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