convert-objects.c

   1 #define _XOPEN_SOURCE 500 /* glibc2 and AIX 5.3L need this */
   2 #define _XOPEN_SOURCE_EXTENDED 1 /* AIX 5.3L needs this */
   3 #include <time.h>
   4 #include "cache.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tree.h"
   8
   9 struct entry {
  10         unsigned char old_sha1[20];
  11         unsigned char new_sha1[20];
  12         int converted;
  13 };
  14
  15 #define MAXOBJECTS (1000000)
  16
  17 static struct entry *convert[MAXOBJECTS];
  18 static int nr_convert;
  19
  20 static struct entry * convert_entry(unsigned char *sha1);
  21
  22 static struct entry *insert_new(unsigned char *sha1, int pos)
  23 {
  24         struct entry *new = xcalloc(1, sizeof(struct entry));
  25         memcpy(new->old_sha1, sha1, 20);
  26         memmove(convert + pos + 1, convert + pos, (nr_convert - pos) * sizeof(struct entry *));
  27         convert[pos] = new;
  28         nr_convert++;
  29         if (nr_convert == MAXOBJECTS)
  30                 die("you're kidding me - hit maximum object limit");
  31         return new;
  32 }
  33
  34 static struct entry *lookup_entry(unsigned char *sha1)
  35 {
  36         int low = 0, high = nr_convert;
  37
  38         while (low < high) {
  39                 int next = (low + high) / 2;
  40                 struct entry *n = convert[next];
  41                 int cmp = memcmp(sha1, n->old_sha1, 20);
  42                 if (!cmp)
  43                         return n;
  44                 if (cmp < 0) {
  45                         high = next;
  46                         continue;
  47                 }
  48                 low = next+1;
  49         }
  50         return insert_new(sha1, low);
  51 }
  52
  53 static void convert_binary_sha1(void *buffer)
  54 {
  55         struct entry *entry = convert_entry(buffer);
  56         memcpy(buffer, entry->new_sha1, 20);
  57 }
  58
  59 static void convert_ascii_sha1(void *buffer)
  60 {
  61         unsigned char sha1[20];
  62         struct entry *entry;
  63
  64         if (get_sha1_hex(buffer, sha1))
  65                 die("expected sha1, got '%s'", (char*) buffer);
  66         entry = convert_entry(sha1);
  67         memcpy(buffer, sha1_to_hex(entry->new_sha1), 40);
  68 }
  69
  70 static unsigned int convert_mode(unsigned int mode)
  71 {
  72         unsigned int newmode;
  73
  74         newmode = mode & S_IFMT;
  75         if (S_ISREG(mode))
  76                 newmode |= (mode & 0100) ? 0755 : 0644;
  77         return newmode;
  78 }
  79
  80 static int write_subdirectory(void *buffer, unsigned long size, const char *base, int baselen, unsigned char *result_sha1)
  81 {
  82         char *new = xmalloc(size);
  83         unsigned long newlen = 0;
  84         unsigned long used;
  85
  86         used = 0;
  87         while (size) {
  88                 int len = 21 + strlen(buffer);
  89                 char *path = strchr(buffer, ' ');
  90                 unsigned char *sha1;
  91                 unsigned int mode;
  92                 char *slash, *origpath;
  93
  94                 if (!path || sscanf(buffer, "%o", &mode) != 1)
  95                         die("bad tree conversion");
  96                 mode = convert_mode(mode);
  97                 path++;
  98                 if (memcmp(path, base, baselen))
  99                         break;
 100                 origpath = path;
 101                 path += baselen;
 102                 slash = strchr(path, '/');
 103                 if (!slash) {
 104                         newlen += sprintf(new + newlen, "%o %s", mode, path);
 105                         new[newlen++] = '\0';
 106                         memcpy(new + newlen, buffer + len - 20, 20);
 107                         newlen += 20;
 108
 109                         used += len;
 110                         size -= len;
 111                         buffer += len;
 112                         continue;
 113                 }
 114
 115                 newlen += sprintf(new + newlen, "%o %.*s", S_IFDIR, (int)(slash - path), path);
 116                 new[newlen++] = 0;
 117                 sha1 = (unsigned char *)(new + newlen);
 118                 newlen += 20;
 119
 120                 len = write_subdirectory(buffer, size, origpath, slash-origpath+1, sha1);
 121
 122                 used += len;
 123                 size -= len;
 124                 buffer += len;
 125         }
 126
 127         write_sha1_file(new, newlen, tree_type, result_sha1);
 128         free(new);
 129         return used;
 130 }
 131
 132 static void convert_tree(void *buffer, unsigned long size, unsigned char *result_sha1)
 133 {
 134         void *orig_buffer = buffer;
 135         unsigned long orig_size = size;
 136
 137         while (size) {
 138                 int len = 1+strlen(buffer);
 139
 140                 convert_binary_sha1(buffer + len);
 141
 142                 len += 20;
 143                 if (len > size)
 144                         die("corrupt tree object");
 145                 size -= len;
 146                 buffer += len;
 147         }
 148
 149         write_subdirectory(orig_buffer, orig_size, "", 0, result_sha1);
 150 }
 151
 152 static unsigned long parse_oldstyle_date(const char *buf)
 153 {
 154         char c, *p;
 155         char buffer[100];
 156         struct tm tm;
 157         const char *formats[] = {
 158                 "%c",
 159                 "%a %b %d %T",
 160                 "%Z",
 161                 "%Y",
 162                 " %Y",
 163                 NULL
 164         };
 165         /* We only ever did two timezones in the bad old format .. */
 166         const char *timezones[] = {
 167                 "PDT", "PST", "CEST", NULL
 168         };
 169         const char **fmt = formats;
 170
 171         p = buffer;
 172         while (isspace(c = *buf))
 173                 buf++;
 174         while ((c = *buf++) != '\n')
 175                 *p++ = c;
 176         *p++ = 0;
 177         buf = buffer;
 178         memset(&tm, 0, sizeof(tm));
 179         do {
 180                 const char *next = strptime(buf, *fmt, &tm);
 181                 if (next) {
 182                         if (!*next)
 183                                 return mktime(&tm);
 184                         buf = next;
 185                 } else {
 186                         const char **p = timezones;
 187                         while (isspace(*buf))
 188                                 buf++;
 189                         while (*p) {
 190                                 if (!memcmp(buf, *p, strlen(*p))) {
 191                                         buf += strlen(*p);
 192                                         break;
 193                                 }
 194                                 p++;
 195                         }
 196                 }
 197                 fmt++;
 198         } while (*buf && *fmt);
 199         printf("left: %s\n", buf);
 200         return mktime(&tm);
 201 }
 202
 203 static int convert_date_line(char *dst, void **buf, unsigned long *sp)
 204 {
 205         unsigned long size = *sp;
 206         char *line = *buf;
 207         char *next = strchr(line, '\n');
 208         char *date = strchr(line, '>');
 209         int len;
 210
 211         if (!next || !date)
 212                 die("missing or bad author/committer line %s", line);
 213         next++; date += 2;
 214
 215         *buf = next;
 216         *sp = size - (next - line);
 217
 218         len = date - line;
 219         memcpy(dst, line, len);
 220         dst += len;
 221
 222         /* Is it already in new format? */
 223         if (isdigit(*date)) {
 224                 int datelen = next - date;
 225                 memcpy(dst, date, datelen);
 226                 return len + datelen;
 227         }
 228
 229         /*
 230          * Hacky hacky: one of the sparse old-style commits does not have
 231          * any date at all, but we can fake it by using the committer date.
 232          */
 233         if (*date == '\n' && strchr(next, '>'))
 234                 date = strchr(next, '>')+2;
 235
 236         return len + sprintf(dst, "%lu -0700\n", parse_oldstyle_date(date));
 237 }
 238
 239 static void convert_date(void *buffer, unsigned long size, unsigned char *result_sha1)
 240 {
 241         char *new = xmalloc(size + 100);
 242         unsigned long newlen = 0;
 243
 244         // "tree <sha1>\n"
 245         memcpy(new + newlen, buffer, 46);
 246         newlen += 46;
 247         buffer += 46;
 248         size -= 46;
 249
 250         // "parent <sha1>\n"
 251         while (!memcmp(buffer, "parent ", 7)) {
 252                 memcpy(new + newlen, buffer, 48);
 253                 newlen += 48;
 254                 buffer += 48;
 255                 size -= 48;
 256         }
 257
 258         // "author xyz <xyz> date"
 259         newlen += convert_date_line(new + newlen, &buffer, &size);
 260         // "committer xyz <xyz> date"
 261         newlen += convert_date_line(new + newlen, &buffer, &size);
 262
 263         // Rest
 264         memcpy(new + newlen, buffer, size);
 265         newlen += size;
 266
 267         write_sha1_file(new, newlen, commit_type, result_sha1);
 268         free(new);
 269 }
 270
 271 static void convert_commit(void *buffer, unsigned long size, unsigned char *result_sha1)
 272 {
 273         void *orig_buffer = buffer;
 274         unsigned long orig_size = size;
 275
 276         if (memcmp(buffer, "tree ", 5))
 277                 die("Bad commit '%s'", (char*) buffer);
 278         convert_ascii_sha1(buffer+5);
 279         buffer += 46;    /* "tree " + "hex sha1" + "\n" */
 280         while (!memcmp(buffer, "parent ", 7)) {
 281                 convert_ascii_sha1(buffer+7);
 282                 buffer += 48;
 283         }
 284         convert_date(orig_buffer, orig_size, result_sha1);
 285 }
 286
 287 static struct entry * convert_entry(unsigned char *sha1)
 288 {
 289         struct entry *entry = lookup_entry(sha1);
 290         char type[20];
 291         void *buffer, *data;
 292         unsigned long size;
 293
 294         if (entry->converted)
 295                 return entry;
 296         data = read_sha1_file(sha1, type, &size);
 297         if (!data)
 298                 die("unable to read object %s", sha1_to_hex(sha1));
 299
 300         buffer = xmalloc(size);
 301         memcpy(buffer, data, size);
 302
 303         if (!strcmp(type, blob_type)) {
 304                 write_sha1_file(buffer, size, blob_type, entry->new_sha1);
 305         } else if (!strcmp(type, tree_type))
 306                 convert_tree(buffer, size, entry->new_sha1);
 307         else if (!strcmp(type, commit_type))
 308                 convert_commit(buffer, size, entry->new_sha1);
 309         else
 310                 die("unknown object type '%s' in %s", type, sha1_to_hex(sha1));
 311         entry->converted = 1;
 312         free(buffer);
 313         free(data);
 314         return entry;
 315 }
 316
 317 int main(int argc, char **argv)
 318 {
 319         unsigned char sha1[20];
 320         struct entry *entry;
 321
 322         setup_git_directory();
 323
 324         if (argc != 2 || get_sha1(argv[1], sha1))
 325                 usage("git-convert-objects <sha1>");
 326
 327         entry = convert_entry(sha1);
 328         printf("new sha1: %s\n", sha1_to_hex(entry->new_sha1));
 329         return 0;
 330 }