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 "serv_fulltext.h"
  59 #include "ft_wordbreaker.h"
  60 #include "threads.h"
  61 #include "context.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         StrBuf *msgtext;
 127         char *txt;
 128         int tok;
 129         struct CtdlMessage *msg = NULL;
 130
 131         msg = CtdlFetchMessage(msgnum, 1);
 132         if (msg == NULL) {
 133                 CtdlLogPrintf(CTDL_ERR, "ft_index_message() could not load msg %ld\n", msgnum);
 134                 return;
 135         }
 136
 137         if (msg->cm_fields['1'] != NULL) {
 138                 CtdlLogPrintf(CTDL_DEBUG, "ft_index_message() excluded msg %ld\n", msgnum);
 139                 CtdlFreeMessage(msg);
 140                 return;
 141         }
 142
 143         CtdlLogPrintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
 144                 (op ? "adding" : "removing") , msgnum
 145         );
 146
 147         /* Output the message as text before indexing it, so we don't end up
 148          * indexing a bunch of encoded base64, etc.
 149          */
 150         CC->redirect_buffer = NewStrBufPlain(NULL, SIZ);
 151         CtdlOutputPreLoadedMsg(msg, MT_CITADEL, HEADERS_ALL, 0, 1, 0);
 152         CtdlFreeMessage(msg);
 153         msgtext = CC->redirect_buffer;
 154         CC->redirect_buffer = NULL;
 155         CtdlLogPrintf(CTDL_DEBUG, "Wordbreaking message %ld...\n", msgnum);
 156         txt = SmashStrBuf(&msgtext);
 157         wordbreaker(txt, &num_tokens, &tokens);
 158         free(txt);
 159
 160         CtdlLogPrintf(CTDL_DEBUG, "Indexing message %ld [%d tokens]\n", msgnum, num_tokens);
 161         if (num_tokens > 0) {
 162                 for (i=0; i<num_tokens; ++i) {
 163
 164                         /* Add the message to the relevant token bucket */
 165
 166                         /* search for tokens[i] */
 167                         tok = tokens[i];
 168
 169                         if ( (tok >= 0) && (tok <= 65535) ) {
 170                                 /* fetch the bucket, Liza */
 171                                 if (ftc_msgs[tok] == NULL) {
 172                                         cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
 173                                         if (cdb_bucket != NULL) {
 174                                                 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
 175                                                 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
 176                                                 cdb_bucket->ptr = NULL;
 177                                                 cdb_free(cdb_bucket);
 178                                         }
 179                                         else {
 180                                                 ftc_num_msgs[tok] = 0;
 181                                                 ftc_msgs[tok] = malloc(sizeof(long));
 182                                         }
 183                                 }
 184
 185
 186                                 if (op == 1) {  /* add to index */
 187                                         ++ftc_num_msgs[tok];
 188                                         ftc_msgs[tok] = realloc(ftc_msgs[tok],
 189                                                                 ftc_num_msgs[tok]*sizeof(long));
 190                                         ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
 191                                 }
 192
 193                                 if (op == 0) {  /* remove from index */
 194                                         if (ftc_num_msgs[tok] >= 1) {
 195                                                 for (j=0; j<ftc_num_msgs[tok]; ++j) {
 196                                                         if (ftc_msgs[tok][j] == msgnum) {
 197                                                                 memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
 198                                                                 --ftc_num_msgs[tok];
 199                                                                 --j;
 200                                                         }
 201                                                 }
 202                                         }
 203                                 }
 204                         }
 205                         else {
 206                                 CtdlLogPrintf(CTDL_ALERT, "Invalid token %d !!\n", tok);
 207                         }
 208                 }
 209
 210                 free(tokens);
 211         }
 212 }
 213
 214
 215
 216 /*
 217  * Add a message to the list of those to be indexed.
 218  */
 219 void ft_index_msg(long msgnum, void *userdata) {
 220
 221         if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
 222                 ++ft_num_msgs;
 223                 if (ft_num_msgs > ft_num_alloc) {
 224                         ft_num_alloc += 1024;
 225                         ft_newmsgs = realloc(ft_newmsgs,
 226                                 (ft_num_alloc * sizeof(long)));
 227                 }
 228                 ft_newmsgs[ft_num_msgs - 1] = msgnum;
 229         }
 230
 231 }
 232
 233 /*
 234  * Scan a room for messages to index.
 235  */
 236 void ft_index_room(struct ctdlroom *qrbuf, void *data)
 237 {
 238         if (CtdlThreadCheckStop())
 239                 return;
 240
 241         CtdlGetRoom(&CC->room, qrbuf->QRname);
 242         CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, NULL, ft_index_msg, NULL);
 243 }
 244
 245
 246 /*
 247  * Begin the fulltext indexing process.
 248  */
 249 void do_fulltext_indexing(void) {
 250         int i;
 251         static time_t last_index = 0L;
 252         static time_t last_progress = 0L;
 253         time_t run_time = 0L;
 254         time_t end_time = 0L;
 255
 256         /*
 257          * Don't do this if the site doesn't have it enabled.
 258          */
 259         if (!config.c_enable_fulltext) {
 260                 return;
 261         }
 262
 263         /*
 264          * Make sure we don't run the indexer too frequently.
 265          * FIXME move the setting into config
 266          */
 267 /*
 268  * The thread sleeps for 300 seconds so we don't need this here any more
 269
 270         if ( (time(NULL) - last_index) < 300L) {
 271                 return;
 272         }
 273 */
 274
 275         /*
 276          * Check to see whether the fulltext index is up to date; if there
 277          * are no messages to index, don't waste any more time trying.
 278          */
 279         if ((CitControl.MMfulltext >= CitControl.MMhighest) && (CitControl.fulltext_wordbreaker == FT_WORDBREAKER_ID)) {
 280                 return;         /* nothing to do! */
 281         }
 282
 283         run_time = time(NULL);
 284         CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() started (%ld)\n", run_time);
 285
 286         /*
 287          * If we've switched wordbreaker modules, burn the index and start
 288          * over.
 289          */
 290         begin_critical_section(S_CONTROL);
 291         if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
 292                 CtdlLogPrintf(CTDL_DEBUG, "wb ver on disk = %d, code ver = %d\n",
 293                         CitControl.fulltext_wordbreaker, FT_WORDBREAKER_ID);
 294                 CtdlLogPrintf(CTDL_INFO, "(re)initializing full text index\n");
 295                 cdb_trunc(CDB_FULLTEXT);
 296                 CitControl.MMfulltext = 0L;
 297                 put_control();
 298         }
 299         end_critical_section(S_CONTROL);
 300
 301         /*
 302          * Now go through each room and find messages to index.
 303          */
 304         ft_newhighest = CitControl.MMhighest;
 305         CtdlForEachRoom(ft_index_room, NULL);   /* load all msg pointers */
 306
 307         if (ft_num_msgs > 0) {
 308                 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
 309                 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
 310                         if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
 311                                 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
 312                                         ((ft_num_msgs - i - 1)*sizeof(long)));
 313                                 --ft_num_msgs;
 314                                 --i;
 315                         }
 316                 }
 317
 318                 /* Here it is ... do each message! */
 319                 for (i=0; i<ft_num_msgs; ++i) {
 320                         if (time(NULL) != last_progress) {
 321                                 CtdlLogPrintf(CTDL_DEBUG,
 322                                         "Indexed %d of %d messages (%d%%)\n",
 323                                                 i, ft_num_msgs,
 324                                                 ((i*100) / ft_num_msgs)
 325                                 );
 326                                 last_progress = time(NULL);
 327                         }
 328                         ft_index_message(ft_newmsgs[i], 1);
 329
 330                         /* Check to see if we need to quit early */
 331                         if (CtdlThreadCheckStop()) {
 332                                 CtdlLogPrintf(CTDL_DEBUG, "Indexer quitting early\n");
 333                                 ft_newhighest = ft_newmsgs[i];
 334                                 break;
 335                         }
 336
 337                         /* Check to see if we have to maybe flush to disk */
 338                         if (i >= FT_MAX_CACHE) {
 339                                 CtdlLogPrintf(CTDL_DEBUG, "Time to flush.\n");
 340                                 ft_newhighest = ft_newmsgs[i];
 341                                 break;
 342                         }
 343
 344                 }
 345
 346                 free(ft_newmsgs);
 347                 ft_num_msgs = 0;
 348                 ft_num_alloc = 0;
 349                 ft_newmsgs = NULL;
 350         }
 351         end_time = time(NULL);
 352
 353         if (CtdlThreadCheckStop())
 354                 return;
 355
 356         CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() duration (%ld)\n", end_time - run_time);
 357
 358         /* Save our place so we don't have to do this again */
 359         ft_flush_cache();
 360         begin_critical_section(S_CONTROL);
 361         CitControl.MMfulltext = ft_newhighest;
 362         CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
 363         put_control();
 364         end_critical_section(S_CONTROL);
 365         last_index = time(NULL);
 366
 367         CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
 368         return;
 369 }
 370
 371 /*
 372  * Main loop for the indexer thread.
 373  */
 374 void *indexer_thread(void *arg) {
 375         struct CitContext indexerCC;
 376
 377
 378         CtdlFillSystemContext(&indexerCC, "indexer");
 379         citthread_setspecific(MyConKey, (void *)&indexerCC );
 380         CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() initializing\n");
 381
 382         while (!CtdlThreadCheckStop()) {
 383                 do_fulltext_indexing();
 384                 CtdlThreadSleep(300);
 385         }
 386
 387         CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() exiting\n");
 388         CtdlClearSystemContext();
 389         return NULL;
 390 }
 391
 392
 393
 394 /*
 395  * API call to perform searches.
 396  * (This one does the "all of these words" search.)
 397  * Caller is responsible for freeing the message list.
 398  */
 399 void ft_search(int *fts_num_msgs, long **fts_msgs, const char *search_string) {
 400         int num_tokens = 0;
 401         int *tokens = NULL;
 402         int i, j;
 403         struct cdbdata *cdb_bucket;
 404         int num_all_msgs = 0;
 405         long *all_msgs = NULL;
 406         int num_ret_msgs = 0;
 407         int num_ret_alloc = 0;
 408         long *ret_msgs = NULL;
 409         int tok;
 410
 411         wordbreaker(search_string, &num_tokens, &tokens);
 412         if (num_tokens > 0) {
 413                 for (i=0; i<num_tokens; ++i) {
 414
 415                         /* search for tokens[i] */
 416                         tok = tokens[i];
 417
 418                         /* fetch the bucket, Liza */
 419                         if (ftc_msgs[tok] == NULL) {
 420                                 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
 421                                 if (cdb_bucket != NULL) {
 422                                         ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
 423                                         ftc_msgs[tok] = (long *)cdb_bucket->ptr;
 424                                         cdb_bucket->ptr = NULL;
 425                                         cdb_free(cdb_bucket);
 426                                 }
 427                                 else {
 428                                         ftc_num_msgs[tok] = 0;
 429                                         ftc_msgs[tok] = malloc(sizeof(long));
 430                                 }
 431                         }
 432
 433                         num_all_msgs += ftc_num_msgs[tok];
 434                         if (num_all_msgs > 0) {
 435                                 all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
 436                                 memcpy(&all_msgs[num_all_msgs-ftc_num_msgs[tok]],
 437                                         ftc_msgs[tok], ftc_num_msgs[tok]*sizeof(long) );
 438                         }
 439
 440                 }
 441                 free(tokens);
 442                 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
 443
 444                 /*
 445                  * At this point, if a message appears num_tokens times in the
 446                  * list, then it contains all of the search tokens.
 447                  */
 448                 if (num_all_msgs >= num_tokens)
 449                    for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
 450                         if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
 451
 452                                 ++num_ret_msgs;
 453                                 if (num_ret_msgs > num_ret_alloc) {
 454                                         num_ret_alloc += 64;
 455                                         ret_msgs = realloc(ret_msgs,
 456                                                 (num_ret_alloc*sizeof(long)) );
 457                                 }
 458                                 ret_msgs[num_ret_msgs - 1] = all_msgs[j];
 459
 460                         }
 461                 }
 462
 463                 free(all_msgs);
 464         }
 465
 466         *fts_num_msgs = num_ret_msgs;
 467         *fts_msgs = ret_msgs;
 468 }
 469
 470
 471 /*
 472  * This search command is for diagnostic purposes and may be removed or replaced.
 473  */
 474 void cmd_srch(char *argbuf) {
 475         int num_msgs = 0;
 476         long *msgs = NULL;
 477         int i;
 478         char search_string[256];
 479
 480         if (CtdlAccessCheck(ac_logged_in)) return;
 481
 482         if (!config.c_enable_fulltext) {
 483                 cprintf("%d Full text index is not enabled on this server.\n",
 484                         ERROR + CMD_NOT_SUPPORTED);
 485                 return;
 486         }
 487
 488         extract_token(search_string, argbuf, 0, '|', sizeof search_string);
 489         ft_search(&num_msgs, &msgs, search_string);
 490
 491         cprintf("%d %d msgs match all search words:\n",
 492                 LISTING_FOLLOWS, num_msgs);
 493         if (num_msgs > 0) {
 494                 for (i=0; i<num_msgs; ++i) {
 495                         cprintf("%ld\n", msgs[i]);
 496                 }
 497         }
 498         if (msgs != NULL) free(msgs);
 499         cprintf("000\n");
 500 }
 501
 502 /*
 503  * Zero out our index cache.
 504  */
 505 void initialize_ft_cache(void) {
 506         memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
 507         memset(ftc_msgs, 0, (65536 * sizeof(long *)));
 508 }
 509
 510
 511 void ft_delete_remove(char *room, long msgnum)
 512 {
 513         if (room) return;
 514
 515         /* Remove from fulltext index */
 516         if (config.c_enable_fulltext) {
 517                 ft_index_message(msgnum, 0);
 518         }
 519 }
 520
 521 /*****************************************************************************/
 522
 523 CTDL_MODULE_INIT(fulltext)
 524 {
 525         if (!threading)
 526         {
 527                 initialize_ft_cache();
 528                 initialize_noise_words();
 529                 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
 530                 CtdlRegisterDeleteHook(ft_delete_remove);
 531                 CtdlRegisterSearchFuncHook(ft_search, "fulltext");
 532                 CtdlRegisterCleanupHook(noise_word_cleanup);
 533         }
 534         else
 535         {
 536                 CtdlThreadCreate("Indexer", CTDLTHREAD_BIGSTACK, indexer_thread, NULL);
 537         }
 538         /* return our Subversion id for the Log */
 539         return "$Id$";
 540 }