| 1 | /* mdb_load.c - memory-mapped database load tool */ |
| 2 | /* |
| 3 | * Copyright 2011-2018 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 mdb_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 | static MDB_val k0buf; |
| 41 | |
| 42 | #define Yu MDB_PRIy(u) |
| 43 | |
| 44 | #define STRLENOF(s) (sizeof(s)-1) |
| 45 | |
| 46 | typedef struct flagbit { |
| 47 | int bit; |
| 48 | char *name; |
| 49 | int len; |
| 50 | } flagbit; |
| 51 | |
| 52 | #define S(s) s, STRLENOF(s) |
| 53 | |
| 54 | flagbit dbflags[] = { |
| 55 | { MDB_REVERSEKEY, S("reversekey" ) }, |
| 56 | { MDB_DUPSORT, S("dupsort" ) }, |
| 57 | { MDB_INTEGERKEY, S("integerkey" ) }, |
| 58 | { MDB_DUPFIXED, S("dupfixed" ) }, |
| 59 | { MDB_INTEGERDUP, S("integerdup" ) }, |
| 60 | { MDB_REVERSEDUP, S("reversedup" ) }, |
| 61 | { 0, NULL, 0 } |
| 62 | }; |
| 63 | |
| 64 | static void readhdr(void) |
| 65 | { |
| 66 | char *ptr; |
| 67 | |
| 68 | flags = 0; |
| 69 | while (fgets(dbuf.mv_data, dbuf.mv_size, stdin) != NULL) { |
| 70 | lineno++; |
| 71 | if (!strncmp(dbuf.mv_data, "VERSION=" , STRLENOF("VERSION=" ))) { |
| 72 | version=atoi((char *)dbuf.mv_data+STRLENOF("VERSION=" )); |
| 73 | if (version > 3) { |
| 74 | fprintf(stderr, "%s: line %" Yu": unsupported VERSION %d\n" , |
| 75 | prog, lineno, version); |
| 76 | exit(EXIT_FAILURE); |
| 77 | } |
| 78 | } else if (!strncmp(dbuf.mv_data, "HEADER=END" , STRLENOF("HEADER=END" ))) { |
| 79 | break; |
| 80 | } else if (!strncmp(dbuf.mv_data, "format=" , STRLENOF("format=" ))) { |
| 81 | if (!strncmp((char *)dbuf.mv_data+STRLENOF("FORMAT=" ), "print" , STRLENOF("print" ))) |
| 82 | mode |= PRINT; |
| 83 | else if (strncmp((char *)dbuf.mv_data+STRLENOF("FORMAT=" ), "bytevalue" , STRLENOF("bytevalue" ))) { |
| 84 | fprintf(stderr, "%s: line %" Yu": unsupported FORMAT %s\n" , |
| 85 | prog, lineno, (char *)dbuf.mv_data+STRLENOF("FORMAT=" )); |
| 86 | exit(EXIT_FAILURE); |
| 87 | } |
| 88 | } else if (!strncmp(dbuf.mv_data, "database=" , STRLENOF("database=" ))) { |
| 89 | ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size); |
| 90 | if (ptr) *ptr = '\0'; |
| 91 | if (subname) free(subname); |
| 92 | subname = strdup((char *)dbuf.mv_data+STRLENOF("database=" )); |
| 93 | } else if (!strncmp(dbuf.mv_data, "type=" , STRLENOF("type=" ))) { |
| 94 | if (strncmp((char *)dbuf.mv_data+STRLENOF("type=" ), "btree" , STRLENOF("btree" ))) { |
| 95 | fprintf(stderr, "%s: line %" Yu": unsupported type %s\n" , |
| 96 | prog, lineno, (char *)dbuf.mv_data+STRLENOF("type=" )); |
| 97 | exit(EXIT_FAILURE); |
| 98 | } |
| 99 | } else if (!strncmp(dbuf.mv_data, "mapaddr=" , STRLENOF("mapaddr=" ))) { |
| 100 | int i; |
| 101 | ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size); |
| 102 | if (ptr) *ptr = '\0'; |
| 103 | i = sscanf((char *)dbuf.mv_data+STRLENOF("mapaddr=" ), "%p" , &info.me_mapaddr); |
| 104 | if (i != 1) { |
| 105 | fprintf(stderr, "%s: line %" Yu": invalid mapaddr %s\n" , |
| 106 | prog, lineno, (char *)dbuf.mv_data+STRLENOF("mapaddr=" )); |
| 107 | exit(EXIT_FAILURE); |
| 108 | } |
| 109 | } else if (!strncmp(dbuf.mv_data, "mapsize=" , STRLENOF("mapsize=" ))) { |
| 110 | int i; |
| 111 | ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size); |
| 112 | if (ptr) *ptr = '\0'; |
| 113 | i = sscanf((char *)dbuf.mv_data+STRLENOF("mapsize=" ), |
| 114 | "%" MDB_SCNy(u), &info.me_mapsize); |
| 115 | if (i != 1) { |
| 116 | fprintf(stderr, "%s: line %" Yu": invalid mapsize %s\n" , |
| 117 | prog, lineno, (char *)dbuf.mv_data+STRLENOF("mapsize=" )); |
| 118 | exit(EXIT_FAILURE); |
| 119 | } |
| 120 | } else if (!strncmp(dbuf.mv_data, "maxreaders=" , STRLENOF("maxreaders=" ))) { |
| 121 | int i; |
| 122 | ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size); |
| 123 | if (ptr) *ptr = '\0'; |
| 124 | i = sscanf((char *)dbuf.mv_data+STRLENOF("maxreaders=" ), "%u" , &info.me_maxreaders); |
| 125 | if (i != 1) { |
| 126 | fprintf(stderr, "%s: line %" Yu": invalid maxreaders %s\n" , |
| 127 | prog, lineno, (char *)dbuf.mv_data+STRLENOF("maxreaders=" )); |
| 128 | exit(EXIT_FAILURE); |
| 129 | } |
| 130 | } else { |
| 131 | int i; |
| 132 | for (i=0; dbflags[i].bit; i++) { |
| 133 | if (!strncmp(dbuf.mv_data, dbflags[i].name, dbflags[i].len) && |
| 134 | ((char *)dbuf.mv_data)[dbflags[i].len] == '=') { |
| 135 | flags |= dbflags[i].bit; |
| 136 | break; |
| 137 | } |
| 138 | } |
| 139 | if (!dbflags[i].bit) { |
| 140 | ptr = memchr(dbuf.mv_data, '=', dbuf.mv_size); |
| 141 | if (!ptr) { |
| 142 | fprintf(stderr, "%s: line %" Yu": unexpected format\n" , |
| 143 | prog, lineno); |
| 144 | exit(EXIT_FAILURE); |
| 145 | } else { |
| 146 | *ptr = '\0'; |
| 147 | fprintf(stderr, "%s: line %" Yu": unrecognized keyword ignored: %s\n" , |
| 148 | prog, lineno, (char *)dbuf.mv_data); |
| 149 | } |
| 150 | } |
| 151 | } |
| 152 | } |
| 153 | } |
| 154 | |
| 155 | static void badend(void) |
| 156 | { |
| 157 | fprintf(stderr, "%s: line %" Yu": unexpected end of input\n" , |
| 158 | prog, lineno); |
| 159 | } |
| 160 | |
| 161 | static int unhex(unsigned char *c2) |
| 162 | { |
| 163 | int x, c; |
| 164 | x = *c2++ & 0x4f; |
| 165 | if (x & 0x40) |
| 166 | x -= 55; |
| 167 | c = x << 4; |
| 168 | x = *c2 & 0x4f; |
| 169 | if (x & 0x40) |
| 170 | x -= 55; |
| 171 | c |= x; |
| 172 | return c; |
| 173 | } |
| 174 | |
| 175 | static int readline(MDB_val *out, MDB_val *buf) |
| 176 | { |
| 177 | unsigned char *c1, *c2, *end; |
| 178 | size_t len, l2; |
| 179 | int c; |
| 180 | |
| 181 | if (!(mode & NOHDR)) { |
| 182 | c = fgetc(stdin); |
| 183 | if (c == EOF) { |
| 184 | Eof = 1; |
| 185 | return EOF; |
| 186 | } |
| 187 | if (c != ' ') { |
| 188 | lineno++; |
| 189 | if (fgets(buf->mv_data, buf->mv_size, stdin) == NULL) { |
| 190 | badend: |
| 191 | Eof = 1; |
| 192 | badend(); |
| 193 | return EOF; |
| 194 | } |
| 195 | if (c == 'D' && !strncmp(buf->mv_data, "ATA=END" , STRLENOF("ATA=END" ))) |
| 196 | return EOF; |
| 197 | goto badend; |
| 198 | } |
| 199 | } |
| 200 | if (fgets(buf->mv_data, buf->mv_size, stdin) == NULL) { |
| 201 | Eof = 1; |
| 202 | return EOF; |
| 203 | } |
| 204 | lineno++; |
| 205 | |
| 206 | c1 = buf->mv_data; |
| 207 | len = strlen((char *)c1); |
| 208 | l2 = len; |
| 209 | |
| 210 | /* Is buffer too short? */ |
| 211 | while (c1[len-1] != '\n') { |
| 212 | buf->mv_data = realloc(buf->mv_data, buf->mv_size*2); |
| 213 | if (!buf->mv_data) { |
| 214 | Eof = 1; |
| 215 | fprintf(stderr, "%s: line %" Yu": out of memory, line too long\n" , |
| 216 | prog, lineno); |
| 217 | return EOF; |
| 218 | } |
| 219 | c1 = buf->mv_data; |
| 220 | c1 += l2; |
| 221 | if (fgets((char *)c1, buf->mv_size+1, stdin) == NULL) { |
| 222 | Eof = 1; |
| 223 | badend(); |
| 224 | return EOF; |
| 225 | } |
| 226 | buf->mv_size *= 2; |
| 227 | len = strlen((char *)c1); |
| 228 | l2 += len; |
| 229 | } |
| 230 | c1 = c2 = buf->mv_data; |
| 231 | len = l2; |
| 232 | c1[--len] = '\0'; |
| 233 | end = c1 + len; |
| 234 | |
| 235 | if (mode & PRINT) { |
| 236 | while (c2 < end) { |
| 237 | if (*c2 == '\\') { |
| 238 | if (c2[1] == '\\') { |
| 239 | c1++; c2 += 2; |
| 240 | } else { |
| 241 | if (c2+3 > end || !isxdigit(c2[1]) || !isxdigit(c2[2])) { |
| 242 | Eof = 1; |
| 243 | badend(); |
| 244 | return EOF; |
| 245 | } |
| 246 | *c1++ = unhex(++c2); |
| 247 | c2 += 2; |
| 248 | } |
| 249 | } else { |
| 250 | /* copies are redundant when no escapes were used */ |
| 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 [-V] [-a] [-f input] [-n] [-s name] [-N] [-T] dbpath\n" , prog); |
| 280 | exit(EXIT_FAILURE); |
| 281 | } |
| 282 | |
| 283 | static int greater(const MDB_val *a, const MDB_val *b) |
| 284 | { |
| 285 | return 1; |
| 286 | } |
| 287 | |
| 288 | int main(int argc, char *argv[]) |
| 289 | { |
| 290 | int i, rc; |
| 291 | MDB_env *env; |
| 292 | MDB_txn *txn; |
| 293 | MDB_cursor *mc; |
| 294 | MDB_dbi dbi; |
| 295 | char *envname; |
| 296 | int envflags = MDB_NOSYNC, putflags = 0; |
| 297 | int dohdr = 0, append = 0; |
| 298 | MDB_val prevk; |
| 299 | |
| 300 | prog = argv[0]; |
| 301 | |
| 302 | if (argc < 2) { |
| 303 | usage(); |
| 304 | } |
| 305 | |
| 306 | /* -a: append records in input order |
| 307 | * -f: load file instead of stdin |
| 308 | * -n: use NOSUBDIR flag on env_open |
| 309 | * -s: load into named subDB |
| 310 | * -N: use NOOVERWRITE on puts |
| 311 | * -T: read plaintext |
| 312 | * -V: print version and exit |
| 313 | */ |
| 314 | while ((i = getopt(argc, argv, "af:ns:NTV" )) != EOF) { |
| 315 | switch(i) { |
| 316 | case 'V': |
| 317 | printf("%s\n" , MDB_VERSION_STRING); |
| 318 | exit(0); |
| 319 | break; |
| 320 | case 'a': |
| 321 | append = 1; |
| 322 | break; |
| 323 | case 'f': |
| 324 | if (freopen(optarg, "r" , stdin) == NULL) { |
| 325 | fprintf(stderr, "%s: %s: reopen: %s\n" , |
| 326 | prog, optarg, strerror(errno)); |
| 327 | exit(EXIT_FAILURE); |
| 328 | } |
| 329 | break; |
| 330 | case 'n': |
| 331 | envflags |= MDB_NOSUBDIR; |
| 332 | break; |
| 333 | case 's': |
| 334 | subname = strdup(optarg); |
| 335 | break; |
| 336 | case 'N': |
| 337 | putflags = MDB_NOOVERWRITE|MDB_NODUPDATA; |
| 338 | break; |
| 339 | case 'T': |
| 340 | mode |= NOHDR | PRINT; |
| 341 | break; |
| 342 | default: |
| 343 | usage(); |
| 344 | } |
| 345 | } |
| 346 | |
| 347 | if (optind != argc - 1) |
| 348 | usage(); |
| 349 | |
| 350 | dbuf.mv_size = 4096; |
| 351 | dbuf.mv_data = malloc(dbuf.mv_size); |
| 352 | |
| 353 | if (!(mode & NOHDR)) |
| 354 | readhdr(); |
| 355 | |
| 356 | envname = argv[optind]; |
| 357 | rc = mdb_env_create(&env); |
| 358 | if (rc) { |
| 359 | fprintf(stderr, "mdb_env_create failed, error %d %s\n" , rc, mdb_strerror(rc)); |
| 360 | return EXIT_FAILURE; |
| 361 | } |
| 362 | |
| 363 | mdb_env_set_maxdbs(env, 2); |
| 364 | |
| 365 | if (info.me_maxreaders) |
| 366 | mdb_env_set_maxreaders(env, info.me_maxreaders); |
| 367 | |
| 368 | if (info.me_mapsize) |
| 369 | mdb_env_set_mapsize(env, info.me_mapsize); |
| 370 | |
| 371 | if (info.me_mapaddr) |
| 372 | envflags |= MDB_FIXEDMAP; |
| 373 | |
| 374 | rc = mdb_env_open(env, envname, envflags, 0664); |
| 375 | if (rc) { |
| 376 | fprintf(stderr, "mdb_env_open failed, error %d %s\n" , rc, mdb_strerror(rc)); |
| 377 | goto env_close; |
| 378 | } |
| 379 | |
| 380 | kbuf.mv_size = mdb_env_get_maxkeysize(env) * 2 + 2; |
| 381 | kbuf.mv_data = malloc(kbuf.mv_size * 2); |
| 382 | k0buf.mv_size = kbuf.mv_size; |
| 383 | k0buf.mv_data = (char *)kbuf.mv_data + kbuf.mv_size; |
| 384 | prevk.mv_data = k0buf.mv_data; |
| 385 | |
| 386 | while(!Eof) { |
| 387 | MDB_val key, data; |
| 388 | int batch = 0; |
| 389 | int appflag; |
| 390 | |
| 391 | if (!dohdr) { |
| 392 | dohdr = 1; |
| 393 | } else if (!(mode & NOHDR)) |
| 394 | readhdr(); |
| 395 | |
| 396 | rc = mdb_txn_begin(env, NULL, 0, &txn); |
| 397 | if (rc) { |
| 398 | fprintf(stderr, "mdb_txn_begin failed, error %d %s\n" , rc, mdb_strerror(rc)); |
| 399 | goto env_close; |
| 400 | } |
| 401 | |
| 402 | rc = mdb_open(txn, subname, flags|MDB_CREATE, &dbi); |
| 403 | if (rc) { |
| 404 | fprintf(stderr, "mdb_open failed, error %d %s\n" , rc, mdb_strerror(rc)); |
| 405 | goto txn_abort; |
| 406 | } |
| 407 | prevk.mv_size = 0; |
| 408 | if (append) { |
| 409 | mdb_set_compare(txn, dbi, greater); |
| 410 | if (flags & MDB_DUPSORT) |
| 411 | mdb_set_dupsort(txn, dbi, greater); |
| 412 | } |
| 413 | |
| 414 | rc = mdb_cursor_open(txn, dbi, &mc); |
| 415 | if (rc) { |
| 416 | fprintf(stderr, "mdb_cursor_open failed, error %d %s\n" , rc, mdb_strerror(rc)); |
| 417 | goto txn_abort; |
| 418 | } |
| 419 | |
| 420 | while(1) { |
| 421 | rc = readline(&key, &kbuf); |
| 422 | if (rc) /* rc == EOF */ |
| 423 | break; |
| 424 | |
| 425 | rc = readline(&data, &dbuf); |
| 426 | if (rc) { |
| 427 | fprintf(stderr, "%s: line %" Yu": failed to read key value\n" , prog, lineno); |
| 428 | goto txn_abort; |
| 429 | } |
| 430 | |
| 431 | if (append) { |
| 432 | appflag = MDB_APPEND; |
| 433 | if (flags & MDB_DUPSORT) { |
| 434 | if (prevk.mv_size == key.mv_size && !memcmp(prevk.mv_data, key.mv_data, key.mv_size)) |
| 435 | appflag = MDB_CURRENT|MDB_APPENDDUP; |
| 436 | else { |
| 437 | memcpy(prevk.mv_data, key.mv_data, key.mv_size); |
| 438 | prevk.mv_size = key.mv_size; |
| 439 | } |
| 440 | } |
| 441 | } else { |
| 442 | appflag = 0; |
| 443 | } |
| 444 | rc = mdb_cursor_put(mc, &key, &data, putflags|appflag); |
| 445 | if (rc == MDB_KEYEXIST && putflags) |
| 446 | continue; |
| 447 | if (rc) { |
| 448 | fprintf(stderr, "mdb_cursor_put failed, error %d %s\n" , rc, mdb_strerror(rc)); |
| 449 | goto txn_abort; |
| 450 | } |
| 451 | batch++; |
| 452 | if (batch == 100) { |
| 453 | rc = mdb_txn_commit(txn); |
| 454 | if (rc) { |
| 455 | fprintf(stderr, "%s: line %" Yu": txn_commit: %s\n" , |
| 456 | prog, lineno, mdb_strerror(rc)); |
| 457 | goto env_close; |
| 458 | } |
| 459 | rc = mdb_txn_begin(env, NULL, 0, &txn); |
| 460 | if (rc) { |
| 461 | fprintf(stderr, "mdb_txn_begin failed, error %d %s\n" , rc, mdb_strerror(rc)); |
| 462 | goto env_close; |
| 463 | } |
| 464 | rc = mdb_cursor_open(txn, dbi, &mc); |
| 465 | if (rc) { |
| 466 | fprintf(stderr, "mdb_cursor_open failed, error %d %s\n" , rc, mdb_strerror(rc)); |
| 467 | goto txn_abort; |
| 468 | } |
| 469 | if (appflag & MDB_APPENDDUP) { |
| 470 | MDB_val k, d; |
| 471 | mdb_cursor_get(mc, &k, &d, MDB_LAST); |
| 472 | } |
| 473 | batch = 0; |
| 474 | } |
| 475 | } |
| 476 | rc = mdb_txn_commit(txn); |
| 477 | txn = NULL; |
| 478 | if (rc) { |
| 479 | fprintf(stderr, "%s: line %" Yu": txn_commit: %s\n" , |
| 480 | prog, lineno, mdb_strerror(rc)); |
| 481 | goto env_close; |
| 482 | } |
| 483 | mdb_dbi_close(env, dbi); |
| 484 | } |
| 485 | |
| 486 | txn_abort: |
| 487 | mdb_txn_abort(txn); |
| 488 | env_close: |
| 489 | mdb_env_close(env); |
| 490 | |
| 491 | return rc ? EXIT_FAILURE : EXIT_SUCCESS; |
| 492 | } |
| 493 | |