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         char *msgtext;
 127         int tok;
 128         struct CtdlMessage *msg = NULL;
 129
 130         msg = CtdlFetchMessage(msgnum, 1);
 131         if (msg == NULL) {
 132                 CtdlLogPrintf(CTDL_ERR, "ft_index_message() could not load msg %ld\n", msgnum);
 133                 return;
 134         }
 135
 136         if (msg->cm_fields['1'] != NULL) {
 137                 CtdlLogPrintf(CTDL_DEBUG, "ft_index_message() excluded msg %ld\n", msgnum);
 138                 CtdlFreeMessage(msg);
 139                 return;
 140         }
 141
 142         CtdlLogPrintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
 143                 (op ? "adding" : "removing") , msgnum
 144         );
 145
 146         /* Output the message as text before indexing it, so we don't end up
 147          * indexing a bunch of encoded base64, etc.
 148          */
 149         CC->redirect_buffer = malloc(SIZ);
 150         CC->redirect_len = 0;
 151         CC->redirect_alloc = SIZ;
 152         CtdlOutputPreLoadedMsg(msg, MT_CITADEL, HEADERS_ALL, 0, 1, 0);
 153         CtdlFreeMessage(msg);
 154         msgtext = CC->redirect_buffer;
 155         CC->redirect_buffer = NULL;
 156         CC->redirect_len = 0;
 157         CC->redirect_alloc = 0;
 158         CtdlLogPrintf(CTDL_DEBUG, "Wordbreaking message %ld...\n", msgnum);
 159         wordbreaker(msgtext, &num_tokens, &tokens);
 160         free(msgtext);
 161
 162         CtdlLogPrintf(CTDL_DEBUG, "Indexing message %ld [%d tokens]\n", msgnum, num_tokens);
 163         if (num_tokens > 0) {
 164                 for (i=0; i<num_tokens; ++i) {
 165
 166                         /* Add the message to the relevant token bucket */
 167
 168                         /* search for tokens[i] */
 169                         tok = tokens[i];
 170
 171                         if ( (tok >= 0) && (tok <= 65535) ) {
 172                                 /* fetch the bucket, Liza */
 173                                 if (ftc_msgs[tok] == NULL) {
 174                                         cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
 175                                         if (cdb_bucket != NULL) {
 176                                                 ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
 177                                                 ftc_msgs[tok] = (long *)cdb_bucket->ptr;
 178                                                 cdb_bucket->ptr = NULL;
 179                                                 cdb_free(cdb_bucket);
 180                                         }
 181                                         else {
 182                                                 ftc_num_msgs[tok] = 0;
 183                                                 ftc_msgs[tok] = malloc(sizeof(long));
 184                                         }
 185                                 }
 186
 187
 188                                 if (op == 1) {  /* add to index */
 189                                         ++ftc_num_msgs[tok];
 190                                         ftc_msgs[tok] = realloc(ftc_msgs[tok],
 191                                                                 ftc_num_msgs[tok]*sizeof(long));
 192                                         ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
 193                                 }
 194
 195                                 if (op == 0) {  /* remove from index */
 196                                         if (ftc_num_msgs[tok] >= 1) {
 197                                                 for (j=0; j<ftc_num_msgs[tok]; ++j) {
 198                                                         if (ftc_msgs[tok][j] == msgnum) {
 199                                                                 memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
 200                                                                 --ftc_num_msgs[tok];
 201                                                                 --j;
 202                                                         }
 203                                                 }
 204                                         }
 205                                 }
 206                         }
 207                         else {
 208                                 CtdlLogPrintf(CTDL_ALERT, "Invalid token %d !!\n", tok);
 209                         }
 210                 }
 211
 212                 free(tokens);
 213         }
 214 }
 215
 216
 217
 218 /*
 219  * Add a message to the list of those to be indexed.
 220  */
 221 void ft_index_msg(long msgnum, void *userdata) {
 222
 223         if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
 224                 ++ft_num_msgs;
 225                 if (ft_num_msgs > ft_num_alloc) {
 226                         ft_num_alloc += 1024;
 227                         ft_newmsgs = realloc(ft_newmsgs,
 228                                 (ft_num_alloc * sizeof(long)));
 229                 }
 230                 ft_newmsgs[ft_num_msgs - 1] = msgnum;
 231         }
 232
 233 }
 234
 235 /*
 236  * Scan a room for messages to index.
 237  */
 238 void ft_index_room(struct ctdlroom *qrbuf, void *data)
 239 {
 240         if (CtdlThreadCheckStop())
 241                 return;
 242
 243         CtdlGetRoom(&CC->room, qrbuf->QRname);
 244         CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, NULL, ft_index_msg, NULL);
 245 }
 246
 247
 248 /*
 249  * Begin the fulltext indexing process.
 250  */
 251 void do_fulltext_indexing(void) {
 252         int i;
 253         static time_t last_index = 0L;
 254         static time_t last_progress = 0L;
 255         time_t run_time = 0L;
 256         time_t end_time = 0L;
 257
 258         /*
 259          * Don't do this if the site doesn't have it enabled.
 260          */
 261         if (!config.c_enable_fulltext) {
 262                 return;
 263         }
 264
 265         /*
 266          * Make sure we don't run the indexer too frequently.
 267          * FIXME move the setting into config
 268          */
 269 /*
 270  * The thread sleeps for 300 seconds so we don't need this here any more
 271
 272         if ( (time(NULL) - last_index) < 300L) {
 273                 return;
 274         }
 275 */
 276
 277         /*
 278          * Check to see whether the fulltext index is up to date; if there
 279          * are no messages to index, don't waste any more time trying.
 280          */
 281         if ((CitControl.MMfulltext >= CitControl.MMhighest) && (CitControl.fulltext_wordbreaker == FT_WORDBREAKER_ID)) {
 282                 return;         /* nothing to do! */
 283         }
 284
 285         run_time = time(NULL);
 286         CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() started (%ld)\n", run_time);
 287
 288         /*
 289          * If we've switched wordbreaker modules, burn the index and start
 290          * over.
 291          */
 292         begin_critical_section(S_CONTROL);
 293         if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
 294                 CtdlLogPrintf(CTDL_DEBUG, "wb ver on disk = %d, code ver = %d\n",
 295                         CitControl.fulltext_wordbreaker, FT_WORDBREAKER_ID);
 296                 CtdlLogPrintf(CTDL_INFO, "(re)initializing full text index\n");
 297                 cdb_trunc(CDB_FULLTEXT);
 298                 CitControl.MMfulltext = 0L;
 299                 put_control();
 300         }
 301         end_critical_section(S_CONTROL);
 302
 303         /*
 304          * Now go through each room and find messages to index.
 305          */
 306         ft_newhighest = CitControl.MMhighest;
 307         CtdlForEachRoom(ft_index_room, NULL);   /* load all msg pointers */
 308
 309         if (ft_num_msgs > 0) {
 310                 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
 311                 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
 312                         if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
 313                                 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
 314                                         ((ft_num_msgs - i - 1)*sizeof(long)));
 315                                 --ft_num_msgs;
 316                                 --i;
 317                         }
 318                 }
 319
 320                 /* Here it is ... do each message! */
 321                 for (i=0; i<ft_num_msgs; ++i) {
 322                         if (time(NULL) != last_progress) {
 323                                 CtdlLogPrintf(CTDL_DEBUG,
 324                                         "Indexed %d of %d messages (%d%%)\n",
 325                                                 i, ft_num_msgs,
 326                                                 ((i*100) / ft_num_msgs)
 327                                 );
 328                                 last_progress = time(NULL);
 329                         }
 330                         ft_index_message(ft_newmsgs[i], 1);
 331
 332                         /* Check to see if we need to quit early */
 333                         if (CtdlThreadCheckStop()) {
 334                                 CtdlLogPrintf(CTDL_DEBUG, "Indexer quitting early\n");
 335                                 ft_newhighest = ft_newmsgs[i];
 336                                 break;
 337                         }
 338
 339                         /* Check to see if we have to maybe flush to disk */
 340                         if (i >= FT_MAX_CACHE) {
 341                                 CtdlLogPrintf(CTDL_DEBUG, "Time to flush.\n");
 342                                 ft_newhighest = ft_newmsgs[i];
 343                                 break;
 344                         }
 345
 346                 }
 347
 348                 free(ft_newmsgs);
 349                 ft_num_msgs = 0;
 350                 ft_num_alloc = 0;
 351                 ft_newmsgs = NULL;
 352         }
 353         end_time = time(NULL);
 354
 355         if (CtdlThreadCheckStop())
 356                 return;
 357
 358         CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() duration (%ld)\n", end_time - run_time);
 359
 360         /* Save our place so we don't have to do this again */
 361         ft_flush_cache();
 362         begin_critical_section(S_CONTROL);
 363         CitControl.MMfulltext = ft_newhighest;
 364         CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
 365         put_control();
 366         end_critical_section(S_CONTROL);
 367         last_index = time(NULL);
 368
 369         CtdlLogPrintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
 370         return;
 371 }
 372
 373 /*
 374  * Main loop for the indexer thread.
 375  */
 376 void *indexer_thread(void *arg) {
 377         struct CitContext indexerCC;
 378
 379         CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() initializing\n");
 380
 381         CtdlFillSystemContext(&indexerCC, "indexer");
 382         citthread_setspecific(MyConKey, (void *)&indexerCC );
 383
 384         while (!CtdlThreadCheckStop()) {
 385                 do_fulltext_indexing();
 386                 CtdlThreadSleep(300);
 387         }
 388
 389         CtdlLogPrintf(CTDL_DEBUG, "indexer_thread() exiting\n");
 390         return NULL;
 391 }
 392
 393
 394
 395 /*
 396  * API call to perform searches.
 397  * (This one does the "all of these words" search.)
 398  * Caller is responsible for freeing the message list.
 399  */
 400 void ft_search(int *fts_num_msgs, long **fts_msgs, char *search_string) {
 401         int num_tokens = 0;
 402         int *tokens = NULL;
 403         int i, j;
 404         struct cdbdata *cdb_bucket;
 405         int num_all_msgs = 0;
 406         long *all_msgs = NULL;
 407         int num_ret_msgs = 0;
 408         int num_ret_alloc = 0;
 409         long *ret_msgs = NULL;
 410         int tok;
 411
 412         wordbreaker(search_string, &num_tokens, &tokens);
 413         if (num_tokens > 0) {
 414                 for (i=0; i<num_tokens; ++i) {
 415
 416                         /* search for tokens[i] */
 417                         tok = tokens[i];
 418
 419                         /* fetch the bucket, Liza */
 420                         if (ftc_msgs[tok] == NULL) {
 421                                 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
 422                                 if (cdb_bucket != NULL) {
 423                                         ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
 424                                         ftc_msgs[tok] = (long *)cdb_bucket->ptr;
 425                                         cdb_bucket->ptr = NULL;
 426                                         cdb_free(cdb_bucket);
 427                                 }
 428                                 else {
 429                                         ftc_num_msgs[tok] = 0;
 430                                         ftc_msgs[tok] = malloc(sizeof(long));
 431                                 }
 432                         }
 433
 434                         num_all_msgs += ftc_num_msgs[tok];
 435                         if (num_all_msgs > 0) {
 436                                 all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
 437                                 memcpy(&all_msgs[num_all_msgs-ftc_num_msgs[tok]],
 438                                         ftc_msgs[tok], ftc_num_msgs[tok]*sizeof(long) );
 439                         }
 440
 441                 }
 442                 free(tokens);
 443                 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
 444
 445                 /*
 446                  * At this point, if a message appears num_tokens times in the
 447                  * list, then it contains all of the search tokens.
 448                  */
 449                 if (num_all_msgs >= num_tokens)
 450                    for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
 451                         if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
 452
 453                                 ++num_ret_msgs;
 454                                 if (num_ret_msgs > num_ret_alloc) {
 455                                         num_ret_alloc += 64;
 456                                         ret_msgs = realloc(ret_msgs,
 457                                                 (num_ret_alloc*sizeof(long)) );
 458                                 }
 459                                 ret_msgs[num_ret_msgs - 1] = all_msgs[j];
 460
 461                         }
 462                 }
 463
 464                 free(all_msgs);
 465         }
 466
 467         *fts_num_msgs = num_ret_msgs;
 468         *fts_msgs = ret_msgs;
 469 }
 470
 471
 472 /*
 473  * This search command is for diagnostic purposes and may be removed or replaced.
 474  */
 475 void cmd_srch(char *argbuf) {
 476         int num_msgs = 0;
 477         long *msgs = NULL;
 478         int i;
 479         char search_string[256];
 480
 481         if (CtdlAccessCheck(ac_logged_in)) return;
 482
 483         if (!config.c_enable_fulltext) {
 484                 cprintf("%d Full text index is not enabled on this server.\n",
 485                         ERROR + CMD_NOT_SUPPORTED);
 486                 return;
 487         }
 488
 489         extract_token(search_string, argbuf, 0, '|', sizeof search_string);
 490         ft_search(&num_msgs, &msgs, search_string);
 491
 492         cprintf("%d %d msgs match all search words:\n",
 493                 LISTING_FOLLOWS, num_msgs);
 494         if (num_msgs > 0) {
 495                 for (i=0; i<num_msgs; ++i) {
 496                         cprintf("%ld\n", msgs[i]);
 497                 }
 498         }
 499         if (msgs != NULL) free(msgs);
 500         cprintf("000\n");
 501 }
 502
 503 /*
 504  * Zero out our index cache.
 505  */
 506 void initialize_ft_cache(void) {
 507         memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
 508         memset(ftc_msgs, 0, (65536 * sizeof(long *)));
 509 }
 510
 511
 512 void ft_delete_remove(char *room, long msgnum)
 513 {
 514         if (room) return;
 515
 516         /* Remove from fulltext index */
 517         if (config.c_enable_fulltext) {
 518                 ft_index_message(msgnum, 0);
 519         }
 520 }
 521
 522 /*****************************************************************************/
 523
 524 CTDL_MODULE_INIT(fulltext)
 525 {
 526         if (!threading)
 527         {
 528                 initialize_ft_cache();
 529                 initialize_noise_words();
 530                 CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
 531                 CtdlRegisterDeleteHook(ft_delete_remove);
 532                 CtdlRegisterSearchFuncHook(ft_search, "fulltext");
 533                 CtdlRegisterCleanupHook(noise_word_cleanup);
 534         }
 535         else
 536         {
 537                 CtdlThreadCreate("Indexer", CTDLTHREAD_BIGSTACK, indexer_thread, NULL);
 538         }
 539         /* return our Subversion id for the Log */
 540         return "$Id$";
 541 }