cat-file: use "type" and "size" from outer scope
[git/git-svn.git] / builtin / cat-file.c
blob5e24e811ee93d33180d4e3a19a21989b31169c56
1 /*
2 * GIT - The information manager from hell
4 * Copyright (C) Linus Torvalds, 2005
5 */
6 #include "cache.h"
7 #include "exec_cmd.h"
8 #include "tag.h"
9 #include "tree.h"
10 #include "builtin.h"
11 #include "parse-options.h"
12 #include "diff.h"
13 #include "userdiff.h"
14 #include "streaming.h"
16 static int cat_one_file(int opt, const char *exp_type, const char *obj_name)
18 unsigned char sha1[20];
19 enum object_type type;
20 char *buf;
21 unsigned long size;
22 struct object_context obj_context;
24 if (get_sha1_with_context(obj_name, 0, sha1, &obj_context))
25 die("Not a valid object name %s", obj_name);
27 buf = NULL;
28 switch (opt) {
29 case 't':
30 type = sha1_object_info(sha1, NULL);
31 if (type > 0) {
32 printf("%s\n", typename(type));
33 return 0;
35 break;
37 case 's':
38 type = sha1_object_info(sha1, &size);
39 if (type > 0) {
40 printf("%lu\n", size);
41 return 0;
43 break;
45 case 'e':
46 return !has_sha1_file(sha1);
48 case 'c':
49 if (!obj_context.path[0])
50 die("git cat-file --textconv %s: <object> must be <sha1:path>",
51 obj_name);
53 if (textconv_object(obj_context.path, obj_context.mode, sha1, 1, &buf, &size))
54 break;
56 case 'p':
57 type = sha1_object_info(sha1, NULL);
58 if (type < 0)
59 die("Not a valid object name %s", obj_name);
61 /* custom pretty-print here */
62 if (type == OBJ_TREE) {
63 const char *ls_args[3] = { NULL };
64 ls_args[0] = "ls-tree";
65 ls_args[1] = obj_name;
66 return cmd_ls_tree(2, ls_args, NULL);
69 if (type == OBJ_BLOB)
70 return stream_blob_to_fd(1, sha1, NULL, 0);
71 buf = read_sha1_file(sha1, &type, &size);
72 if (!buf)
73 die("Cannot read object %s", obj_name);
75 /* otherwise just spit out the data */
76 break;
78 case 0:
79 if (type_from_string(exp_type) == OBJ_BLOB) {
80 unsigned char blob_sha1[20];
81 if (sha1_object_info(sha1, NULL) == OBJ_TAG) {
82 char *buffer = read_sha1_file(sha1, &type, &size);
83 const char *target;
84 if (!skip_prefix(buffer, "object ", &target) ||
85 get_sha1_hex(target, blob_sha1))
86 die("%s not a valid tag", sha1_to_hex(sha1));
87 free(buffer);
88 } else
89 hashcpy(blob_sha1, sha1);
91 if (sha1_object_info(blob_sha1, NULL) == OBJ_BLOB)
92 return stream_blob_to_fd(1, blob_sha1, NULL, 0);
94 * we attempted to dereference a tag to a blob
95 * and failed; there may be new dereference
96 * mechanisms this code is not aware of.
97 * fall-back to the usual case.
100 buf = read_object_with_reference(sha1, exp_type, &size, NULL);
101 break;
103 default:
104 die("git cat-file: unknown option: %s", exp_type);
107 if (!buf)
108 die("git cat-file %s: bad file", obj_name);
110 write_or_die(1, buf, size);
111 return 0;
114 struct expand_data {
115 unsigned char sha1[20];
116 enum object_type type;
117 unsigned long size;
118 unsigned long disk_size;
119 const char *rest;
120 unsigned char delta_base_sha1[20];
123 * If mark_query is true, we do not expand anything, but rather
124 * just mark the object_info with items we wish to query.
126 int mark_query;
129 * Whether to split the input on whitespace before feeding it to
130 * get_sha1; this is decided during the mark_query phase based on
131 * whether we have a %(rest) token in our format.
133 int split_on_whitespace;
136 * After a mark_query run, this object_info is set up to be
137 * passed to sha1_object_info_extended. It will point to the data
138 * elements above, so you can retrieve the response from there.
140 struct object_info info;
143 static int is_atom(const char *atom, const char *s, int slen)
145 int alen = strlen(atom);
146 return alen == slen && !memcmp(atom, s, alen);
149 static void expand_atom(struct strbuf *sb, const char *atom, int len,
150 void *vdata)
152 struct expand_data *data = vdata;
154 if (is_atom("objectname", atom, len)) {
155 if (!data->mark_query)
156 strbuf_addstr(sb, sha1_to_hex(data->sha1));
157 } else if (is_atom("objecttype", atom, len)) {
158 if (data->mark_query)
159 data->info.typep = &data->type;
160 else
161 strbuf_addstr(sb, typename(data->type));
162 } else if (is_atom("objectsize", atom, len)) {
163 if (data->mark_query)
164 data->info.sizep = &data->size;
165 else
166 strbuf_addf(sb, "%lu", data->size);
167 } else if (is_atom("objectsize:disk", atom, len)) {
168 if (data->mark_query)
169 data->info.disk_sizep = &data->disk_size;
170 else
171 strbuf_addf(sb, "%lu", data->disk_size);
172 } else if (is_atom("rest", atom, len)) {
173 if (data->mark_query)
174 data->split_on_whitespace = 1;
175 else if (data->rest)
176 strbuf_addstr(sb, data->rest);
177 } else if (is_atom("deltabase", atom, len)) {
178 if (data->mark_query)
179 data->info.delta_base_sha1 = data->delta_base_sha1;
180 else
181 strbuf_addstr(sb, sha1_to_hex(data->delta_base_sha1));
182 } else
183 die("unknown format element: %.*s", len, atom);
186 static size_t expand_format(struct strbuf *sb, const char *start, void *data)
188 const char *end;
190 if (*start != '(')
191 return 0;
192 end = strchr(start + 1, ')');
193 if (!end)
194 die("format element '%s' does not end in ')'", start);
196 expand_atom(sb, start + 1, end - start - 1, data);
198 return end - start + 1;
201 static void print_object_or_die(int fd, struct expand_data *data)
203 const unsigned char *sha1 = data->sha1;
205 assert(data->info.typep);
207 if (data->type == OBJ_BLOB) {
208 if (stream_blob_to_fd(fd, sha1, NULL, 0) < 0)
209 die("unable to stream %s to stdout", sha1_to_hex(sha1));
211 else {
212 enum object_type type;
213 unsigned long size;
214 void *contents;
216 contents = read_sha1_file(sha1, &type, &size);
217 if (!contents)
218 die("object %s disappeared", sha1_to_hex(sha1));
219 if (type != data->type)
220 die("object %s changed type!?", sha1_to_hex(sha1));
221 if (data->info.sizep && size != data->size)
222 die("object %s changed size!?", sha1_to_hex(sha1));
224 write_or_die(fd, contents, size);
225 free(contents);
229 struct batch_options {
230 int enabled;
231 int print_contents;
232 const char *format;
235 static int batch_one_object(const char *obj_name, struct batch_options *opt,
236 struct expand_data *data)
238 struct strbuf buf = STRBUF_INIT;
240 if (!obj_name)
241 return 1;
243 if (get_sha1(obj_name, data->sha1)) {
244 printf("%s missing\n", obj_name);
245 fflush(stdout);
246 return 0;
249 if (sha1_object_info_extended(data->sha1, &data->info, LOOKUP_REPLACE_OBJECT) < 0) {
250 printf("%s missing\n", obj_name);
251 fflush(stdout);
252 return 0;
255 strbuf_expand(&buf, opt->format, expand_format, data);
256 strbuf_addch(&buf, '\n');
257 write_or_die(1, buf.buf, buf.len);
258 strbuf_release(&buf);
260 if (opt->print_contents) {
261 print_object_or_die(1, data);
262 write_or_die(1, "\n", 1);
264 return 0;
267 static int batch_objects(struct batch_options *opt)
269 struct strbuf buf = STRBUF_INIT;
270 struct expand_data data;
271 int save_warning;
272 int retval = 0;
274 if (!opt->format)
275 opt->format = "%(objectname) %(objecttype) %(objectsize)";
278 * Expand once with our special mark_query flag, which will prime the
279 * object_info to be handed to sha1_object_info_extended for each
280 * object.
282 memset(&data, 0, sizeof(data));
283 data.mark_query = 1;
284 strbuf_expand(&buf, opt->format, expand_format, &data);
285 data.mark_query = 0;
288 * If we are printing out the object, then always fill in the type,
289 * since we will want to decide whether or not to stream.
291 if (opt->print_contents)
292 data.info.typep = &data.type;
295 * We are going to call get_sha1 on a potentially very large number of
296 * objects. In most large cases, these will be actual object sha1s. The
297 * cost to double-check that each one is not also a ref (just so we can
298 * warn) ends up dwarfing the actual cost of the object lookups
299 * themselves. We can work around it by just turning off the warning.
301 save_warning = warn_on_object_refname_ambiguity;
302 warn_on_object_refname_ambiguity = 0;
304 while (strbuf_getline(&buf, stdin, '\n') != EOF) {
305 if (data.split_on_whitespace) {
307 * Split at first whitespace, tying off the beginning
308 * of the string and saving the remainder (or NULL) in
309 * data.rest.
311 char *p = strpbrk(buf.buf, " \t");
312 if (p) {
313 while (*p && strchr(" \t", *p))
314 *p++ = '\0';
316 data.rest = p;
319 retval = batch_one_object(buf.buf, opt, &data);
320 if (retval)
321 break;
324 strbuf_release(&buf);
325 warn_on_object_refname_ambiguity = save_warning;
326 return retval;
329 static const char * const cat_file_usage[] = {
330 N_("git cat-file (-t|-s|-e|-p|<type>|--textconv) <object>"),
331 N_("git cat-file (--batch|--batch-check) < <list_of_objects>"),
332 NULL
335 static int git_cat_file_config(const char *var, const char *value, void *cb)
337 if (userdiff_config(var, value) < 0)
338 return -1;
340 return git_default_config(var, value, cb);
343 static int batch_option_callback(const struct option *opt,
344 const char *arg,
345 int unset)
347 struct batch_options *bo = opt->value;
349 if (unset) {
350 memset(bo, 0, sizeof(*bo));
351 return 0;
354 bo->enabled = 1;
355 bo->print_contents = !strcmp(opt->long_name, "batch");
356 bo->format = arg;
358 return 0;
361 int cmd_cat_file(int argc, const char **argv, const char *prefix)
363 int opt = 0;
364 const char *exp_type = NULL, *obj_name = NULL;
365 struct batch_options batch = {0};
367 const struct option options[] = {
368 OPT_GROUP(N_("<type> can be one of: blob, tree, commit, tag")),
369 OPT_SET_INT('t', NULL, &opt, N_("show object type"), 't'),
370 OPT_SET_INT('s', NULL, &opt, N_("show object size"), 's'),
371 OPT_SET_INT('e', NULL, &opt,
372 N_("exit with zero when there's no error"), 'e'),
373 OPT_SET_INT('p', NULL, &opt, N_("pretty-print object's content"), 'p'),
374 OPT_SET_INT(0, "textconv", &opt,
375 N_("for blob objects, run textconv on object's content"), 'c'),
376 { OPTION_CALLBACK, 0, "batch", &batch, "format",
377 N_("show info and content of objects fed from the standard input"),
378 PARSE_OPT_OPTARG, batch_option_callback },
379 { OPTION_CALLBACK, 0, "batch-check", &batch, "format",
380 N_("show info about objects fed from the standard input"),
381 PARSE_OPT_OPTARG, batch_option_callback },
382 OPT_END()
385 git_config(git_cat_file_config, NULL);
387 if (argc != 3 && argc != 2)
388 usage_with_options(cat_file_usage, options);
390 argc = parse_options(argc, argv, prefix, options, cat_file_usage, 0);
392 if (opt) {
393 if (argc == 1)
394 obj_name = argv[0];
395 else
396 usage_with_options(cat_file_usage, options);
398 if (!opt && !batch.enabled) {
399 if (argc == 2) {
400 exp_type = argv[0];
401 obj_name = argv[1];
402 } else
403 usage_with_options(cat_file_usage, options);
405 if (batch.enabled && (opt || argc)) {
406 usage_with_options(cat_file_usage, options);
409 if (batch.enabled)
410 return batch_objects(&batch);
412 return cat_one_file(opt, exp_type, obj_name);