convert-objects.c

   1 #define _XOPEN_SOURCE /* glibc2 needs this */
   2 #define __EXTENSIONS__ /* solaris needs this */
   3 #include <time.h>
   4 #include <ctype.h>
   5 #include "cache.h"
   6
   7 struct entry {
   8         unsigned char old_sha1[20];
   9         unsigned char new_sha1[20];
  10         int converted;
  11 };
  12
  13 #define MAXOBJECTS (1000000)
  14
  15 static struct entry *convert[MAXOBJECTS];
  16 static int nr_convert;
  17
  18 static struct entry * convert_entry(unsigned char *sha1);
  19
  20 static struct entry *insert_new(unsigned char *sha1, int pos)
  21 {
  22         struct entry *new = xmalloc(sizeof(struct entry));
  23         memset(new, 0, sizeof(*new));
  24         memcpy(new->old_sha1, sha1, 20);
  25         memmove(convert + pos + 1, convert + pos, (nr_convert - pos) * sizeof(struct entry *));
  26         convert[pos] = new;
  27         nr_convert++;
  28         if (nr_convert == MAXOBJECTS)
  29                 die("you're kidding me - hit maximum object limit");
  30         return new;
  31 }
  32
  33 static struct entry *lookup_entry(unsigned char *sha1)
  34 {
  35         int low = 0, high = nr_convert;
  36
  37         while (low < high) {
  38                 int next = (low + high) / 2;
  39                 struct entry *n = convert[next];
  40                 int cmp = memcmp(sha1, n->old_sha1, 20);
  41                 if (!cmp)
  42                         return n;
  43                 if (cmp < 0) {
  44                         high = next;
  45                         continue;
  46                 }
  47                 low = next+1;
  48         }
  49         return insert_new(sha1, low);
  50 }
  51
  52 static void convert_binary_sha1(void *buffer)
  53 {
  54         struct entry *entry = convert_entry(buffer);
  55         memcpy(buffer, entry->new_sha1, 20);
  56 }
  57
  58 static void convert_ascii_sha1(void *buffer)
  59 {
  60         unsigned char sha1[20];
  61         struct entry *entry;
  62
  63         if (get_sha1_hex(buffer, sha1))
  64                 die("expected sha1, got '%s'", (char*) buffer);
  65         entry = convert_entry(sha1);
  66         memcpy(buffer, sha1_to_hex(entry->new_sha1), 40);
  67 }
  68
  69 static unsigned int convert_mode(unsigned int mode)
  70 {
  71         unsigned int newmode;
  72
  73         newmode = mode & S_IFMT;
  74         if (S_ISREG(mode))
  75                 newmode |= (mode & 0100) ? 0755 : 0644;
  76         return newmode;
  77 }
  78
  79 static int write_subdirectory(void *buffer, unsigned long size, const char *base, int baselen, unsigned char *result_sha1)
  80 {
  81         char *new = xmalloc(size);
  82         unsigned long newlen = 0;
  83         unsigned long used;
  84
  85         used = 0;
  86         while (size) {
  87                 int len = 21 + strlen(buffer);
  88                 char *path = strchr(buffer, ' ');
  89                 unsigned char *sha1;
  90                 unsigned int mode;
  91                 char *slash, *origpath;
  92
  93                 if (!path || sscanf(buffer, "%o", &mode) != 1)
  94                         die("bad tree conversion");
  95                 mode = convert_mode(mode);
  96                 path++;
  97                 if (memcmp(path, base, baselen))
  98                         break;
  99                 origpath = path;
 100                 path += baselen;
 101                 slash = strchr(path, '/');
 102                 if (!slash) {
 103                         newlen += sprintf(new + newlen, "%o %s", mode, path);
 104                         new[newlen++] = '\0';
 105                         memcpy(new + newlen, buffer + len - 20, 20);
 106                         newlen += 20;
 107
 108                         used += len;
 109                         size -= len;
 110                         buffer += len;
 111                         continue;
 112                 }
 113
 114                 newlen += sprintf(new + newlen, "%o %.*s", S_IFDIR, (int)(slash - path), path);
 115                 new[newlen++] = 0;
 116                 sha1 = (unsigned char *)(new + newlen);
 117                 newlen += 20;
 118
 119                 len = write_subdirectory(buffer, size, origpath, slash-origpath+1, sha1);
 120
 121                 used += len;
 122                 size -= len;
 123                 buffer += len;
 124         }
 125
 126         write_sha1_file(new, newlen, "tree", result_sha1);
 127         free(new);
 128         return used;
 129 }
 130
 131 static void convert_tree(void *buffer, unsigned long size, unsigned char *result_sha1)
 132 {
 133         void *orig_buffer = buffer;
 134         unsigned long orig_size = size;
 135
 136         while (size) {
 137                 int len = 1+strlen(buffer);
 138
 139                 convert_binary_sha1(buffer + len);
 140
 141                 len += 20;
 142                 if (len > size)
 143                         die("corrupt tree object");
 144                 size -= len;
 145                 buffer += len;
 146         }
 147
 148         write_subdirectory(orig_buffer, orig_size, "", 0, result_sha1);
 149 }
 150
 151 static unsigned long parse_oldstyle_date(const char *buf)
 152 {
 153         char c, *p;
 154         char buffer[100];
 155         struct tm tm;
 156         const char *formats[] = {
 157                 "%c",
 158                 "%a %b %d %T",
 159                 "%Z",
 160                 "%Y",
 161                 " %Y",
 162                 NULL
 163         };
 164         /* We only ever did two timezones in the bad old format .. */
 165         const char *timezones[] = {
 166                 "PDT", "PST", "CEST", NULL
 167         };
 168         const char **fmt = formats;
 169
 170         p = buffer;
 171         while (isspace(c = *buf))
 172                 buf++;
 173         while ((c = *buf++) != '\n')
 174                 *p++ = c;
 175         *p++ = 0;
 176         buf = buffer;
 177         memset(&tm, 0, sizeof(tm));
 178         do {
 179                 const char *next = strptime(buf, *fmt, &tm);
 180                 if (next) {
 181                         if (!*next)
 182                                 return mktime(&tm);
 183                         buf = next;
 184                 } else {
 185                         const char **p = timezones;
 186                         while (isspace(*buf))
 187                                 buf++;
 188                         while (*p) {
 189                                 if (!memcmp(buf, *p, strlen(*p))) {
 190                                         buf += strlen(*p);
 191                                         break;
 192                                 }
 193                                 p++;
 194                         }
 195                 }
 196                 fmt++;
 197         } while (*buf && *fmt);
 198         printf("left: %s\n", buf);
 199         return mktime(&tm);
 200 }
 201
 202 static int convert_date_line(char *dst, void **buf, unsigned long *sp)
 203 {
 204         unsigned long size = *sp;
 205         char *line = *buf;
 206         char *next = strchr(line, '\n');
 207         char *date = strchr(line, '>');
 208         int len;
 209
 210         if (!next || !date)
 211                 die("missing or bad author/committer line %s", line);
 212         next++; date += 2;
 213
 214         *buf = next;
 215         *sp = size - (next - line);
 216
 217         len = date - line;
 218         memcpy(dst, line, len);
 219         dst += len;
 220
 221         /* Is it already in new format? */
 222         if (isdigit(*date)) {
 223                 int datelen = next - date;
 224                 memcpy(dst, date, datelen);
 225                 return len + datelen;
 226         }
 227
 228         /*
 229          * Hacky hacky: one of the sparse old-style commits does not have
 230          * any date at all, but we can fake it by using the committer date.
 231          */
 232         if (*date == '\n' && strchr(next, '>'))
 233                 date = strchr(next, '>')+2;
 234
 235         return len + sprintf(dst, "%lu -0700\n", parse_oldstyle_date(date));
 236 }
 237
 238 static void convert_date(void *buffer, unsigned long size, unsigned char *result_sha1)
 239 {
 240         char *new = xmalloc(size + 100);
 241         unsigned long newlen = 0;
 242
 243         // "tree <sha1>\n"
 244         memcpy(new + newlen, buffer, 46);
 245         newlen += 46;
 246         buffer += 46;
 247         size -= 46;
 248
 249         // "parent <sha1>\n"
 250         while (!memcmp(buffer, "parent ", 7)) {
 251                 memcpy(new + newlen, buffer, 48);
 252                 newlen += 48;
 253                 buffer += 48;
 254                 size -= 48;
 255         }
 256
 257         // "author xyz <xyz> date"
 258         newlen += convert_date_line(new + newlen, &buffer, &size);
 259         // "committer xyz <xyz> date"
 260         newlen += convert_date_line(new + newlen, &buffer, &size);
 261
 262         // Rest
 263         memcpy(new + newlen, buffer, size);
 264         newlen += size;
 265
 266         write_sha1_file(new, newlen, "commit", result_sha1);
 267         free(new);
 268 }
 269
 270 static void convert_commit(void *buffer, unsigned long size, unsigned char *result_sha1)
 271 {
 272         void *orig_buffer = buffer;
 273         unsigned long orig_size = size;
 274
 275         if (memcmp(buffer, "tree ", 5))
 276                 die("Bad commit '%s'", (char*) buffer);
 277         convert_ascii_sha1(buffer+5);
 278         buffer += 46;    /* "tree " + "hex sha1" + "\n" */
 279         while (!memcmp(buffer, "parent ", 7)) {
 280                 convert_ascii_sha1(buffer+7);
 281                 buffer += 48;
 282         }
 283         convert_date(orig_buffer, orig_size, result_sha1);
 284 }
 285
 286 static struct entry * convert_entry(unsigned char *sha1)
 287 {
 288         struct entry *entry = lookup_entry(sha1);
 289         char type[20];
 290         void *buffer, *data;
 291         unsigned long size;
 292
 293         if (entry->converted)
 294                 return entry;
 295         data = read_sha1_file(sha1, type, &size);
 296         if (!data)
 297                 die("unable to read object %s", sha1_to_hex(sha1));
 298
 299         buffer = xmalloc(size);
 300         memcpy(buffer, data, size);
 301
 302         if (!strcmp(type, "blob")) {
 303                 write_sha1_file(buffer, size, "blob", entry->new_sha1);
 304         } else if (!strcmp(type, "tree"))
 305                 convert_tree(buffer, size, entry->new_sha1);
 306         else if (!strcmp(type, "commit"))
 307                 convert_commit(buffer, size, entry->new_sha1);
 308         else
 309                 die("unknown object type '%s' in %s", type, sha1_to_hex(sha1));
 310         entry->converted = 1;
 311         free(buffer);
 312         free(data);
 313         return entry;
 314 }
 315
 316 int main(int argc, char **argv)
 317 {
 318         unsigned char sha1[20];
 319         struct entry *entry;
 320
 321         if (argc != 2 || get_sha1(argv[1], sha1))
 322                 usage("git-convert-objects <sha1>");
 323
 324         entry = convert_entry(sha1);
 325         printf("new sha1: %s\n", sha1_to_hex(entry->new_sha1));
 326         return 0;
 327 }