convert-objects.c

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