]> git.sur5r.net Git - openldap/blob - libraries/liblmdb/mdb_load.c
e0d95e13c3f6f006888e19b55d4f35a4d0c7eb80
[openldap] / 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 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;
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
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 %" Z "d: out of memory, line too long\n",
216                                 prog, lineno);
217                         return EOF;
218                 }
219                 c1 = buf->mv_data;
220                 c1 += buf->mv_size;
221                 if (fgets((char *)c1, buf->mv_size, stdin) == NULL) {
222                         Eof = 1;
223                         badend();
224                         return EOF;
225                 }
226                 buf->mv_size *= 2;
227                 len = strlen((char *)c1);
228         }
229         c1 = c2 = buf->mv_data;
230         len = strlen((char *)c1);
231         c1[--len] = '\0';
232         end = c1 + len;
233
234         if (mode & PRINT) {
235                 while (c2 < end) {
236                         if (*c2 == '\\') {
237                                 if (c2[1] == '\\') {
238                                         c1++; c2 += 2;
239                                 } else {
240                                         if (c2+3 > end || !isxdigit(c2[1]) || !isxdigit(c2[2])) {
241                                                 Eof = 1;
242                                                 badend();
243                                                 return EOF;
244                                         }
245                                         *c1++ = unhex(++c2);
246                                         c2 += 2;
247                                 }
248                         } else {
249                                 c1++; c2++;
250                         }
251                 }
252         } else {
253                 /* odd length not allowed */
254                 if (len & 1) {
255                         Eof = 1;
256                         badend();
257                         return EOF;
258                 }
259                 while (c2 < end) {
260                         if (!isxdigit(*c2) || !isxdigit(c2[1])) {
261                                 Eof = 1;
262                                 badend();
263                                 return EOF;
264                         }
265                         *c1++ = unhex(c2);
266                         c2 += 2;
267                 }
268         }
269         c2 = out->mv_data = buf->mv_data;
270         out->mv_size = c1 - c2;
271
272         return 0;
273 }
274
275 static void usage(void)
276 {
277         fprintf(stderr, "usage: %s dbpath [-V] [-f input] [-n] [-s name] [-N] [-T]\n", prog);
278         exit(EXIT_FAILURE);
279 }
280
281 int main(int argc, char *argv[])
282 {
283         int i, rc;
284         MDB_env *env;
285         MDB_txn *txn;
286         MDB_cursor *mc;
287         MDB_dbi dbi;
288         char *envname;
289         int envflags = 0, putflags = 0;
290         int dohdr = 0;
291
292         prog = argv[0];
293
294         if (argc < 2) {
295                 usage();
296         }
297
298         /* -f: load file instead of stdin
299          * -n: use NOSUBDIR flag on env_open
300          * -s: load into named subDB
301          * -N: use NOOVERWRITE on puts
302          * -T: read plaintext
303          * -V: print version and exit
304          */
305         while ((i = getopt(argc, argv, "f:ns:NTV")) != EOF) {
306                 switch(i) {
307                 case 'V':
308                         printf("%s\n", MDB_VERSION_STRING);
309                         exit(0);
310                         break;
311                 case 'f':
312                         if (freopen(optarg, "r", stdin) == NULL) {
313                                 fprintf(stderr, "%s: %s: reopen: %s\n",
314                                         prog, optarg, strerror(errno));
315                                 exit(EXIT_FAILURE);
316                         }
317                         break;
318                 case 'n':
319                         envflags |= MDB_NOSUBDIR;
320                         break;
321                 case 's':
322                         subname = strdup(optarg);
323                         break;
324                 case 'N':
325                         putflags = MDB_NOOVERWRITE|MDB_NODUPDATA;
326                         break;
327                 case 'T':
328                         mode |= NOHDR;
329                         break;
330                 default:
331                         usage();
332                 }
333         }
334
335         if (optind != argc - 1)
336                 usage();
337
338         dbuf.mv_size = 4096;
339         dbuf.mv_data = malloc(dbuf.mv_size);
340
341         if (!(mode & NOHDR))
342                 readhdr();
343
344         envname = argv[optind];
345         rc = mdb_env_create(&env);
346         if (rc) {
347                 fprintf(stderr, "mdb_env_create failed, error %d %s\n", rc, mdb_strerror(rc));
348                 return EXIT_FAILURE;
349         }
350
351         mdb_env_set_maxdbs(env, 2);
352
353         if (info.me_maxreaders)
354                 mdb_env_set_maxreaders(env, info.me_maxreaders);
355
356         if (info.me_mapsize)
357                 mdb_env_set_mapsize(env, info.me_mapsize);
358
359         if (info.me_mapaddr)
360                 envflags |= MDB_FIXEDMAP;
361
362         rc = mdb_env_open(env, envname, envflags, 0664);
363         if (rc) {
364                 fprintf(stderr, "mdb_env_open failed, error %d %s\n", rc, mdb_strerror(rc));
365                 goto env_close;
366         }
367
368         kbuf.mv_size = mdb_env_get_maxkeysize(env) * 2 + 2;
369         kbuf.mv_data = malloc(kbuf.mv_size);
370
371         while(!Eof) {
372                 MDB_val key, data;
373                 int batch = 0;
374                 flags = 0;
375
376                 if (!dohdr) {
377                         dohdr = 1;
378                 } else if (!(mode & NOHDR))
379                         readhdr();
380                 
381                 rc = mdb_txn_begin(env, NULL, 0, &txn);
382                 if (rc) {
383                         fprintf(stderr, "mdb_txn_begin failed, error %d %s\n", rc, mdb_strerror(rc));
384                         goto env_close;
385                 }
386
387                 rc = mdb_open(txn, subname, flags|MDB_CREATE, &dbi);
388                 if (rc) {
389                         fprintf(stderr, "mdb_open failed, error %d %s\n", rc, mdb_strerror(rc));
390                         goto txn_abort;
391                 }
392
393                 rc = mdb_cursor_open(txn, dbi, &mc);
394                 if (rc) {
395                         fprintf(stderr, "mdb_cursor_open failed, error %d %s\n", rc, mdb_strerror(rc));
396                         goto txn_abort;
397                 }
398
399                 while(1) {
400                         rc = readline(&key, &kbuf);
401                         if (rc == EOF)
402                                 break;
403                         if (rc)
404                                 goto txn_abort;
405
406                         rc = readline(&data, &dbuf);
407                         if (rc)
408                                 goto txn_abort;
409                         
410                         rc = mdb_cursor_put(mc, &key, &data, putflags);
411                         if (rc == MDB_KEYEXIST && putflags)
412                                 continue;
413                         if (rc)
414                                 goto txn_abort;
415                         batch++;
416                         if (batch == 100) {
417                                 rc = mdb_txn_commit(txn);
418                                 if (rc) {
419                                         fprintf(stderr, "%s: line %" Z "d: txn_commit: %s\n",
420                                                 prog, lineno, mdb_strerror(rc));
421                                         goto env_close;
422                                 }
423                                 rc = mdb_txn_begin(env, NULL, 0, &txn);
424                                 if (rc) {
425                                         fprintf(stderr, "mdb_txn_begin failed, error %d %s\n", rc, mdb_strerror(rc));
426                                         goto env_close;
427                                 }
428                                 rc = mdb_cursor_open(txn, dbi, &mc);
429                                 if (rc) {
430                                         fprintf(stderr, "mdb_cursor_open failed, error %d %s\n", rc, mdb_strerror(rc));
431                                         goto txn_abort;
432                                 }
433                                 batch = 0;
434                         }
435                 }
436                 rc = mdb_txn_commit(txn);
437                 txn = NULL;
438                 if (rc) {
439                         fprintf(stderr, "%s: line %" Z "d: txn_commit: %s\n",
440                                 prog, lineno, mdb_strerror(rc));
441                         goto env_close;
442                 }
443                 mdb_dbi_close(env, dbi);
444         }
445
446 txn_abort:
447         mdb_txn_abort(txn);
448 env_close:
449         mdb_env_close(env);
450
451         return rc ? EXIT_FAILURE : EXIT_SUCCESS;
452 }