cat-file: provide %(deltabase) batch format
[git] / builtin / cat-file.c
1 /*
2  * GIT - The information manager from hell
3  *
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"
15
16 static int cat_one_file(int opt, const char *exp_type, const char *obj_name)
17 {
18         unsigned char sha1[20];
19         enum object_type type;
20         char *buf;
21         unsigned long size;
22         struct object_context obj_context;
23
24         if (get_sha1_with_context(obj_name, 0, sha1, &obj_context))
25                 die("Not a valid object name %s", obj_name);
26
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;
34                 }
35                 break;
36
37         case 's':
38                 type = sha1_object_info(sha1, &size);
39                 if (type > 0) {
40                         printf("%lu\n", size);
41                         return 0;
42                 }
43                 break;
44
45         case 'e':
46                 return !has_sha1_file(sha1);
47
48         case 'c':
49                 if (!obj_context.path[0])
50                         die("git cat-file --textconv %s: <object> must be <sha1:path>",
51                             obj_name);
52
53                 if (textconv_object(obj_context.path, obj_context.mode, sha1, 1, &buf, &size))
54                         break;
55
56         case 'p':
57                 type = sha1_object_info(sha1, NULL);
58                 if (type < 0)
59                         die("Not a valid object name %s", obj_name);
60
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);
67                 }
68
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);
74
75                 /* otherwise just spit out the data */
76                 break;
77
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                                 enum object_type type;
83                                 unsigned long size;
84                                 char *buffer = read_sha1_file(sha1, &type, &size);
85                                 if (memcmp(buffer, "object ", 7) ||
86                                     get_sha1_hex(buffer + 7, blob_sha1))
87                                         die("%s not a valid tag", sha1_to_hex(sha1));
88                                 free(buffer);
89                         } else
90                                 hashcpy(blob_sha1, sha1);
91
92                         if (sha1_object_info(blob_sha1, NULL) == OBJ_BLOB)
93                                 return stream_blob_to_fd(1, blob_sha1, NULL, 0);
94                         /*
95                          * we attempted to dereference a tag to a blob
96                          * and failed; there may be new dereference
97                          * mechanisms this code is not aware of.
98                          * fall-back to the usual case.
99                          */
100                 }
101                 buf = read_object_with_reference(sha1, exp_type, &size, NULL);
102                 break;
103
104         default:
105                 die("git cat-file: unknown option: %s", exp_type);
106         }
107
108         if (!buf)
109                 die("git cat-file %s: bad file", obj_name);
110
111         write_or_die(1, buf, size);
112         return 0;
113 }
114
115 struct expand_data {
116         unsigned char sha1[20];
117         enum object_type type;
118         unsigned long size;
119         unsigned long disk_size;
120         const char *rest;
121         unsigned char delta_base_sha1[20];
122
123         /*
124          * If mark_query is true, we do not expand anything, but rather
125          * just mark the object_info with items we wish to query.
126          */
127         int mark_query;
128
129         /*
130          * Whether to split the input on whitespace before feeding it to
131          * get_sha1; this is decided during the mark_query phase based on
132          * whether we have a %(rest) token in our format.
133          */
134         int split_on_whitespace;
135
136         /*
137          * After a mark_query run, this object_info is set up to be
138          * passed to sha1_object_info_extended. It will point to the data
139          * elements above, so you can retrieve the response from there.
140          */
141         struct object_info info;
142 };
143
144 static int is_atom(const char *atom, const char *s, int slen)
145 {
146         int alen = strlen(atom);
147         return alen == slen && !memcmp(atom, s, alen);
148 }
149
150 static void expand_atom(struct strbuf *sb, const char *atom, int len,
151                         void *vdata)
152 {
153         struct expand_data *data = vdata;
154
155         if (is_atom("objectname", atom, len)) {
156                 if (!data->mark_query)
157                         strbuf_addstr(sb, sha1_to_hex(data->sha1));
158         } else if (is_atom("objecttype", atom, len)) {
159                 if (data->mark_query)
160                         data->info.typep = &data->type;
161                 else
162                         strbuf_addstr(sb, typename(data->type));
163         } else if (is_atom("objectsize", atom, len)) {
164                 if (data->mark_query)
165                         data->info.sizep = &data->size;
166                 else
167                         strbuf_addf(sb, "%lu", data->size);
168         } else if (is_atom("objectsize:disk", atom, len)) {
169                 if (data->mark_query)
170                         data->info.disk_sizep = &data->disk_size;
171                 else
172                         strbuf_addf(sb, "%lu", data->disk_size);
173         } else if (is_atom("rest", atom, len)) {
174                 if (data->mark_query)
175                         data->split_on_whitespace = 1;
176                 else if (data->rest)
177                         strbuf_addstr(sb, data->rest);
178         } else if (is_atom("deltabase", atom, len)) {
179                 if (data->mark_query)
180                         data->info.delta_base_sha1 = data->delta_base_sha1;
181                 else
182                         strbuf_addstr(sb, sha1_to_hex(data->delta_base_sha1));
183         } else
184                 die("unknown format element: %.*s", len, atom);
185 }
186
187 static size_t expand_format(struct strbuf *sb, const char *start, void *data)
188 {
189         const char *end;
190
191         if (*start != '(')
192                 return 0;
193         end = strchr(start + 1, ')');
194         if (!end)
195                 die("format element '%s' does not end in ')'", start);
196
197         expand_atom(sb, start + 1, end - start - 1, data);
198
199         return end - start + 1;
200 }
201
202 static void print_object_or_die(int fd, const unsigned char *sha1,
203                                 enum object_type type, unsigned long size)
204 {
205         if (type == OBJ_BLOB) {
206                 if (stream_blob_to_fd(fd, sha1, NULL, 0) < 0)
207                         die("unable to stream %s to stdout", sha1_to_hex(sha1));
208         }
209         else {
210                 enum object_type rtype;
211                 unsigned long rsize;
212                 void *contents;
213
214                 contents = read_sha1_file(sha1, &rtype, &rsize);
215                 if (!contents)
216                         die("object %s disappeared", sha1_to_hex(sha1));
217                 if (rtype != type)
218                         die("object %s changed type!?", sha1_to_hex(sha1));
219                 if (rsize != size)
220                         die("object %s change size!?", sha1_to_hex(sha1));
221
222                 write_or_die(fd, contents, size);
223                 free(contents);
224         }
225 }
226
227 struct batch_options {
228         int enabled;
229         int print_contents;
230         const char *format;
231 };
232
233 static int batch_one_object(const char *obj_name, struct batch_options *opt,
234                             struct expand_data *data)
235 {
236         struct strbuf buf = STRBUF_INIT;
237
238         if (!obj_name)
239            return 1;
240
241         if (get_sha1(obj_name, data->sha1)) {
242                 printf("%s missing\n", obj_name);
243                 fflush(stdout);
244                 return 0;
245         }
246
247         if (sha1_object_info_extended(data->sha1, &data->info) < 0) {
248                 printf("%s missing\n", obj_name);
249                 fflush(stdout);
250                 return 0;
251         }
252
253         strbuf_expand(&buf, opt->format, expand_format, data);
254         strbuf_addch(&buf, '\n');
255         write_or_die(1, buf.buf, buf.len);
256         strbuf_release(&buf);
257
258         if (opt->print_contents) {
259                 print_object_or_die(1, data->sha1, data->type, data->size);
260                 write_or_die(1, "\n", 1);
261         }
262         return 0;
263 }
264
265 static int batch_objects(struct batch_options *opt)
266 {
267         struct strbuf buf = STRBUF_INIT;
268         struct expand_data data;
269
270         if (!opt->format)
271                 opt->format = "%(objectname) %(objecttype) %(objectsize)";
272
273         /*
274          * Expand once with our special mark_query flag, which will prime the
275          * object_info to be handed to sha1_object_info_extended for each
276          * object.
277          */
278         memset(&data, 0, sizeof(data));
279         data.mark_query = 1;
280         strbuf_expand(&buf, opt->format, expand_format, &data);
281         data.mark_query = 0;
282
283         /*
284          * We are going to call get_sha1 on a potentially very large number of
285          * objects. In most large cases, these will be actual object sha1s. The
286          * cost to double-check that each one is not also a ref (just so we can
287          * warn) ends up dwarfing the actual cost of the object lookups
288          * themselves. We can work around it by just turning off the warning.
289          */
290         warn_on_object_refname_ambiguity = 0;
291
292         while (strbuf_getline(&buf, stdin, '\n') != EOF) {
293                 int error;
294
295                 if (data.split_on_whitespace) {
296                         /*
297                          * Split at first whitespace, tying off the beginning
298                          * of the string and saving the remainder (or NULL) in
299                          * data.rest.
300                          */
301                         char *p = strpbrk(buf.buf, " \t");
302                         if (p) {
303                                 while (*p && strchr(" \t", *p))
304                                         *p++ = '\0';
305                         }
306                         data.rest = p;
307                 }
308
309                 error = batch_one_object(buf.buf, opt, &data);
310                 if (error)
311                         return error;
312         }
313
314         return 0;
315 }
316
317 static const char * const cat_file_usage[] = {
318         N_("git cat-file (-t|-s|-e|-p|<type>|--textconv) <object>"),
319         N_("git cat-file (--batch|--batch-check) < <list_of_objects>"),
320         NULL
321 };
322
323 static int git_cat_file_config(const char *var, const char *value, void *cb)
324 {
325         if (userdiff_config(var, value) < 0)
326                 return -1;
327
328         return git_default_config(var, value, cb);
329 }
330
331 static int batch_option_callback(const struct option *opt,
332                                  const char *arg,
333                                  int unset)
334 {
335         struct batch_options *bo = opt->value;
336
337         if (unset) {
338                 memset(bo, 0, sizeof(*bo));
339                 return 0;
340         }
341
342         bo->enabled = 1;
343         bo->print_contents = !strcmp(opt->long_name, "batch");
344         bo->format = arg;
345
346         return 0;
347 }
348
349 int cmd_cat_file(int argc, const char **argv, const char *prefix)
350 {
351         int opt = 0;
352         const char *exp_type = NULL, *obj_name = NULL;
353         struct batch_options batch = {0};
354
355         const struct option options[] = {
356                 OPT_GROUP(N_("<type> can be one of: blob, tree, commit, tag")),
357                 OPT_SET_INT('t', NULL, &opt, N_("show object type"), 't'),
358                 OPT_SET_INT('s', NULL, &opt, N_("show object size"), 's'),
359                 OPT_SET_INT('e', NULL, &opt,
360                             N_("exit with zero when there's no error"), 'e'),
361                 OPT_SET_INT('p', NULL, &opt, N_("pretty-print object's content"), 'p'),
362                 OPT_SET_INT(0, "textconv", &opt,
363                             N_("for blob objects, run textconv on object's content"), 'c'),
364                 { OPTION_CALLBACK, 0, "batch", &batch, "format",
365                         N_("show info and content of objects fed from the standard input"),
366                         PARSE_OPT_OPTARG, batch_option_callback },
367                 { OPTION_CALLBACK, 0, "batch-check", &batch, "format",
368                         N_("show info about objects fed from the standard input"),
369                         PARSE_OPT_OPTARG, batch_option_callback },
370                 OPT_END()
371         };
372
373         git_config(git_cat_file_config, NULL);
374
375         if (argc != 3 && argc != 2)
376                 usage_with_options(cat_file_usage, options);
377
378         argc = parse_options(argc, argv, prefix, options, cat_file_usage, 0);
379
380         if (opt) {
381                 if (argc == 1)
382                         obj_name = argv[0];
383                 else
384                         usage_with_options(cat_file_usage, options);
385         }
386         if (!opt && !batch.enabled) {
387                 if (argc == 2) {
388                         exp_type = argv[0];
389                         obj_name = argv[1];
390                 } else
391                         usage_with_options(cat_file_usage, options);
392         }
393         if (batch.enabled && (opt || argc)) {
394                 usage_with_options(cat_file_usage, options);
395         }
396
397         if (batch.enabled)
398                 return batch_objects(&batch);
399
400         return cat_one_file(opt, exp_type, obj_name);
401 }