X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Ftrunc.c;h=19ba2293a22e6438d63e026f5c60649a48219971;hb=3335cd5f7e7af06ac6ed943859c473e092d559a1;hp=f9ecdc650dc2f984afc104f4c4943c43bdb8f2c5;hpb=3c5f6226f97612c0d6ac40591f600587c5ffa858;p=idzebra-moved-to-github.git diff --git a/index/trunc.c b/index/trunc.c index f9ecdc6..19ba229 100644 --- a/index/trunc.c +++ b/index/trunc.c @@ -1,66 +1,41 @@ -/* - * Copyright (C) 1994-1999, Index Data - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: trunc.c,v $ - * Revision 1.12 1999-02-02 14:51:10 adam - * Updated WIN32 code specific sections. Changed header. - * - * 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 1.35 2004-08-19 12:49:14 heikki Exp $ + Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004 + 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 #include -#include "zserver.h" -#include -#include -#include -#include - #define NEW_TRUNC 1 +#include "index.h" +#include +#include +#include +#include +#include #if NEW_TRUNC #include #endif +#include struct trunc_info { int *ptr; @@ -126,19 +101,19 @@ static struct trunc_info *heap_init (int size, int key_size, 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; @@ -159,16 +134,19 @@ static void heap_close (struct trunc_info *ti) } 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; + int nn = 0; + 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); @@ -182,17 +160,21 @@ static RSET rset_trunc_r (ZebraHandle zi, const char *term, int length, 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); @@ -212,6 +194,7 @@ static RSET rset_trunc_r (ZebraHandle zi, const char *term, int length, int n = ti->indx[ti->ptr[1]]; rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]); + nn++; while (1) { @@ -234,96 +217,91 @@ static RSET rset_trunc_r (ZebraHandle zi, const char *term, int length, xfree (rsfd); heap_close (ti); } - else if (zi->isam) + else if (zi->reg->isamc) { - ISPT *ispt; + ISAMC_PP *ispt; int i; struct trunc_info *ti; - ispt = xmalloc (sizeof(*ispt) * (to-from)); + 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] = is_position (zi->isam, isam_p[from+i]); - if (is_readkey (ispt[i], ti->tmpbuf)) + 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 - is_pt_free (ispt[i]); + 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 1 -/* section that preserve all keys */ - heap_delete (ti); - if (is_readkey (ispt[n], ti->tmpbuf)) - heap_insert (ti, ti->tmpbuf, n); + nn++; + 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 - is_pt_free (ispt[n]); -#else -/* section that preserve all keys with unique sysnos */ - while (1) { - if (!is_readkey (ispt[n], ti->tmpbuf)) + while (1) { - 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; + 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; + } } } -#endif } heap_close (ti); xfree (ispt); } - else + else if (zi->reg->isams) { - ISAMC_PP *ispt; + ISAMS_PP *ispt; int i; struct trunc_info *ti; + int nn = 0; - 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 */ + nn++; 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) @@ -333,93 +311,208 @@ static RSET rset_trunc_r (ZebraHandle zi, const char *term, int length, 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; ) + { + if (isam_p[from+i]) { + 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]]); + nn++; + + 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"); + + parms.rset_term->nn = nn; rset_close (result, result_rsfd); return result; } -static int isam_trunc_cmp (const void *p1, const void *p2) +static int isams_trunc_cmp (const void *p1, const void *p2) { - ISAM_P i1 = *(ISAM_P*) p1; - ISAM_P i2 = *(ISAM_P*) p2; - int d; - - d = is_type (i1) - is_type (i2); - if (d) - return d; - return is_block (i1) - is_block (i2); + ISAMS_P i1 = *(ISAMS_P*) p1; + ISAMS_P i2 = *(ISAMS_P*) p2; + + if (i1 > i2) + return 1; + else if (i1 < i2) + return -1; + return 0; } static int isamc_trunc_cmp (const void *p1, const void *p2) { ISAMC_P i1 = *(ISAMC_P*) p1; ISAMC_P i2 = *(ISAMC_P*) p2; - int d; - - d = isc_type (i1) - isc_type (i2); - if (d) - return d; - return isc_block (i1) - isc_block (i2); + zint d; + + d = (isc_type (i1) - isc_type (i2)); + if (d == 0) + d = isc_block (i1) - isc_block (i2); + if (d > 0) + return 1; + else if (d < 0) + return -1; + return 0; } -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) { - if (no < 1) - return rset_create (rset_kind_null, NULL); - else 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_isam_parms parms; + rset_isams_parms parms; parms.pos = *isam_p; - parms.is = zi->isam; - parms.rset_term = rset_term_create (term, length, flags); - return rset_create (rset_kind_isam, &parms); + 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), isam_trunc_cmp); + qsort (isam_p, no, sizeof(*isam_p), isams_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->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); + } +#if 1 + else if (no <10000 ) /* FIXME - hardcoded number */ + { + rset_multior_parms m_parms; + rset_isamb_parms b_parms; + int i; + m_parms.key_size = sizeof(struct it_key); + m_parms.cmp = key_compare_it; + m_parms.no_rsets=no; + m_parms.rsets=xmalloc(sizeof(*m_parms.rsets)*no); + m_parms.rset_term = rset_term_create (term, length, flags, + term_type); + b_parms.key_size = sizeof(struct it_key); + b_parms.cmp = key_compare_it; + b_parms.is = zi->reg->isamb; + /* FIXME - make it so that we can pass a null ptr to term */ + /* needs changes in all rsets, here and there! */ + for (i=0;i