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(5, "Flushing index cache to disk (%d%% complete)\n",
  83                                 (i * 100 / 65536)
  84                         );
  85                         last_update = time(NULL);
  86                 }
  87                 if (ftc_msgs[i] != NULL) {
  88                         cdb_store(CDB_FULLTEXT, &i, sizeof(int), ftc_msgs[i],
  89                                 (ftc_num_msgs[i] * sizeof(long)));
  90                         ftc_num_msgs[i] = 0;
  91                         free(ftc_msgs[i]);
  92                         ftc_msgs[i] = NULL;
  93                 }
  94         }
  95         lprintf(5, "Cache flush has been completed\n");
  96 }
  97
  98
  99 /*
 100  * Index or de-index a message.  (op == 1 to index, 0 to de-index)
 101  */
 102 void ft_index_message(long msgnum, int op) {
 103         int num_tokens = 0;
 104         int *tokens = NULL;
 105         int i, j;
 106         struct cdbdata *cdb_bucket;
 107         char *msgtext;
 108         int tok;
 109
 110         lprintf(CTDL_DEBUG, "ft_index_message() %s msg %ld\n",
 111                 (op ? "adding" : "removing") , msgnum
 112         );
 113
 114         /* Output the message as text before indexing it, so we don't end up
 115          * indexing a bunch of encoded base64, etc.
 116          */
 117         CC->redirect_buffer = malloc(SIZ);
 118         CC->redirect_len = 0;
 119         CC->redirect_alloc = SIZ;
 120         CtdlOutputMsg(msgnum, MT_CITADEL, HEADERS_ALL, 0, 1);
 121         msgtext = CC->redirect_buffer;
 122         CC->redirect_buffer = NULL;
 123         CC->redirect_len = 0;
 124         CC->redirect_alloc = 0;
 125         lprintf(CTDL_DEBUG, "Wordbreaking message %ld...\n", msgnum);
 126         wordbreaker(msgtext, &num_tokens, &tokens);
 127         free(msgtext);
 128
 129         lprintf(CTDL_DEBUG, "Indexing message %ld...\n", msgnum);
 130         if (num_tokens > 0) {
 131                 for (i=0; i<num_tokens; ++i) {
 132
 133                         /* Add the message to the relevant token bucket */
 134
 135                         /* search for tokens[i] */
 136                         tok = tokens[i];
 137
 138                         /* fetch the bucket, Liza */
 139                         if (ftc_msgs[tok] == NULL) {
 140                                 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
 141                                 if (cdb_bucket != NULL) {
 142                                         ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
 143                                         ftc_msgs[tok] = (long *)cdb_bucket->ptr;
 144                                         cdb_bucket->ptr = NULL;
 145                                         cdb_free(cdb_bucket);
 146                                 }
 147                                 else {
 148                                         ftc_num_msgs[tok] = 0;
 149                                         ftc_msgs[tok] = malloc(sizeof(long));
 150                                 }
 151                         }
 152
 153
 154                         if (op == 1) {  /* add to index */
 155                                 ++ftc_num_msgs[tok];
 156                                 ftc_msgs[tok] = realloc(ftc_msgs[tok],
 157                                                         ftc_num_msgs[tok]*sizeof(long));
 158                                 ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
 159                         }
 160
 161                         if (op == 0) {  /* remove from index */
 162                                 if (ftc_num_msgs[tok] >= 1) {
 163                                         for (j=0; j<ftc_num_msgs[tok]; ++j) {
 164                                                 if (ftc_msgs[tok][j] == msgnum) {
 165                                                         memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
 166                                                         --ftc_num_msgs[tok];
 167                                                         --j;
 168                                                 }
 169                                         }
 170                                 }
 171                         }
 172
 173                         /* FIXME do we need to unlock the file here? */
 174                 }
 175
 176                 free(tokens);
 177         }
 178 }
 179
 180
 181
 182 /*
 183  * Add a message to the list of those to be indexed.
 184  */
 185 void ft_index_msg(long msgnum, void *userdata) {
 186
 187         if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
 188                 ++ft_num_msgs;
 189                 if (ft_num_msgs > ft_num_alloc) {
 190                         ft_num_alloc += 1024;
 191                         ft_newmsgs = realloc(ft_newmsgs,
 192                                 (ft_num_alloc * sizeof(long)));
 193                 }
 194                 ft_newmsgs[ft_num_msgs - 1] = msgnum;
 195         }
 196
 197 }
 198
 199 /*
 200  * Scan a room for messages to index.
 201  */
 202 void ft_index_room(struct ctdlroom *qrbuf, void *data)
 203 {
 204         getroom(&CC->room, qrbuf->QRname);
 205         CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, ft_index_msg, NULL);
 206 }
 207
 208
 209 /*
 210  * Begin the fulltext indexing process.  (Called as an EVT_TIMER event)
 211  */
 212 void do_fulltext_indexing(void) {
 213         int i;
 214         static time_t last_index = 0L;
 215         static time_t last_progress = 0L;
 216
 217         /*
 218          * Don't do this if the site doesn't have it enabled.
 219          */
 220         if (!config.c_enable_fulltext) {
 221                 return;
 222         }
 223
 224         /*
 225          * Make sure we don't run the indexer too frequently.
 226          * FIXME move the setting into config
 227          */
 228         if ( (time(NULL) - last_index) < 300L) {
 229                 return;
 230         }
 231
 232         /*
 233          * Check to see whether the fulltext index is up to date; if there
 234          * are no messages to index, don't waste any more time trying.
 235          */
 236         if (CitControl.MMfulltext >= CitControl.MMhighest) {
 237                 return;         /* nothing to do! */
 238         }
 239
 240         lprintf(CTDL_DEBUG, "do_fulltext_indexing() started\n");
 241
 242         /*
 243          * If we've switched wordbreaker modules, burn the index and start
 244          * over.
 245          */
 246         if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
 247                 lprintf(CTDL_INFO, "(re)initializing full text index\n");
 248                 cdb_trunc(CDB_FULLTEXT);
 249                 CitControl.MMfulltext = 0L;
 250                 put_control();
 251         }
 252
 253         /*
 254          * Now go through each room and find messages to index.
 255          */
 256         ft_newhighest = CitControl.MMhighest;
 257         ForEachRoom(ft_index_room, NULL);       /* load all msg pointers */
 258
 259         if (ft_num_msgs > 0) {
 260                 qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
 261                 for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
 262                         if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
 263                                 memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
 264                                         ((ft_num_msgs - i - 1)*sizeof(long)));
 265                                 --ft_num_msgs;
 266                                 --i;
 267                         }
 268                 }
 269
 270                 /* Here it is ... do each message! */
 271                 for (i=0; i<ft_num_msgs; ++i) {
 272                         if ((time(NULL) - last_progress) > 10) {
 273                                 lprintf(CTDL_DEBUG,
 274                                         "Indexed %d of %d messages (%d%%)\n",
 275                                                 i, ft_num_msgs,
 276                                                 ((i*100) / ft_num_msgs)
 277                                 );
 278                                 last_progress = time(NULL);
 279                         }
 280                         ft_index_message(ft_newmsgs[i], 1);
 281
 282                         /* Check to see if we need to quit early */
 283                         if (time_to_die) {
 284                                 lprintf(CTDL_DEBUG, "Indexer quitting early\n");
 285                                 ft_newhighest = ft_newmsgs[i];
 286                                 break;
 287                         }
 288
 289                 }
 290
 291                 free(ft_newmsgs);
 292                 ft_num_msgs = 0;
 293                 ft_num_alloc = 0;
 294                 ft_newmsgs = NULL;
 295         }
 296
 297         /* Save our place so we don't have to do this again */
 298         ft_flush_cache();
 299         CitControl.MMfulltext = ft_newhighest;
 300         CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
 301         put_control();
 302         last_index = time(NULL);
 303
 304         lprintf(CTDL_DEBUG, "do_fulltext_indexing() finished\n");
 305         return;
 306 }
 307
 308 /*
 309  * Main loop for the indexer thread.
 310  */
 311 void *indexer_thread(void *arg) {
 312         struct CitContext indexerCC;
 313
 314         lprintf(CTDL_DEBUG, "indexer_thread() initializing\n");
 315
 316         memset(&indexerCC, 0, sizeof(struct CitContext));
 317         indexerCC.internal_pgm = 1;
 318         indexerCC.cs_pid = 0;
 319         pthread_setspecific(MyConKey, (void *)&indexerCC );
 320
 321         cdb_allocate_tsd();
 322
 323         while (!time_to_die) {
 324                 do_fulltext_indexing();
 325                 sleep(1);
 326         }
 327
 328         lprintf(CTDL_DEBUG, "indexer_thread() exiting\n");
 329         pthread_exit(NULL);
 330 }
 331
 332
 333
 334 /*
 335  * API call to perform searches.
 336  * (This one does the "all of these words" search.)
 337  * Caller is responsible for freeing the message list.
 338  */
 339 void ft_search(int *fts_num_msgs, long **fts_msgs, char *search_string) {
 340         int num_tokens = 0;
 341         int *tokens = NULL;
 342         int i, j;
 343         struct cdbdata *cdb_bucket;
 344         int num_all_msgs = 0;
 345         long *all_msgs = NULL;
 346         int num_ret_msgs = 0;
 347         int num_ret_alloc = 0;
 348         long *ret_msgs = NULL;
 349         int tok;
 350
 351         wordbreaker(search_string, &num_tokens, &tokens);
 352         if (num_tokens > 0) {
 353                 for (i=0; i<num_tokens; ++i) {
 354
 355                         /* search for tokens[i] */
 356                         tok = tokens[i];
 357
 358                         /* fetch the bucket, Liza */
 359                         if (ftc_msgs[tok] == NULL) {
 360                                 cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
 361                                 if (cdb_bucket != NULL) {
 362                                         ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
 363                                         ftc_msgs[tok] = (long *)cdb_bucket->ptr;
 364                                         cdb_bucket->ptr = NULL;
 365                                         cdb_free(cdb_bucket);
 366                                 }
 367                                 else {
 368                                         ftc_num_msgs[tok] = 0;
 369                                         ftc_msgs[tok] = malloc(sizeof(long));
 370                                 }
 371                         }
 372
 373                         num_all_msgs += ftc_num_msgs[tok];
 374                         if (num_all_msgs > 0) {
 375                                 all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
 376                                 memcpy(&all_msgs[num_all_msgs-ftc_num_msgs[tok]],
 377                                         ftc_msgs[tok], ftc_num_msgs[tok]*sizeof(long) );
 378                         }
 379
 380                 }
 381                 free(tokens);
 382                 qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);
 383
 384                 /*
 385                  * At this point, if a message appears num_tokens times in the
 386                  * list, then it contains all of the search tokens.
 387                  */
 388                 if (num_all_msgs >= num_tokens)
 389                    for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
 390                         if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
 391
 392                                 ++num_ret_msgs;
 393                                 if (num_ret_msgs > num_ret_alloc) {
 394                                         num_ret_alloc += 64;
 395                                         ret_msgs = realloc(ret_msgs,
 396                                                 (num_ret_alloc*sizeof(long)) );
 397                                 }
 398                                 ret_msgs[num_ret_msgs - 1] = all_msgs[j];
 399
 400                         }
 401                 }
 402
 403                 free(all_msgs);
 404         }
 405
 406         *fts_num_msgs = num_ret_msgs;
 407         *fts_msgs = ret_msgs;
 408 }
 409
 410
 411 /*
 412  * Tentative form of a search command
 413  */
 414 void cmd_srch(char *argbuf) {
 415         int num_msgs = 0;
 416         long *msgs = NULL;
 417         int i;
 418         char search_string[256];
 419
 420         if (CtdlAccessCheck(ac_logged_in)) return;
 421
 422         if (!config.c_enable_fulltext) {
 423                 cprintf("%d Full text index is not enabled on this server.\n",
 424                         ERROR + CMD_NOT_SUPPORTED);
 425         }
 426
 427         extract_token(search_string, argbuf, 0, '|', sizeof search_string);
 428         ft_search(&num_msgs, &msgs, search_string);
 429
 430         cprintf("%d %d msgs match all search words:\n",
 431                 LISTING_FOLLOWS, num_msgs);
 432         if (num_msgs > 0) {
 433                 for (i=0; i<num_msgs; ++i) {
 434                         cprintf("%ld\n", msgs[i]);
 435                 }
 436         }
 437         if (msgs != NULL) free(msgs);
 438         cprintf("000\n");
 439 }
 440
 441 /*
 442  * Zero out our index cache.
 443  */
 444 void initialize_ft_cache(void) {
 445         memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
 446         memset(ftc_msgs, 0, (65536 * sizeof(long *)));
 447 }
 448
 449
 450 /*****************************************************************************/
 451
 452 char *serv_fulltext_init(void)
 453 {
 454         initialize_ft_cache();
 455         CtdlRegisterProtoHook(cmd_srch, "SRCH", "Full text search");
 456         return "$Id$";
 457 }