| 1 |
/*
|
| 2 |
* This file is a part of tinycdb package by Michael Tokarev, mjt@corpit.ru.
|
| 3 |
* Public domain.
|
| 4 |
*/
|
| 5 |
|
| 6 |
#include "common/setup_before.h"
|
| 7 |
#include <stdio.h>
|
| 8 |
#ifdef STDC_HEADERS
|
| 9 |
# include <stdlib.h>
|
| 10 |
#else
|
| 11 |
# ifdef HAVE_MALLOC_H
|
| 12 |
# include <malloc.h>
|
| 13 |
# endif
|
| 14 |
#endif
|
| 15 |
#ifdef HAVE_STRING_H
|
| 16 |
# include <string.h>
|
| 17 |
#else
|
| 18 |
# ifdef HAVE_STRINGS_H
|
| 19 |
# include <strings.h>
|
| 20 |
# endif
|
| 21 |
#endif
|
| 22 |
#include <stdarg.h>
|
| 23 |
#include "compat/getopt.h"
|
| 24 |
#include "compat/strerror.h"
|
| 25 |
#include <errno.h>
|
| 26 |
#include "common/xalloc.h"
|
| 27 |
#include "cdb.h"
|
| 28 |
#include "common/setup_after.h"
|
| 29 |
|
| 30 |
#ifndef EPROTO
|
| 31 |
# define EPROTO EINVAL
|
| 32 |
#endif
|
| 33 |
|
| 34 |
static char *progname;
|
| 35 |
|
| 36 |
#define F_DUPMASK 0x000f
|
| 37 |
#define F_WARNDUP 0x0100
|
| 38 |
#define F_ERRDUP 0x0200
|
| 39 |
#define F_MAP 0x1000 /* map format (or else CDB native format) */
|
| 40 |
|
| 41 |
static char *buf;
|
| 42 |
static unsigned blen;
|
| 43 |
|
| 44 |
static void
|
| 45 |
#ifdef __GNUC__
|
| 46 |
__attribute__((noreturn,format(printf,2,3)))
|
| 47 |
#endif
|
| 48 |
error(int errnum, const char *fmt, ...)
|
| 49 |
{
|
| 50 |
if (fmt) {
|
| 51 |
va_list ap;
|
| 52 |
fprintf(stderr, "%s: ", progname);
|
| 53 |
va_start(ap, fmt);
|
| 54 |
vfprintf(stderr, fmt, ap);
|
| 55 |
va_end(ap);
|
| 56 |
}
|
| 57 |
if (errnum)
|
| 58 |
fprintf(stderr, ": %s\n", pstrerror(errnum));
|
| 59 |
else {
|
| 60 |
if (fmt) putc('\n', stderr);
|
| 61 |
fprintf(stderr, "%s: try `%s -h' for help\n", progname, progname);
|
| 62 |
}
|
| 63 |
fflush(stderr);
|
| 64 |
exit(errnum ? 111 : 2);
|
| 65 |
}
|
| 66 |
|
| 67 |
static void allocbuf(unsigned len) {
|
| 68 |
if (blen < len) {
|
| 69 |
if (buf) buf = (char*)xrealloc(buf, len);
|
| 70 |
else buf = (char*)xmalloc(len);
|
| 71 |
if (!buf) error(ENOMEM, "unable to allocate %u bytes", len);
|
| 72 |
blen = len;
|
| 73 |
}
|
| 74 |
}
|
| 75 |
|
| 76 |
static int qmode(char *dbname, char *key, int num, int flags)
|
| 77 |
{
|
| 78 |
struct cdb c;
|
| 79 |
struct cdb_find cf;
|
| 80 |
FILE *fd;
|
| 81 |
int r;
|
| 82 |
int n, found;
|
| 83 |
|
| 84 |
fd = fopen(dbname, "rb");
|
| 85 |
if (fd == NULL || cdb_init(&c, fd) != 0)
|
| 86 |
error(errno, "unable to open database `%s'", dbname);
|
| 87 |
|
| 88 |
r = cdb_findinit(&cf, &c, key, strlen(key));
|
| 89 |
if (!r)
|
| 90 |
return 100;
|
| 91 |
else if (r < 0)
|
| 92 |
error(errno, "%s", key);
|
| 93 |
n = 0; found = 0;
|
| 94 |
while((r = cdb_findnext(&cf)) > 0) {
|
| 95 |
++n;
|
| 96 |
if (num && num != n) continue;
|
| 97 |
++found;
|
| 98 |
allocbuf(cdb_datalen(&c));
|
| 99 |
if (cdb_read(&c, buf, cdb_datalen(&c), cdb_datapos(&c)) != 0)
|
| 100 |
error(errno, "unable to read value");
|
| 101 |
fwrite(buf, 1, cdb_datalen(&c), stdout);
|
| 102 |
if (flags & F_MAP) putchar('\n');
|
| 103 |
if (num)
|
| 104 |
break;
|
| 105 |
}
|
| 106 |
if (r < 0)
|
| 107 |
error(0, "%s", key);
|
| 108 |
return found ? 0 : 100;
|
| 109 |
}
|
| 110 |
|
| 111 |
static void
|
| 112 |
fget(FILE *f, unsigned char *b, unsigned len, unsigned *posp, unsigned limit)
|
| 113 |
{
|
| 114 |
if (posp && limit - *posp < len)
|
| 115 |
error(EPROTO, "invalid database format");
|
| 116 |
if (fread(b, 1, len, f) != len) {
|
| 117 |
if (ferror(f)) error(errno, "unable to read");
|
| 118 |
fprintf(stderr, "%s: unable to read: short file\n", progname);
|
| 119 |
exit(2);
|
| 120 |
}
|
| 121 |
if (posp) *posp += len;
|
| 122 |
}
|
| 123 |
|
| 124 |
static int
|
| 125 |
fcpy(FILE *fi, FILE *fo, unsigned len, unsigned *posp, unsigned limit)
|
| 126 |
{
|
| 127 |
while(len > blen) {
|
| 128 |
fget(fi, buf, blen, posp, limit);
|
| 129 |
if (fo && fwrite(buf, 1, blen, fo) != blen) return -1;
|
| 130 |
len -= blen;
|
| 131 |
}
|
| 132 |
if (len) {
|
| 133 |
fget(fi, buf, len, posp, limit);
|
| 134 |
if (fo && fwrite(buf, 1, len, fo) != len) return -1;
|
| 135 |
}
|
| 136 |
return 0;
|
| 137 |
}
|
| 138 |
|
| 139 |
static int
|
| 140 |
dmode(char *dbname, char mode, int flags)
|
| 141 |
{
|
| 142 |
unsigned eod, klen, vlen;
|
| 143 |
unsigned pos = 0;
|
| 144 |
FILE *f;
|
| 145 |
if (strcmp(dbname, "-") == 0)
|
| 146 |
f = stdin;
|
| 147 |
else if ((f = fopen(dbname, "rb")) == NULL)
|
| 148 |
error(errno, "open %s", dbname);
|
| 149 |
allocbuf(2048);
|
| 150 |
fget(f, buf, 2048, &pos, 2048);
|
| 151 |
eod = cdb_unpack(buf);
|
| 152 |
while(pos < eod) {
|
| 153 |
fget(f, buf, 8, &pos, eod);
|
| 154 |
klen = cdb_unpack(buf);
|
| 155 |
vlen = cdb_unpack(buf + 4);
|
| 156 |
if (!(flags & F_MAP))
|
| 157 |
if (printf(mode == 'd' ? "+%u,%u:" : "+%u:", klen, vlen) < 0) return -1;
|
| 158 |
if (fcpy(f, stdout, klen, &pos, eod) != 0) return -1;
|
| 159 |
if (mode == 'd')
|
| 160 |
if (fputs(flags & F_MAP ? " " : "->", stdout) < 0)
|
| 161 |
return -1;
|
| 162 |
if (fcpy(f, mode == 'd' ? stdout : NULL, vlen, &pos, eod) != 0)
|
| 163 |
return -1;
|
| 164 |
if (putc('\n', stdout) < 0)
|
| 165 |
return -1;
|
| 166 |
}
|
| 167 |
if (pos != eod)
|
| 168 |
error(EPROTO, "invalid cdb file format");
|
| 169 |
if (!(flags & F_MAP))
|
| 170 |
if (putc('\n', stdout) < 0)
|
| 171 |
return -1;
|
| 172 |
return 0;
|
| 173 |
}
|
| 174 |
|
| 175 |
static int smode(char *dbname) {
|
| 176 |
FILE *f;
|
| 177 |
unsigned pos, eod;
|
| 178 |
unsigned cnt = 0;
|
| 179 |
unsigned kmin = 0, kmax = 0, ktot = 0;
|
| 180 |
unsigned vmin = 0, vmax = 0, vtot = 0;
|
| 181 |
unsigned hmin = 0, hmax = 0, htot = 0, hcnt = 0;
|
| 182 |
#define NDIST 11
|
| 183 |
unsigned dist[NDIST];
|
| 184 |
unsigned char toc[2048];
|
| 185 |
unsigned k;
|
| 186 |
|
| 187 |
if (strcmp(dbname, "-") == 0)
|
| 188 |
f = stdin;
|
| 189 |
else if ((f = fopen(dbname, "rb")) == NULL)
|
| 190 |
error(errno, "open %s", dbname);
|
| 191 |
|
| 192 |
pos = 0;
|
| 193 |
fget(f, toc, 2048, &pos, 2048);
|
| 194 |
|
| 195 |
allocbuf(2048);
|
| 196 |
|
| 197 |
eod = cdb_unpack(toc);
|
| 198 |
while(pos < eod) {
|
| 199 |
unsigned klen, vlen;
|
| 200 |
fget(f, buf, 8, &pos, eod);
|
| 201 |
klen = cdb_unpack(buf);
|
| 202 |
vlen = cdb_unpack(buf + 4);
|
| 203 |
fcpy(f, NULL, klen, &pos, eod);
|
| 204 |
fcpy(f, NULL, vlen, &pos, eod);
|
| 205 |
++cnt;
|
| 206 |
ktot += klen;
|
| 207 |
if (!kmin || kmin > klen) kmin = klen;
|
| 208 |
if (kmax < klen) kmax = klen;
|
| 209 |
vtot += vlen;
|
| 210 |
if (!vmin || vmin > vlen) vmin = vlen;
|
| 211 |
if (vmax < vlen) vmax = vlen;
|
| 212 |
vlen += klen;
|
| 213 |
}
|
| 214 |
if (pos != eod) error(EPROTO, "invalid cdb file format");
|
| 215 |
|
| 216 |
for (k = 0; k < NDIST; ++k)
|
| 217 |
dist[k] = 0;
|
| 218 |
for (k = 0; k < 256; ++k) {
|
| 219 |
unsigned i = cdb_unpack(toc + (k << 3));
|
| 220 |
unsigned hlen = cdb_unpack(toc + (k << 3) + 4);
|
| 221 |
if (i != pos) error(EPROTO, "invalid cdb hash table");
|
| 222 |
if (!hlen) continue;
|
| 223 |
for (i = 0; i < hlen; ++i) {
|
| 224 |
unsigned h;
|
| 225 |
fget(f, buf, 8, &pos, 0xffffffff);
|
| 226 |
if (!cdb_unpack(buf + 4)) continue;
|
| 227 |
h = (cdb_unpack(buf) >> 8) % hlen;
|
| 228 |
if (h == i) h = 0;
|
| 229 |
else {
|
| 230 |
if (h < i) h = i - h;
|
| 231 |
else h = hlen - h + i;
|
| 232 |
if (h >= NDIST) h = NDIST - 1;
|
| 233 |
}
|
| 234 |
++dist[h];
|
| 235 |
}
|
| 236 |
if (!hmin || hmin > hlen) hmin = hlen;
|
| 237 |
if (hmax < hlen) hmax = hlen;
|
| 238 |
htot += hlen;
|
| 239 |
++hcnt;
|
| 240 |
}
|
| 241 |
printf("number of records: %u\n", cnt);
|
| 242 |
printf("key min/avg/max length: %u/%u/%u\n",
|
| 243 |
kmin, cnt ? (ktot + cnt / 2) / cnt : 0, kmax);
|
| 244 |
printf("val min/avg/max length: %u/%u/%u\n",
|
| 245 |
vmin, cnt ? (vtot + cnt / 2) / cnt : 0, vmax);
|
| 246 |
printf("hash tables/entries/collisions: %u/%u/%u\n",
|
| 247 |
hcnt, htot, cnt - dist[0]);
|
| 248 |
printf("hash table min/avg/max length: %u/%u/%u\n",
|
| 249 |
hmin, hcnt ? (htot + hcnt / 2) / hcnt : 0, hmax);
|
| 250 |
printf("hash table distances:\n");
|
| 251 |
for(k = 0; k < NDIST; ++k)
|
| 252 |
printf(" %c%u: %6u %2u%%\n",
|
| 253 |
k == NDIST - 1 ? '>' : 'd', k == NDIST - 1 ? k - 1 : k,
|
| 254 |
dist[k], cnt ? dist[k] * 100 / cnt : 0);
|
| 255 |
return 0;
|
| 256 |
}
|
| 257 |
|
| 258 |
static void badinput(const char *fn) {
|
| 259 |
fprintf(stderr, "%s: %s: bad format\n", progname, fn);
|
| 260 |
exit(2);
|
| 261 |
}
|
| 262 |
|
| 263 |
static int getnum(FILE *f, unsigned *np, const char *fn) {
|
| 264 |
unsigned n;
|
| 265 |
int c = getc(f);
|
| 266 |
if (c < '0' || c > '9') badinput(fn);
|
| 267 |
n = c - '0';
|
| 268 |
while((c = getc(f)) >= '0' && c <= '9') {
|
| 269 |
c -= '0';
|
| 270 |
if (0xffffffff / 10 - c < n) badinput(fn);
|
| 271 |
n = n * 10 + c;
|
| 272 |
}
|
| 273 |
*np = n;
|
| 274 |
return c;
|
| 275 |
}
|
| 276 |
|
| 277 |
static void
|
| 278 |
addrec(struct cdb_make *cdbmp,
|
| 279 |
char *key, unsigned klen,
|
| 280 |
char *val, unsigned vlen,
|
| 281 |
int flags)
|
| 282 |
{
|
| 283 |
int r = cdb_make_put(cdbmp, key, klen, val, vlen, flags & F_DUPMASK);
|
| 284 |
if (r < 0)
|
| 285 |
error(errno, "cdb_make_put");
|
| 286 |
else if (r && (flags & F_WARNDUP)) {
|
| 287 |
fprintf(stderr, "%s: key `", progname);
|
| 288 |
fwrite(key, 1, klen, stderr);
|
| 289 |
fputs("' duplicated\n", stderr);
|
| 290 |
if (flags & F_ERRDUP)
|
| 291 |
exit(1);
|
| 292 |
}
|
| 293 |
}
|
| 294 |
|
| 295 |
static void
|
| 296 |
dofile_cdb(struct cdb_make *cdbmp, FILE *f, const char *fn, int flags)
|
| 297 |
{
|
| 298 |
unsigned klen, vlen;
|
| 299 |
int c;
|
| 300 |
while((c = getc(f)) == '+') {
|
| 301 |
if ((c = getnum(f, &klen, fn)) != ',' ||
|
| 302 |
(c = getnum(f, &vlen, fn)) != ':' ||
|
| 303 |
0xffffffff - klen < vlen)
|
| 304 |
badinput(fn);
|
| 305 |
allocbuf(klen + vlen);
|
| 306 |
fget(f, buf, klen, NULL, 0);
|
| 307 |
if (getc(f) != '-' || getc(f) != '>') badinput(fn);
|
| 308 |
fget(f, buf + klen, vlen, NULL, 0);
|
| 309 |
switch (getc(f))
|
| 310 |
{
|
| 311 |
case '\n': break;
|
| 312 |
case '\r': if (getc(f)=='\n') break;
|
| 313 |
default: badinput(fn);
|
| 314 |
}
|
| 315 |
addrec(cdbmp, buf, klen, buf + klen, vlen, flags);
|
| 316 |
}
|
| 317 |
switch (c)
|
| 318 |
{
|
| 319 |
case '\n': break;
|
| 320 |
case '\r': if (getc(f)=='\n') break;
|
| 321 |
default: badinput(fn);
|
| 322 |
}
|
| 323 |
}
|
| 324 |
|
| 325 |
static void
|
| 326 |
dofile_ln(struct cdb_make *cdbmp, FILE *f, const char *fn, int flags)
|
| 327 |
{
|
| 328 |
char *k, *v;
|
| 329 |
while(fgets(buf, blen, f) != NULL) {
|
| 330 |
unsigned l = 0;
|
| 331 |
for (;;) {
|
| 332 |
l += strlen(buf + l);
|
| 333 |
v = buf + l;
|
| 334 |
if (v > buf && v[-1] == '\n') {
|
| 335 |
v[-1] = '\0';
|
| 336 |
break;
|
| 337 |
}
|
| 338 |
if (l < blen)
|
| 339 |
allocbuf(l + 512);
|
| 340 |
if (!fgets(buf + l, blen - l, f))
|
| 341 |
break;
|
| 342 |
}
|
| 343 |
k = buf;
|
| 344 |
while(*k == ' ' || *k == '\t') ++k;
|
| 345 |
if (!*k || *k == '#')
|
| 346 |
continue;
|
| 347 |
v = k;
|
| 348 |
while(*v && *v != ' ' && *v != '\t') ++v;
|
| 349 |
if (*v) *v++ = '\0';
|
| 350 |
while(*v == ' ' || *v == '\t') ++v;
|
| 351 |
addrec(cdbmp, k, strlen(k), v, strlen(v), flags);
|
| 352 |
}
|
| 353 |
}
|
| 354 |
|
| 355 |
static void
|
| 356 |
dofile(struct cdb_make *cdbmp, FILE *f, const char *fn, int flags)
|
| 357 |
{
|
| 358 |
if (flags & F_MAP)
|
| 359 |
dofile_ln(cdbmp, f, fn, flags);
|
| 360 |
else
|
| 361 |
dofile_cdb(cdbmp, f, fn, flags);
|
| 362 |
if (ferror(f))
|
| 363 |
error(errno, "read error");
|
| 364 |
}
|
| 365 |
|
| 366 |
static int
|
| 367 |
cmode(char *dbname, char *tmpname, int argc, char **argv, int flags)
|
| 368 |
{
|
| 369 |
struct cdb_make cdb;
|
| 370 |
FILE *fd;
|
| 371 |
if (!tmpname) {
|
| 372 |
tmpname = (char*)xmalloc(strlen(dbname) + 5);
|
| 373 |
if (!tmpname)
|
| 374 |
error(ENOMEM, "unable to allocate memory");
|
| 375 |
strcat(strcpy(tmpname, dbname), ".tmp");
|
| 376 |
}
|
| 377 |
fd = fopen(tmpname, "w+b");
|
| 378 |
if (fd == 0)
|
| 379 |
error(errno, "unable to create %s", tmpname);
|
| 380 |
cdb_make_start(&cdb, fd);
|
| 381 |
allocbuf(4096);
|
| 382 |
if (argc) {
|
| 383 |
int i;
|
| 384 |
for (i = 0; i < argc; ++i) {
|
| 385 |
if (strcmp(argv[i], "-") == 0)
|
| 386 |
dofile(&cdb, stdin, "(stdin)", flags);
|
| 387 |
else {
|
| 388 |
FILE *f = fopen(argv[i], "rb");
|
| 389 |
if (!f)
|
| 390 |
error(errno, "%s", argv[i]);
|
| 391 |
dofile(&cdb, f, argv[i], flags);
|
| 392 |
fclose(f);
|
| 393 |
}
|
| 394 |
}
|
| 395 |
}
|
| 396 |
else
|
| 397 |
dofile(&cdb, stdin, "(stdin)", flags);
|
| 398 |
if (cdb_make_finish(&cdb) != 0)
|
| 399 |
error(errno, "cdb_make_finish");
|
| 400 |
fclose(fd);
|
| 401 |
if (rename(tmpname, dbname) != 0)
|
| 402 |
error(errno, "rename %s->%s", tmpname, dbname);
|
| 403 |
return 0;
|
| 404 |
}
|
| 405 |
|
| 406 |
int main(int argc, char **argv)
|
| 407 |
{
|
| 408 |
int c;
|
| 409 |
char mode = 0;
|
| 410 |
char *tmpname = NULL;
|
| 411 |
int flags = 0;
|
| 412 |
int num = 0;
|
| 413 |
int r;
|
| 414 |
extern char *optarg;
|
| 415 |
extern int optind;
|
| 416 |
|
| 417 |
if ((progname = strrchr(argv[0], '/')) != NULL)
|
| 418 |
argv[0] = ++progname;
|
| 419 |
else
|
| 420 |
progname = argv[0];
|
| 421 |
|
| 422 |
if (argc == 1)
|
| 423 |
error(0, "no arguments given");
|
| 424 |
|
| 425 |
while((c = getopt(argc, argv, "qdlcsht:n:mwrue")) != EOF)
|
| 426 |
switch(c) {
|
| 427 |
case 'q': case 'd': case 'l': case 'c': case 's':
|
| 428 |
if (mode && mode != c)
|
| 429 |
error(0, "different modes of operation requested");
|
| 430 |
mode = c;
|
| 431 |
break;
|
| 432 |
case 't': tmpname = optarg; break;
|
| 433 |
case 'w': flags |= F_WARNDUP; break;
|
| 434 |
case 'e': flags |= F_WARNDUP | F_ERRDUP; break;
|
| 435 |
case 'r': flags = (flags & ~F_DUPMASK) | CDB_PUT_REPLACE; break;
|
| 436 |
case 'u': flags = (flags & ~F_DUPMASK) | CDB_PUT_INSERT; break;
|
| 437 |
case 'm': flags |= F_MAP; break;
|
| 438 |
case 'n':
|
| 439 |
if ((num = atoi(optarg)) <= 0)
|
| 440 |
error(0, "invalid record number `%s'", optarg);
|
| 441 |
break;
|
| 442 |
case 'h':
|
| 443 |
printf("\
|
| 444 |
%s: Constant DataBase (CDB) tool. Usage is:\n\
|
| 445 |
query: %s -q [-m] [-n recno|-a] cdbfile key\n\
|
| 446 |
dump: %s -d [-m] [cdbfile|-]\n\
|
| 447 |
list: %s -l [-m] [cdbfile|-]\n\
|
| 448 |
create: %s -c [-m] [-wrue] [-t tempfile] cdbfile [infile...]\n\
|
| 449 |
stats: %s -s [cdbfile|-]\n\
|
| 450 |
help: %s -h\n\
|
| 451 |
", progname, progname, progname, progname, progname, progname, progname);
|
| 452 |
return 0;
|
| 453 |
|
| 454 |
default:
|
| 455 |
error(0, NULL);
|
| 456 |
}
|
| 457 |
|
| 458 |
argv += optind;
|
| 459 |
argc -= optind;
|
| 460 |
switch(mode) {
|
| 461 |
case 'q':
|
| 462 |
if (argc < 2) error(0, "no database or key to query specified");
|
| 463 |
if (argc > 2) error(0, "extra arguments in command line");
|
| 464 |
r = qmode(argv[0], argv[1], num, flags);
|
| 465 |
break;
|
| 466 |
case 'c':
|
| 467 |
if (!argc) error(0, "no database name specified");
|
| 468 |
if ((flags & F_WARNDUP) && !(flags & F_DUPMASK))
|
| 469 |
flags |= CDB_PUT_WARN;
|
| 470 |
r = cmode(argv[0], tmpname, argc - 1, argv + 1, flags);
|
| 471 |
break;
|
| 472 |
case 'd':
|
| 473 |
case 'l':
|
| 474 |
if (argc > 1) error(0, "extra arguments for dump/list");
|
| 475 |
r = dmode(argc ? argv[0] : "-", mode, flags);
|
| 476 |
break;
|
| 477 |
case 's':
|
| 478 |
if (argc > 1) error(0, "extra argument(s) for stats");
|
| 479 |
r = smode(argc ? argv[0] : "-");
|
| 480 |
break;
|
| 481 |
default:
|
| 482 |
error(0, "no -q, -c, -d, -l or -s option specified");
|
| 483 |
}
|
| 484 |
if (r < 0 || fflush(stdout) < 0)
|
| 485 |
error(errno, "unable to write: %d", c);
|
| 486 |
return r;
|
| 487 |
}
|
Parent Directory
| 
Revision Log