gitweb: protect blob and diff output lines from controls.
[git] / archive-tar.c
1 /*
2  * Copyright (c) 2005, 2006 Rene Scharfe
3  */
4 #include <time.h>
5 #include "cache.h"
6 #include "commit.h"
7 #include "strbuf.h"
8 #include "tar.h"
9 #include "builtin.h"
10 #include "archive.h"
11
12 #define RECORDSIZE      (512)
13 #define BLOCKSIZE       (RECORDSIZE * 20)
14
15 static char block[BLOCKSIZE];
16 static unsigned long offset;
17
18 static time_t archive_time;
19 static int tar_umask;
20 static int verbose;
21
22 /* writes out the whole block, but only if it is full */
23 static void write_if_needed(void)
24 {
25         if (offset == BLOCKSIZE) {
26                 write_or_die(1, block, BLOCKSIZE);
27                 offset = 0;
28         }
29 }
30
31 /*
32  * queues up writes, so that all our write(2) calls write exactly one
33  * full block; pads writes to RECORDSIZE
34  */
35 static void write_blocked(const void *data, unsigned long size)
36 {
37         const char *buf = data;
38         unsigned long tail;
39
40         if (offset) {
41                 unsigned long chunk = BLOCKSIZE - offset;
42                 if (size < chunk)
43                         chunk = size;
44                 memcpy(block + offset, buf, chunk);
45                 size -= chunk;
46                 offset += chunk;
47                 buf += chunk;
48                 write_if_needed();
49         }
50         while (size >= BLOCKSIZE) {
51                 write_or_die(1, buf, BLOCKSIZE);
52                 size -= BLOCKSIZE;
53                 buf += BLOCKSIZE;
54         }
55         if (size) {
56                 memcpy(block + offset, buf, size);
57                 offset += size;
58         }
59         tail = offset % RECORDSIZE;
60         if (tail)  {
61                 memset(block + offset, 0, RECORDSIZE - tail);
62                 offset += RECORDSIZE - tail;
63         }
64         write_if_needed();
65 }
66
67 /*
68  * The end of tar archives is marked by 2*512 nul bytes and after that
69  * follows the rest of the block (if any).
70  */
71 static void write_trailer(void)
72 {
73         int tail = BLOCKSIZE - offset;
74         memset(block + offset, 0, tail);
75         write_or_die(1, block, BLOCKSIZE);
76         if (tail < 2 * RECORDSIZE) {
77                 memset(block, 0, offset);
78                 write_or_die(1, block, BLOCKSIZE);
79         }
80 }
81
82 static void strbuf_append_string(struct strbuf *sb, const char *s)
83 {
84         int slen = strlen(s);
85         int total = sb->len + slen;
86         if (total > sb->alloc) {
87                 sb->buf = xrealloc(sb->buf, total);
88                 sb->alloc = total;
89         }
90         memcpy(sb->buf + sb->len, s, slen);
91         sb->len = total;
92 }
93
94 /*
95  * pax extended header records have the format "%u %s=%s\n".  %u contains
96  * the size of the whole string (including the %u), the first %s is the
97  * keyword, the second one is the value.  This function constructs such a
98  * string and appends it to a struct strbuf.
99  */
100 static void strbuf_append_ext_header(struct strbuf *sb, const char *keyword,
101                                      const char *value, unsigned int valuelen)
102 {
103         char *p;
104         int len, total, tmp;
105
106         /* "%u %s=%s\n" */
107         len = 1 + 1 + strlen(keyword) + 1 + valuelen + 1;
108         for (tmp = len; tmp > 9; tmp /= 10)
109                 len++;
110
111         total = sb->len + len;
112         if (total > sb->alloc) {
113                 sb->buf = xrealloc(sb->buf, total);
114                 sb->alloc = total;
115         }
116
117         p = sb->buf;
118         p += sprintf(p, "%u %s=", len, keyword);
119         memcpy(p, value, valuelen);
120         p += valuelen;
121         *p = '\n';
122         sb->len = total;
123 }
124
125 static unsigned int ustar_header_chksum(const struct ustar_header *header)
126 {
127         char *p = (char *)header;
128         unsigned int chksum = 0;
129         while (p < header->chksum)
130                 chksum += *p++;
131         chksum += sizeof(header->chksum) * ' ';
132         p += sizeof(header->chksum);
133         while (p < (char *)header + sizeof(struct ustar_header))
134                 chksum += *p++;
135         return chksum;
136 }
137
138 static int get_path_prefix(const struct strbuf *path, int maxlen)
139 {
140         int i = path->len;
141         if (i > maxlen)
142                 i = maxlen;
143         do {
144                 i--;
145         } while (i > 0 && path->buf[i] != '/');
146         return i;
147 }
148
149 static void write_entry(const unsigned char *sha1, struct strbuf *path,
150                         unsigned int mode, void *buffer, unsigned long size)
151 {
152         struct ustar_header header;
153         struct strbuf ext_header;
154
155         memset(&header, 0, sizeof(header));
156         ext_header.buf = NULL;
157         ext_header.len = ext_header.alloc = 0;
158
159         if (!sha1) {
160                 *header.typeflag = TYPEFLAG_GLOBAL_HEADER;
161                 mode = 0100666;
162                 strcpy(header.name, "pax_global_header");
163         } else if (!path) {
164                 *header.typeflag = TYPEFLAG_EXT_HEADER;
165                 mode = 0100666;
166                 sprintf(header.name, "%s.paxheader", sha1_to_hex(sha1));
167         } else {
168                 if (verbose)
169                         fprintf(stderr, "%.*s\n", path->len, path->buf);
170                 if (S_ISDIR(mode)) {
171                         *header.typeflag = TYPEFLAG_DIR;
172                         mode = (mode | 0777) & ~tar_umask;
173                 } else if (S_ISLNK(mode)) {
174                         *header.typeflag = TYPEFLAG_LNK;
175                         mode |= 0777;
176                 } else if (S_ISREG(mode)) {
177                         *header.typeflag = TYPEFLAG_REG;
178                         mode = (mode | ((mode & 0100) ? 0777 : 0666)) & ~tar_umask;
179                 } else {
180                         error("unsupported file mode: 0%o (SHA1: %s)",
181                               mode, sha1_to_hex(sha1));
182                         return;
183                 }
184                 if (path->len > sizeof(header.name)) {
185                         int plen = get_path_prefix(path, sizeof(header.prefix));
186                         int rest = path->len - plen - 1;
187                         if (plen > 0 && rest <= sizeof(header.name)) {
188                                 memcpy(header.prefix, path->buf, plen);
189                                 memcpy(header.name, path->buf + plen + 1, rest);
190                         } else {
191                                 sprintf(header.name, "%s.data",
192                                         sha1_to_hex(sha1));
193                                 strbuf_append_ext_header(&ext_header, "path",
194                                                          path->buf, path->len);
195                         }
196                 } else
197                         memcpy(header.name, path->buf, path->len);
198         }
199
200         if (S_ISLNK(mode) && buffer) {
201                 if (size > sizeof(header.linkname)) {
202                         sprintf(header.linkname, "see %s.paxheader",
203                                 sha1_to_hex(sha1));
204                         strbuf_append_ext_header(&ext_header, "linkpath",
205                                                  buffer, size);
206                 } else
207                         memcpy(header.linkname, buffer, size);
208         }
209
210         sprintf(header.mode, "%07o", mode & 07777);
211         sprintf(header.size, "%011lo", S_ISREG(mode) ? size : 0);
212         sprintf(header.mtime, "%011lo", archive_time);
213
214         /* XXX: should we provide more meaningful info here? */
215         sprintf(header.uid, "%07o", 0);
216         sprintf(header.gid, "%07o", 0);
217         strlcpy(header.uname, "git", sizeof(header.uname));
218         strlcpy(header.gname, "git", sizeof(header.gname));
219         sprintf(header.devmajor, "%07o", 0);
220         sprintf(header.devminor, "%07o", 0);
221
222         memcpy(header.magic, "ustar", 6);
223         memcpy(header.version, "00", 2);
224
225         sprintf(header.chksum, "%07o", ustar_header_chksum(&header));
226
227         if (ext_header.len > 0) {
228                 write_entry(sha1, NULL, 0, ext_header.buf, ext_header.len);
229                 free(ext_header.buf);
230         }
231         write_blocked(&header, sizeof(header));
232         if (S_ISREG(mode) && buffer && size > 0)
233                 write_blocked(buffer, size);
234 }
235
236 static void write_global_extended_header(const unsigned char *sha1)
237 {
238         struct strbuf ext_header;
239         ext_header.buf = NULL;
240         ext_header.len = ext_header.alloc = 0;
241         strbuf_append_ext_header(&ext_header, "comment", sha1_to_hex(sha1), 40);
242         write_entry(NULL, NULL, 0, ext_header.buf, ext_header.len);
243         free(ext_header.buf);
244 }
245
246 static int git_tar_config(const char *var, const char *value)
247 {
248         if (!strcmp(var, "tar.umask")) {
249                 if (!strcmp(value, "user")) {
250                         tar_umask = umask(0);
251                         umask(tar_umask);
252                 } else {
253                         tar_umask = git_config_int(var, value);
254                 }
255                 return 0;
256         }
257         return git_default_config(var, value);
258 }
259
260 static int write_tar_entry(const unsigned char *sha1,
261                            const char *base, int baselen,
262                            const char *filename, unsigned mode, int stage)
263 {
264         static struct strbuf path;
265         int filenamelen = strlen(filename);
266         void *buffer;
267         char type[20];
268         unsigned long size;
269
270         if (!path.alloc) {
271                 path.buf = xmalloc(PATH_MAX);
272                 path.alloc = PATH_MAX;
273                 path.len = path.eof = 0;
274         }
275         if (path.alloc < baselen + filenamelen) {
276                 free(path.buf);
277                 path.buf = xmalloc(baselen + filenamelen);
278                 path.alloc = baselen + filenamelen;
279         }
280         memcpy(path.buf, base, baselen);
281         memcpy(path.buf + baselen, filename, filenamelen);
282         path.len = baselen + filenamelen;
283         if (S_ISDIR(mode)) {
284                 strbuf_append_string(&path, "/");
285                 buffer = NULL;
286                 size = 0;
287         } else {
288                 buffer = read_sha1_file(sha1, type, &size);
289                 if (!buffer)
290                         die("cannot read %s", sha1_to_hex(sha1));
291         }
292
293         write_entry(sha1, &path, mode, buffer, size);
294         free(buffer);
295
296         return READ_TREE_RECURSIVE;
297 }
298
299 int write_tar_archive(struct archiver_args *args)
300 {
301         int plen = args->base ? strlen(args->base) : 0;
302
303         git_config(git_tar_config);
304
305         archive_time = args->time;
306         verbose = args->verbose;
307
308         if (args->commit_sha1)
309                 write_global_extended_header(args->commit_sha1);
310
311         if (args->base && plen > 0 && args->base[plen - 1] == '/') {
312                 char *base = xstrdup(args->base);
313                 int baselen = strlen(base);
314
315                 while (baselen > 0 && base[baselen - 1] == '/')
316                         base[--baselen] = '\0';
317                 write_tar_entry(args->tree->object.sha1, "", 0, base, 040777, 0);
318                 free(base);
319         }
320         read_tree_recursive(args->tree, args->base, plen, 0,
321                             args->pathspec, write_tar_entry);
322         write_trailer();
323
324         return 0;
325 }