/*
- * Copyright (C) 1994-1996, Index Data I/S
+ * Copyright (C) 1994-1999, Index Data
* All rights reserved.
- * Sebastian Hammer, Adam Dickmeiss
- *
- * $Log: kinput.c,v $
- * Revision 1.19 1996-10-29 14:09:46 adam
- * Use of cisam system - enabled if setting isamc is 1.
- *
- * Revision 1.18 1996/06/04 10:18:59 adam
- * Minor changes - removed include of ctype.h.
- *
- * Revision 1.17 1996/05/14 15:47:07 adam
- * Cleanup of various buffer size entities.
- *
- * Revision 1.16 1996/04/09 10:05:20 adam
- * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
- *
- * Revision 1.15 1996/03/21 14:50:09 adam
- * File update uses modify-time instead of change-time.
- *
- * Revision 1.14 1996/02/07 14:06:37 adam
- * Better progress report during register merge.
- * New command: clean - removes temporary shadow files.
- *
- * Revision 1.13 1996/02/05 12:30:00 adam
- * Logging reduced a bit.
- * The remaining running time is estimated during register merge.
- *
- * Revision 1.12 1995/12/06 17:49:19 adam
- * Uses dict_delete now.
- *
- * Revision 1.11 1995/12/06 16:06:43 adam
- * Better diagnostics. Work on 'real' dictionary deletion.
- *
- * Revision 1.10 1995/12/06 12:41:22 adam
- * New command 'stat' for the index program.
- * Filenames can be read from stdin by specifying '-'.
- * Bug fix/enhancement of the transformation from terms to regular
- * expressons in the search engine.
- *
- * Revision 1.9 1995/10/10 12:24:39 adam
- * Temporary sort files are compressed.
- *
- * Revision 1.8 1995/10/04 16:57:19 adam
- * Key input and merge sort in one pass.
- *
- * Revision 1.7 1995/10/02 15:18:52 adam
- * New member in recRetrieveCtrl: diagnostic.
- *
- * Revision 1.6 1995/09/29 15:51:56 adam
- * First work on multi-way read.
- *
- * Revision 1.5 1995/09/29 14:01:43 adam
- * Bug fixes.
- *
- * Revision 1.4 1995/09/28 14:22:57 adam
- * Sort uses smaller temporary files.
- *
- * Revision 1.3 1995/09/06 16:11:17 adam
- * Option: only one word key per file.
- *
- * Revision 1.2 1995/09/04 12:33:42 adam
- * Various cleanup. YAZ util used instead.
- *
- * Revision 1.1 1995/09/04 09:10:37 adam
- * More work on index add/del/update.
- * Merge sort implemented.
- * Initial work on z39 server.
+ * Sebastian Hammer, Adam Dickmeiss, Heikki Levanto
*
+ * (log at the end)
+
*/
-
+
#include <fcntl.h>
+#ifdef WIN32
+#include <io.h>
+#else
#include <unistd.h>
+#endif
#include <stdlib.h>
#include <string.h>
#include <stdio.h>
void key_file_chunk_read (struct key_file *f)
{
- int nr = 0, r, fd;
+ int nr = 0, r = 0, fd;
char fname[1024];
getFnameTmp (fname, f->no);
- fd = open (fname, O_RDONLY);
+ fd = open (fname, O_BINARY|O_RDONLY);
if (fd == -1)
{
logf (LOG_FATAL|LOG_ERRNO, "cannot open %s", fname);
{
struct key_file *f;
- f = xmalloc (sizeof(*f));
+ f = (struct key_file *) xmalloc (sizeof(*f));
f->sysno = 0;
f->seqno = 0;
f->no = no;
f->offset = 0;
f->length = 0;
f->readHandler = NULL;
- f->buf = xmalloc (f->chunk);
- f->prev_name = xmalloc (INP_NAME_MAX);
+ f->buf = (unsigned char *) xmalloc (f->chunk);
+ f->prev_name = (char *) xmalloc (INP_NAME_MAX);
*f->prev_name = '\0';
key_file_chunk_read (f);
return f;
Dict dict;
ISAM isam;
ISAMC isamc;
+ ISAMS isams;
+ ISAMH isamh;
};
struct heap_info *key_heap_init (int nkeys,
struct heap_info *hi;
int i;
- hi = xmalloc (sizeof(*hi));
- hi->info.file = xmalloc (sizeof(*hi->info.file) * (1+nkeys));
- hi->info.buf = xmalloc (sizeof(*hi->info.buf) * (1+nkeys));
+ hi = (struct heap_info *) xmalloc (sizeof(*hi));
+ hi->info.file = (struct key_file **)
+ xmalloc (sizeof(*hi->info.file) * (1+nkeys));
+ hi->info.buf = (char **) xmalloc (sizeof(*hi->info.buf) * (1+nkeys));
hi->heapnum = 0;
- hi->ptr = xmalloc (sizeof(*hi->ptr) * (1+nkeys));
+ hi->ptr = (int *) xmalloc (sizeof(*hi->ptr) * (1+nkeys));
hi->cmp = cmp;
for (i = 0; i<= nkeys; i++)
{
hi->ptr[i] = i;
- hi->info.buf[i] = xmalloc (INP_NAME_MAX);
+ hi->info.buf[i] = (char *) xmalloc (INP_NAME_MAX);
}
return hi;
}
int heap_cread_item (void *vp, char **dst, int *insertMode)
{
- struct heap_cread_info *p = vp;
+ struct heap_cread_info *p = (struct heap_cread_info *) vp;
struct heap_info *hi = p->hi;
if (p->mode == 1)
int heap_inpc (struct heap_info *hi)
{
struct heap_cread_info hci;
- ISAMC_P isamc_p;
- ISAMC_I isamc_i = xmalloc (sizeof(*isamc_i));
- void *dict_info;
+ ISAMC_I isamc_i = (ISAMC_I) xmalloc (sizeof(*isamc_i));
- hci.key = xmalloc (KEY_SIZE);
+ hci.key = (char *) xmalloc (KEY_SIZE);
hci.mode = 1;
hci.hi = hi;
hci.more = heap_read_one (hi, hci.cur_name, hci.key);
while (hci.more)
{
- logf (LOG_DEBUG, "inserting %s", 1+hci.cur_name);
+ char this_name[INP_NAME_MAX];
+ ISAMC_P isamc_p, isamc_p2;
+ char *dict_info;
+
+ strcpy (this_name, hci.cur_name);
+ assert (hci.cur_name[1]);
+ no_diffs++;
if ((dict_info = dict_lookup (hi->dict, hci.cur_name)))
{
- logf (LOG_FATAL, "Cannot merge really yet! %s", hci.cur_name);
- exit (1);
+ memcpy (&isamc_p, dict_info+1, sizeof(ISAMC_P));
+ isamc_p2 = isc_merge (hi->isamc, isamc_p, isamc_i);
+ if (!isamc_p2)
+ {
+ no_deletions++;
+ if (!dict_delete (hi->dict, this_name))
+ abort();
+ }
+ else
+ {
+ no_updates++;
+ if (isamc_p2 != isamc_p)
+ dict_insert (hi->dict, this_name,
+ sizeof(ISAMC_P), &isamc_p2);
+ }
}
else
{
- char this_name[128];
- strcpy (this_name, hci.cur_name);
isamc_p = isc_merge (hi->isamc, 0, isamc_i);
no_insertions++;
dict_insert (hi->dict, this_name, sizeof(ISAMC_P), &isamc_p);
return 0;
}
+int heap_inps (struct heap_info *hi)
+{
+ struct heap_cread_info hci;
+ ISAMS_I isams_i = (ISAMS_I) xmalloc (sizeof(*isams_i));
+
+ hci.key = (char *) xmalloc (KEY_SIZE);
+ hci.mode = 1;
+ hci.hi = hi;
+ hci.more = heap_read_one (hi, hci.cur_name, hci.key);
+
+ isams_i->clientData = &hci;
+ isams_i->read_item = heap_cread_item;
+
+ while (hci.more)
+ {
+ char this_name[INP_NAME_MAX];
+ ISAMS_P isams_p;
+ char *dict_info;
+
+ strcpy (this_name, hci.cur_name);
+ assert (hci.cur_name[1]);
+ no_diffs++;
+ if (!(dict_info = dict_lookup (hi->dict, hci.cur_name)))
+ {
+ isams_p = isams_merge (hi->isams, isams_i);
+ no_insertions++;
+ dict_insert (hi->dict, this_name, sizeof(ISAMS_P), &isams_p);
+ }
+ else
+ abort();
+ }
+ xfree (isams_i);
+ return 0;
+}
+
+int heap_inph (struct heap_info *hi)
+{
+ struct heap_cread_info hci;
+ ISAMH_I isamh_i = (ISAMH_I) xmalloc (sizeof(*isamh_i));
+
+ hci.key = (char *) xmalloc (KEY_SIZE);
+ hci.mode = 1;
+ hci.hi = hi;
+ hci.more = heap_read_one (hi, hci.cur_name, hci.key);
+
+ isamh_i->clientData = &hci;
+ isamh_i->read_item = heap_cread_item;
+
+ while (hci.more)
+ {
+ char this_name[INP_NAME_MAX];
+ ISAMH_P isamh_p, isamh_p2;
+ char *dict_info;
+
+ strcpy (this_name, hci.cur_name);
+ assert (hci.cur_name[1]);
+ no_diffs++;
+ if ((dict_info = dict_lookup (hi->dict, hci.cur_name)))
+ {
+ memcpy (&isamh_p, dict_info+1, sizeof(ISAMH_P));
+ isamh_p2 = isamh_append (hi->isamh, isamh_p, isamh_i);
+ if (!isamh_p2)
+ {
+ no_deletions++;
+ if (!dict_delete (hi->dict, this_name))
+ abort();
+ }
+ else
+ {
+ no_updates++;
+ if (isamh_p2 != isamh_p)
+ dict_insert (hi->dict, this_name,
+ sizeof(ISAMH_P), &isamh_p2);
+ }
+ }
+ else
+ {
+ isamh_p = isamh_append (hi->isamh, 0, isamh_i);
+ no_insertions++;
+ dict_insert (hi->dict, this_name, sizeof(ISAMH_P), &isamh_p);
+ }
+ }
+ xfree (isamh_i);
+ return 0;
+}
+
int heap_inp (struct heap_info *hi)
{
char *info;
char *key_buf;
int more;
- next_key = xmalloc (KEY_SIZE);
- key_buf = xmalloc (key_buf_size);
+ next_key = (char *) xmalloc (KEY_SIZE);
+ key_buf = (char *) xmalloc (key_buf_size);
more = heap_read_one (hi, cur_name, key_buf);
while (more) /* EOF ? */
{
break;
memcpy (key_buf + key_buf_ptr, next_key, KEY_SIZE);
key_buf_ptr += KEY_SIZE;
- if (key_buf_ptr+KEY_SIZE >= key_buf_size)
+ if (key_buf_ptr+(int) KEY_SIZE >= key_buf_size)
{
char *new_key_buf;
- new_key_buf = xmalloc (key_buf_size + INP_BUF_ADD);
+ new_key_buf = (char *) xmalloc (key_buf_size + INP_BUF_ADD);
memcpy (new_key_buf, key_buf, key_buf_size);
key_buf_size += INP_BUF_ADD;
xfree (key_buf);
}
no_diffs++;
nmemb = key_buf_ptr / KEY_SIZE;
- assert (nmemb*KEY_SIZE == key_buf_ptr);
+ assert (nmemb * (int) KEY_SIZE == key_buf_ptr);
if ((info = dict_lookup (hi->dict, cur_name)))
{
ISAM_P isam_p, isam_p2;
- logf (LOG_DEBUG, "updating %s", 1+cur_name);
memcpy (&isam_p, info+1, sizeof(ISAM_P));
isam_p2 = is_merge (hi->isam, isam_p, nmemb, key_buf);
if (!isam_p2)
else
{
ISAM_P isam_p;
- logf (LOG_DEBUG, "inserting %s", 1+cur_name);
no_insertions++;
isam_p = is_merge (hi->isam, 0, nmemb, key_buf);
dict_insert (hi->dict, cur_name, sizeof(ISAM_P), &isam_p);
void progressFunc (struct key_file *keyp, void *info)
{
- struct progressInfo *p = info;
+ struct progressInfo *p = (struct progressInfo *) info;
time_t now, remaining;
if (keyp->buf_size <= 0 || p->totalBytes <= 0)
if (now >= p->lastTime+10)
{
p->lastTime = now;
- remaining = (now - p->startTime)*
- ((double) p->totalBytes/p->totalOffset - 1.0);
+ remaining = (time_t) ((now - p->startTime)*
+ ((double) p->totalBytes/p->totalOffset - 1.0));
if (remaining <= 130)
logf (LOG_LOG, "Merge %2.1f%% completed; %ld seconds remaining",
(100.0*p->totalOffset) / p->totalBytes, (long) remaining);
p->totalOffset += keyp->buf_size;
}
-void key_input (int nkeys, int cache)
+#ifndef R_OK
+#define R_OK 4
+#endif
+
+void key_input (BFiles bfs, int nkeys, int cache)
{
Dict dict;
ISAM isam = NULL;
ISAMC isamc = NULL;
+ ISAMS isams = NULL;
+ ISAMH isamh = NULL;
struct key_file **kf;
char rbuf[1024];
int i, r;
if (!nkeys)
return ;
}
- dict = dict_open (FNAME_DICT, cache, 1);
+ dict = dict_open (bfs, FNAME_DICT, cache, 1, 0);
if (!dict)
{
logf (LOG_FATAL, "dict_open fail");
exit (1);
}
- if (res_get_match (common_resource, "isam", "c", NULL))
+ if (res_get_match (common_resource, "isam", "s", NULL))
{
- isamc = isc_open (FNAME_ISAMC, 1, key_isamc_m ());
- if (!isamc)
+ isams = isams_open (bfs, FNAME_ISAMS, 1,
+ key_isams_m (common_resource));
+ if (!isams)
{
- logf (LOG_FATAL, "isc_open fail");
- exit (1);
+ logf (LOG_FATAL, "isams_open fail");
+ exit (1);
}
}
- else
+ else if (res_get_match (common_resource, "isam", "i", NULL))
{
- isam = is_open (FNAME_ISAM, key_compare, 1, sizeof(struct it_key));
+ isam = is_open (bfs, FNAME_ISAM, key_compare, 1,
+ sizeof(struct it_key), common_resource);
if (!isam)
{
logf (LOG_FATAL, "is_open fail");
exit (1);
}
}
- kf = xmalloc ((1+nkeys) * sizeof(*kf));
+ else if (res_get_match (common_resource, "isam", "h", NULL))
+ {
+ isamh = isamh_open (bfs, FNAME_ISAMH, 1,
+ key_isamh_m (common_resource));
+ if (!isamh)
+ {
+ logf (LOG_FATAL, "isamh_open fail");
+ exit (1);
+ }
+ }
+ else
+ {
+ isamc = isc_open (bfs, FNAME_ISAMC, 1,
+ key_isamc_m (common_resource));
+ if (!isamc)
+ {
+ logf (LOG_FATAL, "isc_open fail");
+ exit (1);
+ }
+ }
+ kf = (struct key_file **) xmalloc ((1+nkeys) * sizeof(*kf));
progressInfo.totalBytes = 0;
progressInfo.totalOffset = 0;
time (&progressInfo.startTime);
hi->dict = dict;
hi->isam = isam;
hi->isamc = isamc;
-
+ hi->isams = isams;
+ hi->isamh = isamh;
+
for (i = 1; i<=nkeys; i++)
if ((r = key_file_read (kf[i], rbuf)))
key_heap_insert (hi, rbuf, r, kf[i]);
if (isamc)
heap_inpc (hi);
- else
- heap_inp (hi);
+ else if (isams)
+ heap_inps (hi);
+ else if (isam)
+ heap_inp (hi);
+ else if (isamh)
+ heap_inph (hi);
+
dict_close (dict);
if (isam)
is_close (isam);
if (isamc)
isc_close (isamc);
-
+ if (isams)
+ isams_close (isams);
+ if (isamh)
+ isamh_close (isamh);
+
for (i = 1; i<=nkeys; i++)
{
getFnameTmp (rbuf, i);
}
+
+/*
+ * $Log: kinput.c,v $
+ * Revision 1.35 1999-06-30 15:07:23 heikki
+ * Adding isamh stuff
+ *
+ * Revision 1.34 1999/05/26 07:49:13 adam
+ * C++ compilation.
+ *
+ * Revision 1.33 1999/05/15 14:36:38 adam
+ * Updated dictionary. Implemented "compression" of dictionary.
+ *
+ * Revision 1.32 1999/05/12 13:08:06 adam
+ * First version of ISAMS.
+ *
+ * Revision 1.31 1999/02/02 14:50:56 adam
+ * Updated WIN32 code specific sections. Changed header.
+ *
+ * Revision 1.30 1998/10/28 10:53:57 adam
+ * Added type cast to prevent warning.
+ *
+ * Revision 1.29 1998/06/11 15:41:39 adam
+ * Minor changes.
+ *
+ * Revision 1.28 1998/03/05 08:45:12 adam
+ * New result set model and modular ranking system. Moved towards
+ * descent server API. System information stored as "SGML" records.
+ *
+ * Revision 1.27 1998/02/17 10:32:52 adam
+ * Fixed bug: binary files weren't opened with flag b on NT.
+ *
+ * Revision 1.26 1998/01/29 13:39:13 adam
+ * Compress ISAM is default.
+ *
+ * Revision 1.25 1997/09/17 12:19:14 adam
+ * Zebra version corresponds to YAZ version 1.4.
+ * Changed Zebra server so that it doesn't depend on global common_resource.
+ *
+ * Revision 1.24 1997/09/09 13:38:07 adam
+ * Partial port to WIN95/NT.
+ *
+ * Revision 1.23 1997/09/04 13:57:39 adam
+ * Added O_BINARY for open calls.
+ *
+ * Revision 1.22 1997/02/12 20:39:45 adam
+ * Implemented options -f <n> that limits the log to the first <n>
+ * records.
+ * Changed some log messages also.
+ *
+ * Revision 1.21 1996/11/08 11:10:23 adam
+ * Buffers used during file match got bigger.
+ * Compressed ISAM support everywhere.
+ * Bug fixes regarding masking characters in queries.
+ * Redesigned Regexp-2 queries.
+ *
+ * Revision 1.20 1996/11/01 08:58:41 adam
+ * Interface to isamc system now includes update and delete.
+ *
+ * Revision 1.19 1996/10/29 14:09:46 adam
+ * Use of cisam system - enabled if setting isamc is 1.
+ *
+ * Revision 1.18 1996/06/04 10:18:59 adam
+ * Minor changes - removed include of ctype.h.
+ *
+ * Revision 1.17 1996/05/14 15:47:07 adam
+ * Cleanup of various buffer size entities.
+ *
+ * Revision 1.16 1996/04/09 10:05:20 adam
+ * Bug fix: prev_name buffer possibly too small; allocated in key_file_init.
+ *
+ * Revision 1.15 1996/03/21 14:50:09 adam
+ * File update uses modify-time instead of change-time.
+ *
+ * Revision 1.14 1996/02/07 14:06:37 adam
+ * Better progress report during register merge.
+ * New command: clean - removes temporary shadow files.
+ *
+ * Revision 1.13 1996/02/05 12:30:00 adam
+ * Logging reduced a bit.
+ * The remaining running time is estimated during register merge.
+ *
+ * Revision 1.12 1995/12/06 17:49:19 adam
+ * Uses dict_delete now.
+ *
+ * Revision 1.11 1995/12/06 16:06:43 adam
+ * Better diagnostics. Work on 'real' dictionary deletion.
+ *
+ * Revision 1.10 1995/12/06 12:41:22 adam
+ * New command 'stat' for the index program.
+ * Filenames can be read from stdin by specifying '-'.
+ * Bug fix/enhancement of the transformation from terms to regular
+ * expressons in the search engine.
+ *
+ * Revision 1.9 1995/10/10 12:24:39 adam
+ * Temporary sort files are compressed.
+ *
+ * Revision 1.8 1995/10/04 16:57:19 adam
+ * Key input and merge sort in one pass.
+ *
+ * Revision 1.7 1995/10/02 15:18:52 adam
+ * New member in recRetrieveCtrl: diagnostic.
+ *
+ * Revision 1.6 1995/09/29 15:51:56 adam
+ * First work on multi-way read.
+ *
+ * Revision 1.5 1995/09/29 14:01:43 adam
+ * Bug fixes.
+ *
+ * Revision 1.4 1995/09/28 14:22:57 adam
+ * Sort uses smaller temporary files.
+ *
+ * Revision 1.3 1995/09/06 16:11:17 adam
+ * Option: only one word key per file.
+ *
+ * Revision 1.2 1995/09/04 12:33:42 adam
+ * Various cleanup. YAZ util used instead.
+ *
+ * Revision 1.1 1995/09/04 09:10:37 adam
+ * More work on index add/del/update.
+ * Merge sort implemented.
+ * Initial work on z39 server.
+ *
+ */
+