X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Finvstat.c;h=5c9779dd4395bf6be4b4a83bc6ad25963d89338b;hb=3726bf6622da6a8b983bb4cbb7d654e84c3216d7;hp=6cfeab9c3888c5ec331a8172691dd1ec24a6bcc8;hpb=4c6c27a5098e5fd9425051ef97f98e898c098a20;p=idzebra-moved-to-github.git diff --git a/index/invstat.c b/index/invstat.c index 6cfeab9..5c9779d 100644 --- a/index/invstat.c +++ b/index/invstat.c @@ -1,24 +1,24 @@ /* - * Copyright (C) 1994-1996, Index Data I/S + * Copyright (C) 1994-1999, Index Data * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: invstat.c,v $ - * 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 #include #include -#include #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; int isam_bounds[20]; @@ -29,10 +29,9 @@ struct inv_stat_info { static int inv_stat_handle (char *name, const char *info, int pos, void *client) { - int occur; + int occur = 0; int i = 0; struct inv_stat_info *stat_info = (struct inv_stat_info*) client; - ISPT ispt; ISAM_P isam_p; stat_info->no_dict_entries++; @@ -41,23 +40,88 @@ 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)); - ispt = is_position (stat_info->isam, isam_p); - - occur = is_numkeys (ispt); + //printf ("---\n"); + if (stat_info->isam) + { + ISPT ispt; + + ispt = is_position (stat_info->isam, isam_p); + occur = is_numkeys (ispt); + is_pt_free (ispt); + } + if (stat_info->isamc) + { + ISAMC_PP pp; + int occurx = 0; + struct it_key key; + + pp = isc_pp_open (stat_info->isamc, isam_p); + occur = isc_pp_num (pp); + while (isc_pp_read(pp, &key)) + { + printf ("sysno=%d seqno=%d\n", key.sysno, key.seqno); + occurx++; + } + assert (occurx == occur); + 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; + int occurx = 0; + struct it_key key; - is_pt_free (ispt); + pp = isams_pp_open (stat_info->isams, isam_p); + occur = isams_pp_num (pp); + while (isams_pp_read(pp, &key)) + { + //printf ("sysno=%d seqno=%d\n", key.sysno, key.seqno); + occurx++; + } + assert (occurx == occur); + stat_info->no_isam_entries[isc_type(isam_p)] += occur; + isams_pp_close (pp); + } while (occur > stat_info->isam_bounds[i] && stat_info->isam_bounds[i]) i++; ++(stat_info->isam_occurrences[i]); - return 0; } -void inv_prstat (const char *dict_fname, const char *isam_fname) +void inv_prstat (BFiles bfs) { Dict dict; - ISAM isam; + ISAM isam = NULL; + ISAMC isamc = NULL; + ISAMS isams = NULL; + ISAMH isamh = NULL; Records records; int i, prev; int before = 0; @@ -68,27 +132,67 @@ void inv_prstat (const char *dict_fname, const char *isam_fname) term_dict[0] = 1; term_dict[1] = 0; - dict = dict_open (dict_fname, 100, 0); + dict = dict_open (bfs, FNAME_DICT, 100, 0, 0); if (!dict) { - logf (LOG_FATAL, "dict_open fail of `%s'", dict_fname); + logf (LOG_FATAL, "dict_open fail"); exit (1); } - isam = is_open (isam_fname, key_compare, 0, sizeof(struct it_key)); - if (!isam) + if (res_get_match (common_resource, "isam", "i", NULL)) { - logf (LOG_FATAL, "is_open fail of `%s'", isam_fname); - exit (1); + isam = is_open (bfs, FNAME_ISAM, key_compare, 0, + sizeof(struct it_key), common_resource); + if (!isam) + { + logf (LOG_FATAL, "is_open fail"); + exit (1); + } + } + else if (res_get_match (common_resource, "isam", "s", NULL)) + { + struct ISAMS_M_s isams_m; + isams = isams_open (bfs, FNAME_ISAMS, 0, + key_isams_m(common_resource, &isams_m)); + if (!isams) + { + logf (LOG_FATAL, "isams_open fail"); + exit (1); + } } - records = rec_open (0); + 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 + { + struct ISAMC_M_s isamc_m; + isamc = isc_open (bfs, FNAME_ISAMC, 0, + key_isamc_m (common_resource, &isamc_m)); + if (!isamc) + { + logf (LOG_FATAL, "isc_open fail"); + exit (1); + } + } + records = rec_open (bfs, 0, 0); + for (i = 0; i<8; i++) + stat_info.no_isam_entries[i] = 0; stat_info.no_dict_entries = 0; stat_info.no_dict_bytes = 0; 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; - stat_info.isam_bounds[3] = 5; + stat_info.isam_bounds[3] = 6; stat_info.isam_bounds[4] = 10; stat_info.isam_bounds[5] = 20; stat_info.isam_bounds[6] = 30; @@ -110,13 +214,28 @@ void inv_prstat (const char *dict_fname, const char *isam_fname) dict_scan (dict, term_dict, &before, &after, &stat_info, inv_stat_handle); - rec_close (&records); - dict_close (dict); - is_close (isam); + if (isamc) + { + fprintf (stderr, " Blocks Occur Size KB Bytes/Entry\n"); + for (i = 0; isc_block_used (isamc, i) >= 0; i++) + { + fprintf (stderr, " %8d %8d", isc_block_used (isamc, i), + stat_info.no_isam_entries[i]); - fprintf (stderr, "%d dictionary entries. %d bytes for strings\n", + if (stat_info.no_isam_entries[i]) + fprintf (stderr, " %8d %f", + (int) ((1023.0 + (double) isc_block_used(isamc, i) * + isc_block_size(isamc,i))/1024), + ((double) isc_block_used(isamc, i) * + isc_block_size(isamc,i))/ + stat_info.no_isam_entries[i]); + fprintf (stderr, "\n"); + } + } + + fprintf (stderr, "\n%d words using %d bytes\n", stat_info.no_dict_entries, stat_info.no_dict_bytes); - fprintf (stderr, " size occurrences\n"); + fprintf (stderr, " Occurrences Words\n"); prev = 1; for (i = 0; stat_info.isam_bounds[i]; i++) { @@ -125,6 +244,74 @@ void inv_prstat (const char *dict_fname, const char *isam_fname) prev, here, stat_info.isam_occurrences[i]); prev = here+1; } - fprintf (stderr, "%7d- %7d\n", + fprintf (stderr, "%7d- %7d\n", prev, stat_info.isam_occurrences[i]); + + rec_close (&records); + dict_close (dict); + + if (isam) + is_close (isam); + if (isamc) + isc_close (isamc); + if (isams) + isams_close (isams); + if (isamh) + isamh_close (isamh); + } + + +/* + * + * $Log: invstat.c,v $ + * Revision 1.14 1999-07-14 10:59:26 adam + * Changed functions isc_getmethod, isams_getmethod. + * Improved fatal error handling (such as missing EXPLAIN schema). + * + * 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. + */