citadel/modules/fulltext/serv_fulltext.c

   1 /*
   2  * $Id$
   3  *
   4  * This module handles fulltext indexing of the message base.
   5  *
   6  *
   7  * Copyright (c) 2005-2009 by the citadel.org team
   8  *
   9  *  This program is free software; you can redistribute it and/or modify
  10  *  it under the terms of the GNU General Public License as published by
  11  *  the Free Software Foundation; either version 3 of the License, or
  12  *  (at your option) any later version.
  13  *
  14  *  This program is distributed in the hope that it will be useful,
  15  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  *  GNU General Public License for more details.
  18  *
  19  *  You should have received a copy of the GNU General Public License
  20  *  along with this program; if not, write to the Free Software
  21  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  22  */
  23
  24
  25 #include "sysdep.h"
  26 #include <stdlib.h>
  27 #include <unistd.h>
  28 #include <stdio.h>
  29 #include <fcntl.h>
  30 #include <signal.h>
  31 #include <pwd.h>
  32 #include <errno.h>
  33 #include <sys/types.h>
  34
  35 #if TIME_WITH_SYS_TIME
  36 # include <sys/time.h>
  37 # include <time.h>
  38 #else
  39 # if HAVE_SYS_TIME_H
  40 #  include <sys/time.h>
  41 # else
  42 #  include <time.h>
  43 # endif
  44 #endif
  45
  46 #include <sys/wait.h>
  47 #include <string.h>
  48 #include <limits.h>
  49 #include <libcitadel.h>
  50 #include "citadel.h"
  51 #include "server.h"
  52 #include "citserver.h"
  53 #include "support.h"
  54 #include "config.h"
  55 #include "database.h"
  56 #include "msgbase.h"
  57 #include "control.h"
  58 #include "room_ops.h"
  59 #include "serv_fulltext.h"
  60 #include "ft_wordbreaker.h"
  61 #include "threads.h"
  62
  63 #include "ctdl_module.h"
  64
  65
  66
  67 long ft_newhighest = 0L;
  68 long *ft_newmsgs = NULL;
  69 int ft_num_msgs = 0;
  70 int ft_num_alloc = 0;
  71
  72
  73 int ftc_num_msgs[65536];
  74 long *ftc_msgs[65536];
  75
  76
  77 /*
  78  * Compare function
  79  */
  80 int longcmp(const void *rec1, const void *rec2) {
  81         long i1, i2;
  82
  83         i1 = *(const long *)rec1;
  84         i2 = *(const long *)rec2;
  85
  86         if (i1 > i2) return(1);
  87         if (i1 < i2) return(-1);
  88         return(0);
  89 }
  90
  91 /*
  92  * Flush our index cache out to disk.
  93  */
  94 void ft_flush_cache(void) {
  95         int i;
  96         time_t last_update = 0;
  97
  98         for (i=0; i<65536; ++i) {
  99                 if ((time(NULL) - last_update) >= 10) {
 100                         CtdlLogPrintf(CTDL_INFO,
 101                                 "Flushing index cache to disk (%d%% complete)\n",
 102                                 (i * 100 / 65536)
 103                         );
 104                         last_update = time(NULL);
 105                 }
 106                 if (ftc_msgs[i] != NULL) {
 107                         cdb_store(CDB_FULLTEXT, &i, sizeof(int), ftc_msgs[i],
 108                                 (ftc_num_msgs[i] * sizeof(long)));
 109                         ftc_num_msgs[i] = 0;
 110                         free(ftc_msgs[i]);
 111                         ftc_msgs[i] = NULL;
 112                 }
 113         }
 114         CtdlLogPrintf(CTDL_INFO, "Flushed index cache to disk (100%% complete)\n");
 115 }
 116
 117
 118 /*
 119  * Index or de-index a message.  (op == 1 to index, 0 to de-index)
 120  */
 121 void ft_index_message(long msgnum, int op) {
 122         int num_tokens = 0;
 123         int *tokens = NULL;
 124         int i, j;
 125         struct cdbdata *cdb_bucket;
 126         char *msgtext;
 127         int tok;
 128
 129         CtdlLogPrintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
 130                 (op ? "adding" : "removing") , msgnum
 131         );
 132
 133         /* Output the message as text before indexing it, so we don't end up
 134          * indexing a bunch of encoded base64, etc.
 135          */
 136         CC->redirect_buffer = malloc(SIZ);
 137         CC->redirect_len = 0;
 138         CC->redirect_alloc = SIZ;
 139         CtdlOutputMsg(msgnum, MT_CITADEL, HEADERS_ALL, 0, 1, NULL, 0);
 140         msgtext = CC->redirect_buffer;
 141         CC->redirect_buffer = NULL;
 142         CC->redirect_len = 0;
 143         CC->redirect_alloc = 0;
 144         CtdlLogPrintf(CTDL_DEBUG, "Wordbreaking message %ld...\n", msgnum);
 145         wordbreaker(msgtext, &num_tokens, &tokens);
 146         free(msgtext);
 147
 148         CtdlLogPrintf(CTDL_DEBUG, "Indexing message %ld [%d tokens]\n", msgnum, num_tokens);
 149         if (num_tokens > 0) {
 150                 for (i=0; i<num_tokens; ++i) {
 151
 152                         /* Add the message to the relevant token bucket */
 153
 154                         /* search for tokens[i] */
 155                         tok = tokens[i];
 156
 157                         if ( (tok >= 0) && (tok <= 65535) ) {
 158                                 /* fetch the bucket, Liza */
 159                                 if (ftc_msgs[tok] == NULL) {
 160                                         cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
 161                                         if (cdb_bucket != NULL) {
 162                                                 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
 163                                                 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
 164                                                 cdb_bucket->ptr = NULL;
 165                                                 cdb_free(cdb_bucket);
 166                                         }
 167                                         else {
 168                                                 ftc_num_msgs[tok] = 0;
 169                                                 ftc_msgs[tok] = malloc(sizeof(long));
 170                                         }
 171                                 }
 172
 173
 174                                 if (op == 1) {  /* add to index */
 175                                         ++ftc_num_msgs[tok];
 176                                         ftc_msgs[tok] = realloc(ftc_msgs[tok],
 177                                                                 ftc_num_msgs[tok]*sizeof(long));
 178                                         ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
 179                                 }
 180
 181                                 if (op == 0) {  /* remove from index */
 182                                         if (ftc_num_msgs[tok] >= 1) {
 183                                                 for (j=0; j<ftc_num_msgs[tok]; ++j) {
 184                                                         if (ftc_msgs[tok][j] == msgnum) {
 185                                                                 memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
 186                                                                 --ftc_num_msgs[tok];
 187                                                                 --j;
 188                                                         }
 189                                                 }
 190                                         }
 191                                 }
 192                         }
 193                         else {
 194                                 CtdlLogPrintf(CTDL_ALERT, "Invalid token %d !!\n", tok);
 195                         }
 196                 }
 197
 198                 free(tokens);
 199         }
 200 }
 201
 202
 203
 204 /*
 205  * Add a message to the list of those to be indexed.
 206  */
 207 void ft_index_msg(long msgnum, void *userdata) {
 208
 209         if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
 210                 ++ft_num_msgs;
 211                 if (ft_num_msgs > ft_num_alloc) {
 212                         ft_num_alloc += 1024;
 213                         ft_newmsgs = realloc(ft_newmsgs,
 214                                 (ft_num_alloc * sizeof(long)));
 215                 }
 216                 ft_newmsgs[ft_num_msgs - 1] = msgnum;
 217         }
 218
 219 }
 220
 221 /*
 222  * Scan a room for messages to index.
 223  */
 224 void ft_index_room(struct ctdlroom *qrbuf, void *data)
 225 {
 226         if (CtdlThreadCheckStop())
 227                 return;
 228
 229         getroom(&CC->room, qrbuf->QRname);
 230         CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, NULL, ft_index_msg, NULL);
 231 }
 232
 233
 234 /*
 235  * Begin the fulltext indexing process.  (Called as an EVT_TIMER event)
 236  */
 237 void do_fulltext_indexing(void) {
 238         int i;
 239         static time_t last_index = 0L;
 240         static time_t last_progress = 0L;
 241         time_t run_time = 0L;
 242         time_t end_time = 0L;
 243
 244         /*
 245          * Don't do this if the site doesn't have it enabled.
 246          */
 247         if (!config.c_enable_fulltext) {
 248                 return;
 249         }
 250
 251         /*
 252          * Make sure we don't run the indexer too frequently.
 253          * FIXME move the setting into config
 254          */
 255 /*
 256  * The thread sleeps for 300 seconds so we don't need this here any more
 257
 258         if ( (time(NULL) - last_index) < 300L) {
 259                 return;
 260         }
 261 */
 262
 263         /*
 264          * Check to see whether the fulltext index is up to date; if there
 265          * are no messages to index, don't waste any more time trying.
 266          */
 267         if ((CitControl.MMfulltext >= CitControl.MMhighest) && (CitControl.fulltext_wordbreaker == FT_WORDBREAKER_ID)) {
 268                 return;         /* nothing to do! */
 269         }
 270
 271         run_time = time(NULL);
 272         CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() started (%ld)\n", run_time);
 273
 274         /*
 275          * If we've switched wordbreaker modules, burn the index and start
 276          * over.
 277          */
 278         begin_critical_section(S_CONTROL);
 279         if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
 280                 CtdlLogPrintf(CTDL_DEBUG, "wb ver on disk = %d, code ver = %d\n",
 281                         CitControl.fulltext_wordbreaker, FT_WORDBREAKER_ID);
 282                 CtdlLogPrintf(CTDL_INFO, "(re)initializing full text index\n");
 283                 cdb_trunc(CDB_FULLTEXT);
 284                 CitControl.MMfulltext = 0L;
 285                 put_control();
 286         }
 287         end_critical_section(S_CONTROL);
 288
 289         /*
 290          * Now go through each room and find messages to index.
 291          */
 292         ft_newhighest = CitControl.MMhighest;
 293         ForEachRoom(ft_index_room, NULL);       /* load all msg pointers */
 294
 295         if (ft_num_msgs > 0) {
 296                 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
 297                 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
 298                         if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
 299                                 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
 300                                         ((ft_num_msgs - i - 1)*sizeof(long)));
 301                                 --ft_num_msgs;
 302                                 --i;
 303                         }
 304                 }
 305
 306                 /* Here it is ... do each message! */
 307                 for (i=0; i<ft_num_msgs; ++i) {
 308                         if (time(NULL) != last_progress) {
 309                                 CtdlLogPrintf(CTDL_DEBUG,
 310                                         "Indexed %d of %d messages (%d%%)\n",
 311                                                 i, ft_num_msgs,
 312                                                 ((i*100) / ft_num_msgs)
 313                                 );
 314                                 last_progress = time(NULL);
 315                         }
 316                         ft_index_message(ft_newmsgs[i], 1);
 317
 318                         /* Check to see if we need to quit early */
 319                         if (CtdlThreadCheckStop()) {
 320                                 CtdlLogPrintf(CTDL_DEBUG, "Indexer quitting early\n");
 321                                 ft_newhighest = ft_newmsgs[i];
 322                                 break;
 323                         }
 324
 325                         /* Check to see if we have to maybe flush to disk */
 326                         if (i >= FT_MAX_CACHE) {
 327                                 CtdlLogPrintf(CTDL_DEBUG, "Time to flush.\n");
 328                                 ft_newhighest = ft_newmsgs[i];
 329                                 break;
 330                         }
 331
 332                 }
 333
 334                 free(ft_newmsgs);
 335                 ft_num_msgs = 0;
 336                 ft_num_alloc = 0;
 337                 ft_newmsgs = NULL;
 338         }
 339         end_time = time(NULL);
 340
 341         if (CtdlThreadCheckStop())
 342                 return;
 343
 344         CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() duration (%ld)\n", end_time - run_time);
 345
 346         /* Save our place so we don't have to do this again */
 347         ft_flush_cache();
 348         begin_critical_section(S_CONTROL);
 349         CitControl.MMfulltext = ft_newhighest;
 350         CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
 351         put_control();
 352         end_critical_section(S_CONTROL);
 353         last_index = time(NULL);
 354
 355         CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
 356         return;
 357 }
 358
 359 /*
 360  * Main loop for the indexer thread.
 361  */
 362 void *indexer_thread(void *arg) {
 363         struct CitContext indexerCC;
 364
 365         CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() initializing\n");
 366
 367         CtdlFillSystemContext(&indexerCC, "indexer");
 368         citthread_setspecific(MyConKey, (void *)&indexerCC );
 369
 370         while (!CtdlThreadCheckStop()) {
 371                 do_fulltext_indexing();
 372                 CtdlThreadSleep(300);
 373         }
 374
 375         CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() exiting\n");
 376         return NULL;
 377 }
 378
 379
 380
 381 /*
 382  * API call to perform searches.
 383  * (This one does the "all of these words" search.)
 384  * Caller is responsible for freeing the message list.
 385  */
 386 void ft_search(int *fts_num_msgs, long **fts_msgs, char *search_string) {
 387         int num_tokens = 0;
 388         int *tokens = NULL;
 389         int i, j;
 390         struct cdbdata *cdb_bucket;
 391         int num_all_msgs = 0;
 392         long *all_msgs = NULL;
 393         int num_ret_msgs = 0;
 394         int num_ret_alloc = 0;
 395         long *ret_msgs = NULL;
 396         int tok;
 397
 398         wordbreaker(search_string, &num_tokens, &tokens);
 399         if (num_tokens > 0) {
 400                 for (i=0; i<num_tokens; ++i) {
 401
 402                         /* search for tokens[i] */
 403                         tok = tokens[i];
 404
 405                         /* fetch the bucket, Liza */
 406                         if (ftc_msgs[tok] == NULL) {
 407                                 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
 408                                 if (cdb_bucket != NULL) {
 409                                         ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
 410                                         ftc_msgs[tok] = (long *)cdb_bucket->ptr;
 411                                         cdb_bucket->ptr = NULL;
 412                                         cdb_free(cdb_bucket);
 413                                 }
 414                                 else {
 415                                         ftc_num_msgs[tok] = 0;
 416                                         ftc_msgs[tok] = malloc(sizeof(long));
 417                                 }
 418                         }
 419
 420                         num_all_msgs += ftc_num_msgs[tok];
 421                         if (num_all_msgs > 0) {
 422                                 all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
 423                                 memcpy(&all_msgs[num_all_msgs-ftc_num_msgs[tok]],
 424                                         ftc_msgs[tok], ftc_num_msgs[tok]*sizeof(long) );
 425                         }
 426
 427                 }
 428                 free(tokens);
 429                 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
 430
 431                 /*
 432                  * At this point, if a message appears num_tokens times in the
 433                  * list, then it contains all of the search tokens.
 434                  */
 435                 if (num_all_msgs >= num_tokens)
 436                    for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
 437                         if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
 438
 439                                 ++num_ret_msgs;
 440                                 if (num_ret_msgs > num_ret_alloc) {
 441                                         num_ret_alloc += 64;
 442                                         ret_msgs = realloc(ret_msgs,
 443                                                 (num_ret_alloc*sizeof(long)) );
 444                                 }
 445                                 ret_msgs[num_ret_msgs - 1] = all_msgs[j];
 446
 447                         }
 448                 }
 449
 450                 free(all_msgs);
 451         }
 452
 453         *fts_num_msgs = num_ret_msgs;
 454         *fts_msgs = ret_msgs;
 455 }
 456
 457
 458 /*
 459  * This search command is for diagnostic purposes and may be removed or replaced.
 460  */
 461 void cmd_srch(char *argbuf) {
 462         int num_msgs = 0;
 463         long *msgs = NULL;
 464         int i;
 465         char search_string[256];
 466
 467         if (CtdlAccessCheck(ac_logged_in)) return;
 468
 469         if (!config.c_enable_fulltext) {
 470                 cprintf("%d Full text index is not enabled on this server.\n",
 471                         ERROR + CMD_NOT_SUPPORTED);
 472                 return;
 473         }
 474
 475         extract_token(search_string, argbuf, 0, '|', sizeof search_string);
 476         ft_search(&num_msgs, &msgs, search_string);
 477
 478         cprintf("%d %d msgs match all search words:\n",
 479                 LISTING_FOLLOWS, num_msgs);
 480         if (num_msgs > 0) {
 481                 for (i=0; i<num_msgs; ++i) {
 482                         cprintf("%ld\n", msgs[i]);
 483                 }
 484         }
 485         if (msgs != NULL) free(msgs);
 486         cprintf("000\n");
 487 }
 488
 489 /*
 490  * Zero out our index cache.
 491  */
 492 void initialize_ft_cache(void) {
 493         memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
 494         memset(ftc_msgs, 0, (65536 * sizeof(long *)));
 495 }
 496
 497
 498 void ft_delete_remove(char *room, long msgnum)
 499 {
 500         if (room) return;
 501
 502         /* Remove from fulltext index */
 503         if (config.c_enable_fulltext) {
 504                 ft_index_message(msgnum, 0);
 505         }
 506 }
 507
 508 /*****************************************************************************/
 509
 510 CTDL_MODULE_INIT(fulltext)
 511 {
 512         if (!threading)
 513         {
 514                 initialize_ft_cache();
 515                 initialize_noise_words();
 516                 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
 517                 CtdlRegisterDeleteHook(ft_delete_remove);
 518                 CtdlRegisterSearchFuncHook(ft_search, "fulltext");
 519                 CtdlRegisterCleanupHook(noise_word_cleanup);
 520         }
 521         else
 522         {
 523                 CtdlThreadCreate("Indexer", CTDLTHREAD_BIGSTACK, indexer_thread, NULL);
 524         }
 525         /* return our Subversion id for the Log */
 526         return "$Id$";
 527 }