t1300: add more tests for whitespace and inline comments
[alt-git.git] / builtin / cat-file.c
blobea8ad601ecc0b7ee2d7eb9c3da2bb70bcfc55cf8
1 /*
2 * GIT - The information manager from hell
4 * Copyright (C) Linus Torvalds, 2005
5 */
6 #define USE_THE_INDEX_VARIABLE
7 #include "builtin.h"
8 #include "config.h"
9 #include "convert.h"
10 #include "diff.h"
11 #include "environment.h"
12 #include "gettext.h"
13 #include "hex.h"
14 #include "ident.h"
15 #include "parse-options.h"
16 #include "userdiff.h"
17 #include "streaming.h"
18 #include "tree-walk.h"
19 #include "oid-array.h"
20 #include "packfile.h"
21 #include "object-file.h"
22 #include "object-name.h"
23 #include "object-store-ll.h"
24 #include "replace-object.h"
25 #include "promisor-remote.h"
26 #include "mailmap.h"
27 #include "write-or-die.h"
29 enum batch_mode {
30 BATCH_MODE_CONTENTS,
31 BATCH_MODE_INFO,
32 BATCH_MODE_QUEUE_AND_DISPATCH,
35 struct batch_options {
36 int enabled;
37 int follow_symlinks;
38 enum batch_mode batch_mode;
39 int buffer_output;
40 int all_objects;
41 int unordered;
42 int transform_mode; /* may be 'w' or 'c' for --filters or --textconv */
43 char input_delim;
44 char output_delim;
45 const char *format;
48 static const char *force_path;
50 static struct string_list mailmap = STRING_LIST_INIT_NODUP;
51 static int use_mailmap;
53 static char *replace_idents_using_mailmap(char *, size_t *);
55 static char *replace_idents_using_mailmap(char *object_buf, size_t *size)
57 struct strbuf sb = STRBUF_INIT;
58 const char *headers[] = { "author ", "committer ", "tagger ", NULL };
60 strbuf_attach(&sb, object_buf, *size, *size + 1);
61 apply_mailmap_to_header(&sb, headers, &mailmap);
62 *size = sb.len;
63 return strbuf_detach(&sb, NULL);
66 static int filter_object(const char *path, unsigned mode,
67 const struct object_id *oid,
68 char **buf, unsigned long *size)
70 enum object_type type;
72 *buf = repo_read_object_file(the_repository, oid, &type, size);
73 if (!*buf)
74 return error(_("cannot read object %s '%s'"),
75 oid_to_hex(oid), path);
76 if ((type == OBJ_BLOB) && S_ISREG(mode)) {
77 struct strbuf strbuf = STRBUF_INIT;
78 struct checkout_metadata meta;
80 init_checkout_metadata(&meta, NULL, NULL, oid);
81 if (convert_to_working_tree(&the_index, path, *buf, *size, &strbuf, &meta)) {
82 free(*buf);
83 *size = strbuf.len;
84 *buf = strbuf_detach(&strbuf, NULL);
88 return 0;
91 static int stream_blob(const struct object_id *oid)
93 if (stream_blob_to_fd(1, oid, NULL, 0))
94 die("unable to stream %s to stdout", oid_to_hex(oid));
95 return 0;
98 static int cat_one_file(int opt, const char *exp_type, const char *obj_name,
99 int unknown_type)
101 int ret;
102 struct object_id oid;
103 enum object_type type;
104 char *buf;
105 unsigned long size;
106 struct object_context obj_context;
107 struct object_info oi = OBJECT_INFO_INIT;
108 struct strbuf sb = STRBUF_INIT;
109 unsigned flags = OBJECT_INFO_LOOKUP_REPLACE;
110 unsigned get_oid_flags = GET_OID_RECORD_PATH | GET_OID_ONLY_TO_DIE;
111 const char *path = force_path;
112 const int opt_cw = (opt == 'c' || opt == 'w');
113 if (!path && opt_cw)
114 get_oid_flags |= GET_OID_REQUIRE_PATH;
116 if (unknown_type)
117 flags |= OBJECT_INFO_ALLOW_UNKNOWN_TYPE;
119 if (get_oid_with_context(the_repository, obj_name, get_oid_flags, &oid,
120 &obj_context))
121 die("Not a valid object name %s", obj_name);
123 if (!path)
124 path = obj_context.path;
125 if (obj_context.mode == S_IFINVALID)
126 obj_context.mode = 0100644;
128 buf = NULL;
129 switch (opt) {
130 case 't':
131 oi.type_name = &sb;
132 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
133 die("git cat-file: could not get object info");
134 if (sb.len) {
135 printf("%s\n", sb.buf);
136 strbuf_release(&sb);
137 ret = 0;
138 goto cleanup;
140 break;
142 case 's':
143 oi.sizep = &size;
145 if (use_mailmap) {
146 oi.typep = &type;
147 oi.contentp = (void**)&buf;
150 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
151 die("git cat-file: could not get object info");
153 if (use_mailmap && (type == OBJ_COMMIT || type == OBJ_TAG)) {
154 size_t s = size;
155 buf = replace_idents_using_mailmap(buf, &s);
156 size = cast_size_t_to_ulong(s);
159 printf("%"PRIuMAX"\n", (uintmax_t)size);
160 ret = 0;
161 goto cleanup;
163 case 'e':
164 return !repo_has_object_file(the_repository, &oid);
166 case 'w':
168 if (filter_object(path, obj_context.mode,
169 &oid, &buf, &size)) {
170 ret = -1;
171 goto cleanup;
173 break;
175 case 'c':
176 if (textconv_object(the_repository, path, obj_context.mode,
177 &oid, 1, &buf, &size))
178 break;
179 /* else fallthrough */
181 case 'p':
182 type = oid_object_info(the_repository, &oid, NULL);
183 if (type < 0)
184 die("Not a valid object name %s", obj_name);
186 /* custom pretty-print here */
187 if (type == OBJ_TREE) {
188 const char *ls_args[3] = { NULL };
189 ls_args[0] = "ls-tree";
190 ls_args[1] = obj_name;
191 ret = cmd_ls_tree(2, ls_args, NULL);
192 goto cleanup;
195 if (type == OBJ_BLOB) {
196 ret = stream_blob(&oid);
197 goto cleanup;
199 buf = repo_read_object_file(the_repository, &oid, &type,
200 &size);
201 if (!buf)
202 die("Cannot read object %s", obj_name);
204 if (use_mailmap) {
205 size_t s = size;
206 buf = replace_idents_using_mailmap(buf, &s);
207 size = cast_size_t_to_ulong(s);
210 /* otherwise just spit out the data */
211 break;
213 case 0:
215 enum object_type exp_type_id = type_from_string(exp_type);
217 if (exp_type_id == OBJ_BLOB) {
218 struct object_id blob_oid;
219 if (oid_object_info(the_repository, &oid, NULL) == OBJ_TAG) {
220 char *buffer = repo_read_object_file(the_repository,
221 &oid,
222 &type,
223 &size);
224 const char *target;
225 if (!skip_prefix(buffer, "object ", &target) ||
226 get_oid_hex(target, &blob_oid))
227 die("%s not a valid tag", oid_to_hex(&oid));
228 free(buffer);
229 } else
230 oidcpy(&blob_oid, &oid);
232 if (oid_object_info(the_repository, &blob_oid, NULL) == OBJ_BLOB) {
233 ret = stream_blob(&blob_oid);
234 goto cleanup;
237 * we attempted to dereference a tag to a blob
238 * and failed; there may be new dereference
239 * mechanisms this code is not aware of.
240 * fall-back to the usual case.
243 buf = read_object_with_reference(the_repository, &oid,
244 exp_type_id, &size, NULL);
246 if (use_mailmap) {
247 size_t s = size;
248 buf = replace_idents_using_mailmap(buf, &s);
249 size = cast_size_t_to_ulong(s);
251 break;
253 default:
254 die("git cat-file: unknown option: %s", exp_type);
257 if (!buf)
258 die("git cat-file %s: bad file", obj_name);
260 write_or_die(1, buf, size);
261 ret = 0;
262 cleanup:
263 free(buf);
264 free(obj_context.path);
265 return ret;
268 struct expand_data {
269 struct object_id oid;
270 enum object_type type;
271 unsigned long size;
272 off_t disk_size;
273 const char *rest;
274 struct object_id delta_base_oid;
277 * If mark_query is true, we do not expand anything, but rather
278 * just mark the object_info with items we wish to query.
280 int mark_query;
283 * Whether to split the input on whitespace before feeding it to
284 * get_sha1; this is decided during the mark_query phase based on
285 * whether we have a %(rest) token in our format.
287 int split_on_whitespace;
290 * After a mark_query run, this object_info is set up to be
291 * passed to oid_object_info_extended. It will point to the data
292 * elements above, so you can retrieve the response from there.
294 struct object_info info;
297 * This flag will be true if the requested batch format and options
298 * don't require us to call oid_object_info, which can then be
299 * optimized out.
301 unsigned skip_object_info : 1;
304 static int is_atom(const char *atom, const char *s, int slen)
306 int alen = strlen(atom);
307 return alen == slen && !memcmp(atom, s, alen);
310 static void expand_atom(struct strbuf *sb, const char *atom, int len,
311 struct expand_data *data)
313 if (is_atom("objectname", atom, len)) {
314 if (!data->mark_query)
315 strbuf_addstr(sb, oid_to_hex(&data->oid));
316 } else if (is_atom("objecttype", atom, len)) {
317 if (data->mark_query)
318 data->info.typep = &data->type;
319 else
320 strbuf_addstr(sb, type_name(data->type));
321 } else if (is_atom("objectsize", atom, len)) {
322 if (data->mark_query)
323 data->info.sizep = &data->size;
324 else
325 strbuf_addf(sb, "%"PRIuMAX , (uintmax_t)data->size);
326 } else if (is_atom("objectsize:disk", atom, len)) {
327 if (data->mark_query)
328 data->info.disk_sizep = &data->disk_size;
329 else
330 strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
331 } else if (is_atom("rest", atom, len)) {
332 if (data->mark_query)
333 data->split_on_whitespace = 1;
334 else if (data->rest)
335 strbuf_addstr(sb, data->rest);
336 } else if (is_atom("deltabase", atom, len)) {
337 if (data->mark_query)
338 data->info.delta_base_oid = &data->delta_base_oid;
339 else
340 strbuf_addstr(sb,
341 oid_to_hex(&data->delta_base_oid));
342 } else
343 die("unknown format element: %.*s", len, atom);
346 static void expand_format(struct strbuf *sb, const char *start,
347 struct expand_data *data)
349 while (strbuf_expand_step(sb, &start)) {
350 const char *end;
352 if (skip_prefix(start, "%", &start) || *start != '(')
353 strbuf_addch(sb, '%');
354 else if (!(end = strchr(start + 1, ')')))
355 die("format element '%s' does not end in ')'", start);
356 else {
357 expand_atom(sb, start + 1, end - start - 1, data);
358 start = end + 1;
363 static void batch_write(struct batch_options *opt, const void *data, int len)
365 if (opt->buffer_output) {
366 if (fwrite(data, 1, len, stdout) != len)
367 die_errno("unable to write to stdout");
368 } else
369 write_or_die(1, data, len);
372 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
374 const struct object_id *oid = &data->oid;
376 assert(data->info.typep);
378 if (data->type == OBJ_BLOB) {
379 if (opt->buffer_output)
380 fflush(stdout);
381 if (opt->transform_mode) {
382 char *contents;
383 unsigned long size;
385 if (!data->rest)
386 die("missing path for '%s'", oid_to_hex(oid));
388 if (opt->transform_mode == 'w') {
389 if (filter_object(data->rest, 0100644, oid,
390 &contents, &size))
391 die("could not convert '%s' %s",
392 oid_to_hex(oid), data->rest);
393 } else if (opt->transform_mode == 'c') {
394 enum object_type type;
395 if (!textconv_object(the_repository,
396 data->rest, 0100644, oid,
397 1, &contents, &size))
398 contents = repo_read_object_file(the_repository,
399 oid,
400 &type,
401 &size);
402 if (!contents)
403 die("could not convert '%s' %s",
404 oid_to_hex(oid), data->rest);
405 } else
406 BUG("invalid transform_mode: %c", opt->transform_mode);
407 batch_write(opt, contents, size);
408 free(contents);
409 } else {
410 stream_blob(oid);
413 else {
414 enum object_type type;
415 unsigned long size;
416 void *contents;
418 contents = repo_read_object_file(the_repository, oid, &type,
419 &size);
421 if (use_mailmap) {
422 size_t s = size;
423 contents = replace_idents_using_mailmap(contents, &s);
424 size = cast_size_t_to_ulong(s);
427 if (!contents)
428 die("object %s disappeared", oid_to_hex(oid));
429 if (type != data->type)
430 die("object %s changed type!?", oid_to_hex(oid));
431 if (data->info.sizep && size != data->size && !use_mailmap)
432 die("object %s changed size!?", oid_to_hex(oid));
434 batch_write(opt, contents, size);
435 free(contents);
439 static void print_default_format(struct strbuf *scratch, struct expand_data *data,
440 struct batch_options *opt)
442 strbuf_addf(scratch, "%s %s %"PRIuMAX"%c", oid_to_hex(&data->oid),
443 type_name(data->type),
444 (uintmax_t)data->size, opt->output_delim);
448 * If "pack" is non-NULL, then "offset" is the byte offset within the pack from
449 * which the object may be accessed (though note that we may also rely on
450 * data->oid, too). If "pack" is NULL, then offset is ignored.
452 static void batch_object_write(const char *obj_name,
453 struct strbuf *scratch,
454 struct batch_options *opt,
455 struct expand_data *data,
456 struct packed_git *pack,
457 off_t offset)
459 if (!data->skip_object_info) {
460 int ret;
462 if (use_mailmap)
463 data->info.typep = &data->type;
465 if (pack)
466 ret = packed_object_info(the_repository, pack, offset,
467 &data->info);
468 else
469 ret = oid_object_info_extended(the_repository,
470 &data->oid, &data->info,
471 OBJECT_INFO_LOOKUP_REPLACE);
472 if (ret < 0) {
473 printf("%s missing%c",
474 obj_name ? obj_name : oid_to_hex(&data->oid), opt->output_delim);
475 fflush(stdout);
476 return;
479 if (use_mailmap && (data->type == OBJ_COMMIT || data->type == OBJ_TAG)) {
480 size_t s = data->size;
481 char *buf = NULL;
483 buf = repo_read_object_file(the_repository, &data->oid, &data->type,
484 &data->size);
485 buf = replace_idents_using_mailmap(buf, &s);
486 data->size = cast_size_t_to_ulong(s);
488 free(buf);
492 strbuf_reset(scratch);
494 if (!opt->format) {
495 print_default_format(scratch, data, opt);
496 } else {
497 expand_format(scratch, opt->format, data);
498 strbuf_addch(scratch, opt->output_delim);
501 batch_write(opt, scratch->buf, scratch->len);
503 if (opt->batch_mode == BATCH_MODE_CONTENTS) {
504 print_object_or_die(opt, data);
505 batch_write(opt, &opt->output_delim, 1);
509 static void batch_one_object(const char *obj_name,
510 struct strbuf *scratch,
511 struct batch_options *opt,
512 struct expand_data *data)
514 struct object_context ctx;
515 int flags = opt->follow_symlinks ? GET_OID_FOLLOW_SYMLINKS : 0;
516 enum get_oid_result result;
518 result = get_oid_with_context(the_repository, obj_name,
519 flags, &data->oid, &ctx);
520 if (result != FOUND) {
521 switch (result) {
522 case MISSING_OBJECT:
523 printf("%s missing%c", obj_name, opt->output_delim);
524 break;
525 case SHORT_NAME_AMBIGUOUS:
526 printf("%s ambiguous%c", obj_name, opt->output_delim);
527 break;
528 case DANGLING_SYMLINK:
529 printf("dangling %"PRIuMAX"%c%s%c",
530 (uintmax_t)strlen(obj_name),
531 opt->output_delim, obj_name, opt->output_delim);
532 break;
533 case SYMLINK_LOOP:
534 printf("loop %"PRIuMAX"%c%s%c",
535 (uintmax_t)strlen(obj_name),
536 opt->output_delim, obj_name, opt->output_delim);
537 break;
538 case NOT_DIR:
539 printf("notdir %"PRIuMAX"%c%s%c",
540 (uintmax_t)strlen(obj_name),
541 opt->output_delim, obj_name, opt->output_delim);
542 break;
543 default:
544 BUG("unknown get_sha1_with_context result %d\n",
545 result);
546 break;
548 fflush(stdout);
549 return;
552 if (ctx.mode == 0) {
553 printf("symlink %"PRIuMAX"%c%s%c",
554 (uintmax_t)ctx.symlink_path.len,
555 opt->output_delim, ctx.symlink_path.buf, opt->output_delim);
556 fflush(stdout);
557 return;
560 batch_object_write(obj_name, scratch, opt, data, NULL, 0);
563 struct object_cb_data {
564 struct batch_options *opt;
565 struct expand_data *expand;
566 struct oidset *seen;
567 struct strbuf *scratch;
570 static int batch_object_cb(const struct object_id *oid, void *vdata)
572 struct object_cb_data *data = vdata;
573 oidcpy(&data->expand->oid, oid);
574 batch_object_write(NULL, data->scratch, data->opt, data->expand,
575 NULL, 0);
576 return 0;
579 static int collect_loose_object(const struct object_id *oid,
580 const char *path UNUSED,
581 void *data)
583 oid_array_append(data, oid);
584 return 0;
587 static int collect_packed_object(const struct object_id *oid,
588 struct packed_git *pack UNUSED,
589 uint32_t pos UNUSED,
590 void *data)
592 oid_array_append(data, oid);
593 return 0;
596 static int batch_unordered_object(const struct object_id *oid,
597 struct packed_git *pack, off_t offset,
598 void *vdata)
600 struct object_cb_data *data = vdata;
602 if (oidset_insert(data->seen, oid))
603 return 0;
605 oidcpy(&data->expand->oid, oid);
606 batch_object_write(NULL, data->scratch, data->opt, data->expand,
607 pack, offset);
608 return 0;
611 static int batch_unordered_loose(const struct object_id *oid,
612 const char *path UNUSED,
613 void *data)
615 return batch_unordered_object(oid, NULL, 0, data);
618 static int batch_unordered_packed(const struct object_id *oid,
619 struct packed_git *pack,
620 uint32_t pos,
621 void *data)
623 return batch_unordered_object(oid, pack,
624 nth_packed_object_offset(pack, pos),
625 data);
628 typedef void (*parse_cmd_fn_t)(struct batch_options *, const char *,
629 struct strbuf *, struct expand_data *);
631 struct queued_cmd {
632 parse_cmd_fn_t fn;
633 char *line;
636 static void parse_cmd_contents(struct batch_options *opt,
637 const char *line,
638 struct strbuf *output,
639 struct expand_data *data)
641 opt->batch_mode = BATCH_MODE_CONTENTS;
642 batch_one_object(line, output, opt, data);
645 static void parse_cmd_info(struct batch_options *opt,
646 const char *line,
647 struct strbuf *output,
648 struct expand_data *data)
650 opt->batch_mode = BATCH_MODE_INFO;
651 batch_one_object(line, output, opt, data);
654 static void dispatch_calls(struct batch_options *opt,
655 struct strbuf *output,
656 struct expand_data *data,
657 struct queued_cmd *cmd,
658 int nr)
660 int i;
662 if (!opt->buffer_output)
663 die(_("flush is only for --buffer mode"));
665 for (i = 0; i < nr; i++)
666 cmd[i].fn(opt, cmd[i].line, output, data);
668 fflush(stdout);
671 static void free_cmds(struct queued_cmd *cmd, size_t *nr)
673 size_t i;
675 for (i = 0; i < *nr; i++)
676 FREE_AND_NULL(cmd[i].line);
678 *nr = 0;
682 static const struct parse_cmd {
683 const char *name;
684 parse_cmd_fn_t fn;
685 unsigned takes_args;
686 } commands[] = {
687 { "contents", parse_cmd_contents, 1},
688 { "info", parse_cmd_info, 1},
689 { "flush", NULL, 0},
692 static void batch_objects_command(struct batch_options *opt,
693 struct strbuf *output,
694 struct expand_data *data)
696 struct strbuf input = STRBUF_INIT;
697 struct queued_cmd *queued_cmd = NULL;
698 size_t alloc = 0, nr = 0;
700 while (strbuf_getdelim_strip_crlf(&input, stdin, opt->input_delim) != EOF) {
701 int i;
702 const struct parse_cmd *cmd = NULL;
703 const char *p = NULL, *cmd_end;
704 struct queued_cmd call = {0};
706 if (!input.len)
707 die(_("empty command in input"));
708 if (isspace(*input.buf))
709 die(_("whitespace before command: '%s'"), input.buf);
711 for (i = 0; i < ARRAY_SIZE(commands); i++) {
712 if (!skip_prefix(input.buf, commands[i].name, &cmd_end))
713 continue;
715 cmd = &commands[i];
716 if (cmd->takes_args) {
717 if (*cmd_end != ' ')
718 die(_("%s requires arguments"),
719 commands[i].name);
721 p = cmd_end + 1;
722 } else if (*cmd_end) {
723 die(_("%s takes no arguments"),
724 commands[i].name);
727 break;
730 if (!cmd)
731 die(_("unknown command: '%s'"), input.buf);
733 if (!strcmp(cmd->name, "flush")) {
734 dispatch_calls(opt, output, data, queued_cmd, nr);
735 free_cmds(queued_cmd, &nr);
736 } else if (!opt->buffer_output) {
737 cmd->fn(opt, p, output, data);
738 } else {
739 ALLOC_GROW(queued_cmd, nr + 1, alloc);
740 call.fn = cmd->fn;
741 call.line = xstrdup_or_null(p);
742 queued_cmd[nr++] = call;
746 if (opt->buffer_output &&
747 nr &&
748 !git_env_bool("GIT_TEST_CAT_FILE_NO_FLUSH_ON_EXIT", 0)) {
749 dispatch_calls(opt, output, data, queued_cmd, nr);
750 free_cmds(queued_cmd, &nr);
753 free_cmds(queued_cmd, &nr);
754 free(queued_cmd);
755 strbuf_release(&input);
758 #define DEFAULT_FORMAT "%(objectname) %(objecttype) %(objectsize)"
760 static int batch_objects(struct batch_options *opt)
762 struct strbuf input = STRBUF_INIT;
763 struct strbuf output = STRBUF_INIT;
764 struct expand_data data;
765 int save_warning;
766 int retval = 0;
769 * Expand once with our special mark_query flag, which will prime the
770 * object_info to be handed to oid_object_info_extended for each
771 * object.
773 memset(&data, 0, sizeof(data));
774 data.mark_query = 1;
775 expand_format(&output,
776 opt->format ? opt->format : DEFAULT_FORMAT,
777 &data);
778 data.mark_query = 0;
779 strbuf_release(&output);
780 if (opt->transform_mode)
781 data.split_on_whitespace = 1;
783 if (opt->format && !strcmp(opt->format, DEFAULT_FORMAT))
784 opt->format = NULL;
786 * If we are printing out the object, then always fill in the type,
787 * since we will want to decide whether or not to stream.
789 if (opt->batch_mode == BATCH_MODE_CONTENTS)
790 data.info.typep = &data.type;
792 if (opt->all_objects) {
793 struct object_cb_data cb;
794 struct object_info empty = OBJECT_INFO_INIT;
796 if (!memcmp(&data.info, &empty, sizeof(empty)))
797 data.skip_object_info = 1;
799 if (repo_has_promisor_remote(the_repository))
800 warning("This repository uses promisor remotes. Some objects may not be loaded.");
802 disable_replace_refs();
804 cb.opt = opt;
805 cb.expand = &data;
806 cb.scratch = &output;
808 if (opt->unordered) {
809 struct oidset seen = OIDSET_INIT;
811 cb.seen = &seen;
813 for_each_loose_object(batch_unordered_loose, &cb, 0);
814 for_each_packed_object(batch_unordered_packed, &cb,
815 FOR_EACH_OBJECT_PACK_ORDER);
817 oidset_clear(&seen);
818 } else {
819 struct oid_array sa = OID_ARRAY_INIT;
821 for_each_loose_object(collect_loose_object, &sa, 0);
822 for_each_packed_object(collect_packed_object, &sa, 0);
824 oid_array_for_each_unique(&sa, batch_object_cb, &cb);
826 oid_array_clear(&sa);
829 strbuf_release(&output);
830 return 0;
834 * We are going to call get_sha1 on a potentially very large number of
835 * objects. In most large cases, these will be actual object sha1s. The
836 * cost to double-check that each one is not also a ref (just so we can
837 * warn) ends up dwarfing the actual cost of the object lookups
838 * themselves. We can work around it by just turning off the warning.
840 save_warning = warn_on_object_refname_ambiguity;
841 warn_on_object_refname_ambiguity = 0;
843 if (opt->batch_mode == BATCH_MODE_QUEUE_AND_DISPATCH) {
844 batch_objects_command(opt, &output, &data);
845 goto cleanup;
848 while (strbuf_getdelim_strip_crlf(&input, stdin, opt->input_delim) != EOF) {
849 if (data.split_on_whitespace) {
851 * Split at first whitespace, tying off the beginning
852 * of the string and saving the remainder (or NULL) in
853 * data.rest.
855 char *p = strpbrk(input.buf, " \t");
856 if (p) {
857 while (*p && strchr(" \t", *p))
858 *p++ = '\0';
860 data.rest = p;
863 batch_one_object(input.buf, &output, opt, &data);
866 cleanup:
867 strbuf_release(&input);
868 strbuf_release(&output);
869 warn_on_object_refname_ambiguity = save_warning;
870 return retval;
873 static int git_cat_file_config(const char *var, const char *value,
874 const struct config_context *ctx, void *cb)
876 if (userdiff_config(var, value) < 0)
877 return -1;
879 return git_default_config(var, value, ctx, cb);
882 static int batch_option_callback(const struct option *opt,
883 const char *arg,
884 int unset)
886 struct batch_options *bo = opt->value;
888 BUG_ON_OPT_NEG(unset);
890 if (bo->enabled) {
891 return error(_("only one batch option may be specified"));
894 bo->enabled = 1;
896 if (!strcmp(opt->long_name, "batch"))
897 bo->batch_mode = BATCH_MODE_CONTENTS;
898 else if (!strcmp(opt->long_name, "batch-check"))
899 bo->batch_mode = BATCH_MODE_INFO;
900 else if (!strcmp(opt->long_name, "batch-command"))
901 bo->batch_mode = BATCH_MODE_QUEUE_AND_DISPATCH;
902 else
903 BUG("%s given to batch-option-callback", opt->long_name);
905 bo->format = arg;
907 return 0;
910 int cmd_cat_file(int argc, const char **argv, const char *prefix)
912 int opt = 0;
913 int opt_cw = 0;
914 int opt_epts = 0;
915 const char *exp_type = NULL, *obj_name = NULL;
916 struct batch_options batch = {0};
917 int unknown_type = 0;
918 int input_nul_terminated = 0;
919 int nul_terminated = 0;
921 const char * const usage[] = {
922 N_("git cat-file <type> <object>"),
923 N_("git cat-file (-e | -p) <object>"),
924 N_("git cat-file (-t | -s) [--allow-unknown-type] <object>"),
925 N_("git cat-file (--textconv | --filters)\n"
926 " [<rev>:<path|tree-ish> | --path=<path|tree-ish> <rev>]"),
927 N_("git cat-file (--batch | --batch-check | --batch-command) [--batch-all-objects]\n"
928 " [--buffer] [--follow-symlinks] [--unordered]\n"
929 " [--textconv | --filters] [-Z]"),
930 NULL
932 const struct option options[] = {
933 /* Simple queries */
934 OPT_GROUP(N_("Check object existence or emit object contents")),
935 OPT_CMDMODE('e', NULL, &opt,
936 N_("check if <object> exists"), 'e'),
937 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print <object> content"), 'p'),
939 OPT_GROUP(N_("Emit [broken] object attributes")),
940 OPT_CMDMODE('t', NULL, &opt, N_("show object type (one of 'blob', 'tree', 'commit', 'tag', ...)"), 't'),
941 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
942 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
943 N_("allow -s and -t to work with broken/corrupt objects")),
944 OPT_BOOL(0, "use-mailmap", &use_mailmap, N_("use mail map file")),
945 OPT_ALIAS(0, "mailmap", "use-mailmap"),
946 /* Batch mode */
947 OPT_GROUP(N_("Batch objects requested on stdin (or --batch-all-objects)")),
948 OPT_CALLBACK_F(0, "batch", &batch, N_("format"),
949 N_("show full <object> or <rev> contents"),
950 PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
951 batch_option_callback),
952 OPT_CALLBACK_F(0, "batch-check", &batch, N_("format"),
953 N_("like --batch, but don't emit <contents>"),
954 PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
955 batch_option_callback),
956 OPT_BOOL_F('z', NULL, &input_nul_terminated, N_("stdin is NUL-terminated"),
957 PARSE_OPT_HIDDEN),
958 OPT_BOOL('Z', NULL, &nul_terminated, N_("stdin and stdout is NUL-terminated")),
959 OPT_CALLBACK_F(0, "batch-command", &batch, N_("format"),
960 N_("read commands from stdin"),
961 PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
962 batch_option_callback),
963 OPT_CMDMODE(0, "batch-all-objects", &opt,
964 N_("with --batch[-check]: ignores stdin, batches all known objects"), 'b'),
965 /* Batch-specific options */
966 OPT_GROUP(N_("Change or optimize batch output")),
967 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
968 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
969 N_("follow in-tree symlinks")),
970 OPT_BOOL(0, "unordered", &batch.unordered,
971 N_("do not order objects before emitting them")),
972 /* Textconv options, stand-ole*/
973 OPT_GROUP(N_("Emit object (blob or tree) with conversion or filter (stand-alone, or with batch)")),
974 OPT_CMDMODE(0, "textconv", &opt,
975 N_("run textconv on object's content"), 'c'),
976 OPT_CMDMODE(0, "filters", &opt,
977 N_("run filters on object's content"), 'w'),
978 OPT_STRING(0, "path", &force_path, N_("blob|tree"),
979 N_("use a <path> for (--textconv | --filters); Not with 'batch'")),
980 OPT_END()
983 git_config(git_cat_file_config, NULL);
985 batch.buffer_output = -1;
987 argc = parse_options(argc, argv, prefix, options, usage, 0);
988 opt_cw = (opt == 'c' || opt == 'w');
989 opt_epts = (opt == 'e' || opt == 'p' || opt == 't' || opt == 's');
991 if (use_mailmap)
992 read_mailmap(&mailmap);
994 /* --batch-all-objects? */
995 if (opt == 'b')
996 batch.all_objects = 1;
998 /* Option compatibility */
999 if (force_path && !opt_cw)
1000 usage_msg_optf(_("'%s=<%s>' needs '%s' or '%s'"),
1001 usage, options,
1002 "--path", _("path|tree-ish"), "--filters",
1003 "--textconv");
1005 /* Option compatibility with batch mode */
1006 if (batch.enabled)
1008 else if (batch.follow_symlinks)
1009 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1010 "--follow-symlinks");
1011 else if (batch.buffer_output >= 0)
1012 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1013 "--buffer");
1014 else if (batch.all_objects)
1015 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1016 "--batch-all-objects");
1017 else if (input_nul_terminated)
1018 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1019 "-z");
1020 else if (nul_terminated)
1021 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1022 "-Z");
1024 batch.input_delim = batch.output_delim = '\n';
1025 if (input_nul_terminated)
1026 batch.input_delim = '\0';
1027 if (nul_terminated)
1028 batch.input_delim = batch.output_delim = '\0';
1030 /* Batch defaults */
1031 if (batch.buffer_output < 0)
1032 batch.buffer_output = batch.all_objects;
1034 /* Return early if we're in batch mode? */
1035 if (batch.enabled) {
1036 if (opt_cw)
1037 batch.transform_mode = opt;
1038 else if (opt && opt != 'b')
1039 usage_msg_optf(_("'-%c' is incompatible with batch mode"),
1040 usage, options, opt);
1041 else if (argc)
1042 usage_msg_opt(_("batch modes take no arguments"), usage,
1043 options);
1045 return batch_objects(&batch);
1048 if (opt) {
1049 if (!argc && opt == 'c')
1050 usage_msg_optf(_("<rev> required with '%s'"),
1051 usage, options, "--textconv");
1052 else if (!argc && opt == 'w')
1053 usage_msg_optf(_("<rev> required with '%s'"),
1054 usage, options, "--filters");
1055 else if (!argc && opt_epts)
1056 usage_msg_optf(_("<object> required with '-%c'"),
1057 usage, options, opt);
1058 else if (argc == 1)
1059 obj_name = argv[0];
1060 else
1061 usage_msg_opt(_("too many arguments"), usage, options);
1062 } else if (!argc) {
1063 usage_with_options(usage, options);
1064 } else if (argc != 2) {
1065 usage_msg_optf(_("only two arguments allowed in <type> <object> mode, not %d"),
1066 usage, options, argc);
1067 } else if (argc) {
1068 exp_type = argv[0];
1069 obj_name = argv[1];
1072 if (unknown_type && opt != 't' && opt != 's')
1073 die("git cat-file --allow-unknown-type: use with -s or -t");
1074 return cat_one_file(opt, exp_type, obj_name, unknown_type);