From f7afb145661ac2f567c60f203bf0dbe32309402f Mon Sep 17 00:00:00 2001 From: Heikki Levanto Date: Thu, 8 Jul 1999 14:23:27 +0000 Subject: [PATCH] Fixed a bug in isamh_pp_read and cleaned up a bit --- index/invstat.c | 144 ++++++++++++++++++++++++++++++++++----------------- isamc/isamh.c | 28 ++++++++-- isamc/merge.c | 13 +++-- test/gils/zebra.cfg | 10 +++- 4 files changed, 137 insertions(+), 58 deletions(-) diff --git a/index/invstat.c b/index/invstat.c index 30a5e90..fdab6ba 100644 --- a/index/invstat.c +++ b/index/invstat.c @@ -1,51 +1,8 @@ /* * Copyright (C) 1994-1999, Index Data * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: invstat.c,v $ - * Revision 1.12 1999-07-06 12:28:04 adam - * Updated record index structure. Format includes version ID. Compression - * algorithm ID is stored for each record block. - * - * Revision 1.11 1999/05/15 14:36:38 adam - * Updated dictionary. Implemented "compression" of dictionary. - * - * Revision 1.10 1999/05/12 13:08:06 adam - * First version of ISAMS. - * - * Revision 1.9 1999/02/12 13:29:23 adam - * Implemented position-flag for registers. - * - * Revision 1.8 1999/02/02 14:50:53 adam - * Updated WIN32 code specific sections. Changed header. - * - * Revision 1.7 1998/03/13 15:30:50 adam - * New functions isc_block_used and isc_block_size. Fixed 'leak' - * in isc_alloc_block. - * - * Revision 1.6 1998/03/06 13:54:02 adam - * Fixed two nasty bugs in isc_merge. - * - * Revision 1.5 1997/09/17 12:19:13 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 1996/11/08 11:10:21 adam - * Buffers used during file match got bigger. - * Compressed ISAM support everywhere. - * Bug fixes regarding masking characters in queries. - * Redesigned Regexp-2 queries. - * - * Revision 1.3 1996/06/04 10:18:58 adam - * Minor changes - removed include of ctype.h. - * - * Revision 1.2 1996/05/22 08:25:56 adam - * Minor change. - * - * Revision 1.1 1996/05/14 14:04:34 adam - * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT - * is collected. + * Sebastian Hammer, Adam Dickmeiss, Heikki Levanto + * log at eof * */ #include @@ -54,11 +11,13 @@ #include "index.h" #include "recindex.h" +#include "../isamc/isamh-p.h" struct inv_stat_info { ISAM isam; ISAMC isamc; ISAMS isams; + ISAMH isamh; int no_isam_entries[8]; int no_dict_entries; int no_dict_bytes; @@ -81,7 +40,7 @@ static int inv_stat_handle (char *name, const char *info, int pos, assert (*info == sizeof(ISAM_P)); memcpy (&isam_p, info+1, sizeof(ISAM_P)); - printf ("---\n"); + //printf ("---\n"); if (stat_info->isam) { ISPT ispt; @@ -107,6 +66,31 @@ static int inv_stat_handle (char *name, const char *info, int pos, stat_info->no_isam_entries[isc_type(isam_p)] += occur; isc_pp_close (pp); } + if (stat_info->isamh) + { + ISAMH_PP pp; + int occurx = 0; + struct it_key key; + + pp = isamh_pp_open (stat_info->isamh, isam_p); + + occur = isamh_pp_num (pp); + // printf (" opening item %d=%d:%d \n", + // isam_p, isamh_type(isam_p),isamh_block(isam_p)); + while (isamh_pp_read(pp, &key)) + { + occurx++; + logf (LOG_LOG,"sysno=%d seqno=%d (%x/%x) oc=%d/%d ofs=%d ", + key.sysno, key.seqno, + key.sysno, key.seqno, + occur,occurx, pp->offset); + } + if (occurx != occur) + logf(LOG_LOG,"Count error!!! read %d, counted %d", occur, occurx); + assert (occurx == occur); + stat_info->no_isam_entries[isamh_type(isam_p)] += occur; + isamh_pp_close (pp); + } if (stat_info->isams) { ISAMS_PP pp; @@ -117,7 +101,7 @@ static int inv_stat_handle (char *name, const char *info, int pos, occur = isams_pp_num (pp); while (isams_pp_read(pp, &key)) { - printf ("sysno=%d seqno=%d\n", key.sysno, key.seqno); + //printf ("sysno=%d seqno=%d\n", key.sysno, key.seqno); occurx++; } assert (occurx == occur); @@ -134,9 +118,10 @@ static int inv_stat_handle (char *name, const char *info, int pos, void inv_prstat (BFiles bfs) { Dict dict; - ISAM isam = NULL; + ISAM isam = NULL; ISAMC isamc = NULL; ISAMS isams = NULL; + ISAMH isamh = NULL; Records records; int i, prev; int before = 0; @@ -172,6 +157,15 @@ void inv_prstat (BFiles bfs) exit (1); } } + else if (res_get_match (common_resource, "isam", "h", NULL)) + { + isamh = isamh_open (bfs, FNAME_ISAMH, 0, key_isamh_m(common_resource)); + if (!isamh) + { + logf (LOG_FATAL, "isamh_open fail"); + exit (1); + } + } else { isamc = isc_open (bfs, FNAME_ISAMC, 0, key_isamc_m (common_resource)); @@ -190,6 +184,7 @@ void inv_prstat (BFiles bfs) stat_info.isam = isam; stat_info.isamc = isamc; stat_info.isams = isams; + stat_info.isamh = isamh; stat_info.isam_bounds[0] = 1; stat_info.isam_bounds[1] = 2; stat_info.isam_bounds[2] = 3; @@ -257,5 +252,58 @@ void inv_prstat (BFiles bfs) isc_close (isamc); if (isams) isams_close (isams); + if (isamh) + isamh_close (isamh); } + + +/* + * + * $Log: invstat.c,v $ + * Revision 1.13 1999-07-08 14:23:27 heikki + * Fixed a bug in isamh_pp_read and cleaned up a bit + * + * Revision 1.12 1999/07/06 12:28:04 adam + * Updated record index structure. Format includes version ID. Compression + * algorithm ID is stored for each record block. + * + * Revision 1.11 1999/05/15 14:36:38 adam + * Updated dictionary. Implemented "compression" of dictionary. + * + * Revision 1.10 1999/05/12 13:08:06 adam + * First version of ISAMS. + * + * Revision 1.9 1999/02/12 13:29:23 adam + * Implemented position-flag for registers. + * + * Revision 1.8 1999/02/02 14:50:53 adam + * Updated WIN32 code specific sections. Changed header. + * + * Revision 1.7 1998/03/13 15:30:50 adam + * New functions isc_block_used and isc_block_size. Fixed 'leak' + * in isc_alloc_block. + * + * Revision 1.6 1998/03/06 13:54:02 adam + * Fixed two nasty bugs in isc_merge. + * + * Revision 1.5 1997/09/17 12:19:13 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 1996/11/08 11:10:21 adam + * Buffers used during file match got bigger. + * Compressed ISAM support everywhere. + * Bug fixes regarding masking characters in queries. + * Redesigned Regexp-2 queries. + * + * Revision 1.3 1996/06/04 10:18:58 adam + * Minor changes - removed include of ctype.h. + * + * Revision 1.2 1996/05/22 08:25:56 adam + * Minor change. + * + * Revision 1.1 1996/05/14 14:04:34 adam + * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT + * is collected. + */ diff --git a/isamc/isamh.c b/isamc/isamh.c index d3cab74..c6c35de 100644 --- a/isamc/isamh.c +++ b/isamc/isamh.c @@ -5,6 +5,12 @@ * * Isamh - append-only isam * + * todo + * (get invstat to work) + * implement dirty bit + * implement direct address bit + * (result set stuff) + * */ @@ -64,6 +70,7 @@ ISAMH_M isamh_getmethod (void) } + ISAMH isamh_open (BFiles bfs, const char *name, int writeflag, ISAMH_M method) { ISAMH is; @@ -80,7 +87,7 @@ ISAMH isamh_open (BFiles bfs, const char *name, int writeflag, ISAMH_M method) /* determine number of block categories */ if (is->method->debug) - logf (LOG_LOG, "isc: bsize ifill mfill mblocks"); + logf (LOG_LOG, "isc: bsize maxkeys"); do { if (is->method->debug) @@ -548,6 +555,7 @@ int isamh_read_item (ISAMH_PP pp, char **dst) { ISAMH is = pp->is; char *src = pp->buf + pp->offset; + int newcat; if (pp->offset >= pp->size) { @@ -577,7 +585,13 @@ int isamh_read_item (ISAMH_PP pp, char **dst) } } /* out new block position */ - pp->pos = pp->next; + newcat = isamh_type(pp->next); + if (pp->cat != newcat ) { + pp->buf = xrealloc(pp->buf, is->method->filecat[newcat].bsize); + } + pp->pos = isamh_block(pp->next); + pp->cat = isamh_type(pp->next); + src = pp->buf; /* read block and save 'next' and 'size' entry */ isamh_read_block (is, pp->cat, pp->pos, src); @@ -587,9 +601,10 @@ int isamh_read_item (ISAMH_PP pp, char **dst) src += sizeof(pp->size); /* assume block is non-empty */ assert (src - pp->buf == ISAMH_BLOCK_OFFSET_N); - assert (pp->next != pp->pos); + assert (pp->next != isamh_addr(pp->pos,pp->cat)); if (pp->deleteFlag) isamh_release_block (is, pp->cat, pp->pos); + (*is->method->code_reset)(pp->decodeClientData); (*is->method->code_item)(ISAMH_DECODE, pp->decodeClientData, dst, &src); pp->offset = src - pp->buf; if (is->method->debug > 2) @@ -607,9 +622,14 @@ int isamh_pp_num (ISAMH_PP pp) return pp->numKeys; } + + /* * $Log: isamh.c,v $ - * Revision 1.4 1999-07-07 09:36:04 heikki + * Revision 1.5 1999-07-08 14:23:27 heikki + * Fixed a bug in isamh_pp_read and cleaned up a bit + * + * Revision 1.4 1999/07/07 09:36:04 heikki * Fixed an assertion in isamh * * Revision 1.2 1999/07/06 09:37:05 heikki diff --git a/isamc/merge.c b/isamc/merge.c index a2653c6..3a7e80f 100644 --- a/isamc/merge.c +++ b/isamc/merge.c @@ -576,8 +576,8 @@ ISAMC_P isamh_append (ISAMH is, ISAMH_P ipos, ISAMH_I data) assert( (codelen < 128) && (codelen>0)); if (pp->is->method->debug > 3) - logf(LOG_LOG,"isamh_append: coded into %d:%s", - codelen,hexdump(codebuffer,codelen,0)); + logf(LOG_LOG,"isamh_append: coded into %d:%s (nk=%d)", + codelen,hexdump(codebuffer,codelen,0),firstpp->numKeys); if ( pp->offset + codelen > maxsize ) { /* oops, block full, do something */ @@ -622,8 +622,8 @@ ISAMC_P isamh_append (ISAMH is, ISAMH_P ipos, ISAMH_I data) (*is->method->code_item)(ISAMH_ENCODE, r_clientData, &codeptr, &i_item_ptr); codelen = codeptr-codebuffer; if (pp->is->method->debug > 3) - logf(LOG_LOG,"isamh_append: coded again %d:%s", - codelen,hexdump(codebuffer,codelen,0)); + logf(LOG_LOG,"isamh_append: coded again %d:%s (nk=%d)", + codelen,hexdump(codebuffer,codelen,0),firstpp->numKeys); } /* block full */ @@ -672,7 +672,10 @@ ISAMC_P isamh_append (ISAMH is, ISAMH_P ipos, ISAMH_I data) /* * $Log: merge.c,v $ - * Revision 1.15 1999-07-07 09:36:04 heikki + * Revision 1.16 1999-07-08 14:23:27 heikki + * Fixed a bug in isamh_pp_read and cleaned up a bit + * + * Revision 1.15 1999/07/07 09:36:04 heikki * Fixed an assertion in isamh * * Revision 1.13 1999/07/06 09:37:05 heikki diff --git a/test/gils/zebra.cfg b/test/gils/zebra.cfg index 68882a1..b471228 100644 --- a/test/gils/zebra.cfg +++ b/test/gils/zebra.cfg @@ -1,5 +1,5 @@ # Simple Zebra configuration file -# $Id: zebra.cfg,v 1.4 1998-11-03 14:38:44 adam Exp $ +# $Id: zebra.cfg,v 1.5 1999-07-08 14:23:27 heikki Exp $ # # Where are the YAZ / Zebra tables located? profilePath: .:../../tab:../../../yaz/tab @@ -8,3 +8,11 @@ profilePath: .:../../tab:../../../yaz/tab attset: bib1.att attset: gils.att attset: explain.att + +isam: h +#isam: c + +isamhDebug: 7 + +recordtype: grs.sgml + -- 1.7.10.4