]> git.sur5r.net Git - openldap/blob - libraries/liblmdb/mdb_load.c
053cc88d256058a2169e52d585c3b37860a3579e
[openldap] / libraries / liblmdb / mdb_load.c
1 /* mdb_load.c - memory-mapped database load tool */
2 /*
3  * Copyright 2011-2016 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 void readhdr(void)
68 {
69         char *ptr;
70
71         while (fgets(dbuf.mv_data, dbuf.mv_size, stdin) != NULL) {
72                 lineno++;
73                 if (!strncmp(dbuf.mv_data, "VERSION=", STRLENOF("VERSION="))) {
74                         version=atoi((char *)dbuf.mv_data+STRLENOF("VERSION="));
75                         if (version > 3) {
76                                 fprintf(stderr, "%s: line %" Z "d: unsupported VERSION %d\n",
77                                         prog, lineno, version);
78                                 exit(EXIT_FAILURE);
79                         }
80                 } else if (!strncmp(dbuf.mv_data, "HEADER=END", STRLENOF("HEADER=END"))) {
81                         break;
82                 } else if (!strncmp(dbuf.mv_data, "format=", STRLENOF("format="))) {
83                         if (!strncmp((char *)dbuf.mv_data+STRLENOF("FORMAT="), "print", STRLENOF("print")))
84                                 mode |= PRINT;
85                         else if (strncmp((char *)dbuf.mv_data+STRLENOF("FORMAT="), "bytevalue", STRLENOF("bytevalue"))) {
86                                 fprintf(stderr, "%s: line %" Z "d: unsupported FORMAT %s\n",
87                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("FORMAT="));
88                                 exit(EXIT_FAILURE);
89                         }
90                 } else if (!strncmp(dbuf.mv_data, "database=", STRLENOF("database="))) {
91                         ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size);
92                         if (ptr) *ptr = '\0';
93                         if (subname) free(subname);
94                         subname = strdup((char *)dbuf.mv_data+STRLENOF("database="));
95                 } else if (!strncmp(dbuf.mv_data, "type=", STRLENOF("type="))) {
96                         if (strncmp((char *)dbuf.mv_data+STRLENOF("type="), "btree", STRLENOF("btree")))  {
97                                 fprintf(stderr, "%s: line %" Z "d: unsupported type %s\n",
98                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("type="));
99                                 exit(EXIT_FAILURE);
100                         }
101                 } else if (!strncmp(dbuf.mv_data, "mapaddr=", STRLENOF("mapaddr="))) {
102                         int i;
103                         ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size);
104                         if (ptr) *ptr = '\0';
105                         i = sscanf((char *)dbuf.mv_data+STRLENOF("mapaddr="), "%p", &info.me_mapaddr);
106                         if (i != 1) {
107                                 fprintf(stderr, "%s: line %" Z "d: invalid mapaddr %s\n",
108                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("mapaddr="));
109                                 exit(EXIT_FAILURE);
110                         }
111                 } else if (!strncmp(dbuf.mv_data, "mapsize=", STRLENOF("mapsize="))) {
112                         int i;
113                         ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size);
114                         if (ptr) *ptr = '\0';
115                         i = sscanf((char *)dbuf.mv_data+STRLENOF("mapsize="), "%" Z "u", &info.me_mapsize);
116                         if (i != 1) {
117                                 fprintf(stderr, "%s: line %" Z "d: invalid mapsize %s\n",
118                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("mapsize="));
119                                 exit(EXIT_FAILURE);
120                         }
121                 } else if (!strncmp(dbuf.mv_data, "maxreaders=", STRLENOF("maxreaders="))) {
122                         int i;
123                         ptr = memchr(dbuf.mv_data, '\n', dbuf.mv_size);
124                         if (ptr) *ptr = '\0';
125                         i = sscanf((char *)dbuf.mv_data+STRLENOF("maxreaders="), "%u", &info.me_maxreaders);
126                         if (i != 1) {
127                                 fprintf(stderr, "%s: line %" Z "d: invalid maxreaders %s\n",
128                                         prog, lineno, (char *)dbuf.mv_data+STRLENOF("maxreaders="));
129                                 exit(EXIT_FAILURE);
130                         }
131                 } else {
132                         int i;
133                         for (i=0; dbflags[i].bit; i++) {
134                                 if (!strncmp(dbuf.mv_data, dbflags[i].name, dbflags[i].len) &&
135                                         ((char *)dbuf.mv_data)[dbflags[i].len] == '=') {
136                                         flags |= dbflags[i].bit;
137                                         break;
138                                 }
139                         }
140                         if (!dbflags[i].bit) {
141                                 ptr = memchr(dbuf.mv_data, '=', dbuf.mv_size);
142                                 if (!ptr) {
143                                         fprintf(stderr, "%s: line %" Z "d: unexpected format\n",
144                                                 prog, lineno);
145                                         exit(EXIT_FAILURE);
146                                 } else {
147                                         *ptr = '\0';
148                                         fprintf(stderr, "%s: line %" Z "d: unrecognized keyword ignored: %s\n",
149                                                 prog, lineno, (char *)dbuf.mv_data);
150                                 }
151                         }
152                 }
153         }
154 }
155
156 static void badend(void)
157 {
158         fprintf(stderr, "%s: line %" Z "d: unexpected end of input\n",
159                 prog, lineno);
160 }
161
162 static int unhex(unsigned char *c2)
163 {
164         int x, c;
165         x = *c2++ & 0x4f;
166         if (x & 0x40)
167                 x -= 55;
168         c = x << 4;
169         x = *c2 & 0x4f;
170         if (x & 0x40)
171                 x -= 55;
172         c |= x;
173         return c;
174 }
175
176 static int readline(MDB_val *out, MDB_val *buf)
177 {
178         unsigned char *c1, *c2, *end;
179         size_t len, l2;
180         int c;
181
182         if (!(mode & NOHDR)) {
183                 c = fgetc(stdin);
184                 if (c == EOF) {
185                         Eof = 1;
186                         return EOF;
187                 }
188                 if (c != ' ') {
189                         lineno++;
190                         if (fgets(buf->mv_data, buf->mv_size, stdin) == NULL) {
191 badend:
192                                 Eof = 1;
193                                 badend();
194                                 return EOF;
195                         }
196                         if (c == 'D' && !strncmp(buf->mv_data, "ATA=END", STRLENOF("ATA=END")))
197                                 return EOF;
198                         goto badend;
199                 }
200         }
201         if (fgets(buf->mv_data, buf->mv_size, stdin) == NULL) {
202                 Eof = 1;
203                 return EOF;
204         }
205         lineno++;
206
207         c1 = buf->mv_data;
208         len = strlen((char *)c1);
209         l2 = len;
210
211         /* Is buffer too short? */
212         while (c1[len-1] != '\n') {
213                 buf->mv_data = realloc(buf->mv_data, buf->mv_size*2);
214                 if (!buf->mv_data) {
215                         Eof = 1;
216                         fprintf(stderr, "%s: line %" Z "d: out of memory, line too long\n",
217                                 prog, lineno);
218                         return EOF;
219                 }
220                 c1 = buf->mv_data;
221                 c1 += l2;
222                 if (fgets((char *)c1, buf->mv_size+1, stdin) == NULL) {
223                         Eof = 1;
224                         badend();
225                         return EOF;
226                 }
227                 buf->mv_size *= 2;
228                 len = strlen((char *)c1);
229                 l2 += len;
230         }
231         c1 = c2 = buf->mv_data;
232         len = l2;
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 [-V] [-f input] [-n] [-s name] [-N] [-T] dbpath\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 | PRINT;
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         if (rc) {
349                 fprintf(stderr, "mdb_env_create failed, error %d %s\n", rc, mdb_strerror(rc));
350                 return EXIT_FAILURE;
351         }
352
353         mdb_env_set_maxdbs(env, 2);
354
355         if (info.me_maxreaders)
356                 mdb_env_set_maxreaders(env, info.me_maxreaders);
357
358         if (info.me_mapsize)
359                 mdb_env_set_mapsize(env, info.me_mapsize);
360
361         if (info.me_mapaddr)
362                 envflags |= MDB_FIXEDMAP;
363
364         rc = mdb_env_open(env, envname, envflags, 0664);
365         if (rc) {
366                 fprintf(stderr, "mdb_env_open failed, error %d %s\n", rc, mdb_strerror(rc));
367                 goto env_close;
368         }
369
370         kbuf.mv_size = mdb_env_get_maxkeysize(env) * 2 + 2;
371         kbuf.mv_data = malloc(kbuf.mv_size);
372
373         while(!Eof) {
374                 MDB_val key, data;
375                 int batch = 0;
376                 flags = 0;
377
378                 if (!dohdr) {
379                         dohdr = 1;
380                 } else if (!(mode & NOHDR))
381                         readhdr();
382                 
383                 rc = mdb_txn_begin(env, NULL, 0, &txn);
384                 if (rc) {
385                         fprintf(stderr, "mdb_txn_begin failed, error %d %s\n", rc, mdb_strerror(rc));
386                         goto env_close;
387                 }
388
389                 rc = mdb_open(txn, subname, flags|MDB_CREATE, &dbi);
390                 if (rc) {
391                         fprintf(stderr, "mdb_open failed, error %d %s\n", rc, mdb_strerror(rc));
392                         goto txn_abort;
393                 }
394
395                 rc = mdb_cursor_open(txn, dbi, &mc);
396                 if (rc) {
397                         fprintf(stderr, "mdb_cursor_open failed, error %d %s\n", rc, mdb_strerror(rc));
398                         goto txn_abort;
399                 }
400
401                 while(1) {
402                         rc = readline(&key, &kbuf);
403                         if (rc)  /* rc == EOF */
404                                 break;
405
406                         rc = readline(&data, &dbuf);
407                         if (rc) {
408                                 fprintf(stderr, "%s: line %" Z "d: failed to read key value\n", prog, lineno);
409                                 goto txn_abort;
410                         }
411
412                         rc = mdb_cursor_put(mc, &key, &data, putflags);
413                         if (rc == MDB_KEYEXIST && putflags)
414                                 continue;
415                         if (rc) {
416                                 fprintf(stderr, "mdb_cursor_put failed, error %d %s\n", rc, mdb_strerror(rc));
417                                 goto txn_abort;
418                         }
419                         batch++;
420                         if (batch == 100) {
421                                 rc = mdb_txn_commit(txn);
422                                 if (rc) {
423                                         fprintf(stderr, "%s: line %" Z "d: txn_commit: %s\n",
424                                                 prog, lineno, mdb_strerror(rc));
425                                         goto env_close;
426                                 }
427                                 rc = mdb_txn_begin(env, NULL, 0, &txn);
428                                 if (rc) {
429                                         fprintf(stderr, "mdb_txn_begin failed, error %d %s\n", rc, mdb_strerror(rc));
430                                         goto env_close;
431                                 }
432                                 rc = mdb_cursor_open(txn, dbi, &mc);
433                                 if (rc) {
434                                         fprintf(stderr, "mdb_cursor_open failed, error %d %s\n", rc, mdb_strerror(rc));
435                                         goto txn_abort;
436                                 }
437                                 batch = 0;
438                         }
439                 }
440                 rc = mdb_txn_commit(txn);
441                 txn = NULL;
442                 if (rc) {
443                         fprintf(stderr, "%s: line %" Z "d: txn_commit: %s\n",
444                                 prog, lineno, mdb_strerror(rc));
445                         goto env_close;
446                 }
447                 mdb_dbi_close(env, dbi);
448         }
449
450 txn_abort:
451         mdb_txn_abort(txn);
452 env_close:
453         mdb_env_close(env);
454
455         return rc ? EXIT_FAILURE : EXIT_SUCCESS;
456 }