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_val kbuf, dbuf;
  38
  39 #define STRLENOF(s)     (sizeof(s)-1)
  40
  41 typedef struct flagbit {
  42         int bit;
  43         char *name;
  44         int len;
  45 } flagbit;
  46
  47 #define S(s)    s, STRLENOF(s)
  48
  49 flagbit dbflags[] = {
  50         { MDB_REVERSEKEY, S("reversekey") },
  51         { MDB_DUPSORT, S("dupsort") },
  52         { MDB_INTEGERKEY, S("integerkey") },
  53         { MDB_DUPFIXED, S("dupfixed") },
  54         { MDB_INTEGERDUP, S("integerdup") },
  55         { MDB_REVERSEDUP, S("reversedup") },
  56         { 0, NULL, 0 }
  57 };
  58
  59 static const char hexc[] = "0123456789abcdef";
  60
  61 static void readhdr()
  62 {
  63         char *ptr;
  64
  65         while (fgets(dbuf.mv_data, dbuf.mv_size, stdin) != NULL) {
  66                 lineno++;
  67                 if (!strncmp(dbuf.mv_data, "VERSION=", STRLENOF("VERSION="))) {
  68                         version=atoi(dbuf.mv_data+STRLENOF("VERSION="));
  69                         if (version > 3) {
  70                                 fprintf(stderr, "%s: line %zd: unsupported VERSION %d\n",
  71                                         prog, lineno, version);
  72                                 exit(EXIT_FAILURE);
  73                         }
  74                 } else if (!strncmp(dbuf.mv_data, "HEADER=END", STRLENOF("HEADER=END"))) {
  75                         break;
  76                 } else if (!strncmp(dbuf.mv_data, "format=", STRLENOF("format="))) {
  77                         if (!strncmp(dbuf.mv_data+STRLENOF("FORMAT="), "print", STRLENOF("print")))
  78                                 mode |= PRINT;
  79                         else if (strncmp(dbuf.mv_data+STRLENOF("FORMAT="), "bytevalue", STRLENOF("bytevalue"))) {
  80                                 fprintf(stderr, "%s: line %zd: unsupported FORMAT %s\n",
  81                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("FORMAT="));
  82                                 exit(EXIT_FAILURE);
  83                         }
  84                 } else if (!strncmp(dbuf.mv_data, "database=", STRLENOF("database="))) {
  85                         ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size);
  86                         if (ptr) *ptr = '\0';
  87                         if (subname) free(subname);
  88                         subname = strdup(dbuf.mv_data+STRLENOF("database="));
  89                 } else if (!strncmp(dbuf.mv_data, "type=", STRLENOF("type="))) {
  90                         if (strncmp(dbuf.mv_data+STRLENOF("type="), "btree", STRLENOF("btree")))  {
  91                                 fprintf(stderr, "%s: line %zd: unsupported type %s\n",
  92                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("type="));
  93                                 exit(EXIT_FAILURE);
  94                         }
  95                 } else {
  96                         int i;
  97                         for (i=0; dbflags[i].bit; i++) {
  98                                 if (!strncmp(dbuf.mv_data, dbflags[i].name, dbflags[i].len) &&
  99                                         ((char *)dbuf.mv_data)[dbflags[i].len] == '=') {
 100                                         flags |= dbflags[i].bit;
 101                                         break;
 102                                 }
 103                         }
 104                         if (!dbflags[i].bit) {
 105                                 ptr = memchr(dbuf.mv_data, '=', dbuf.mv_size);
 106                                 if (!ptr) {
 107                                         fprintf(stderr, "%s: line %zd: unexpected format\n",
 108                                                 prog, lineno);
 109                                         exit(EXIT_FAILURE);
 110                                 } else {
 111                                         *ptr = '\0';
 112                                         fprintf(stderr, "%s: line %zd: unrecognized keyword ignored: %s\n",
 113                                                 prog, lineno, (char *)dbuf.mv_data);
 114                                 }
 115                         }
 116                 }
 117         }
 118 }
 119
 120 static void badend()
 121 {
 122         fprintf(stderr, "%s: line %zd: unexpected end of input\n",
 123                 prog, lineno);
 124 }
 125
 126 static int unhex(unsigned char *c2)
 127 {
 128         int x, c;
 129         x = *c2++ & 0x4f;
 130         if (x & 0x40)
 131                 x -= 55;
 132         c = x << 4;
 133         x = *c2 & 0x4f;
 134         if (x & 0x40)
 135                 x -= 55;
 136         c |= x;
 137         return c;
 138 }
 139
 140 static int readline(MDB_val *out, MDB_val *buf)
 141 {
 142         unsigned char *c1, *c2, *end;
 143         size_t len;
 144         int c;
 145
 146         if (!(mode & NOHDR)) {
 147                 c = fgetc(stdin);
 148                 if (c == EOF) {
 149                         eof = 1;
 150                         return EOF;
 151                 }
 152                 if (c != ' ') {
 153                         lineno++;
 154                         if (fgets(buf->mv_data, buf->mv_size, stdin) == NULL) {
 155 badend:
 156                                 eof = 1;
 157                                 badend();
 158                                 return EOF;
 159                         }
 160                         if (c == 'D' && !strncmp(buf->mv_data, "ATA=END", STRLENOF("ATA=END")))
 161                                 return EOF;
 162                         goto badend;
 163                 }
 164         }
 165         if (fgets(buf->mv_data, buf->mv_size, stdin) == NULL) {
 166                 eof = 1;
 167                 return EOF;
 168         }
 169         lineno++;
 170
 171         c1 = buf->mv_data;
 172         len = strlen((char *)c1);
 173
 174         /* Is buffer too short? */
 175         while (c1[len-1] != '\n') {
 176                 buf->mv_data = realloc(buf->mv_data, buf->mv_size*2);
 177                 if (!buf->mv_data) {
 178                         eof = 1;
 179                         fprintf(stderr, "%s: line %zd: out of memory, line too long\n",
 180                                 prog, lineno);
 181                         return EOF;
 182                 }
 183                 c1 = buf->mv_data;
 184                 c1 += buf->mv_size;
 185                 if (fgets((char *)c1, buf->mv_size, stdin) == NULL) {
 186                         eof = 1;
 187                         badend();
 188                         return EOF;
 189                 }
 190                 buf->mv_size *= 2;
 191                 len = strlen((char *)c1);
 192         }
 193         c1 = c2 = buf->mv_data;
 194         len = strlen((char *)c1);
 195         c1[--len] = '\0';
 196         end = c1 + len;
 197
 198         if (mode & PRINT) {
 199                 while (c2 < end) {
 200                         if (*c2 == '\\') {
 201                                 if (c2[1] == '\\') {
 202                                         c1++; c2 += 2;
 203                                 } else {
 204                                         if (c2+3 >= end || !isxdigit(c2[1]) || !isxdigit(c2[2])) {
 205                                                 eof = 1;
 206                                                 badend();
 207                                                 return EOF;
 208                                         }
 209                                         *c1++ = unhex(++c2);
 210                                         c2 += 2;
 211                                 }
 212                         } else {
 213                                 c1++; c2++;
 214                         }
 215                 }
 216         } else {
 217                 /* odd length not allowed */
 218                 if (len & 1) {
 219                         eof = 1;
 220                         badend();
 221                         return EOF;
 222                 }
 223                 while (c2 < end) {
 224                         if (!isxdigit(*c2) || !isxdigit(c2[1])) {
 225                                 eof = 1;
 226                                 badend();
 227                                 return EOF;
 228                         }
 229                         *c1++ = unhex(c2);
 230                         c2 += 2;
 231                 }
 232         }
 233         c2 = out->mv_data = buf->mv_data;
 234         out->mv_size = c1 - c2;
 235
 236         return 0;
 237 }
 238
 239 static void usage()
 240 {
 241         fprintf(stderr, "usage: %s dbpath [-V] [-f input] [-n] [-s name] [-N] [-T]\n", prog);
 242         exit(EXIT_FAILURE);
 243 }
 244
 245 int main(int argc, char *argv[])
 246 {
 247         int i, rc;
 248         MDB_env *env;
 249         MDB_txn *txn;
 250         MDB_cursor *mc;
 251         MDB_dbi dbi;
 252         char *envname;
 253         int envflags = 0, putflags = 0;
 254
 255         prog = argv[0];
 256
 257         if (argc < 2) {
 258                 usage(prog);
 259         }
 260
 261         /* -f: load file instead of stdin
 262          * -n: use NOSUBDIR flag on env_open
 263          * -s: load into named subDB
 264          * -N: use NOOVERWRITE on puts
 265          * -T: read plaintext
 266          * -V: print version and exit
 267          */
 268         while ((i = getopt(argc, argv, "f:ns:NTV")) != EOF) {
 269                 switch(i) {
 270                 case 'V':
 271                         printf("%s\n", MDB_VERSION_STRING);
 272                         exit(0);
 273                         break;
 274                 case 'f':
 275                         if (freopen(optarg, "r", stdin) == NULL) {
 276                                 fprintf(stderr, "%s: %s: reopen: %s\n",
 277                                         prog, optarg, strerror(errno));
 278                                 exit(EXIT_FAILURE);
 279                         }
 280                         break;
 281                 case 'n':
 282                         envflags |= MDB_NOSUBDIR;
 283                         break;
 284                 case 's':
 285                         subname = strdup(optarg);
 286                         break;
 287                 case 'N':
 288                         putflags = MDB_NOOVERWRITE|MDB_NODUPDATA;
 289                         break;
 290                 case 'T':
 291                         mode |= NOHDR;
 292                         break;
 293                 default:
 294                         usage(prog);
 295                 }
 296         }
 297
 298         if (optind != argc - 1)
 299                 usage(prog);
 300
 301         envname = argv[optind];
 302         rc = mdb_env_create(&env);
 303
 304         mdb_env_set_maxdbs(env, 2);
 305
 306         rc = mdb_env_open(env, envname, envflags, 0664);
 307         if (rc) {
 308                 printf("mdb_env_open failed, error %d %s\n", rc, mdb_strerror(rc));
 309                 goto env_close;
 310         }
 311
 312         kbuf.mv_size = mdb_env_get_maxkeysize(env) * 2 + 2;
 313         kbuf.mv_data = malloc(kbuf.mv_size);
 314         dbuf.mv_size = 4096;
 315         dbuf.mv_data = malloc(dbuf.mv_size);
 316
 317         while(!eof) {
 318                 MDB_val key, data;
 319                 int batch = 0;
 320                 flags = 0;
 321
 322                 if (!(mode & NOHDR))
 323                         readhdr();
 324
 325                 rc = mdb_txn_begin(env, NULL, 0, &txn);
 326                 if (rc) {
 327                         printf("mdb_txn_begin failed, error %d %s\n", rc, mdb_strerror(rc));
 328                         goto env_close;
 329                 }
 330
 331                 rc = mdb_open(txn, subname, flags|MDB_CREATE, &dbi);
 332                 if (rc) {
 333                         printf("mdb_open failed, error %d %s\n", rc, mdb_strerror(rc));
 334                         goto txn_abort;
 335                 }
 336
 337                 rc = mdb_cursor_open(txn, dbi, &mc);
 338                 if (rc) {
 339                         printf("mdb_cursor_open failed, error %d %s\n", rc, mdb_strerror(rc));
 340                         goto txn_abort;
 341                 }
 342
 343                 while(1) {
 344                         rc = readline(&key, &kbuf);
 345                         if (rc == EOF)
 346                                 break;
 347                         if (rc)
 348                                 goto txn_abort;
 349
 350                         rc = readline(&data, &dbuf);
 351                         if (rc)
 352                                 goto txn_abort;
 353
 354                         rc = mdb_cursor_put(mc, &key, &data, putflags);
 355                         if (rc == MDB_KEYEXIST && putflags)
 356                                 continue;
 357                         if (rc)
 358                                 goto txn_abort;
 359                         batch++;
 360                         if (batch == 100) {
 361                                 rc = mdb_txn_commit(txn);
 362                                 if (rc) {
 363                                         fprintf(stderr, "%s: line %zd: txn_commit: %s\n",
 364                                                 prog, lineno, mdb_strerror(rc));
 365                                         goto env_close;
 366                                 }
 367                                 rc = mdb_txn_begin(env, NULL, 0, &txn);
 368                                 if (rc) {
 369                                         printf("mdb_txn_begin failed, error %d %s\n", rc, mdb_strerror(rc));
 370                                         goto env_close;
 371                                 }
 372                                 rc = mdb_cursor_open(txn, dbi, &mc);
 373                                 if (rc) {
 374                                         printf("mdb_cursor_open failed, error %d %s\n", rc, mdb_strerror(rc));
 375                                         goto txn_abort;
 376                                 }
 377                                 batch = 0;
 378                         }
 379                 }
 380                 rc = mdb_txn_commit(txn);
 381                 txn = NULL;
 382                 if (rc) {
 383                         fprintf(stderr, "%s: line %zd: txn_commit: %s\n",
 384                                 prog, lineno, mdb_strerror(rc));
 385                         goto env_close;
 386                 }
 387                 mdb_dbi_close(env, dbi);
 388         }
 389
 390 txn_abort:
 391         mdb_txn_abort(txn);
 392 env_close:
 393         mdb_env_close(env);
 394
 395         return rc ? EXIT_FAILURE : EXIT_SUCCESS;
 396 }