| 1 |
/* |
| 2 |
** |
| 3 |
** This program and library is free software; you can redistribute it and/or |
| 4 |
** modify it under the terms of the GNU (Library) General Public License |
| 5 |
** as published by the Free Software Foundation; either version 2 |
| 6 |
** of the License, or any later version. |
| 7 |
** |
| 8 |
** This program is distributed in the hope that it will be useful, |
| 9 |
** but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 10 |
** MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 11 |
** GNU (Library) General Public License for more details. |
| 12 |
** |
| 13 |
** You should have received a copy of the GNU (Library) General Public License |
| 14 |
** along with this program; if not, write to the Free Software |
| 15 |
** Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. |
| 16 |
** |
| 17 |
** db_berkeley_db module |
| 18 |
** |
| 19 |
** 2001-07 jose ruiz initial coding |
| 20 |
** |
| 21 |
** Tested and Built with db-3.29 (See www.sleepycat.com) |
| 22 |
** |
| 23 |
** Must be linked with the correspondant libdb.a |
| 24 |
** |
| 25 |
*/ |
| 26 |
|
| 27 |
#include "swish.h" |
| 28 |
#include "file.h" |
| 29 |
#include "error.h" |
| 30 |
#include "string.h" |
| 31 |
#include "mem.h" |
| 32 |
#include "compress.h" |
| 33 |
#include "hash.h" |
| 34 |
#include "db.h" |
| 35 |
#include "db_berkeley_db.h" |
| 36 |
|
| 37 |
|
| 38 |
/* |
| 39 |
-- init structures for this module |
| 40 |
*/ |
| 41 |
|
| 42 |
void initModule_DB_db(SWISH * sw) |
| 43 |
{ |
| 44 |
struct MOD_DB *Db; |
| 45 |
|
| 46 |
Db = (struct MOD_DB *) emalloc(sizeof(struct MOD_DB)); |
| 47 |
|
| 48 |
Db->DB_name = (char *) estrdup("gdbm"); |
| 49 |
|
| 50 |
Db->DB_Create = DB_Create_db; |
| 51 |
Db->DB_Open = DB_Open_db; |
| 52 |
Db->DB_Close = DB_Close_db; |
| 53 |
Db->DB_Remove = DB_Remove_db; |
| 54 |
|
| 55 |
Db->DB_InitWriteHeader = DB_InitWriteHeader_db; |
| 56 |
Db->DB_WriteHeaderData = DB_WriteHeaderData_db; |
| 57 |
Db->DB_EndWriteHeader = DB_EndWriteHeader_db; |
| 58 |
|
| 59 |
Db->DB_InitReadHeader = DB_InitReadHeader_db; |
| 60 |
Db->DB_ReadHeaderData = DB_ReadHeaderData_db; |
| 61 |
Db->DB_EndReadHeader = DB_EndReadHeader_db; |
| 62 |
|
| 63 |
Db->DB_InitWriteWords = DB_InitWriteWords_db; |
| 64 |
Db->DB_GetWordID = DB_GetWordID_db; |
| 65 |
Db->DB_WriteWord = DB_WriteWord_db; |
| 66 |
Db->DB_WriteWordHash = DB_WriteWordHash_db; |
| 67 |
Db->DB_WriteWordData = DB_WriteWordData_db; |
| 68 |
Db->DB_EndWriteWords = DB_EndWriteWords_db; |
| 69 |
|
| 70 |
Db->DB_InitReadWords = DB_InitReadWords_db; |
| 71 |
Db->DB_ReadWordHash = DB_ReadWordHash_db; |
| 72 |
Db->DB_ReadFirstWordInvertedIndex = DB_ReadFirstWordInvertedIndex_db; |
| 73 |
Db->DB_ReadNextWordInvertedIndex = DB_ReadNextWordInvertedIndex_db; |
| 74 |
Db->DB_ReadWordData = DB_ReadWordData_db; |
| 75 |
Db->DB_EndReadWords = DB_EndReadWords_db; |
| 76 |
|
| 77 |
Db->DB_InitWriteFiles = DB_InitWriteFiles_db; |
| 78 |
Db->DB_WriteFile = DB_WriteFile_db; |
| 79 |
Db->DB_EndWriteFiles = DB_EndWriteFiles_db; |
| 80 |
|
| 81 |
Db->DB_InitReadFiles = DB_InitReadFiles_db; |
| 82 |
Db->DB_ReadFile = DB_ReadFile_db; |
| 83 |
Db->DB_EndReadFiles = DB_EndReadFiles_db; |
| 84 |
|
| 85 |
Db->DB_InitWriteSortedIndex = DB_InitWriteSortedIndex_db; |
| 86 |
Db->DB_WriteSortedIndex = DB_WriteSortedIndex_db; |
| 87 |
Db->DB_EndWriteSortedIndex = DB_EndWriteSortedIndex_db; |
| 88 |
|
| 89 |
Db->DB_InitReadSortedIndex = DB_InitReadSortedIndex_db; |
| 90 |
Db->DB_ReadSortedIndex = DB_ReadSortedIndex_db; |
| 91 |
Db->DB_EndReadSortedIndex = DB_EndReadSortedIndex_db; |
| 92 |
|
| 93 |
|
| 94 |
#ifdef PROPFILE |
| 95 |
Db->DB_WriteProperty = DB_WriteProperty_db; |
| 96 |
Db->DB_ReadProperty = DB_ReadProperty_db; |
| 97 |
Db->DB_Reopen_PropertiesForRead = DB_Reopen_PropertiesForRead_db; |
| 98 |
#endif |
| 99 |
|
| 100 |
sw->Db = Db; |
| 101 |
|
| 102 |
return; |
| 103 |
} |
| 104 |
|
| 105 |
|
| 106 |
/* |
| 107 |
-- release all wired memory for this module |
| 108 |
*/ |
| 109 |
|
| 110 |
void freeModule_DB_db(SWISH * sw) |
| 111 |
{ |
| 112 |
efree(sw->Db->DB_name); |
| 113 |
efree(sw->Db); |
| 114 |
return; |
| 115 |
} |
| 116 |
|
| 117 |
|
| 118 |
|
| 119 |
/* ---------------------------------------------- */ |
| 120 |
|
| 121 |
|
| 122 |
|
| 123 |
/* |
| 124 |
-- Config Directives |
| 125 |
-- Configuration directives for this Module |
| 126 |
-- return: 0/1 = none/config applied |
| 127 |
*/ |
| 128 |
|
| 129 |
int configModule_DB_db(SWISH * sw, StringList * sl) |
| 130 |
{ |
| 131 |
// struct MOD_DB_db *md = sw->DB_db; |
| 132 |
// char *w0 = sl->word[0]; |
| 133 |
int retval = 1; |
| 134 |
|
| 135 |
|
| 136 |
retval = 0; // tmp due to empty routine |
| 137 |
|
| 138 |
return retval; |
| 139 |
} |
| 140 |
|
| 141 |
|
| 142 |
|
| 143 |
struct Handle_DB_db *open_db_files(char *dbname, u_int32_t flags) |
| 144 |
{ |
| 145 |
char *tmp; |
| 146 |
DB *dbp; |
| 147 |
struct Handle_DB_db *DB; |
| 148 |
int ret; |
| 149 |
|
| 150 |
/* Allocate structure */ |
| 151 |
DB = (struct Handle_DB_db *) emalloc(sizeof(struct Handle_DB_db)); |
| 152 |
|
| 153 |
DB->wordcounter = 0; |
| 154 |
|
| 155 |
tmp = emalloc(strlen(dbname) + 7 + 1); |
| 156 |
|
| 157 |
|
| 158 |
/* Create index DB */ |
| 159 |
|
| 160 |
/* db file to store header */ |
| 161 |
/* we will use a hash for the pair values */ |
| 162 |
|
| 163 |
sprintf(tmp, "%s_hdr.db", dbname); |
| 164 |
|
| 165 |
|
| 166 |
if ((ret = db_create(&dbp, NULL, 0)) != 0) |
| 167 |
{ |
| 168 |
progerr("db_berkeley_db.c error in db_create: %s\n", db_strerror(ret)); |
| 169 |
} |
| 170 |
|
| 171 |
if ((ret = dbp->open(dbp, tmp, NULL, DB_HASH, flags, 0664)) != 0) |
| 172 |
{ |
| 173 |
dbp->err(dbp, ret, "%s", tmp); |
| 174 |
progerr("db_berkeley_db.c error in db->open: %s\n", db_strerror(ret)); |
| 175 |
} |
| 176 |
DB->dbf_header = dbp; |
| 177 |
|
| 178 |
|
| 179 |
|
| 180 |
/* db file to store words */ |
| 181 |
|
| 182 |
sprintf(tmp, "%s_wds.db", dbname); |
| 183 |
if ((ret = db_create(&dbp, NULL, 0)) != 0) |
| 184 |
{ |
| 185 |
progerr("db_berkeley_db.c error in db_create: %s\n", db_strerror(ret)); |
| 186 |
} |
| 187 |
if ((ret = dbp->open(dbp, tmp, NULL, DB_HASH, flags, 0664)) != 0) |
| 188 |
{ |
| 189 |
dbp->err(dbp, ret, "%s", tmp); |
| 190 |
progerr("db_berkeley_db.c error in db->open: %s\n", db_strerror(ret)); |
| 191 |
} |
| 192 |
|
| 193 |
DB->dbf_words = dbp; |
| 194 |
|
| 195 |
|
| 196 |
|
| 197 |
/* db file to store properties */ |
| 198 |
|
| 199 |
sprintf(tmp, "%s_prp.db", dbname); |
| 200 |
if ((ret = db_create(&dbp, NULL, 0)) != 0) |
| 201 |
{ |
| 202 |
progerr("db_berkeley_db.c error in db_create: %s\n", db_strerror(ret)); |
| 203 |
} |
| 204 |
if ((ret = dbp->open(dbp, tmp, NULL, DB_HASH, flags, 0664)) != 0) |
| 205 |
{ |
| 206 |
dbp->err(dbp, ret, "%s", tmp); |
| 207 |
progerr("db_berkeley_db.c error in db->open: %s\n", db_strerror(ret)); |
| 208 |
} |
| 209 |
|
| 210 |
DB->dbf_properties = dbp; |
| 211 |
|
| 212 |
|
| 213 |
|
| 214 |
/* db file to store word's data */ |
| 215 |
|
| 216 |
sprintf(tmp, "%s_wdd.db", dbname); |
| 217 |
if ((ret = db_create(&dbp, NULL, 0)) != 0) |
| 218 |
{ |
| 219 |
progerr("db_berkeley_db.c error in db_create: %s\n", db_strerror(ret)); |
| 220 |
} |
| 221 |
if ((ret = dbp->open(dbp, tmp, NULL, DB_HASH, flags, 0664)) != 0) |
| 222 |
{ |
| 223 |
dbp->err(dbp, ret, "%s", tmp); |
| 224 |
progerr("db_berkeley_db.c error in db->open: %s\n", db_strerror(ret)); |
| 225 |
} |
| 226 |
DB->dbf_worddata = dbp; |
| 227 |
|
| 228 |
|
| 229 |
|
| 230 |
/* db file to store word's inverted index */ |
| 231 |
|
| 232 |
sprintf(tmp, "%s_wii.db", dbname); |
| 233 |
if ((ret = db_create(&dbp, NULL, 0)) != 0) |
| 234 |
{ |
| 235 |
progerr("db_berkeley_db.c error in db_create: %s\n", db_strerror(ret)); |
| 236 |
} |
| 237 |
if ((ret = dbp->open(dbp, tmp, NULL, DB_BTREE, flags, 0664)) != 0) |
| 238 |
{ |
| 239 |
dbp->err(dbp, ret, "%s", tmp); |
| 240 |
progerr("db_berkeley_db.c error in db->open: %s\n", db_strerror(ret)); |
| 241 |
} |
| 242 |
DB->dbf_invertedindex = dbp; |
| 243 |
|
| 244 |
|
| 245 |
|
| 246 |
/* db file to store docs's data */ |
| 247 |
|
| 248 |
sprintf(tmp, "%s_doc.db", dbname); |
| 249 |
if ((ret = db_create(&dbp, NULL, 0)) != 0) |
| 250 |
{ |
| 251 |
progerr("db_berkeley_db.c error in db_create: %s\n", db_strerror(ret)); |
| 252 |
} |
| 253 |
if ((ret = dbp->open(dbp, tmp, NULL, DB_HASH, flags, 0664)) != 0) |
| 254 |
{ |
| 255 |
dbp->err(dbp, ret, "%s", tmp); |
| 256 |
progerr("db_berkeley_db.c error in db->open: %s\n", db_strerror(ret)); |
| 257 |
} |
| 258 |
DB->dbf_docs = dbp; |
| 259 |
|
| 260 |
|
| 261 |
|
| 262 |
/* db file to store sorted indexes */ |
| 263 |
|
| 264 |
sprintf(tmp, "%s_srt.db", dbname); |
| 265 |
if ((ret = db_create(&dbp, NULL, 0)) != 0) |
| 266 |
{ |
| 267 |
progerr("db_berkeley_db.c error in db_create: %s\n", db_strerror(ret)); |
| 268 |
} |
| 269 |
if ((ret = dbp->open(dbp, tmp, NULL, DB_HASH, flags, 0664)) != 0) |
| 270 |
{ |
| 271 |
dbp->err(dbp, ret, "%s", tmp); |
| 272 |
progerr("db_berkeley_db.c error in db->open: %s\n", db_strerror(ret)); |
| 273 |
} |
| 274 |
DB->dbf_sorted_indexes = dbp; |
| 275 |
|
| 276 |
efree(tmp); |
| 277 |
|
| 278 |
return DB; |
| 279 |
} |
| 280 |
|
| 281 |
void *DB_Create_db(SWISH *sw, char *dbname) |
| 282 |
{ |
| 283 |
return (void *) open_db_files(dbname, DB_CREATE); |
| 284 |
} |
| 285 |
|
| 286 |
void *DB_Open_db(SWISH *sw, char *dbname) |
| 287 |
{ |
| 288 |
return (void *) open_db_files(dbname, DB_RDONLY); |
| 289 |
} |
| 290 |
|
| 291 |
void DB_Close_db(void *db) |
| 292 |
{ |
| 293 |
DB *dbp; |
| 294 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 295 |
int ret; |
| 296 |
|
| 297 |
dbp = DB->dbf_header; |
| 298 |
if ((ret = dbp->close(dbp, 0)) != 0) |
| 299 |
{ |
| 300 |
progerr("db_berkeley_db.c error in db->close\n"); |
| 301 |
} |
| 302 |
|
| 303 |
dbp = DB->dbf_words; |
| 304 |
if ((ret = dbp->close(dbp, 0)) != 0) |
| 305 |
{ |
| 306 |
progerr("db_berkeley_db.c error in db->close\n"); |
| 307 |
} |
| 308 |
|
| 309 |
dbp = DB->dbf_worddata; |
| 310 |
if ((ret = dbp->close(dbp, 0)) != 0) |
| 311 |
{ |
| 312 |
progerr("db_berkeley_db.c error in db->close\n"); |
| 313 |
} |
| 314 |
|
| 315 |
dbp = DB->dbf_invertedindex; |
| 316 |
if ((ret = dbp->close(dbp, 0)) != 0) |
| 317 |
{ |
| 318 |
progerr("db_berkeley_db.c error in db->close\n"); |
| 319 |
} |
| 320 |
|
| 321 |
dbp = DB->dbf_docs; |
| 322 |
if ((ret = dbp->close(dbp, 0)) != 0) |
| 323 |
{ |
| 324 |
progerr("db_berkeley_db.c error in db->close\n"); |
| 325 |
} |
| 326 |
|
| 327 |
dbp = DB->dbf_sorted_indexes; |
| 328 |
if ((ret = dbp->close(dbp, 0)) != 0) |
| 329 |
{ |
| 330 |
progerr("db_berkeley_db.c error in db->close\n"); |
| 331 |
} |
| 332 |
|
| 333 |
|
| 334 |
dbp = DB->dbf_properties; |
| 335 |
if ((ret = dbp->close(dbp, 0)) != 0) |
| 336 |
{ |
| 337 |
progerr("db_berkeley_db.c error in db->close\n"); |
| 338 |
} |
| 339 |
|
| 340 |
|
| 341 |
efree(DB); |
| 342 |
} |
| 343 |
|
| 344 |
void DB_Remove_db(void *db) |
| 345 |
{ |
| 346 |
DB *dbp; |
| 347 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 348 |
|
| 349 |
dbp = DB->dbf_header; |
| 350 |
dbp = DB->dbf_words; |
| 351 |
dbp = DB->dbf_worddata; |
| 352 |
dbp = DB->dbf_invertedindex; |
| 353 |
dbp = DB->dbf_docs; |
| 354 |
dbp = DB->dbf_sorted_indexes; |
| 355 |
dbp = DB->dbf_properties; |
| 356 |
|
| 357 |
efree(DB); |
| 358 |
} |
| 359 |
|
| 360 |
|
| 361 |
/*--------------------------------------------*/ |
| 362 |
/*--------------------------------------------*/ |
| 363 |
/* Header stuff */ |
| 364 |
/*--------------------------------------------*/ |
| 365 |
/*--------------------------------------------*/ |
| 366 |
|
| 367 |
int DB_InitWriteHeader_db(void *db) |
| 368 |
{ |
| 369 |
return 0; |
| 370 |
} |
| 371 |
|
| 372 |
|
| 373 |
int DB_EndWriteHeader_db(void *db) |
| 374 |
{ |
| 375 |
return 0; |
| 376 |
} |
| 377 |
|
| 378 |
int DB_WriteHeaderData_db(int id, unsigned char *s, int len, void *db) |
| 379 |
{ |
| 380 |
DB *dbp; |
| 381 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 382 |
|
| 383 |
DBT key, |
| 384 |
content; |
| 385 |
int ret; |
| 386 |
|
| 387 |
memset(&key, 0, sizeof(key)); |
| 388 |
memset(&content, 0, sizeof(content)); |
| 389 |
|
| 390 |
key.size = sizeof(int); |
| 391 |
|
| 392 |
key.data = (char *) &id; |
| 393 |
|
| 394 |
content.size = len; |
| 395 |
content.data = s; |
| 396 |
|
| 397 |
dbp = DB->dbf_header; |
| 398 |
ret = dbp->put(dbp, NULL, &key, &content, 0); |
| 399 |
|
| 400 |
if (ret) |
| 401 |
progerr("Berkeley DB error writing header"); |
| 402 |
|
| 403 |
return 0; |
| 404 |
} |
| 405 |
|
| 406 |
|
| 407 |
int DB_InitReadHeader_db(void *db) |
| 408 |
{ |
| 409 |
DB *dbp; |
| 410 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 411 |
|
| 412 |
int ret; |
| 413 |
|
| 414 |
dbp = DB->dbf_header; |
| 415 |
|
| 416 |
/* Acquire a cursor for the database. */ |
| 417 |
if ((ret = dbp->cursor(dbp, NULL, &DB->cursor_header, 0)) != 0) |
| 418 |
{ |
| 419 |
DB->cursor_header = NULL; |
| 420 |
} |
| 421 |
|
| 422 |
return 0; |
| 423 |
} |
| 424 |
|
| 425 |
int DB_ReadHeaderData_db(int *id, unsigned char **s, int *len, void *db) |
| 426 |
{ |
| 427 |
DB *dbp; |
| 428 |
DBC *dbcp; |
| 429 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 430 |
DBT key, |
| 431 |
content; |
| 432 |
int ret; |
| 433 |
|
| 434 |
dbp = DB->dbf_header; |
| 435 |
dbcp = DB->cursor_header; |
| 436 |
|
| 437 |
|
| 438 |
if (!dbcp) |
| 439 |
{ |
| 440 |
*id = 0; |
| 441 |
*len = 0; |
| 442 |
*s = NULL; |
| 443 |
} |
| 444 |
else |
| 445 |
{ |
| 446 |
/* Initialize the key/data pair so the flags aren't set. */ |
| 447 |
memset(&key, 0, sizeof(key)); |
| 448 |
memset(&content, 0, sizeof(content)); |
| 449 |
|
| 450 |
ret = dbcp->c_get(dbcp, &key, &content, DB_NEXT); |
| 451 |
if (!ret) |
| 452 |
{ |
| 453 |
*id = *(int *) key.data; |
| 454 |
*len = content.size; |
| 455 |
*s = emalloc(content.size); |
| 456 |
memcpy(*s, content.data, content.size); |
| 457 |
//*s = content.data; |
| 458 |
} |
| 459 |
else |
| 460 |
{ |
| 461 |
*id = 0; |
| 462 |
*len = 0; |
| 463 |
*s = NULL; |
| 464 |
} |
| 465 |
} |
| 466 |
return 0; |
| 467 |
} |
| 468 |
|
| 469 |
|
| 470 |
int DB_EndReadHeader_db(void *db) |
| 471 |
{ |
| 472 |
DB *dbp; |
| 473 |
DBC *dbcp; |
| 474 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 475 |
|
| 476 |
dbp = DB->dbf_header; |
| 477 |
dbcp = DB->cursor_header; |
| 478 |
|
| 479 |
if (dbcp) |
| 480 |
dbcp->c_close(dbcp); |
| 481 |
|
| 482 |
return 0; |
| 483 |
} |
| 484 |
|
| 485 |
/*--------------------------------------------*/ |
| 486 |
/*--------------------------------------------*/ |
| 487 |
/* Word Stuff */ |
| 488 |
/*--------------------------------------------*/ |
| 489 |
/*--------------------------------------------*/ |
| 490 |
|
| 491 |
int DB_InitWriteWords_db(void *db) |
| 492 |
{ |
| 493 |
return 0; |
| 494 |
} |
| 495 |
|
| 496 |
|
| 497 |
int DB_EndWriteWords_db(void *db) |
| 498 |
{ |
| 499 |
return 0; |
| 500 |
} |
| 501 |
|
| 502 |
long DB_GetWordID_db(void *db) |
| 503 |
{ |
| 504 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 505 |
|
| 506 |
return (long) ++(DB->wordcounter); |
| 507 |
} |
| 508 |
|
| 509 |
int DB_WriteWord_db(char *word, long wordID, void *db) |
| 510 |
{ |
| 511 |
DB *dbp; |
| 512 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 513 |
|
| 514 |
DBT key, |
| 515 |
content; |
| 516 |
int ret; |
| 517 |
|
| 518 |
dbp = DB->dbf_invertedindex; |
| 519 |
|
| 520 |
memset(&key, 0, sizeof(key)); |
| 521 |
memset(&content, 0, sizeof(content)); |
| 522 |
|
| 523 |
key.size = strlen(word) + 1; |
| 524 |
key.data = (char *) word; |
| 525 |
|
| 526 |
content.size = sizeof(int); |
| 527 |
|
| 528 |
content.data = (char *) &wordID; |
| 529 |
|
| 530 |
ret = dbp->put(dbp, NULL, &key, &content, 0); |
| 531 |
|
| 532 |
if (ret) |
| 533 |
progerr("DB error writing word"); |
| 534 |
|
| 535 |
return 0; |
| 536 |
} |
| 537 |
|
| 538 |
|
| 539 |
long DB_WriteWordData_db(long wordID, unsigned char *worddata, int lendata, void *db) |
| 540 |
{ |
| 541 |
DB *dbp; |
| 542 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 543 |
|
| 544 |
DBT key, |
| 545 |
content; |
| 546 |
int ret; |
| 547 |
|
| 548 |
dbp = DB->dbf_worddata; |
| 549 |
|
| 550 |
memset(&key, 0, sizeof(key)); |
| 551 |
memset(&content, 0, sizeof(content)); |
| 552 |
|
| 553 |
key.size = sizeof(long); |
| 554 |
|
| 555 |
key.data = (char *) &wordID; |
| 556 |
|
| 557 |
content.size = lendata; |
| 558 |
content.data = worddata; |
| 559 |
|
| 560 |
ret = dbp->put(dbp, NULL, &key, &content, 0); |
| 561 |
|
| 562 |
if (ret) |
| 563 |
progerr("DB error writing word's data"); |
| 564 |
|
| 565 |
return 0; |
| 566 |
} |
| 567 |
|
| 568 |
|
| 569 |
|
| 570 |
int DB_WriteWordHash_db(char *word, long wordID, void *db) |
| 571 |
{ |
| 572 |
DB *dbp; |
| 573 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 574 |
|
| 575 |
DBT key, |
| 576 |
content; |
| 577 |
int ret; |
| 578 |
|
| 579 |
dbp = DB->dbf_words; |
| 580 |
|
| 581 |
memset(&key, 0, sizeof(key)); |
| 582 |
memset(&content, 0, sizeof(content)); |
| 583 |
|
| 584 |
key.size = strlen(word) + 1; |
| 585 |
key.data = (char *) word; |
| 586 |
|
| 587 |
content.size = sizeof(int); |
| 588 |
|
| 589 |
content.data = (char *) &wordID; |
| 590 |
|
| 591 |
ret = dbp->put(dbp, NULL, &key, &content, 0); |
| 592 |
|
| 593 |
if (ret) |
| 594 |
progerr("DB error writing word"); |
| 595 |
|
| 596 |
return 0; |
| 597 |
} |
| 598 |
|
| 599 |
int DB_InitReadWords_db(void *db) |
| 600 |
{ |
| 601 |
return 0; |
| 602 |
} |
| 603 |
|
| 604 |
int DB_EndReadWords_db(void *db) |
| 605 |
{ |
| 606 |
return 0; |
| 607 |
} |
| 608 |
|
| 609 |
|
| 610 |
int DB_ReadWordHash_db(char *word, long *wordID, void *db) |
| 611 |
{ |
| 612 |
DB *dbp; |
| 613 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 614 |
|
| 615 |
DBT key, |
| 616 |
content; |
| 617 |
int ret; |
| 618 |
|
| 619 |
dbp = DB->dbf_words; |
| 620 |
|
| 621 |
memset(&key, 0, sizeof(key)); |
| 622 |
memset(&content, 0, sizeof(content)); |
| 623 |
|
| 624 |
key.size = strlen(word) + 1; |
| 625 |
key.data = (char *) word; |
| 626 |
|
| 627 |
ret = dbp->get(dbp, NULL, &key, &content, 0); |
| 628 |
|
| 629 |
if (ret) |
| 630 |
{ |
| 631 |
*wordID = 0; |
| 632 |
} |
| 633 |
else |
| 634 |
{ |
| 635 |
*wordID = (long) *(int *) content.data; |
| 636 |
} |
| 637 |
|
| 638 |
return 0; |
| 639 |
} |
| 640 |
|
| 641 |
|
| 642 |
|
| 643 |
int DB_ReadFirstWordInvertedIndex_db(char *word, char **resultword, long *wordID, void *db) |
| 644 |
{ |
| 645 |
DB *dbp; |
| 646 |
DBC *dbcp; |
| 647 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 648 |
DBT key, |
| 649 |
content; |
| 650 |
int ret; |
| 651 |
int len; |
| 652 |
|
| 653 |
len = strlen(word); |
| 654 |
|
| 655 |
dbp = DB->dbf_invertedindex; |
| 656 |
dbcp = NULL; |
| 657 |
|
| 658 |
DB->cursor_inverted = NULL; |
| 659 |
|
| 660 |
/* Acquire a cursor for the database. */ |
| 661 |
if ((ret = dbp->cursor(dbp, NULL, &DB->cursor_inverted, 0)) != 0) |
| 662 |
{ |
| 663 |
DB->cursor_inverted = NULL; |
| 664 |
} |
| 665 |
dbcp = DB->cursor_inverted; |
| 666 |
|
| 667 |
/* Initialize the key/data pair so the flags aren't set. */ |
| 668 |
memset(&key, 0, sizeof(key)); |
| 669 |
memset(&content, 0, sizeof(content)); |
| 670 |
|
| 671 |
key.size = strlen(word); |
| 672 |
key.data = word; |
| 673 |
|
| 674 |
ret = dbcp->c_get(dbcp, &key, &content, DB_SET_RANGE); |
| 675 |
if (!ret) |
| 676 |
{ |
| 677 |
if (strncmp(word, key.data, len) == 0) |
| 678 |
{ |
| 679 |
*resultword = emalloc(key.size + 1); |
| 680 |
memcpy(*resultword, key.data, key.size); |
| 681 |
(*resultword)[key.size] = '\0'; |
| 682 |
*wordID = *(long *) content.data; |
| 683 |
return 0; |
| 684 |
} |
| 685 |
} |
| 686 |
*resultword = NULL; |
| 687 |
*wordID = 0; |
| 688 |
dbcp->c_close(dbcp); |
| 689 |
DB->cursor_inverted = NULL; |
| 690 |
|
| 691 |
return 0; |
| 692 |
} |
| 693 |
|
| 694 |
int DB_ReadNextWordInvertedIndex_db(char *word, char **resultword, long *wordID, void *db) |
| 695 |
{ |
| 696 |
DB *dbp; |
| 697 |
DBC *dbcp; |
| 698 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 699 |
DBT key, |
| 700 |
content; |
| 701 |
int ret; |
| 702 |
int len; |
| 703 |
|
| 704 |
len = strlen(word); |
| 705 |
|
| 706 |
dbp = DB->dbf_invertedindex; |
| 707 |
dbcp = DB->cursor_inverted; |
| 708 |
|
| 709 |
|
| 710 |
if (!dbcp) |
| 711 |
{ |
| 712 |
*resultword = NULL; |
| 713 |
*wordID = 0; |
| 714 |
} |
| 715 |
else |
| 716 |
{ |
| 717 |
/* Initialize the key/data pair so the flags aren't set. */ |
| 718 |
memset(&key, 0, sizeof(key)); |
| 719 |
memset(&content, 0, sizeof(content)); |
| 720 |
|
| 721 |
ret = dbcp->c_get(dbcp, &key, &content, DB_NEXT); |
| 722 |
if (!ret) |
| 723 |
{ |
| 724 |
if (strncmp(word, key.data, len) == 0) |
| 725 |
{ |
| 726 |
|
| 727 |
*resultword = emalloc(key.size + 1); |
| 728 |
memcpy(*resultword, key.data, key.size); |
| 729 |
(*resultword)[key.size] = '\0'; |
| 730 |
*wordID = *(long *) content.data; |
| 731 |
return 0; |
| 732 |
} |
| 733 |
} |
| 734 |
*resultword = NULL; |
| 735 |
*wordID = 0; |
| 736 |
dbcp->c_close(dbcp); |
| 737 |
DB->cursor_inverted = NULL; |
| 738 |
} |
| 739 |
return 0; |
| 740 |
} |
| 741 |
|
| 742 |
|
| 743 |
long DB_ReadWordData_db(long wordID, unsigned char **worddata, int *lendata, void *db) |
| 744 |
{ |
| 745 |
DB *dbp; |
| 746 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 747 |
|
| 748 |
DBT key, |
| 749 |
content; |
| 750 |
int ret; |
| 751 |
|
| 752 |
dbp = DB->dbf_worddata; |
| 753 |
|
| 754 |
memset(&key, 0, sizeof(key)); |
| 755 |
memset(&content, 0, sizeof(content)); |
| 756 |
|
| 757 |
key.size = sizeof(long); |
| 758 |
|
| 759 |
key.data = (char *) &wordID; |
| 760 |
|
| 761 |
ret = dbp->get(dbp, NULL, &key, &content, 0); |
| 762 |
|
| 763 |
if (ret) |
| 764 |
{ |
| 765 |
*lendata = 0; |
| 766 |
*worddata = NULL; |
| 767 |
} |
| 768 |
else |
| 769 |
{ |
| 770 |
*lendata = content.size; |
| 771 |
*worddata = emalloc(content.size); |
| 772 |
memcpy(*worddata, content.data, content.size); |
| 773 |
//*worddata = content.data; |
| 774 |
} |
| 775 |
|
| 776 |
return 0; |
| 777 |
} |
| 778 |
|
| 779 |
/*--------------------------------------------*/ |
| 780 |
/*--------------------------------------------*/ |
| 781 |
/* FileList Stuff */ |
| 782 |
/*--------------------------------------------*/ |
| 783 |
/*--------------------------------------------*/ |
| 784 |
|
| 785 |
int DB_InitWriteFiles_db(void *db) |
| 786 |
{ |
| 787 |
return 0; |
| 788 |
} |
| 789 |
|
| 790 |
|
| 791 |
int DB_EndWriteFiles_db(void *db) |
| 792 |
{ |
| 793 |
return 0; |
| 794 |
} |
| 795 |
|
| 796 |
int DB_WriteFile_db(int filenum, unsigned char *filedata, int sz_filedata, void *db) |
| 797 |
{ |
| 798 |
DB *dbp; |
| 799 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 800 |
|
| 801 |
DBT key, |
| 802 |
content; |
| 803 |
int ret; |
| 804 |
|
| 805 |
dbp = DB->dbf_docs; |
| 806 |
|
| 807 |
memset(&key, 0, sizeof(key)); |
| 808 |
memset(&content, 0, sizeof(content)); |
| 809 |
|
| 810 |
key.size = sizeof(int); |
| 811 |
|
| 812 |
key.data = (char *) &filenum; |
| 813 |
|
| 814 |
content.size = sz_filedata; |
| 815 |
content.data = (char *) filedata; |
| 816 |
|
| 817 |
ret = dbp->put(dbp, NULL, &key, &content, 0); |
| 818 |
|
| 819 |
if (ret) |
| 820 |
progerr("DB error writing docs"); |
| 821 |
|
| 822 |
return 0; |
| 823 |
} |
| 824 |
|
| 825 |
int DB_InitReadFiles_db(void *db) |
| 826 |
{ |
| 827 |
return 0; |
| 828 |
} |
| 829 |
|
| 830 |
int DB_ReadFile_db(int filenum, unsigned char **filedata, int *sz_filedata, void *db) |
| 831 |
{ |
| 832 |
DB *dbp; |
| 833 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 834 |
|
| 835 |
DBT key, |
| 836 |
content; |
| 837 |
int ret; |
| 838 |
|
| 839 |
dbp = DB->dbf_docs; |
| 840 |
|
| 841 |
memset(&key, 0, sizeof(key)); |
| 842 |
memset(&content, 0, sizeof(content)); |
| 843 |
|
| 844 |
// This needs to be fixed --> The old compress problem with zeroes |
| 845 |
filenum--; |
| 846 |
|
| 847 |
key.size = sizeof(int); |
| 848 |
|
| 849 |
key.data = (char *) &filenum; |
| 850 |
|
| 851 |
ret = dbp->get(dbp, NULL, &key, &content, 0); |
| 852 |
|
| 853 |
if (ret) |
| 854 |
{ |
| 855 |
*sz_filedata = 0; |
| 856 |
*filedata = NULL; |
| 857 |
} |
| 858 |
else |
| 859 |
{ |
| 860 |
*sz_filedata = content.size; |
| 861 |
*filedata = emalloc(content.size); |
| 862 |
memcpy(*filedata, content.data, content.size); |
| 863 |
//*filedata = content.data; |
| 864 |
} |
| 865 |
return 0; |
| 866 |
} |
| 867 |
|
| 868 |
|
| 869 |
int DB_EndReadFiles_db(void *db) |
| 870 |
{ |
| 871 |
return 0; |
| 872 |
} |
| 873 |
|
| 874 |
|
| 875 |
/*--------------------------------------------*/ |
| 876 |
/*--------------------------------------------*/ |
| 877 |
/* Sorted data Stuff */ |
| 878 |
/*--------------------------------------------*/ |
| 879 |
/*--------------------------------------------*/ |
| 880 |
|
| 881 |
|
| 882 |
|
| 883 |
|
| 884 |
|
| 885 |
int DB_InitWriteSortedIndex_db(void *db) |
| 886 |
{ |
| 887 |
return 0; |
| 888 |
} |
| 889 |
|
| 890 |
int DB_WriteSortedIndex_db(int propID, unsigned char *data, int sz_data, void *db) |
| 891 |
{ |
| 892 |
DB *dbp; |
| 893 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 894 |
|
| 895 |
DBT key, |
| 896 |
content; |
| 897 |
int ret; |
| 898 |
|
| 899 |
dbp = DB->dbf_sorted_indexes; |
| 900 |
|
| 901 |
memset(&key, 0, sizeof(key)); |
| 902 |
memset(&content, 0, sizeof(content)); |
| 903 |
|
| 904 |
key.size = sizeof(int); |
| 905 |
|
| 906 |
key.data = (char *) &propID; |
| 907 |
|
| 908 |
content.size = sz_data; |
| 909 |
content.data = (char *) data; |
| 910 |
|
| 911 |
ret = dbp->put(dbp, NULL, &key, &content, 0); |
| 912 |
|
| 913 |
if (ret) |
| 914 |
progerr("DB error writing sorted indexes"); |
| 915 |
|
| 916 |
return 0; |
| 917 |
} |
| 918 |
|
| 919 |
int DB_EndWriteSortedIndex_db(void *db) |
| 920 |
{ |
| 921 |
return 0; |
| 922 |
} |
| 923 |
|
| 924 |
|
| 925 |
int DB_InitReadSortedIndex_db(void *db) |
| 926 |
{ |
| 927 |
return 0; |
| 928 |
} |
| 929 |
|
| 930 |
int DB_ReadSortedIndex_db(int propID, unsigned char **data, int *sz_data, void *db) |
| 931 |
{ |
| 932 |
DB *dbp; |
| 933 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 934 |
|
| 935 |
DBT key, |
| 936 |
content; |
| 937 |
int ret; |
| 938 |
|
| 939 |
dbp = DB->dbf_sorted_indexes; |
| 940 |
|
| 941 |
memset(&key, 0, sizeof(key)); |
| 942 |
memset(&content, 0, sizeof(content)); |
| 943 |
|
| 944 |
key.size = sizeof(int); |
| 945 |
|
| 946 |
key.data = (char *) &propID; |
| 947 |
|
| 948 |
ret = dbp->get(dbp, NULL, &key, &content, 0); |
| 949 |
|
| 950 |
if (ret) |
| 951 |
{ |
| 952 |
*sz_data = 0; |
| 953 |
*data = NULL; |
| 954 |
} |
| 955 |
else |
| 956 |
{ |
| 957 |
*sz_data = content.size; |
| 958 |
*data = emalloc(content.size); |
| 959 |
memcpy(*data, content.data, content.size); |
| 960 |
//*data = content.data; |
| 961 |
} |
| 962 |
|
| 963 |
return 0; |
| 964 |
} |
| 965 |
|
| 966 |
int DB_EndReadSortedIndex_db(void *db) |
| 967 |
{ |
| 968 |
return 0; |
| 969 |
} |
| 970 |
|
| 971 |
|
| 972 |
#ifdef PROPFILE |
| 973 |
/*--------------------------------------------*/ |
| 974 |
/*--------------------------------------------*/ |
| 975 |
/* Properties */ |
| 976 |
/*--------------------------------------------*/ |
| 977 |
/*--------------------------------------------*/ |
| 978 |
|
| 979 |
|
| 980 |
typedef union |
| 981 |
{ |
| 982 |
char skey; |
| 983 |
struct { |
| 984 |
int filenum; |
| 985 |
int propID; |
| 986 |
} key; |
| 987 |
} PropKEY; |
| 988 |
|
| 989 |
|
| 990 |
void DB_WriteProperty_db( FileRec *fi, int propID, char *buffer, int datalen, void *db ) |
| 991 |
{ |
| 992 |
DB *dbp; |
| 993 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 994 |
DBT key, |
| 995 |
content; |
| 996 |
PropKEY propkey; |
| 997 |
int ret; |
| 998 |
|
| 999 |
|
| 1000 |
/*** Just until I can get a flag to see if this data is needed ***/ |
| 1001 |
/*** create empty space using arrays! ***/ |
| 1002 |
|
| 1003 |
/* Create places to store the seek positions and lengths if first time */ |
| 1004 |
if ( !fi->propSize ) |
| 1005 |
{ |
| 1006 |
int i; |
| 1007 |
|
| 1008 |
fi->propLocationsCount = fi->docProperties->n; |
| 1009 |
fi->propLocations = (long *) emalloc( fi->propLocationsCount * sizeof( int *) ); |
| 1010 |
fi->propSize = (long *) emalloc( fi->propLocationsCount * sizeof( int *) ); |
| 1011 |
|
| 1012 |
/* Zero array */ |
| 1013 |
for( i = 0; i < fi->propLocationsCount; i++ ) |
| 1014 |
{ |
| 1015 |
fi->propSize[ i ] = 0; // here's the flag! |
| 1016 |
fi->propLocations[ i ] = 0; // not here! |
| 1017 |
} |
| 1018 |
} |
| 1019 |
|
| 1020 |
|
| 1021 |
|
| 1022 |
dbp = DB->dbf_properties; |
| 1023 |
|
| 1024 |
memset(&key, 0, sizeof(key)); |
| 1025 |
memset(&content, 0, sizeof(content)); |
| 1026 |
|
| 1027 |
propkey.key.filenum = fi->filenum; |
| 1028 |
propkey.key.propID = propID; |
| 1029 |
key.size = sizeof(propkey); |
| 1030 |
key.data = (char *) &propkey; |
| 1031 |
|
| 1032 |
content.size = datalen; |
| 1033 |
content.data = (char *) buffer; |
| 1034 |
|
| 1035 |
ret = dbp->put(dbp, NULL, &key, &content, 0); |
| 1036 |
|
| 1037 |
/* error message? */ |
| 1038 |
if (ret) |
| 1039 |
progerr("DB error writing properties filenum: %d property %d", fi->filenum, propID); |
| 1040 |
} |
| 1041 |
|
| 1042 |
|
| 1043 |
char * DB_ReadProperty_db( FileRec *fi, int propID, void *db ) |
| 1044 |
{ |
| 1045 |
DB *dbp; |
| 1046 |
struct Handle_DB_db *DB = (struct Handle_DB_db *) db; |
| 1047 |
DBT key, |
| 1048 |
content; |
| 1049 |
PropKEY propkey; |
| 1050 |
char *buffer; |
| 1051 |
int ret; |
| 1052 |
|
| 1053 |
|
| 1054 |
dbp = DB->dbf_properties; |
| 1055 |
|
| 1056 |
memset(&key, 0, sizeof(key)); |
| 1057 |
memset(&content, 0, sizeof(content)); |
| 1058 |
|
| 1059 |
propkey.key.filenum = fi->filenum; |
| 1060 |
propkey.key.propID = propID; |
| 1061 |
key.size = sizeof(propkey); |
| 1062 |
key.data = (char *) &propkey; |
| 1063 |
|
| 1064 |
ret = dbp->get(dbp, NULL, &key, &content, 0); |
| 1065 |
|
| 1066 |
/* No property for this file:id? */ |
| 1067 |
if (ret) |
| 1068 |
return NULL; |
| 1069 |
|
| 1070 |
buffer = (char *)emalloc(content.size); |
| 1071 |
memcpy(buffer, content.data, content.size); |
| 1072 |
|
| 1073 |
return buffer; |
| 1074 |
} |
| 1075 |
|
| 1076 |
|
| 1077 |
|
| 1078 |
|
| 1079 |
void DB_Reopen_PropertiesForRead_db(void *db) |
| 1080 |
{ |
| 1081 |
return; |
| 1082 |
} |
| 1083 |
#endif |
Parent Directory
| 
Revision Log
| 
Revision Graph