X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;ds=sidebyside;f=index%2Finvstat.c;h=69973eb0bb65e24426a89f48a884eb77f11af639;hb=dbf7ead21f60be189f0d532175c92924a8eb73fd;hp=ca44bda61b1bb6247bffb998be591356517a7e44;hpb=175aced6e1fba324771b361822e8a27a38e567bb;p=idzebra-moved-to-github.git diff --git a/index/invstat.c b/index/invstat.c index ca44bda..69973eb 100644 --- a/index/invstat.c +++ b/index/invstat.c @@ -23,11 +23,36 @@ struct inv_stat_info { int isamb_levels[10][5]; int isamb_sizes[10]; int isamb_blocks[10]; + unsigned long cksum; + int dumpwords; }; #define SINGLETON_TYPE 8 /* the type to use for singletons that */ /* have no block and no block type */ +static void print_dict_item (ZebraMaps zm, const char *s, int count, + int firstsys, int firstseq, int lastsys, int lastseq ) +{ + int reg_type = s[1]; + char keybuf[IT_MAX_WORD+1]; + char *to = keybuf; + const char *from = s + 2; + + while (*from) + { + const char *res = zebra_maps_output (zm, reg_type, &from); + if (!res) + *to++ = *from++; + else + while (*res) + *to++ = *res++; + } + *to = '\0'; + /* yaz_log (LOG_LOG, "%s", keybuf); */ + printf("%10d %s %d.%d - %d.%d\n",count, keybuf, + firstsys,firstseq, lastsys,lastseq); +} + static int inv_stat_handle (char *name, const char *info, int pos, void *client) { @@ -35,12 +60,19 @@ static int inv_stat_handle (char *name, const char *info, int pos, int i = 0; struct inv_stat_info *stat_info = (struct inv_stat_info*) client; ISAMS_P isam_p; + int firstsys=-1; + int firstseq=-1; + int lastsys=-1; + int lastseq=-1; stat_info->no_dict_entries++; stat_info->no_dict_bytes += strlen(name); - assert (*info == sizeof(ISAMS_P)); - memcpy (&isam_p, info+1, sizeof(ISAMS_P)); + if (!stat_info->zh->reg->isamd) + { + assert (*info == sizeof(ISAMS_P)); + memcpy (&isam_p, info+1, sizeof(ISAMS_P)); + } if (stat_info->zh->reg->isams) { @@ -52,8 +84,16 @@ 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); + stat_info->cksum = stat_info->cksum * 65509 + + key.sysno + 11 * key.seqno; occurx++; + if (-1==firstsys) + { + firstseq=key.seqno; + firstsys=key.sysno; + } + lastsys=key.sysno; + lastseq=key.seqno; } assert (occurx == occur); stat_info->no_isam_entries[0] += occur; @@ -65,6 +105,7 @@ static int inv_stat_handle (char *name, const char *info, int pos, 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->zh->reg->isamc) @@ -77,8 +118,16 @@ static int inv_stat_handle (char *name, const char *info, int pos, 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++; + if (-1==firstsys) + { + firstseq=key.seqno; + firstsys=key.sysno; + } + lastsys=key.sysno; + lastseq=key.seqno; } assert (occurx == occur); stat_info->no_isam_entries[isc_type(isam_p)] += occur; @@ -89,30 +138,44 @@ static int inv_stat_handle (char *name, const char *info, int pos, ISAMD_PP pp; int occurx = 0; struct it_key key; - - pp = isamd_pp_open (stat_info->zh->reg->isamd, isam_p); + /* printf("[%d: %d %d %d %d %d %d] ", */ + /* info[0], info[1], info[2], info[3], info[4], info[5], info[7]);*/ + pp = isamd_pp_open (stat_info->zh->reg->isamd, info+1, info[0]); occur = isamd_pp_num (pp); while (isamd_pp_read(pp, &key)) { + stat_info->cksum = stat_info->cksum * 65509 + + key.sysno + 11 * key.seqno; occurx++; + /* printf("%d.%d ", key.sysno, key.seqno); */ /*!*/ + if (-1==firstsys) + { + firstseq=key.seqno; + firstsys=key.sysno; + } + lastsys=key.sysno; + lastseq=key.seqno; if ( pp->is->method->debug >8 ) 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); } + /* printf("\n"); */ /*!*/ +#ifdef SKIPTHIS if ( pp->is->method->debug >7 ) logf(LOG_LOG,"item %d=%d:%d says %d keys, counted %d", isam_p, isamd_type(isam_p), isamd_block(isam_p), occur, occurx); +#endif if (occurx != occur) logf(LOG_LOG,"Count error!!! read %d, counted %d", occur, occurx); assert (occurx == occur); - if ( is_singleton(isam_p) ) - stat_info->no_isam_entries[SINGLETON_TYPE] += occur; - else - stat_info->no_isam_entries[isamd_type(isam_p)] += occur; + i = pp->cat; + if (info[1]) + i=SINGLETON_TYPE; + stat_info->no_isam_entries[i] += occur; isamd_pp_close (pp); } if (stat_info->zh->reg->isamb) @@ -127,28 +190,44 @@ static int inv_stat_handle (char *name, const char *info, int pos, pp = isamb_pp_open_x(stat_info->zh->reg->isamb, isam_p, &level); while (isamb_pp_read(pp, &key)) + { + stat_info->cksum = stat_info->cksum * 65509 + + key.sysno + 11 * key.seqno; occur++; + if (-1==firstsys) + { + firstseq=key.seqno; + firstsys=key.sysno; + } + lastsys=key.sysno; + lastseq=key.seqno; + } 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; } - + i=0; while (occur > stat_info->isam_bounds[i] && stat_info->isam_bounds[i]) i++; ++(stat_info->isam_occurrences[i]); + if (stat_info->dumpwords) + print_dict_item(stat_info->zh->reg->zebra_maps, name, occur, + firstsys,firstseq, lastsys, lastseq); return 0; } -void zebra_register_statistics (ZebraHandle zh) +void zebra_register_statistics (ZebraHandle zh, int dumpdict) { 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]; @@ -157,6 +236,7 @@ void zebra_register_statistics (ZebraHandle zh) return; stat_info.zh = zh; + stat_info.dumpwords=dumpdict; term_dict[0] = 1; term_dict[1] = 0; @@ -185,6 +265,8 @@ void zebra_register_statistics (ZebraHandle zh) 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; @@ -202,26 +284,26 @@ void zebra_register_statistics (ZebraHandle zh) if (zh->reg->isamc) { - fprintf (stderr, " Blocks Occur Size KB Bytes/Entry\n"); + 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 (zh->reg->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", + 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 (zh->reg->isamd) { - fprintf (stderr, " Blocks Occur KB Bytes/Entry\n"); + 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++) @@ -233,7 +315,7 @@ void zebra_register_statistics (ZebraHandle zh) 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, @@ -250,7 +332,7 @@ void zebra_register_statistics (ZebraHandle zh) } /* for */ } /* isamd */ if ( (zh->reg->isamd) && (zh->reg->isamd->method->debug>0)) - fprintf (stderr, "\n%d words using %d bytes\n", + fprintf (stdout, "\n%d words using %d bytes\n", stat_info.no_dict_entries, stat_info.no_dict_bytes); if (zh->reg->isamb) @@ -261,29 +343,37 @@ void zebra_register_statistics (ZebraHandle zh) int bsize = isamb_block_info(zh->reg->isamb, i); if (bsize < 0) break; - fprintf (stderr, "Category %d\n", i); - fprintf (stderr, "Block size %d\n", bsize); - fprintf (stderr, "Blocks: %d\n", stat_info.isamb_blocks[i]); - fprintf (stderr, "Size: %d\n", stat_info.isamb_sizes[i]); - fprintf (stderr, "Total %d\n", stat_info.isamb_blocks[i]* + 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 (stderr, "Level%d %d\n", 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 (stderr, " Occurrences Words\n"); + 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", + 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]); xmalloc_trav("unfreed"); /*! while hunting memory leaks */ zebra_end_read (zh); @@ -293,7 +383,25 @@ void zebra_register_statistics (ZebraHandle zh) /* * * $Log: invstat.c,v $ - * Revision 1.27 2002-04-30 08:28:37 adam + * Revision 1.32 2002-07-12 18:12:22 heikki + * Isam-D now stores small entries directly in the dictionary. + * Needs more tuning and cleaning... + * + * Revision 1.31 2002/07/11 16:16:00 heikki + * Fixed a bug in isamd, failed to store a single key when its bits + * did not fit into a singleton. + * + * Revision 1.30 2002/07/11 13:03:01 heikki + * Added dumpdict command line option to dump the + * dictionary before doing the usual stats + * + * 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