convert-cache.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 = malloc(sizeof(struct entry));
  22
  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("bad sha1");
  65         entry = convert_entry(sha1);
  66         memcpy(buffer, sha1_to_hex(entry->new_sha1), 40);
  67 }
  68
  69 #define ORIG_OFFSET (40)
  70
  71 static int prepend_integer(char *buffer, unsigned val, int i)
  72 {
  73         buffer[--i] = '\0';
  74         do {
  75                 buffer[--i] = '0' + (val % 10);
  76                 val /= 10;
  77         } while (val);
  78         return i;
  79 }
  80
  81
  82 static int write_subdirectory(void *buffer, unsigned long size, const char *base, int baselen, unsigned char *result_sha1)
  83 {
  84         char *new = malloc(size + ORIG_OFFSET);
  85         unsigned long newlen = ORIG_OFFSET;
  86         unsigned long used;
  87         int i;
  88
  89         used = 0;
  90         while (size) {
  91                 int len = 21 + strlen(buffer);
  92                 char *path = strchr(buffer, ' ');
  93                 unsigned char *sha1;
  94                 unsigned int mode;
  95                 char *slash, *origpath;
  96
  97                 if (!path || sscanf(buffer, "%o", &mode) != 1)
  98                         die("bad tree conversion");
  99                 path++;
 100                 if (memcmp(path, base, baselen))
 101                         break;
 102                 origpath = path;
 103                 path += baselen;
 104                 slash = strchr(path, '/');
 105                 if (!slash) {
 106                         newlen += sprintf(new + newlen, "%o %s", mode, path);
 107                         new[newlen++] = '\0';
 108                         memcpy(new + newlen, buffer + len - 20, 20);
 109                         newlen += 20;
 110
 111                         used += len;
 112                         size -= len;
 113                         buffer += len;
 114                         continue;
 115                 }
 116
 117                 newlen += sprintf(new + newlen, "%o %.*s", S_IFDIR, slash - path, path);
 118                 new[newlen++] = 0;
 119                 sha1 = (unsigned char *)(new + newlen);
 120                 newlen += 20;
 121
 122                 len = write_subdirectory(buffer, size, origpath, slash-origpath+1, sha1);
 123
 124                 used += len;
 125                 size -= len;
 126                 buffer += len;
 127         }
 128
 129         i = prepend_integer(new, newlen - ORIG_OFFSET, ORIG_OFFSET);
 130         i -= 5;
 131         memcpy(new + i, "tree ", 5);
 132
 133         write_sha1_file(new + i, newlen - i, result_sha1);
 134         free(new);
 135         return used;
 136 }
 137
 138 static void convert_tree(void *buffer, unsigned long size, unsigned char *result_sha1)
 139 {
 140         void *orig_buffer = buffer;
 141         unsigned long orig_size = size;
 142
 143         while (size) {
 144                 int len = 1+strlen(buffer);
 145
 146                 convert_binary_sha1(buffer + len);
 147
 148                 len += 20;
 149                 if (len > size)
 150                         die("corrupt tree object");
 151                 size -= len;
 152                 buffer += len;
 153         }
 154
 155         write_subdirectory(orig_buffer, orig_size, "", 0, result_sha1);
 156 }
 157
 158 static unsigned long parse_oldstyle_date(const char *buf)
 159 {
 160         char c, *p;
 161         char buffer[100];
 162         struct tm tm;
 163         const char *formats[] = {
 164                 "%c",
 165                 "%a %b %d %T",
 166                 "%Z",
 167                 "%Y",
 168                 " %Y",
 169                 NULL
 170         };
 171         /* We only ever did two timezones in the bad old format .. */
 172         const char *timezones[] = {
 173                 "PDT", "PST", NULL
 174         };
 175         const char **fmt = formats;
 176
 177         p = buffer;
 178         while (isspace(c = *buf))
 179                 buf++;
 180         while ((c = *buf++) != '\n')
 181                 *p++ = c;
 182         *p++ = 0;
 183         buf = buffer;
 184         memset(&tm, 0, sizeof(tm));
 185         do {
 186                 const char *next = strptime(buf, *fmt, &tm);
 187                 if (next) {
 188                         if (!*next)
 189                                 return mktime(&tm);
 190                         buf = next;
 191                 } else {
 192                         const char **p = timezones;
 193                         while (isspace(*buf))
 194                                 buf++;
 195                         while (*p) {
 196                                 if (!memcmp(buf, *p, strlen(*p))) {
 197                                         buf += strlen(*p);
 198                                         break;
 199                                 }
 200                                 p++;
 201                         }
 202                 }
 203                 fmt++;
 204         } while (*buf && *fmt);
 205         printf("left: %s\n", buf);
 206         return mktime(&tm);
 207 }
 208
 209 static int convert_date_line(char *dst, void **buf, unsigned long *sp)
 210 {
 211         unsigned long size = *sp;
 212         char *line = *buf;
 213         char *next = strchr(line, '\n');
 214         char *date = strchr(line, '>');
 215         int len;
 216
 217         if (!next || !date)
 218                 die("missing or bad author/committer line %s", line);
 219         next++; date += 2;
 220
 221         *buf = next;
 222         *sp = size - (next - line);
 223
 224         len = date - line;
 225         memcpy(dst, line, len);
 226         dst += len;
 227
 228         /* Is it already in new format? */
 229         if (isdigit(*date)) {
 230                 int datelen = next - date;
 231                 memcpy(dst, date, datelen);
 232                 printf("new format date '%.*s'?\n", datelen-1, date);
 233                 return len + datelen;
 234         }
 235
 236         /*
 237          * Hacky hacky: one of the sparse old-style commits does not have
 238          * any date at all, but we can fake it by using the committer date.
 239          */
 240         if (*date == '\n' && strchr(next, '>'))
 241                 date = strchr(next, '>')+2;
 242
 243         return len + sprintf(dst, "%lu -0700\n", parse_oldstyle_date(date));
 244 }
 245
 246 static void convert_date(void *buffer, unsigned long size, unsigned char *result_sha1)
 247 {
 248         char *new = malloc(size + ORIG_OFFSET + 100);
 249         unsigned long newlen = ORIG_OFFSET;
 250         int i;
 251
 252         // "tree <sha1>\n"
 253         memcpy(new + newlen, buffer, 46);
 254         newlen += 46;
 255         buffer += 46;
 256         size -= 46;
 257
 258         // "parent <sha1>\n"
 259         while (!memcmp(buffer, "parent ", 7)) {
 260                 memcpy(new + newlen, buffer, 48);
 261                 newlen += 48;
 262                 buffer += 48;
 263                 size -= 48;
 264         }
 265
 266         // "author xyz <xyz> date"
 267         newlen += convert_date_line(new + newlen, &buffer, &size);
 268         // "committer xyz <xyz> date"
 269         newlen += convert_date_line(new + newlen, &buffer, &size);
 270
 271         // Rest
 272         memcpy(new + newlen, buffer, size);
 273         newlen += size;
 274
 275         i = prepend_integer(new, newlen - ORIG_OFFSET, ORIG_OFFSET);
 276         i -= 7;
 277         memcpy(new + i, "commit ", 7);
 278
 279         write_sha1_file(new + i, newlen - i, result_sha1);
 280         free(new);
 281 }
 282
 283 static void convert_commit(void *buffer, unsigned long size, unsigned char *result_sha1)
 284 {
 285         void *orig_buffer = buffer;
 286         unsigned long orig_size = size;
 287
 288         convert_ascii_sha1(buffer+5);
 289         buffer += 46;    /* "tree " + "hex sha1" + "\n" */
 290         while (!memcmp(buffer, "parent ", 7)) {
 291                 convert_ascii_sha1(buffer+7);
 292                 buffer += 48;
 293         }
 294         convert_date(orig_buffer, orig_size, result_sha1);
 295 }
 296
 297 static struct entry * convert_entry(unsigned char *sha1)
 298 {
 299         struct entry *entry = lookup_entry(sha1);
 300         char type[20];
 301         void *buffer, *data;
 302         unsigned long size, offset;
 303
 304         if (entry->converted)
 305                 return entry;
 306         data = read_sha1_file(sha1, type, &size);
 307         if (!data)
 308                 die("unable to read object %s", sha1_to_hex(sha1));
 309
 310         buffer = malloc(size + 100);
 311         offset = sprintf(buffer, "%s %lu", type, size)+1;
 312         memcpy(buffer + offset, data, size);
 313
 314         if (!strcmp(type, "blob")) {
 315                 write_sha1_file(buffer, size + offset, entry->new_sha1);
 316         } else if (!strcmp(type, "tree"))
 317                 convert_tree(buffer + offset, size, entry->new_sha1);
 318         else if (!strcmp(type, "commit"))
 319                 convert_commit(buffer + offset, size, entry->new_sha1);
 320         else
 321                 die("unknown object type '%s' in %s", type, sha1_to_hex(sha1));
 322         entry->converted = 1;
 323         free(buffer);
 324         return entry;
 325 }
 326
 327 int main(int argc, char **argv)
 328 {
 329         unsigned char sha1[20];
 330         struct entry *entry;
 331
 332         if (argc != 2 || get_sha1_hex(argv[1], sha1))
 333                 usage("convert-cache <sha1>");
 334
 335         entry = convert_entry(sha1);
 336         printf("new sha1: %s\n", sha1_to_hex(entry->new_sha1));
 337         return 0;
 338 }