citadel/modules/fulltext/serv_fulltext.c

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