-/*
- * Copyright (C) 1994-1998, Index Data I/S
- * All rights reserved.
- * Sebastian Hammer, Adam Dickmeiss
- *
- * $Log: trunc.c,v $
- * Revision 1.11 1998-03-25 13:48:02 adam
- * Fixed bug in rset_trunc_r.
- *
- * Revision 1.10 1998/03/05 08:45:13 adam
- * New result set model and modular ranking system. Moved towards
- * descent server API. System information stored as "SGML" records.
- *
- * Revision 1.9 1998/01/12 15:04:09 adam
- * The test option (-s) only uses read-lock (and not write lock).
- *
- * Revision 1.8 1997/10/31 12:34:27 adam
- * Bug fix: memory leak.
- *
- * Revision 1.7 1997/09/29 09:07:29 adam
- * Minor change.
- *
- * Revision 1.6 1997/09/22 12:39:06 adam
- * Added get_pos method for the ranked result sets.
- *
- * Revision 1.5 1997/09/17 12:19:17 adam
- * Zebra version corresponds to YAZ version 1.4.
- * Changed Zebra server so that it doesn't depend on global common_resource.
- *
- * Revision 1.4 1996/12/23 15:30:44 adam
- * Work on truncation.
- * Bug fix: result sets weren't deleted after server shut down.
- *
- * Revision 1.3 1996/12/20 11:07:14 adam
- * Multi-or result set.
- *
- * Revision 1.2 1996/11/08 11:10:28 adam
- * Buffers used during file match got bigger.
- * Compressed ISAM support everywhere.
- * Bug fixes regarding masking characters in queries.
- * Redesigned Regexp-2 queries.
- *
- * Revision 1.1 1996/11/04 14:07:40 adam
- * Moved truncation code to trunc.c.
- *
- */
+/* $Id: trunc.c,v 1.27 2002-08-02 19:26:55 adam Exp $
+ Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
+ Index Data Aps
+
+This file is part of the Zebra server.
+
+Zebra is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 2, or (at your option) any later
+version.
+
+Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with Zebra; see the file LICENSE.zebra. If not, write to the
+Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
+02111-1307, USA.
+*/
+
+
#include <stdio.h>
#include <assert.h>
-#include "zserver.h"
+#define NEW_TRUNC 1
+
+#include "index.h"
#include <rstemp.h>
+#include <rsnull.h>
+#include <rsisams.h>
#include <rsisam.h>
#include <rsisamc.h>
-#include <rsnull.h>
-
-#define NEW_TRUNC 1
-
+#include <rsisamd.h>
+#include <rsisamb.h>
#if NEW_TRUNC
#include <rsm_or.h>
#endif
int (*cmp)(const void *p1,
const void *p2))
{
- struct trunc_info *ti = xmalloc (sizeof(*ti));
+ struct trunc_info *ti = (struct trunc_info *) xmalloc (sizeof(*ti));
int i;
++size;
ti->heapnum = 0;
ti->keysize = key_size;
ti->cmp = cmp;
- ti->indx = xmalloc (size * sizeof(*ti->indx));
- ti->heap = xmalloc (size * sizeof(*ti->heap));
- ti->ptr = xmalloc (size * sizeof(*ti->ptr));
- ti->swapbuf = xmalloc (ti->keysize);
- ti->tmpbuf = xmalloc (ti->keysize);
- ti->buf = xmalloc (size * ti->keysize);
+ ti->indx = (int *) xmalloc (size * sizeof(*ti->indx));
+ ti->heap = (char **) xmalloc (size * sizeof(*ti->heap));
+ ti->ptr = (int *) xmalloc (size * sizeof(*ti->ptr));
+ ti->swapbuf = (char *) xmalloc (ti->keysize);
+ ti->tmpbuf = (char *) xmalloc (ti->keysize);
+ ti->buf = (char *) xmalloc (size * ti->keysize);
for (i = size; --i >= 0; )
{
ti->ptr[i] = i;
}
static RSET rset_trunc_r (ZebraHandle zi, const char *term, int length,
- const char *flags, ISAM_P *isam_p, int from, int to,
- int merge_chunk)
+ const char *flags, ISAMS_P *isam_p, int from, int to,
+ int merge_chunk, int preserve_position,
+ int term_type)
{
RSET result;
RSFD result_rsfd;
rset_temp_parms parms;
+ parms.cmp = key_compare_it;
parms.key_size = sizeof(struct it_key);
parms.temp_path = res_get (zi->res, "setTmpDir");
- parms.rset_term = rset_term_create (term, length, flags);
+ parms.rset_term = rset_term_create (term, length, flags, term_type);
result = rset_create (rset_kind_temp, &parms);
result_rsfd = rset_open (result, RSETF_WRITE);
int rscur = 0;
int rsmax = (to-from)/i_add + 1;
- rset = xmalloc (sizeof(*rset) * rsmax);
- rsfd = xmalloc (sizeof(*rsfd) * rsmax);
+ rset = (RSET *) xmalloc (sizeof(*rset) * rsmax);
+ rsfd = (RSFD *) xmalloc (sizeof(*rsfd) * rsmax);
for (i = from; i < to; i += i_add)
{
if (i_add <= to - i)
rset[rscur] = rset_trunc_r (zi, term, length, flags,
- isam_p, i, i+i_add, merge_chunk);
+ isam_p, i, i+i_add,
+ merge_chunk, preserve_position,
+ term_type);
else
rset[rscur] = rset_trunc_r (zi, term, length, flags,
- isam_p, i, to, merge_chunk);
+ isam_p, i, to,
+ merge_chunk, preserve_position,
+ term_type);
rscur++;
}
ti = heap_init (rscur, sizeof(struct it_key), key_compare_it);
xfree (rsfd);
heap_close (ti);
}
- else if (zi->isam)
+ else if (zi->reg->isam)
{
ISPT *ispt;
int i;
struct trunc_info *ti;
- ispt = xmalloc (sizeof(*ispt) * (to-from));
+ ispt = (ISPT *) xmalloc (sizeof(*ispt) * (to-from));
ti = heap_init (to-from, sizeof(struct it_key),
key_compare_it);
for (i = to-from; --i >= 0; )
{
- ispt[i] = is_position (zi->isam, isam_p[from+i]);
+ ispt[i] = is_position (zi->reg->isam, isam_p[from+i]);
if (is_readkey (ispt[i], ti->tmpbuf))
heap_insert (ti, ti->tmpbuf, i);
else
int n = ti->indx[ti->ptr[1]];
rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
-#if 1
+ if (preserve_position)
+ {
+/* section that preserve all keys */
+ heap_delete (ti);
+ if (is_readkey (ispt[n], ti->tmpbuf))
+ heap_insert (ti, ti->tmpbuf, n);
+ else
+ is_pt_free (ispt[n]);
+ }
+ else
+ {
+/* section that preserve all keys with unique sysnos */
+ while (1)
+ {
+ if (!is_readkey (ispt[n], ti->tmpbuf))
+ {
+ heap_delete (ti);
+ is_pt_free (ispt[n]);
+ break;
+ }
+ if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
+ {
+ heap_delete (ti);
+ heap_insert (ti, ti->tmpbuf, n);
+ break;
+ }
+ }
+ }
+ }
+ heap_close (ti);
+ xfree (ispt);
+ }
+ else if (zi->reg->isamc)
+ {
+ ISAMC_PP *ispt;
+ int i;
+ struct trunc_info *ti;
+
+ ispt = (ISAMC_PP *) xmalloc (sizeof(*ispt) * (to-from));
+
+ ti = heap_init (to-from, sizeof(struct it_key),
+ key_compare_it);
+ for (i = to-from; --i >= 0; )
+ {
+ ispt[i] = isc_pp_open (zi->reg->isamc, isam_p[from+i]);
+ if (isc_pp_read (ispt[i], ti->tmpbuf))
+ heap_insert (ti, ti->tmpbuf, i);
+ else
+ isc_pp_close (ispt[i]);
+ }
+ while (ti->heapnum)
+ {
+ int n = ti->indx[ti->ptr[1]];
+
+ rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
+ if (preserve_position)
+ {
+ heap_delete (ti);
+ if (isc_pp_read (ispt[n], ti->tmpbuf))
+ heap_insert (ti, ti->tmpbuf, n);
+ else
+ isc_pp_close (ispt[n]);
+ }
+ else
+ {
+ while (1)
+ {
+ if (!isc_pp_read (ispt[n], ti->tmpbuf))
+ {
+ heap_delete (ti);
+ isc_pp_close (ispt[n]);
+ break;
+ }
+ if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
+ {
+ heap_delete (ti);
+ heap_insert (ti, ti->tmpbuf, n);
+ break;
+ }
+ }
+ }
+ }
+ heap_close (ti);
+ xfree (ispt);
+ }
+
+ else if (zi->reg->isamd)
+ {
+ ISAMD_PP *ispt;
+ int i;
+ struct trunc_info *ti;
+
+ ispt = (ISAMD_PP *) xmalloc (sizeof(*ispt) * (to-from));
+
+ ti = heap_init (to-from, sizeof(struct it_key),
+ key_compare_it);
+ for (i = to-from; --i >= 0; )
+ {
+ logf(LOG_FATAL, "isam_d does not (currently) support truncs");
+ abort();
+ /*ispt[i] = isamd_pp_open (zi->reg->isamd, isam_p[from+i]); */
+ if (isamd_pp_read (ispt[i], ti->tmpbuf))
+ heap_insert (ti, ti->tmpbuf, i);
+ else
+ isamd_pp_close (ispt[i]);
+ }
+ while (ti->heapnum)
+ {
+ int n = ti->indx[ti->ptr[1]];
+
+ rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
+#if 0
/* section that preserve all keys */
heap_delete (ti);
- if (is_readkey (ispt[n], ti->tmpbuf))
+ if (isamd_pp_read (ispt[n], ti->tmpbuf))
heap_insert (ti, ti->tmpbuf, n);
else
- is_pt_free (ispt[n]);
+ isamd_pp_close (ispt[n]);
#else
/* section that preserve all keys with unique sysnos */
while (1)
{
- if (!is_readkey (ispt[n], ti->tmpbuf))
+ if (!isamd_pp_read (ispt[n], ti->tmpbuf))
{
heap_delete (ti);
- is_pt_free (ispt[n]);
+ isamd_pp_close (ispt[n]);
break;
}
if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
heap_close (ti);
xfree (ispt);
}
- else
+ else if (zi->reg->isams)
{
- ISAMC_PP *ispt;
+ ISAMS_PP *ispt;
int i;
struct trunc_info *ti;
- ispt = xmalloc (sizeof(*ispt) * (to-from));
+ ispt = (ISAMS_PP *) xmalloc (sizeof(*ispt) * (to-from));
ti = heap_init (to-from, sizeof(struct it_key),
key_compare_it);
for (i = to-from; --i >= 0; )
{
- ispt[i] = isc_pp_open (zi->isamc, isam_p[from+i]);
- if (isc_pp_read (ispt[i], ti->tmpbuf))
+ ispt[i] = isams_pp_open (zi->reg->isams, isam_p[from+i]);
+ if (isams_pp_read (ispt[i], ti->tmpbuf))
heap_insert (ti, ti->tmpbuf, i);
else
- isc_pp_close (ispt[i]);
+ isams_pp_close (ispt[i]);
}
while (ti->heapnum)
{
int n = ti->indx[ti->ptr[1]];
rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
-#if 0
-/* section that preserve all keys */
- heap_delete (ti);
- if (is_readkey (ispt[n], ti->tmpbuf))
- heap_insert (ti, ti->tmpbuf, n);
- else
- isc_pp_close (ispt[n]);
-#else
-/* section that preserve all keys with unique sysnos */
while (1)
{
- if (!isc_pp_read (ispt[n], ti->tmpbuf))
+ if (!isams_pp_read (ispt[n], ti->tmpbuf))
{
heap_delete (ti);
- isc_pp_close (ispt[n]);
+ isams_pp_close (ispt[n]);
break;
}
if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
break;
}
}
-#endif
}
heap_close (ti);
xfree (ispt);
}
+ else if (zi->reg->isamb)
+ {
+ ISAMB_PP *ispt;
+ int i;
+ struct trunc_info *ti;
+
+ ispt = (ISAMB_PP *) xmalloc (sizeof(*ispt) * (to-from));
+
+ ti = heap_init (to-from, sizeof(struct it_key),
+ key_compare_it);
+ for (i = to-from; --i >= 0; )
+ {
+ ispt[i] = isamb_pp_open (zi->reg->isamb, isam_p[from+i]);
+ if (isamb_pp_read (ispt[i], ti->tmpbuf))
+ heap_insert (ti, ti->tmpbuf, i);
+ else
+ isamb_pp_close (ispt[i]);
+ }
+ while (ti->heapnum)
+ {
+ int n = ti->indx[ti->ptr[1]];
+
+ rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
+
+ if (preserve_position)
+ {
+ heap_delete (ti);
+ if (isamb_pp_read (ispt[n], ti->tmpbuf))
+ heap_insert (ti, ti->tmpbuf, n);
+ else
+ isamb_pp_close (ispt[n]);
+ }
+ else
+ {
+ while (1)
+ {
+ if (!isamb_pp_read (ispt[n], ti->tmpbuf))
+ {
+ heap_delete (ti);
+ isamb_pp_close (ispt[n]);
+ break;
+ }
+ if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
+ {
+ heap_delete (ti);
+ heap_insert (ti, ti->tmpbuf, n);
+ break;
+ }
+ }
+ }
+ }
+ heap_close (ti);
+ xfree (ispt);
+ }
+ else
+ logf (LOG_WARN, "Unknown isam set in rset_trunc_r");
+
rset_close (result, result_rsfd);
return result;
}
+static int isams_trunc_cmp (const void *p1, const void *p2)
+{
+ ISAMS_P i1 = *(ISAMS_P*) p1;
+ ISAMS_P i2 = *(ISAMS_P*) p2;
+
+ return i1 - i2;
+}
+
static int isam_trunc_cmp (const void *p1, const void *p2)
{
ISAM_P i1 = *(ISAM_P*) p1;
return d;
return isc_block (i1) - isc_block (i2);
}
+static int isamd_trunc_cmp (const void *p1, const void *p2)
+{
+ ISAMD_P i1 = *(ISAMD_P*) p1;
+ ISAMD_P i2 = *(ISAMD_P*) p2;
+ int d;
+
+ d = isamd_type (i1) - isamd_type (i2);
+ if (d)
+ return d;
+ return isamd_block (i1) - isamd_block (i2);
+}
-RSET rset_trunc (ZebraHandle zi, ISAM_P *isam_p, int no,
- const char *term, int length, const char *flags)
+RSET rset_trunc (ZebraHandle zi, ISAMS_P *isam_p, int no,
+ const char *term, int length, const char *flags,
+ int preserve_position, int term_type)
{
logf (LOG_DEBUG, "rset_trunc no=%d", no);
- if (zi->isam)
+ if (no < 1)
+ {
+ rset_null_parms parms;
+ parms.rset_term = rset_term_create (term, length, flags, term_type);
+ return rset_create (rset_kind_null, &parms);
+ }
+ if (zi->reg->isams)
+ {
+ if (no == 1)
+ {
+ rset_isams_parms parms;
+
+ parms.pos = *isam_p;
+ parms.is = zi->reg->isams;
+ parms.rset_term = rset_term_create (term, length, flags,
+ term_type);
+ return rset_create (rset_kind_isams, &parms);
+ }
+ qsort (isam_p, no, sizeof(*isam_p), isams_trunc_cmp);
+ }
+ else if (zi->reg->isam)
{
- if (no < 1)
- return rset_create (rset_kind_null, NULL);
- else if (no == 1)
+ if (no == 1)
{
rset_isam_parms parms;
parms.pos = *isam_p;
- parms.is = zi->isam;
- parms.rset_term = rset_term_create (term, length, flags);
+ parms.is = zi->reg->isam;
+ parms.rset_term = rset_term_create (term, length, flags,
+ term_type);
return rset_create (rset_kind_isam, &parms);
}
qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
}
- else if (zi->isamc)
+ else if (zi->reg->isamc)
{
- if (no < 1)
- return rset_create (rset_kind_null, NULL);
- else if (no == 1)
+ if (no == 1)
{
rset_isamc_parms parms;
+ parms.key_size = sizeof(struct it_key);
+ parms.cmp = key_compare_it;
parms.pos = *isam_p;
- parms.is = zi->isamc;
- parms.rset_term = rset_term_create (term, length, flags);
+ parms.is = zi->reg->isamc;
+ parms.rset_term = rset_term_create (term, length, flags,
+ term_type);
return rset_create (rset_kind_isamc, &parms);
}
#if NEW_TRUNC
- else if (no < 200)
+ else if (no < 10000)
{
rset_m_or_parms parms;
parms.key_size = sizeof(struct it_key);
parms.cmp = key_compare_it;
- parms.isc = zi->isamc;
+ parms.isc = zi->reg->isamc;
parms.isam_positions = isam_p;
parms.no_isam_positions = no;
parms.no_save_positions = 100000;
- parms.rset_term = rset_term_create (term, length, flags);
+ parms.rset_term = rset_term_create (term, length, flags,
+ term_type);
return rset_create (rset_kind_m_or, &parms);
}
#endif
qsort (isam_p, no, sizeof(*isam_p), isamc_trunc_cmp);
}
+ else if (zi->reg->isamd)
+ {
+ if (no == 1)
+ {
+ rset_isamd_parms parms;
+
+ logf(LOG_FATAL, "isam_d does not (currently) support truncs");
+ abort();
+ /* parms.pos = *isam_p; */
+ parms.is = zi->reg->isamd;
+ parms.rset_term = rset_term_create (term, length, flags,
+ term_type);
+ return rset_create (rset_kind_isamd, &parms);
+ }
+#if NEW_TRUNC_NOT_DONE_FOR_ISAM_D
+ else if (no < 10000)
+ {
+ rset_m_or_parms parms;
+
+ parms.key_size = sizeof(struct it_key);
+ parms.cmp = key_compare_it;
+ parms.isc = 0;
+ parms.isamd=zi->reg->isamd;
+ parms.isam_positions = isam_p;
+ parms.no_isam_positions = no;
+ parms.no_save_positions = 100000;
+ parms.rset_term = rset_term_create (term, length, flags);
+ return rset_create (rset_kind_m_or, &parms);
+ }
+#endif
+ qsort (isam_p, no, sizeof(*isam_p), isamd_trunc_cmp);
+ }
+ else if (zi->reg->isamb)
+ {
+ if (no == 1)
+ {
+ rset_isamb_parms parms;
+
+ parms.key_size = sizeof(struct it_key);
+ parms.cmp = key_compare_it;
+ parms.pos = *isam_p;
+ parms.is = zi->reg->isamb;
+ parms.rset_term = rset_term_create (term, length, flags,
+ term_type);
+ return rset_create (rset_kind_isamb, &parms);
+ }
+ qsort (isam_p, no, sizeof(*isam_p), isamd_trunc_cmp);
+ }
else
{
- logf (LOG_WARN, "Neither isam nor isamc set in rset_trunc");
+ logf (LOG_WARN, "Unknown isam set in rset_trunc");
return rset_create (rset_kind_null, NULL);
}
- return rset_trunc_r (zi, term, length, flags, isam_p, 0, no, 100);
+ return rset_trunc_r (zi, term, length, flags, isam_p, 0, no, 100,
+ preserve_position, term_type);
}