X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Finvstat.c;h=dd808dae03ed68aded73721c32e4bdbcdfb959c7;hb=1dec188ccbdafb0ec7b7c98857d701a7b77d46a1;hp=7f7cca3963c3eebd2fa3ead479be3ee50a304061;hpb=500505a8413687c67b6be99ff1ae9c1460d35e25;p=idzebra-moved-to-github.git diff --git a/index/invstat.c b/index/invstat.c index 7f7cca3..dd808da 100644 --- a/index/invstat.c +++ b/index/invstat.c @@ -10,22 +10,20 @@ #include #include "index.h" -#include "recindex.h" -#include "../isamc/isamh-p.h" #include "../isamc/isamd-p.h" struct inv_stat_info { - ISAM isam; - ISAMC isamc; - ISAMS isams; - ISAMH isamh; - ISAMD isamd; + ZebraHandle zh; int no_isam_entries[9]; int no_dict_entries; int no_dict_bytes; int isam_bounds[20]; int isam_occurrences[20]; char tmp[128]; + int isamb_levels[10][5]; + int isamb_sizes[10]; + int isamb_blocks[10]; + unsigned long cksum; }; #define SINGLETON_TYPE 8 /* the type to use for singletons that */ @@ -37,75 +35,72 @@ static int inv_stat_handle (char *name, const char *info, int pos, int occur = 0; int i = 0; struct inv_stat_info *stat_info = (struct inv_stat_info*) client; - ISAM_P isam_p; + ISAMS_P isam_p; stat_info->no_dict_entries++; stat_info->no_dict_bytes += strlen(name); - assert (*info == sizeof(ISAM_P)); - memcpy (&isam_p, info+1, sizeof(ISAM_P)); + assert (*info == sizeof(ISAMS_P)); + memcpy (&isam_p, info+1, sizeof(ISAMS_P)); - //printf ("---\n"); - if (stat_info->isam) + if (stat_info->zh->reg->isams) + { + ISAMS_PP pp; + int occurx = 0; + struct it_key key; + + pp = isams_pp_open (stat_info->zh->reg->isams, isam_p); + occur = isams_pp_num (pp); + while (isams_pp_read(pp, &key)) + { + stat_info->cksum = stat_info->cksum * 65509 + + key.sysno + 11 * key.seqno; + occurx++; + } + assert (occurx == occur); + stat_info->no_isam_entries[0] += occur; + isams_pp_close (pp); + } + if (stat_info->zh->reg->isam) { ISPT ispt; - ispt = is_position (stat_info->isam, isam_p); + ispt = is_position (stat_info->zh->reg->isam, isam_p); occur = is_numkeys (ispt); + stat_info->no_isam_entries[is_type(isam_p)] += occur; is_pt_free (ispt); } - if (stat_info->isamc) + if (stat_info->zh->reg->isamc) { ISAMC_PP pp; int occurx = 0; struct it_key key; - pp = isc_pp_open (stat_info->isamc, isam_p); + pp = isc_pp_open (stat_info->zh->reg->isamc, isam_p); occur = isc_pp_num (pp); while (isc_pp_read(pp, &key)) { - //printf ("sysno=%d seqno=%d\n", key.sysno, key.seqno); + stat_info->cksum = stat_info->cksum * 65509 + + key.sysno + 11 * 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); - 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); - //isamh_pp_dump(stat_info->isamh, isam_p); - } - stat_info->no_isam_entries[isamh_type(isam_p)] += occur; - isamh_pp_close (pp); - } - if (stat_info->isamd) + if (stat_info->zh->reg->isamd) { ISAMD_PP pp; int occurx = 0; struct it_key key; - pp = isamd_pp_open (stat_info->isamd, isam_p); + pp = isamd_pp_open (stat_info->zh->reg->isamd, isam_p); occur = isamd_pp_num (pp); while (isamd_pp_read(pp, &key)) { + stat_info->cksum = stat_info->cksum * 65509 + + key.sysno + 11 * key.seqno; occurx++; if ( pp->is->method->debug >8 ) logf (LOG_LOG,"sysno=%d seqno=%d (%x/%x) oc=%d/%d ofs=%d ", @@ -126,22 +121,30 @@ static int inv_stat_handle (char *name, const char *info, int pos, stat_info->no_isam_entries[isamd_type(isam_p)] += occur; isamd_pp_close (pp); } - if (stat_info->isams) + if (stat_info->zh->reg->isamb) { - ISAMS_PP pp; - int occurx = 0; - struct it_key key; + ISAMB_PP pp; + struct it_key key; + int cat = isam_p & 3; + int level; + int size; + int blocks; + + pp = isamb_pp_open_x(stat_info->zh->reg->isamb, isam_p, &level); - 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 (isamb_pp_read(pp, &key)) + { + stat_info->cksum = stat_info->cksum * 65509 + + key.sysno + 11 * key.seqno; + occur++; + } + isamb_pp_close_x (pp, &size, &blocks); + stat_info->isamb_blocks[cat] += blocks; + stat_info->isamb_sizes[cat] += size; + if (level > 4) + level = 4; + stat_info->isamb_levels[cat][level] ++; + stat_info->no_isam_entries[cat] += occur; } while (occur > stat_info->isam_bounds[i] && stat_info->isam_bounds[i]) @@ -150,96 +153,30 @@ static int inv_stat_handle (char *name, const char *info, int pos, return 0; } -void inv_prstat (BFiles bfs) +void zebra_register_statistics (ZebraHandle zh) { - Dict dict; - ISAM isam = NULL; - ISAMC isamc = NULL; - ISAMS isams = NULL; - ISAMH isamh = NULL; - ISAMD isamd = NULL; - Records records; + int blocks; + int size; + int count; int i, prev; int before = 0; + int occur; int after = 1000000000; struct inv_stat_info stat_info; char term_dict[2*IT_MAX_WORD+2]; - int blocks; - int size; - int count; - + + if (zebra_begin_read (zh)) + return; + + stat_info.zh = zh; + term_dict[0] = 1; term_dict[1] = 0; - dict = dict_open (bfs, FNAME_DICT, 100, 0, 0); - if (!dict) - { - logf (LOG_FATAL, "dict_open fail"); - exit (1); - } - if (res_get_match (common_resource, "isam", "i", NULL)) - { - 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); - } - } - 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 if (res_get_match (common_resource, "isam", "d", NULL)) - { - struct ISAMD_M_s isamd_m; - isamd = isamd_open (bfs, FNAME_ISAMD, 0, - key_isamd_m(common_resource,&isamd_m)); - if (!isamd) - { - logf (LOG_FATAL, "isamd_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<=SINGLETON_TYPE; 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.isamd = isamd; stat_info.isam_bounds[0] = 1; stat_info.isam_bounds[1] = 2; stat_info.isam_bounds[2] = 3; @@ -260,50 +197,63 @@ void inv_prstat (BFiles bfs) stat_info.isam_bounds[17] = 1000000; stat_info.isam_bounds[18] = 0; + stat_info.cksum = 0; + for (i = 0; i<20; i++) stat_info.isam_occurrences[i] = 0; - dict_scan (dict, term_dict, &before, &after, &stat_info, inv_stat_handle); + for (i = 0; i<10; i++) + { + int j; + for (j = 0; j<5; j++) + stat_info.isamb_levels[i][j] = 0; + stat_info.isamb_sizes[i] = 0; + stat_info.isamb_blocks[i] = 0; + } + + dict_scan (zh->reg->dict, term_dict, &before, &after, &stat_info, + inv_stat_handle); - if (isamc) + if (zh->reg->isamc) { - fprintf (stderr, " Blocks Occur Size KB Bytes/Entry\n"); - for (i = 0; isc_block_used (isamc, i) >= 0; i++) + fprintf (stdout, " Blocks Occur Size KB Bytes/Entry\n"); + for (i = 0; isc_block_used (zh->reg->isamc, i) >= 0; i++) { - fprintf (stderr, " %8d %8d", isc_block_used (isamc, i), + fprintf (stdout, " %8d %8d", isc_block_used (zh->reg->isamc, i), stat_info.no_isam_entries[i]); 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))/ + fprintf (stdout, " %8d %f", + (int) ((1023.0 + (double) + isc_block_used(zh->reg->isamc, i) * + isc_block_size(zh->reg->isamc,i))/1024), + ((double) isc_block_used(zh->reg->isamc, i) * + isc_block_size(zh->reg->isamc,i))/ stat_info.no_isam_entries[i]); - fprintf (stderr, "\n"); + fprintf (stdout, "\n"); } } - if (isamd) + if (zh->reg->isamd) { - fprintf (stderr, " Blocks Occur KB Bytes/Entry\n"); - if (isamd->method->debug >0) + fprintf (stdout, " Blocks Occur KB Bytes/Entry\n"); + if (zh->reg->isamd->method->debug >0) logf(LOG_LOG," Blocks Occur KB Bytes/Entry"); for (i = 0; i<=SINGLETON_TYPE; i++) { - blocks= isamd_block_used(isamd,i); - size= isamd_block_size(isamd,i); + blocks= isamd_block_used(zh->reg->isamd,i); + size= isamd_block_size(zh->reg->isamd,i); count=stat_info.no_isam_entries[i]; if (i==SINGLETON_TYPE) blocks=size=0; if (stat_info.no_isam_entries[i]) { - fprintf (stderr, "%c %7d %7d %7d %5.2f\n", + fprintf (stdout, "%c %7d %7d %7d %5.2f\n", (i==SINGLETON_TYPE)?('z'):('A'+i), blocks, count, (int) ((1023.0 + (double) blocks * size)/1024), ((double) blocks * size)/count); - if (isamd->method->debug >0) + if (zh->reg->isamd->method->debug >0) logf(LOG_LOG, "%c %7d %7d %7d %5.2f", (i==SINGLETON_TYPE)?('z'):('A'+i), blocks, @@ -313,59 +263,92 @@ void inv_prstat (BFiles bfs) } /* entries */ } /* for */ } /* isamd */ - if ( (isamd) && (isamd->method->debug>0)) - fprintf (stderr, "\n%d words using %d bytes\n", + if ( (zh->reg->isamd) && (zh->reg->isamd->method->debug>0)) + fprintf (stdout, "\n%d words using %d bytes\n", stat_info.no_dict_entries, stat_info.no_dict_bytes); - fprintf (stderr, " Occurrences Words\n"); - if ( (isamd) && (isamd->method->debug>0) ) + + if (zh->reg->isamb) { - logf(LOG_LOG, "%d words using %d bytes", - stat_info.no_dict_entries, stat_info.no_dict_bytes); - logf(LOG_LOG, " Occurrences Words"); + for (i = 0; i<4; i++) + { + int j; + int bsize = isamb_block_info(zh->reg->isamb, i); + if (bsize < 0) + break; + fprintf (stdout, "Category %d\n", i); + fprintf (stdout, "Block size %d\n", bsize); + fprintf (stdout, "Blocks: %d\n", stat_info.isamb_blocks[i]); + fprintf (stdout, "Size: %d\n", stat_info.isamb_sizes[i]); + fprintf (stdout, "Entries: %d\n", stat_info.no_isam_entries[i]); + fprintf (stdout, "Total %d\n", stat_info.isamb_blocks[i]* + bsize); + for (j = 0; j<5; j++) + if (stat_info.isamb_levels[i][j]) + fprintf (stdout, "Level%d %d\n", j, + stat_info.isamb_levels[i][j]); + fprintf (stdout, "\n"); + } } + fprintf (stdout, "Checksum %08lX\n", stat_info.cksum); + + fprintf (stdout, "Distinct words %d\n", stat_info.no_dict_entries); + occur = 0; + for (i = 0; i<9; i++) + occur += stat_info.no_isam_entries[i]; + fprintf (stdout, "Word pos %d\n", occur); + fprintf (stdout, " Occurrences Words\n"); prev = 1; for (i = 0; stat_info.isam_bounds[i]; i++) { int here = stat_info.isam_bounds[i]; - fprintf (stderr, "%7d-%-7d %7d\n", - prev, here, stat_info.isam_occurrences[i]); - if ( (isamd) && (isamd->method->debug>0) && - stat_info.isam_occurrences[i] ) - { - logf(LOG_LOG,"%7d-%-7d %7d", + fprintf (stdout, "%7d-%-7d %7d\n", prev, here, stat_info.isam_occurrences[i]); - } prev = here+1; } - fprintf (stderr, "%7d- %7d\n", + fprintf (stdout, "%7d- %7d\n", prev, stat_info.isam_occurrences[i]); - if ( (isamd) && (isamd->method->debug>0) && - ( stat_info.isam_occurrences[i] || - stat_info.isam_occurrences[i+1]) ) - logf(LOG_LOG,"%7d- %7d", - 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); - if (isamd) - isamd_close (isamd); - xmalloc_trav("unfreed"); /*! while hunting memory leaks */ + zebra_end_read (zh); } /* * * $Log: invstat.c,v $ - * Revision 1.18 1999-10-06 11:46:36 heikki + * Revision 1.29 2002-06-19 10:29:17 adam + * align block sizes for isam sys. Better plot for test + * + * Revision 1.28 2002/04/30 19:31:09 adam + * isamb delete; more statistics + * + * Revision 1.27 2002/04/30 08:28:37 adam + * isamb fixes for pp_read. Statistics + * + * Revision 1.26 2002/04/29 18:03:46 adam + * More isamb statistics + * + * Revision 1.25 2002/04/26 08:44:47 adam + * Index statistics working again + * + * Revision 1.24 2002/04/05 08:46:26 adam + * Zebra with full functionality + * + * Revision 1.23 2002/04/04 14:14:13 adam + * Multiple registers (alpha early) + * + * Revision 1.22 2002/02/20 17:30:01 adam + * Work on new API. Locking system re-implemented + * + * Revision 1.21 2000/07/13 10:14:20 heikki + * Removed compiler warnings when making zebra + * + * Revision 1.20 1999/12/01 13:30:30 adam + * Updated configure for Zmbol/Zebra dependent settings. + * + * Revision 1.19 1999/11/30 13:48:03 adam + * Improved installation. Updated for inclusion of YAZ header files. + * + * Revision 1.18 1999/10/06 11:46:36 heikki * mproved statistics on isam-d * * Revision 1.17 1999/08/20 08:28:37 heikki