Back to index

citadel  8.12
Functions | Variables
serv_fulltext.c File Reference
#include "sysdep.h"
#include <stdlib.h>
#include <unistd.h>
#include <stdio.h>
#include <fcntl.h>
#include <signal.h>
#include <pwd.h>
#include <errno.h>
#include <sys/types.h>
#include <time.h>
#include <sys/wait.h>
#include <string.h>
#include <limits.h>
#include <libcitadel.h>
#include "citadel.h"
#include "server.h"
#include "citserver.h"
#include "support.h"
#include "config.h"
#include "database.h"
#include "msgbase.h"
#include "control.h"
#include "serv_fulltext.h"
#include "ft_wordbreaker.h"
#include "threads.h"
#include "context.h"
#include "ctdl_module.h"

Go to the source code of this file.

Functions

int longcmp (const void *rec1, const void *rec2)
void ft_flush_cache (void)
void ft_index_message (long msgnum, int op)
void ft_index_msg (long msgnum, void *userdata)
void ft_index_room (struct ctdlroom *qrbuf, void *data)
void do_fulltext_indexing (void)
void ft_search (int *fts_num_msgs, long **fts_msgs, const char *search_string)
void cmd_srch (char *argbuf)
void initialize_ft_cache (void)
void ft_delete_remove (char *room, long msgnum)
 CTDL_MODULE_INIT (fulltext)

Variables

long ft_newhighest = 0L
long * ft_newmsgs = NULL
int ft_num_msgs = 0
int ft_num_alloc = 0
int ftc_num_msgs [65536]
long * ftc_msgs [65536]

Function Documentation

void cmd_srch ( char *  argbuf)

Definition at line 451 of file serv_fulltext.c.

                            {
       int num_msgs = 0;
       long *msgs = NULL;
       int i;
       char search_string[256];

       if (CtdlAccessCheck(ac_logged_in)) return;

       if (!config.c_enable_fulltext) {
              cprintf("%d Full text index is not enabled on this server.\n",
                     ERROR + CMD_NOT_SUPPORTED);
              return;
       }

       extract_token(search_string, argbuf, 0, '|', sizeof search_string);
       ft_search(&num_msgs, &msgs, search_string);

       cprintf("%d %d msgs match all search words:\n",
              LISTING_FOLLOWS, num_msgs);
       if (num_msgs > 0) {
              for (i=0; i<num_msgs; ++i) {
                     cprintf("%ld\n", msgs[i]);
              }
       }
       if (msgs != NULL) free(msgs);
       cprintf("000\n");
}

Here is the call graph for this function:

Here is the caller graph for this function:

CTDL_MODULE_INIT ( fulltext  )

Definition at line 500 of file serv_fulltext.c.

Here is the call graph for this function:

void do_fulltext_indexing ( void  )

Definition at line 242 of file serv_fulltext.c.

                                {
       int i;
       static time_t last_index = 0L;
       static time_t last_progress = 0L;
       time_t run_time = 0L;
       time_t end_time = 0L;
       static int is_running = 0;
       if (is_running) return;         /* Concurrency check - only one can run */
       is_running = 1;
       
       /*
        * Don't do this if the site doesn't have it enabled.
        */
       if (!config.c_enable_fulltext) {
              return;
       }

       /*
        * Make sure we don't run the indexer too frequently.
        * FIXME move the setting into config
        */
 
       if ( (time(NULL) - last_index) < 300L) {
              return;
       }

       /*
        * Check to see whether the fulltext index is up to date; if there
        * are no messages to index, don't waste any more time trying.
        */
       if ((CitControl.MMfulltext >= CitControl.MMhighest) && (CitControl.fulltext_wordbreaker == FT_WORDBREAKER_ID)) {
              return;              /* nothing to do! */
       }
       
       run_time = time(NULL);
       syslog(LOG_DEBUG, "do_fulltext_indexing() started (%ld)", run_time);
       
       /*
        * If we've switched wordbreaker modules, burn the index and start
        * over.
        */
       begin_critical_section(S_CONTROL);
       if (CitControl.fulltext_wordbreaker != FT_WORDBREAKER_ID) {
              syslog(LOG_DEBUG, "wb ver on disk = %d, code ver = %d",
                     CitControl.fulltext_wordbreaker, FT_WORDBREAKER_ID
              );
              syslog(LOG_INFO, "(re)initializing full text index");
              cdb_trunc(CDB_FULLTEXT);
              CitControl.MMfulltext = 0L;
              put_control();
       }
       end_critical_section(S_CONTROL);

       /*
        * Now go through each room and find messages to index.
        */
       ft_newhighest = CitControl.MMhighest;
       CtdlForEachRoom(ft_index_room, NULL);     /* load all msg pointers */

       if (ft_num_msgs > 0) {
              qsort(ft_newmsgs, ft_num_msgs, sizeof(long), longcmp);
              for (i=0; i<(ft_num_msgs-1); ++i) { /* purge dups */
                     if (ft_newmsgs[i] == ft_newmsgs[i+1]) {
                            memmove(&ft_newmsgs[i], &ft_newmsgs[i+1],
                                   ((ft_num_msgs - i - 1)*sizeof(long)));
                            --ft_num_msgs;
                            --i;
                     }
              }

              /* Here it is ... do each message! */
              for (i=0; i<ft_num_msgs; ++i) {
                     if (time(NULL) != last_progress) {
                            syslog(LOG_DEBUG,
                                   "Indexed %d of %d messages (%d%%)",
                                          i, ft_num_msgs,
                                          ((i*100) / ft_num_msgs)
                            );
                            last_progress = time(NULL);
                     }
                     ft_index_message(ft_newmsgs[i], 1);

                     /* Check to see if we need to quit early */
                     if (server_shutting_down) {
                            syslog(LOG_DEBUG, "Indexer quitting early");
                            ft_newhighest = ft_newmsgs[i];
                            break;
                     }

                     /* Check to see if we have to maybe flush to disk */
                     if (i >= FT_MAX_CACHE) {
                            syslog(LOG_DEBUG, "Time to flush.");
                            ft_newhighest = ft_newmsgs[i];
                            break;
                     }

              }

              free(ft_newmsgs);
              ft_num_msgs = 0;
              ft_num_alloc = 0;
              ft_newmsgs = NULL;
       }
       end_time = time(NULL);

       if (server_shutting_down) {
              is_running = 0;
              return;
       }
       
       syslog(LOG_DEBUG, "do_fulltext_indexing() duration (%ld)", end_time - run_time);
              
       /* Save our place so we don't have to do this again */
       ft_flush_cache();
       begin_critical_section(S_CONTROL);
       CitControl.MMfulltext = ft_newhighest;
       CitControl.fulltext_wordbreaker = FT_WORDBREAKER_ID;
       put_control();
       end_critical_section(S_CONTROL);
       last_index = time(NULL);

       syslog(LOG_DEBUG, "do_fulltext_indexing() finished");
       is_running = 0;
       return;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void ft_delete_remove ( char *  room,
long  msgnum 
)

Definition at line 488 of file serv_fulltext.c.

{
       if (room) return;
       
       /* Remove from fulltext index */
       if (config.c_enable_fulltext) {
              ft_index_message(msgnum, 0);
       }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void ft_flush_cache ( void  )

Definition at line 86 of file serv_fulltext.c.

                          {
       int i;
       time_t last_update = 0;

       for (i=0; i<65536; ++i) {
              if ((time(NULL) - last_update) >= 10) {
                     syslog(LOG_INFO,
                            "Flushing index cache to disk (%d%% complete)",
                            (i * 100 / 65536)
                     );
                     last_update = time(NULL);
              }
              if (ftc_msgs[i] != NULL) {
                     cdb_store(CDB_FULLTEXT, &i, sizeof(int), ftc_msgs[i],
                            (ftc_num_msgs[i] * sizeof(long)));
                     ftc_num_msgs[i] = 0;
                     free(ftc_msgs[i]);
                     ftc_msgs[i] = NULL;
              }
       }
       syslog(LOG_INFO, "Flushed index cache to disk (100%% complete)");
}

Here is the call graph for this function:

Here is the caller graph for this function:

void ft_index_message ( long  msgnum,
int  op 
)

Definition at line 113 of file serv_fulltext.c.

                                           {
       int num_tokens = 0;
       int *tokens = NULL;
       int i, j;
       struct cdbdata *cdb_bucket;
       StrBuf *msgtext;
       char *txt;
       int tok;
       struct CtdlMessage *msg = NULL;

       msg = CtdlFetchMessage(msgnum, 1);
       if (msg == NULL) {
              syslog(LOG_ERR, "ft_index_message() could not load msg %ld", msgnum);
              return;
       }

       if (msg->cm_fields['1'] != NULL) {
              syslog(LOG_DEBUG, "ft_index_message() excluded msg %ld", msgnum);
              CtdlFreeMessage(msg);
              return;
       }

       syslog(LOG_DEBUG, "ft_index_message() %s msg %ld", (op ? "adding" : "removing") , msgnum);

       /* Output the message as text before indexing it, so we don't end up
        * indexing a bunch of encoded base64, etc.
        */
       CC->redirect_buffer = NewStrBufPlain(NULL, SIZ);
       CtdlOutputPreLoadedMsg(msg, MT_CITADEL, HEADERS_ALL, 0, 1, 0);
       CtdlFreeMessage(msg);
       msgtext = CC->redirect_buffer;
       CC->redirect_buffer = NULL;
       syslog(LOG_DEBUG, "Wordbreaking message %ld...", msgnum);
       if ((msgtext == NULL) || (StrLength(msgtext) == 0)) {
              syslog(LOG_ALERT, "This message has a zero length.  Probable data corruption.");
       }
       txt = SmashStrBuf(&msgtext);
       wordbreaker(txt, &num_tokens, &tokens);
       free(txt);

       syslog(LOG_DEBUG, "Indexing message %ld [%d tokens]", msgnum, num_tokens);
       if (num_tokens > 0) {
              for (i=0; i<num_tokens; ++i) {

                     /* Add the message to the relevant token bucket */

                     /* search for tokens[i] */
                     tok = tokens[i];

                     if ( (tok >= 0) && (tok <= 65535) ) {
                            /* fetch the bucket, Liza */
                            if (ftc_msgs[tok] == NULL) {
                                   cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
                                   if (cdb_bucket != NULL) {
                                          ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
                                          ftc_msgs[tok] = (long *)cdb_bucket->ptr;
                                          cdb_bucket->ptr = NULL;
                                          cdb_free(cdb_bucket);
                                   }
                                   else {
                                          ftc_num_msgs[tok] = 0;
                                          ftc_msgs[tok] = malloc(sizeof(long));
                                   }
                            }
       
       
                            if (op == 1) {       /* add to index */
                                   ++ftc_num_msgs[tok];
                                   ftc_msgs[tok] = realloc(ftc_msgs[tok],
                                                        ftc_num_msgs[tok]*sizeof(long));
                                   ftc_msgs[tok][ftc_num_msgs[tok] - 1] = msgnum;
                            }
       
                            if (op == 0) {       /* remove from index */
                                   if (ftc_num_msgs[tok] >= 1) {
                                          for (j=0; j<ftc_num_msgs[tok]; ++j) {
                                                 if (ftc_msgs[tok][j] == msgnum) {
                                                        memmove(&ftc_msgs[tok][j], &ftc_msgs[tok][j+1], ((ftc_num_msgs[tok] - j - 1)*sizeof(long)));
                                                        --ftc_num_msgs[tok];
                                                        --j;
                                                 }
                                          }
                                   }
                            }
                     }
                     else {
                            syslog(LOG_ALERT, "Invalid token %d !!", tok);
                     }
              }

              free(tokens);
       }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void ft_index_msg ( long  msgnum,
void *  userdata 
)

Definition at line 212 of file serv_fulltext.c.

                                               {

       if ((msgnum > CitControl.MMfulltext) && (msgnum <= ft_newhighest)) {
              ++ft_num_msgs;
              if (ft_num_msgs > ft_num_alloc) {
                     ft_num_alloc += 1024;
                     ft_newmsgs = realloc(ft_newmsgs,
                            (ft_num_alloc * sizeof(long)));
              }
              ft_newmsgs[ft_num_msgs - 1] = msgnum;
       }

}

Here is the caller graph for this function:

void ft_index_room ( struct ctdlroom qrbuf,
void *  data 
)

Definition at line 229 of file serv_fulltext.c.

{
       if (server_shutting_down)
              return;
              
       CtdlGetRoom(&CC->room, qrbuf->QRname);
       CtdlForEachMessage(MSGS_ALL, 0L, NULL, NULL, NULL, ft_index_msg, NULL);
}

Here is the call graph for this function:

Here is the caller graph for this function:

void ft_search ( int *  fts_num_msgs,
long **  fts_msgs,
const char *  search_string 
)

Definition at line 375 of file serv_fulltext.c.

                                                                              {
       int num_tokens = 0;
       int *tokens = NULL;
       int i, j;
       struct cdbdata *cdb_bucket;
       int num_all_msgs = 0;
       long *all_msgs = NULL;
       int num_ret_msgs = 0;
       int num_ret_alloc = 0;
       long *ret_msgs = NULL;
       int tok;

       wordbreaker(search_string, &num_tokens, &tokens);
       if (num_tokens > 0) {
              for (i=0; i<num_tokens; ++i) {

                     /* search for tokens[i] */
                     tok = tokens[i];

                     /* fetch the bucket, Liza */
                     if (ftc_msgs[tok] == NULL) {
                            cdb_bucket = cdb_fetch(CDB_FULLTEXT, &tok, sizeof(int));
                            if (cdb_bucket != NULL) {
                                   ftc_num_msgs[tok] = cdb_bucket->len / sizeof(long);
                                   ftc_msgs[tok] = (long *)cdb_bucket->ptr;
                                   cdb_bucket->ptr = NULL;
                                   cdb_free(cdb_bucket);
                            }
                            else {
                                   ftc_num_msgs[tok] = 0;
                                   ftc_msgs[tok] = malloc(sizeof(long));
                            }
                     }

                     num_all_msgs += ftc_num_msgs[tok];
                     if (num_all_msgs > 0) {
                            all_msgs = realloc(all_msgs, num_all_msgs*sizeof(long) );
                            memcpy(&all_msgs[num_all_msgs-ftc_num_msgs[tok]],
                                   ftc_msgs[tok], ftc_num_msgs[tok]*sizeof(long) );
                     }

              }
              free(tokens);
              if (all_msgs != NULL) {
                     qsort(all_msgs, num_all_msgs, sizeof(long), longcmp);

                     /*
                      * At this point, if a message appears num_tokens times in the
                      * list, then it contains all of the search tokens.
                      */
                     if (num_all_msgs >= num_tokens)
                            for (j=0; j<(num_all_msgs-num_tokens+1); ++j) {
                                   if (all_msgs[j] == all_msgs[j+num_tokens-1]) {
                                          
                                          ++num_ret_msgs;
                                          if (num_ret_msgs > num_ret_alloc) {
                                                 num_ret_alloc += 64;
                                                 ret_msgs = realloc(ret_msgs,
                                                                  (num_ret_alloc*sizeof(long)) );
                                          }
                                          ret_msgs[num_ret_msgs - 1] = all_msgs[j];
                                          
                                   }
                            }
                     free(all_msgs);
              }
       }

       *fts_num_msgs = num_ret_msgs;
       *fts_msgs = ret_msgs;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void initialize_ft_cache ( void  )

Definition at line 482 of file serv_fulltext.c.

                               {
       memset(ftc_num_msgs, 0, (65536 * sizeof(int)));
       memset(ftc_msgs, 0, (65536 * sizeof(long *)));
}

Here is the caller graph for this function:

int longcmp ( const void *  rec1,
const void *  rec2 
)

Definition at line 72 of file serv_fulltext.c.

                                                {
       long i1, i2;

       i1 = *(const long *)rec1;
       i2 = *(const long *)rec2;

       if (i1 > i2) return(1);
       if (i1 < i2) return(-1);
       return(0);
}

Here is the caller graph for this function:


Variable Documentation

long ft_newhighest = 0L

Definition at line 60 of file serv_fulltext.c.

long* ft_newmsgs = NULL

Definition at line 61 of file serv_fulltext.c.

int ft_num_alloc = 0

Definition at line 63 of file serv_fulltext.c.

int ft_num_msgs = 0

Definition at line 62 of file serv_fulltext.c.

long* ftc_msgs[65536]

Definition at line 66 of file serv_fulltext.c.

int ftc_num_msgs[65536]

Definition at line 65 of file serv_fulltext.c.