git.sur5r.net Git - openldap/blob - libraries/liblmdb/mdb_load.c

   1 /* mdb_load.c - memory-mapped database load tool */
   2 /*
   3  * Copyright 2011-2014 Howard Chu, Symas Corp.
   4  * All rights reserved.
   5  *
   6  * Redistribution and use in source and binary forms, with or without
   7  * modification, are permitted only as authorized by the OpenLDAP
   8  * Public License.
   9  *
  10  * A copy of this license is available in the file LICENSE in the
  11  * top-level directory of the distribution or, alternatively, at
  12  * <http://www.OpenLDAP.org/license.html>.
  13  */
  14 #include <stdio.h>
  15 #include <stdlib.h>
  16 #include <errno.h>
  17 #include <string.h>
  18 #include <ctype.h>
  19 #include <unistd.h>
  20 #include "lmdb.h"
  21
  22 #define PRINT   1
  23 #define NOHDR   2
  24 static int mode;
  25
  26 static char *subname = NULL;
  27
  28 static size_t lineno;
  29 static int version;
  30
  31 static int flags;
  32
  33 static char *prog;
  34
  35 static int Eof;
  36
  37 static MDB_envinfo info;
  38
  39 static MDB_val kbuf, dbuf;
  40
  41 #ifdef _WIN32
  42 #define Z       "I"
  43 #else
  44 #define Z       "z"
  45 #endif
  46
  47 #define STRLENOF(s)     (sizeof(s)-1)
  48
  49 typedef struct flagbit {
  50         int bit;
  51         char *name;
  52         int len;
  53 } flagbit;
  54
  55 #define S(s)    s, STRLENOF(s)
  56
  57 flagbit dbflags[] = {
  58         { MDB_REVERSEKEY, S("reversekey") },
  59         { MDB_DUPSORT, S("dupsort") },
  60         { MDB_INTEGERKEY, S("integerkey") },
  61         { MDB_DUPFIXED, S("dupfixed") },
  62         { MDB_INTEGERDUP, S("integerdup") },
  63         { MDB_REVERSEDUP, S("reversedup") },
  64         { 0, NULL, 0 }
  65 };
  66
  67 static const char hexc[] = "0123456789abcdef";
  68
  69 static void readhdr(void)
  70 {
  71         char *ptr;
  72
  73         while (fgets(dbuf.mv_data, dbuf.mv_size, stdin) != NULL) {
  74                 lineno++;
  75                 if (!strncmp(dbuf.mv_data, "VERSION=", STRLENOF("VERSION="))) {
  76                         version=atoi((char *)dbuf.mv_data+STRLENOF("VERSION="));
  77                         if (version > 3) {
  78                                 fprintf(stderr, "%s: line %" Z "d: unsupported VERSION %d\n",
  79                                         prog, lineno, version);
  80                                 exit(EXIT_FAILURE);
  81                         }
  82                 } else if (!strncmp(dbuf.mv_data, "HEADER=END", STRLENOF("HEADER=END"))) {
  83                         break;
  84                 } else if (!strncmp(dbuf.mv_data, "format=", STRLENOF("format="))) {
  85                         if (!strncmp((char *)dbuf.mv_data+STRLENOF("FORMAT="), "print", STRLENOF("print")))
  86                                 mode |= PRINT;
  87                         else if (strncmp((char *)dbuf.mv_data+STRLENOF("FORMAT="), "bytevalue", STRLENOF("bytevalue"))) {
  88                                 fprintf(stderr, "%s: line %" Z "d: unsupported FORMAT %s\n",
  89                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("FORMAT="));
  90                                 exit(EXIT_FAILURE);
  91                         }
  92                 } else if (!strncmp(dbuf.mv_data, "database=", STRLENOF("database="))) {
  93                         ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size);
  94                         if (ptr) *ptr = '\0';
  95                         if (subname) free(subname);
  96                         subname = strdup((char *)dbuf.mv_data+STRLENOF("database="));
  97                 } else if (!strncmp(dbuf.mv_data, "type=", STRLENOF("type="))) {
  98                         if (strncmp((char *)dbuf.mv_data+STRLENOF("type="), "btree", STRLENOF("btree")))  {
  99                                 fprintf(stderr, "%s: line %" Z "d: unsupported type %s\n",
 100                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("type="));
 101                                 exit(EXIT_FAILURE);
 102                         }
 103                 } else if (!strncmp(dbuf.mv_data, "mapaddr=", STRLENOF("mapaddr="))) {
 104                         int i;
 105                         ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size);
 106                         if (ptr) *ptr = '\0';
 107                         i = sscanf((char *)dbuf.mv_data+STRLENOF("mapaddr="), "%p", &info.me_mapaddr);
 108                         if (i != 1) {
 109                                 fprintf(stderr, "%s: line %" Z "d: invalid mapaddr %s\n",
 110                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("mapaddr="));
 111                                 exit(EXIT_FAILURE);
 112                         }
 113                 } else if (!strncmp(dbuf.mv_data, "mapsize=", STRLENOF("mapsize="))) {
 114                         int i;
 115                         ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size);
 116                         if (ptr) *ptr = '\0';
 117                         i = sscanf((char *)dbuf.mv_data+STRLENOF("mapsize="), "%" Z "u", &info.me_mapsize);
 118                         if (i != 1) {
 119                                 fprintf(stderr, "%s: line %" Z "d: invalid mapsize %s\n",
 120                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("mapsize="));
 121                                 exit(EXIT_FAILURE);
 122                         }
 123                 } else if (!strncmp(dbuf.mv_data, "maxreaders=", STRLENOF("maxreaders="))) {
 124                         int i;
 125                         ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size);
 126                         if (ptr) *ptr = '\0';
 127                         i = sscanf((char *)dbuf.mv_data+STRLENOF("maxreaders="), "%u", &info.me_maxreaders);
 128                         if (i != 1) {
 129                                 fprintf(stderr, "%s: line %" Z "d: invalid maxreaders %s\n",
 130                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("maxreaders="));
 131                                 exit(EXIT_FAILURE);
 132                         }
 133                 } else {
 134                         int i;
 135                         for (i=0; dbflags[i].bit; i++) {
 136                                 if (!strncmp(dbuf.mv_data, dbflags[i].name, dbflags[i].len) &&
 137                                         ((char *)dbuf.mv_data)[dbflags[i].len] == '=') {
 138                                         flags |= dbflags[i].bit;
 139                                         break;
 140                                 }
 141                         }
 142                         if (!dbflags[i].bit) {
 143                                 ptr = memchr(dbuf.mv_data, '=', dbuf.mv_size);
 144                                 if (!ptr) {
 145                                         fprintf(stderr, "%s: line %" Z "d: unexpected format\n",
 146                                                 prog, lineno);
 147                                         exit(EXIT_FAILURE);
 148                                 } else {
 149                                         *ptr = '\0';
 150                                         fprintf(stderr, "%s: line %" Z "d: unrecognized keyword ignored: %s\n",
 151                                                 prog, lineno, (char *)dbuf.mv_data);
 152                                 }
 153                         }
 154                 }
 155         }
 156 }
 157
 158 static void badend(void)
 159 {
 160         fprintf(stderr, "%s: line %" Z "d: unexpected end of input\n",
 161                 prog, lineno);
 162 }
 163
 164 static int unhex(unsigned char *c2)
 165 {
 166         int x, c;
 167         x = *c2++ & 0x4f;
 168         if (x & 0x40)
 169                 x -= 55;
 170         c = x << 4;
 171         x = *c2 & 0x4f;
 172         if (x & 0x40)
 173                 x -= 55;
 174         c |= x;
 175         return c;
 176 }
 177
 178 static int readline(MDB_val *out, MDB_val *buf)
 179 {
 180         unsigned char *c1, *c2, *end;
 181         size_t len;
 182         int c;
 183
 184         if (!(mode & NOHDR)) {
 185                 c = fgetc(stdin);
 186                 if (c == EOF) {
 187                         Eof = 1;
 188                         return EOF;
 189                 }
 190                 if (c != ' ') {
 191                         lineno++;
 192                         if (fgets(buf->mv_data, buf->mv_size, stdin) == NULL) {
 193 badend:
 194                                 Eof = 1;
 195                                 badend();
 196                                 return EOF;
 197                         }
 198                         if (c == 'D' && !strncmp(buf->mv_data, "ATA=END", STRLENOF("ATA=END")))
 199                                 return EOF;
 200                         goto badend;
 201                 }
 202         }
 203         if (fgets(buf->mv_data, buf->mv_size, stdin) == NULL) {
 204                 Eof = 1;
 205                 return EOF;
 206         }
 207         lineno++;
 208
 209         c1 = buf->mv_data;
 210         len = strlen((char *)c1);
 211
 212         /* Is buffer too short? */
 213         while (c1[len-1] != '\n') {
 214                 buf->mv_data = realloc(buf->mv_data, buf->mv_size*2);
 215                 if (!buf->mv_data) {
 216                         Eof = 1;
 217                         fprintf(stderr, "%s: line %" Z "d: out of memory, line too long\n",
 218                                 prog, lineno);
 219                         return EOF;
 220                 }
 221                 c1 = buf->mv_data;
 222                 c1 += buf->mv_size;
 223                 if (fgets((char *)c1, buf->mv_size, stdin) == NULL) {
 224                         Eof = 1;
 225                         badend();
 226                         return EOF;
 227                 }
 228                 buf->mv_size *= 2;
 229                 len = strlen((char *)c1);
 230         }
 231         c1 = c2 = buf->mv_data;
 232         len = strlen((char *)c1);
 233         c1[--len] = '\0';
 234         end = c1 + len;
 235
 236         if (mode & PRINT) {
 237                 while (c2 < end) {
 238                         if (*c2 == '\\') {
 239                                 if (c2[1] == '\\') {
 240                                         c1++; c2 += 2;
 241                                 } else {
 242                                         if (c2+3 > end || !isxdigit(c2[1]) || !isxdigit(c2[2])) {
 243                                                 Eof = 1;
 244                                                 badend();
 245                                                 return EOF;
 246                                         }
 247                                         *c1++ = unhex(++c2);
 248                                         c2 += 2;
 249                                 }
 250                         } else {
 251                                 c1++; c2++;
 252                         }
 253                 }
 254         } else {
 255                 /* odd length not allowed */
 256                 if (len & 1) {
 257                         Eof = 1;
 258                         badend();
 259                         return EOF;
 260                 }
 261                 while (c2 < end) {
 262                         if (!isxdigit(*c2) || !isxdigit(c2[1])) {
 263                                 Eof = 1;
 264                                 badend();
 265                                 return EOF;
 266                         }
 267                         *c1++ = unhex(c2);
 268                         c2 += 2;
 269                 }
 270         }
 271         c2 = out->mv_data = buf->mv_data;
 272         out->mv_size = c1 - c2;
 273
 274         return 0;
 275 }
 276
 277 static void usage(void)
 278 {
 279         fprintf(stderr, "usage: %s dbpath [-V] [-f input] [-n] [-s name] [-N] [-T]\n", prog);
 280         exit(EXIT_FAILURE);
 281 }
 282
 283 int main(int argc, char *argv[])
 284 {
 285         int i, rc;
 286         MDB_env *env;
 287         MDB_txn *txn;
 288         MDB_cursor *mc;
 289         MDB_dbi dbi;
 290         char *envname;
 291         int envflags = 0, putflags = 0;
 292         int dohdr = 0;
 293
 294         prog = argv[0];
 295
 296         if (argc < 2) {
 297                 usage();
 298         }
 299
 300         /* -f: load file instead of stdin
 301          * -n: use NOSUBDIR flag on env_open
 302          * -s: load into named subDB
 303          * -N: use NOOVERWRITE on puts
 304          * -T: read plaintext
 305          * -V: print version and exit
 306          */
 307         while ((i = getopt(argc, argv, "f:ns:NTV")) != EOF) {
 308                 switch(i) {
 309                 case 'V':
 310                         printf("%s\n", MDB_VERSION_STRING);
 311                         exit(0);
 312                         break;
 313                 case 'f':
 314                         if (freopen(optarg, "r", stdin) == NULL) {
 315                                 fprintf(stderr, "%s: %s: reopen: %s\n",
 316                                         prog, optarg, strerror(errno));
 317                                 exit(EXIT_FAILURE);
 318                         }
 319                         break;
 320                 case 'n':
 321                         envflags |= MDB_NOSUBDIR;
 322                         break;
 323                 case 's':
 324                         subname = strdup(optarg);
 325                         break;
 326                 case 'N':
 327                         putflags = MDB_NOOVERWRITE|MDB_NODUPDATA;
 328                         break;
 329                 case 'T':
 330                         mode |= NOHDR;
 331                         break;
 332                 default:
 333                         usage();
 334                 }
 335         }
 336
 337         if (optind != argc - 1)
 338                 usage();
 339
 340         dbuf.mv_size = 4096;
 341         dbuf.mv_data = malloc(dbuf.mv_size);
 342
 343         if (!(mode & NOHDR))
 344                 readhdr();
 345
 346         envname = argv[optind];
 347         rc = mdb_env_create(&env);
 348
 349         mdb_env_set_maxdbs(env, 2);
 350
 351         if (info.me_maxreaders)
 352                 mdb_env_set_maxreaders(env, info.me_maxreaders);
 353
 354         if (info.me_mapsize)
 355                 mdb_env_set_mapsize(env, info.me_mapsize);
 356
 357         if (info.me_mapaddr)
 358                 envflags |= MDB_FIXEDMAP;
 359
 360         rc = mdb_env_open(env, envname, envflags, 0664);
 361         if (rc) {
 362                 fprintf(stderr, "mdb_env_open failed, error %d %s\n", rc, mdb_strerror(rc));
 363                 goto env_close;
 364         }
 365
 366         kbuf.mv_size = mdb_env_get_maxkeysize(env) * 2 + 2;
 367         kbuf.mv_data = malloc(kbuf.mv_size);
 368
 369         while(!Eof) {
 370                 MDB_val key, data;
 371                 int batch = 0;
 372                 flags = 0;
 373
 374                 if (!dohdr) {
 375                         dohdr = 1;
 376                 } else if (!(mode & NOHDR))
 377                         readhdr();
 378
 379                 rc = mdb_txn_begin(env, NULL, 0, &txn);
 380                 if (rc) {
 381                         fprintf(stderr, "mdb_txn_begin failed, error %d %s\n", rc, mdb_strerror(rc));
 382                         goto env_close;
 383                 }
 384
 385                 rc = mdb_open(txn, subname, flags|MDB_CREATE, &dbi);
 386                 if (rc) {
 387                         fprintf(stderr, "mdb_open failed, error %d %s\n", rc, mdb_strerror(rc));
 388                         goto txn_abort;
 389                 }
 390
 391                 rc = mdb_cursor_open(txn, dbi, &mc);
 392                 if (rc) {
 393                         fprintf(stderr, "mdb_cursor_open failed, error %d %s\n", rc, mdb_strerror(rc));
 394                         goto txn_abort;
 395                 }
 396
 397                 while(1) {
 398                         rc = readline(&key, &kbuf);
 399                         if (rc == EOF)
 400                                 break;
 401                         if (rc)
 402                                 goto txn_abort;
 403
 404                         rc = readline(&data, &dbuf);
 405                         if (rc)
 406                                 goto txn_abort;
 407
 408                         rc = mdb_cursor_put(mc, &key, &data, putflags);
 409                         if (rc == MDB_KEYEXIST && putflags)
 410                                 continue;
 411                         if (rc)
 412                                 goto txn_abort;
 413                         batch++;
 414                         if (batch == 100) {
 415                                 rc = mdb_txn_commit(txn);
 416                                 if (rc) {
 417                                         fprintf(stderr, "%s: line %" Z "d: txn_commit: %s\n",
 418                                                 prog, lineno, mdb_strerror(rc));
 419                                         goto env_close;
 420                                 }
 421                                 rc = mdb_txn_begin(env, NULL, 0, &txn);
 422                                 if (rc) {
 423                                         fprintf(stderr, "mdb_txn_begin failed, error %d %s\n", rc, mdb_strerror(rc));
 424                                         goto env_close;
 425                                 }
 426                                 rc = mdb_cursor_open(txn, dbi, &mc);
 427                                 if (rc) {
 428                                         fprintf(stderr, "mdb_cursor_open failed, error %d %s\n", rc, mdb_strerror(rc));
 429                                         goto txn_abort;
 430                                 }
 431                                 batch = 0;
 432                         }
 433                 }
 434                 rc = mdb_txn_commit(txn);
 435                 txn = NULL;
 436                 if (rc) {
 437                         fprintf(stderr, "%s: line %" Z "d: txn_commit: %s\n",
 438                                 prog, lineno, mdb_strerror(rc));
 439                         goto env_close;
 440                 }
 441                 mdb_dbi_close(env, dbi);
 442         }
 443
 444 txn_abort:
 445         mdb_txn_abort(txn);
 446 env_close:
 447         mdb_env_close(env);
 448
 449         return rc ? EXIT_FAILURE : EXIT_SUCCESS;
 450 }