) && S_ISREG(mode
)) {
43 struct strbuf strbuf
= STRBUF_INIT
;
44 if (convert_to_working_tree(&the_index
, path
, *buf
, *size
, &strbuf
)) {
47 *buf
= strbuf_detach(&strbuf
, NULL
);
54 static int stream_blob(const struct object_id
*oid
)
56 if (stream_blob_to_fd(1, oid
, NULL
, 0))
57 die("unable to stream %s to stdout", oid_to_hex(oid
));
61 static int cat_one_file(int opt
, const char *exp_type
, const char *obj_name
,
65 enum object_type type
;
68 struct object_context obj_context
;
69 struct object_info oi
= OBJECT_INFO_INIT
;
70 struct strbuf sb
= STRBUF_INIT
;
71 unsigned flags
= OBJECT_INFO_LOOKUP_REPLACE
;
72 const char *path
= force_path
;
75 flags
|= OBJECT_INFO_ALLOW_UNKNOWN_TYPE
;
77 if (get_oid_with_context(the_repository
, obj_name
,
80 die("Not a valid object name %s", obj_name
);
83 path
= obj_context
.path
;
84 if (obj_context
.mode
== S_IFINVALID
)
85 obj_context
.mode
= 0100644;
91 if (oid_object_info_extended(the_repository
, &oid
, &oi
, flags
) < 0)
92 die("git cat-file: could not get object info");
94 printf("%s\n", sb
.buf
);
102 if (oid_object_info_extended(the_repository
, &oid
, &oi
, flags
) < 0)
103 die("git cat-file: could not get object info");
104 printf("%"PRIuMAX
"\n", (uintmax_t)size
);
108 return !has_object_file(&oid
);
112 die("git cat-file --filters %s: <object> must be "
113 "<sha1:path>", obj_name
);
115 if (filter_object(path
, obj_context
.mode
,
122 die("git cat-file --textconv %s: <object> must be <sha1:path>",
125 if (textconv_object(the_repository
, path
, obj_context
.mode
,
126 &oid
, 1, &buf
, &size
))
128 /* else fallthrough */
131 type
= oid_object_info(the_repository
, &oid
, NULL
);
133 die("Not a valid object name %s", obj_name
);
135 /* custom pretty-print here */
136 if (type
== OBJ_TREE
) {
137 const char *ls_args
[3] = { NULL
};
138 ls_args
[0] = "ls-tree";
139 ls_args
[1] = obj_name
;
140 return cmd_ls_tree(2, ls_args
, NULL
);
143 if (type
== OBJ_BLOB
)
144 return stream_blob(&oid
);
145 buf
= read_object_file(&oid
, &type
, &size
);
147 die("Cannot read object %s", obj_name
);
149 /* otherwise just spit out the data */
153 if (type_from_string(exp_type
) == OBJ_BLOB
) {
154 struct object_id blob_oid
;
155 if (oid_object_info(the_repository
, &oid
, NULL
) == OBJ_TAG
) {
156 char *buffer
= read_object_file(&oid
, &type
,
159 if (!skip_prefix(buffer
, "object ", &target
) ||
160 get_oid_hex(target
, &blob_oid
))
161 die("%s not a valid tag", oid_to_hex(&oid
));
164 oidcpy(&blob_oid
, &oid
);
166 if (oid_object_info(the_repository
, &blob_oid
, NULL
) == OBJ_BLOB
)
167 return stream_blob(&blob_oid
);
169 * we attempted to dereference a tag to a blob
170 * and failed; there may be new dereference
171 * mechanisms this code is not aware of.
172 * fall-back to the usual case.
175 buf
= read_object_with_reference(the_repository
,
176 &oid
, exp_type
, &size
, NULL
);
180 die("git cat-file: unknown option: %s", exp_type
);
184 die("git cat-file %s: bad file", obj_name
);
186 write_or_die(1, buf
, size
);
188 free(obj_context
.path
);
193 struct object_id oid
;
194 enum object_type type
;
198 struct object_id delta_base_oid
;
201 * If mark_query is true, we do not expand anything, but rather
202 * just mark the object_info with items we wish to query.
207 * Whether to split the input on whitespace before feeding it to
208 * get_sha1; this is decided during the mark_query phase based on
209 * whether we have a %(rest) token in our format.
211 int split_on_whitespace
;
214 * After a mark_query run, this object_info is set up to be
215 * passed to oid_object_info_extended. It will point to the data
216 * elements above, so you can retrieve the response from there.
218 struct object_info info
;
221 * This flag will be true if the requested batch format and options
222 * don't require us to call oid_object_info, which can then be
225 unsigned skip_object_info
: 1;
228 static int is_atom(const char *atom
, const char *s
, int slen
)
230 int alen
= strlen(atom
);
231 return alen
== slen
&& !memcmp(atom
, s
, alen
);
234 static void expand_atom(struct strbuf
*sb
, const char *atom
, int len
,
237 struct expand_data
*data
= vdata
;
239 if (is_atom("objectname", atom
, len
)) {
240 if (!data
->mark_query
)
241 strbuf_addstr(sb
, oid_to_hex(&data
->oid
));
242 } else if (is_atom("objecttype", atom
, len
)) {
243 if (data
->mark_query
)
244 data
->info
.typep
= &data
->type
;
246 strbuf_addstr(sb
, type_name(data
->type
));
247 } else if (is_atom("objectsize", atom
, len
)) {
248 if (data
->mark_query
)
249 data
->info
.sizep
= &data
->size
;
251 strbuf_addf(sb
, "%"PRIuMAX
, (uintmax_t)data
->size
);
252 } else if (is_atom("objectsize:disk", atom
, len
)) {
253 if (data
->mark_query
)
254 data
->info
.disk_sizep
= &data
->disk_size
;
256 strbuf_addf(sb
, "%"PRIuMAX
, (uintmax_t)data
->disk_size
);
257 } else if (is_atom("rest", atom
, len
)) {
258 if (data
->mark_query
)
259 data
->split_on_whitespace
= 1;
261 strbuf_addstr(sb
, data
->rest
);
262 } else if (is_atom("deltabase", atom
, len
)) {
263 if (data
->mark_query
)
264 data
->info
.delta_base_sha1
= data
->delta_base_oid
.hash
;
267 oid_to_hex(&data
->delta_base_oid
));
269 die("unknown format element: %.*s", len
, atom
);
272 static size_t expand_format(struct strbuf
*sb
, const char *start
, void *data
)
278 end
= strchr(start
+ 1, ')');
280 die("format element '%s' does not end in ')'", start
);
282 expand_atom(sb
, start
+ 1, end
- start
- 1, data
);
284 return end
- start
+ 1;
287 static void batch_write(struct batch_options
*opt
, const void *data
, int len
)
289 if (opt
->buffer_output
) {
290 if (fwrite(data
, 1, len
, stdout
) != len
)
291 die_errno("unable to write to stdout");
293 write_or_die(1, data
, len
);
296 static void print_object_or_die(struct batch_options
*opt
, struct expand_data
*data
)
298 const struct object_id
*oid
= &data
->oid
;
300 assert(data
->info
.typep
);
302 if (data
->type
== OBJ_BLOB
) {
303 if (opt
->buffer_output
)
310 die("missing path for '%s'", oid_to_hex(oid
));
312 if (opt
->cmdmode
== 'w') {
313 if (filter_object(data
->rest
, 0100644, oid
,
315 die("could not convert '%s' %s",
316 oid_to_hex(oid
), data
->rest
);
317 } else if (opt
->cmdmode
== 'c') {
318 enum object_type type
;
319 if (!textconv_object(the_repository
,
320 data
->rest
, 0100644, oid
,
321 1, &contents
, &size
))
322 contents
= read_object_file(oid
,
326 die("could not convert '%s' %s",
327 oid_to_hex(oid
), data
->rest
);
329 BUG("invalid cmdmode: %c", opt
->cmdmode
);
330 batch_write(opt
, contents
, size
);
337 enum object_type type
;
341 contents
= read_object_file(oid
, &type
, &size
);
343 die("object %s disappeared", oid_to_hex(oid
));
344 if (type
!= data
->type
)
345 die("object %s changed type!?", oid_to_hex(oid
));
346 if (data
->info
.sizep
&& size
!= data
->size
)
347 die("object %s changed size!?", oid_to_hex(oid
));
349 batch_write(opt
, contents
, size
);
354 static void batch_object_write(const char *obj_name
,
355 struct strbuf
*scratch
,
356 struct batch_options
*opt
,
357 struct expand_data
*data
)
359 if (!data
->skip_object_info
&&
360 oid_object_info_extended(the_repository
, &data
->oid
, &data
->info
,
361 OBJECT_INFO_LOOKUP_REPLACE
) < 0) {
362 printf("%s missing\n",
363 obj_name
? obj_name
: oid_to_hex(&data
->oid
));
368 strbuf_reset(scratch
);
369 strbuf_expand(scratch
, opt
->format
, expand_format
, data
);
370 strbuf_addch(scratch
, '\n');
371 batch_write(opt
, scratch
->buf
, scratch
->len
);
373 if (opt
->print_contents
) {
374 print_object_or_die(opt
, data
);
375 batch_write(opt
, "\n", 1);
379 static void batch_one_object(const char *obj_name
,
380 struct strbuf
*scratch
,
381 struct batch_options
*opt
,
382 struct expand_data
*data
)
384 struct object_context ctx
;
385 int flags
= opt
->follow_symlinks
? GET_OID_FOLLOW_SYMLINKS
: 0;
386 enum get_oid_result result
;
388 result
= get_oid_with_context(the_repository
, obj_name
,
389 flags
, &data
->oid
, &ctx
);
390 if (result
!= FOUND
) {
393 printf("%s missing\n", obj_name
);
395 case SHORT_NAME_AMBIGUOUS
:
396 printf("%s ambiguous\n", obj_name
);
398 case DANGLING_SYMLINK
:
399 printf("dangling %"PRIuMAX
"\n%s\n",
400 (uintmax_t)strlen(obj_name
), obj_name
);
403 printf("loop %"PRIuMAX
"\n%s\n",
404 (uintmax_t)strlen(obj_name
), obj_name
);
407 printf("notdir %"PRIuMAX
"\n%s\n",
408 (uintmax_t)strlen(obj_name
), obj_name
);
411 BUG("unknown get_sha1_with_context result %d\n",
420 printf("symlink %"PRIuMAX
"\n%s\n",
421 (uintmax_t)ctx
.symlink_path
.len
,
422 ctx
.symlink_path
.buf
);
427 batch_object_write(obj_name
, scratch
, opt
, data
);
430 struct object_cb_data
{
431 struct batch_options
*opt
;
432 struct expand_data
*expand
;
434 struct strbuf
*scratch
;
437 static int batch_object_cb(const struct object_id
*oid
, void *vdata
)
439 struct object_cb_data
*data
= vdata
;
440 oidcpy(&data
->expand
->oid
, oid
);
441 batch_object_write(NULL
, data
->scratch
, data
->opt
, data
->expand
);
445 static int collect_loose_object(const struct object_id
*oid
,
449 oid_array_append(data
, oid
);
453 static int collect_packed_object(const struct object_id
*oid
,
454 struct packed_git
*pack
,
458 oid_array_append(data
, oid
);
462 static int batch_unordered_object(const struct object_id
*oid
, void *vdata
)
464 struct object_cb_data
*data
= vdata
;
466 if (oidset_insert(data
->seen
, oid
))
469 return batch_object_cb(oid
, data
);
472 static int batch_unordered_loose(const struct object_id
*oid
,
476 return batch_unordered_object(oid
, data
);
479 static int batch_unordered_packed(const struct object_id
*oid
,
480 struct packed_git
*pack
,
484 return batch_unordered_object(oid
, data
);
487 static int batch_objects(struct batch_options
*opt
)
489 struct strbuf input
= STRBUF_INIT
;
490 struct strbuf output
= STRBUF_INIT
;
491 struct expand_data data
;
496 opt
->format
= "%(objectname) %(objecttype) %(objectsize)";
499 * Expand once with our special mark_query flag, which will prime the
500 * object_info to be handed to oid_object_info_extended for each
503 memset(&data
, 0, sizeof(data
));
505 strbuf_expand(&output
, opt
->format
, expand_format
, &data
);
507 strbuf_release(&output
);
509 data
.split_on_whitespace
= 1;
511 if (opt
->all_objects
) {
512 struct object_info empty
= OBJECT_INFO_INIT
;
513 if (!memcmp(&data
.info
, &empty
, sizeof(empty
)))
514 data
.skip_object_info
= 1;
518 * If we are printing out the object, then always fill in the type,
519 * since we will want to decide whether or not to stream.
521 if (opt
->print_contents
)
522 data
.info
.typep
= &data
.type
;
524 if (opt
->all_objects
) {
525 struct object_cb_data cb
;
527 if (repository_format_partial_clone
)
528 warning("This repository has extensions.partialClone set. Some objects may not be loaded.");
532 cb
.scratch
= &output
;
534 if (opt
->unordered
) {
535 struct oidset seen
= OIDSET_INIT
;
539 for_each_loose_object(batch_unordered_loose
, &cb
, 0);
540 for_each_packed_object(batch_unordered_packed
, &cb
,
541 FOR_EACH_OBJECT_PACK_ORDER
);
545 struct oid_array sa
= OID_ARRAY_INIT
;
547 for_each_loose_object(collect_loose_object
, &sa
, 0);
548 for_each_packed_object(collect_packed_object
, &sa
, 0);
550 oid_array_for_each_unique(&sa
, batch_object_cb
, &cb
);
552 oid_array_clear(&sa
);
555 strbuf_release(&output
);
560 * We are going to call get_sha1 on a potentially very large number of
561 * objects. In most large cases, these will be actual object sha1s. The
562 * cost to double-check that each one is not also a ref (just so we can
563 * warn) ends up dwarfing the actual cost of the object lookups
564 * themselves. We can work around it by just turning off the warning.
566 save_warning
= warn_on_object_refname_ambiguity
;
567 warn_on_object_refname_ambiguity
= 0;
569 while (strbuf_getline(&input
, stdin
) != EOF
) {
570 if (data
.split_on_whitespace
) {
572 * Split at first whitespace, tying off the beginning
573 * of the string and saving the remainder (or NULL) in
576 char *p
= strpbrk(input
.buf
, " \t");
578 while (*p
&& strchr(" \t", *p
))
584 batch_one_object(input
.buf
, &output
, opt
, &data
);
587 strbuf_release(&input
);
588 strbuf_release(&output
);
589 warn_on_object_refname_ambiguity
= save_warning
;
593 static const char * const cat_file_usage
[] = {
594 N_("git cat-file (-t [--allow-unknown-type] | -s [--allow-unknown-type] | -e | -p | <type> | --textconv | --filters) [--path=<path>] <object>"),
595 N_("git cat-file (--batch | --batch-check) [--follow-symlinks] [--textconv | --filters]"),
599 static int git_cat_file_config(const char *var
, const char *value
, void *cb
)
601 if (userdiff_config(var
, value
) < 0)
604 return git_default_config(var
, value
, cb
);
607 static int batch_option_callback(const struct option
*opt
,
611 struct batch_options
*bo
= opt
->value
;
613 BUG_ON_OPT_NEG(unset
);
616 return error(_("only one batch option may be specified"));
620 bo
->print_contents
= !strcmp(opt
->long_name
, "batch");
626 int cmd_cat_file(int argc
, const char **argv
, const char *prefix
)
629 const char *exp_type
= NULL
, *obj_name
= NULL
;
630 struct batch_options batch
= {0};
631 int unknown_type
= 0;
633 const struct option options
[] = {
634 OPT_GROUP(N_("<type> can be one of: blob, tree, commit, tag")),
635 OPT_CMDMODE('t', NULL
, &opt
, N_("show object type"), 't'),
636 OPT_CMDMODE('s', NULL
, &opt
, N_("show object size"), 's'),
637 OPT_CMDMODE('e', NULL
, &opt
,
638 N_("exit with zero when there's no error"), 'e'),
639 OPT_CMDMODE('p', NULL
, &opt
, N_("pretty-print object's content"), 'p'),
640 OPT_CMDMODE(0, "textconv", &opt
,
641 N_("for blob objects, run textconv on object's content"), 'c'),
642 OPT_CMDMODE(0, "filters", &opt
,
643 N_("for blob objects, run filters on object's content"), 'w'),
644 OPT_STRING(0, "path", &force_path
, N_("blob"),
645 N_("use a specific path for --textconv/--filters")),
646 OPT_BOOL(0, "allow-unknown-type", &unknown_type
,
647 N_("allow -s and -t to work with broken/corrupt objects")),
648 OPT_BOOL(0, "buffer", &batch
.buffer_output
, N_("buffer --batch output")),
649 { OPTION_CALLBACK
, 0, "batch", &batch
, "format",
650 N_("show info and content of objects fed from the standard input"),
651 PARSE_OPT_OPTARG
| PARSE_OPT_NONEG
,
652 batch_option_callback
},
653 { OPTION_CALLBACK
, 0, "batch-check", &batch
, "format",
654 N_("show info about objects fed from the standard input"),
655 PARSE_OPT_OPTARG
| PARSE_OPT_NONEG
,
656 batch_option_callback
},
657 OPT_BOOL(0, "follow-symlinks", &batch
.follow_symlinks
,
658 N_("follow in-tree symlinks (used with --batch or --batch-check)")),
659 OPT_BOOL(0, "batch-all-objects", &batch
.all_objects
,
660 N_("show all objects with --batch or --batch-check")),
661 OPT_BOOL(0, "unordered", &batch
.unordered
,
662 N_("do not order --batch-all-objects output")),
666 git_config(git_cat_file_config
, NULL
);
668 batch
.buffer_output
= -1;
669 argc
= parse_options(argc
, argv
, prefix
, options
, cat_file_usage
, 0);
672 if (batch
.enabled
&& (opt
== 'c' || opt
== 'w'))
677 usage_with_options(cat_file_usage
, options
);
679 if (!opt
&& !batch
.enabled
) {
684 usage_with_options(cat_file_usage
, options
);
687 if (batch
.cmdmode
!= opt
|| argc
)
688 usage_with_options(cat_file_usage
, options
);
689 if (batch
.cmdmode
&& batch
.all_objects
)
690 die("--batch-all-objects cannot be combined with "
691 "--textconv nor with --filters");
694 if ((batch
.follow_symlinks
|| batch
.all_objects
) && !batch
.enabled
) {
695 usage_with_options(cat_file_usage
, options
);
698 if (force_path
&& opt
!= 'c' && opt
!= 'w') {
699 error("--path=<path> needs --textconv or --filters");
700 usage_with_options(cat_file_usage
, options
);
703 if (force_path
&& batch
.enabled
) {
704 error("--path=<path> incompatible with --batch");
705 usage_with_options(cat_file_usage
, options
);
708 if (batch
.buffer_output
< 0)
709 batch
.buffer_output
= batch
.all_objects
;
712 return batch_objects(&batch
);
714 if (unknown_type
&& opt
!= 't' && opt
!= 's')
715 die("git cat-file --allow-unknown-type: use with -s or -t");
716 return cat_one_file(opt
, exp_type
, obj_name
, unknown_type
);