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