X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;f=index%2Finvstat.c;h=69973eb0bb65e24426a89f48a884eb77f11af639;hb=dbf7ead21f60be189f0d532175c92924a8eb73fd;hp=f5bc6aa9e31cc0b40d32dfb1aef17963b72b0dbf;hpb=38f321eafd606e8ac91da5c1450393fed73d37f5;p=idzebra-moved-to-github.git diff --git a/index/invstat.c b/index/invstat.c index f5bc6aa..69973eb 100644 --- a/index/invstat.c +++ b/index/invstat.c @@ -30,7 +30,8 @@ struct inv_stat_info { #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 ) +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]; @@ -48,7 +49,8 @@ static void print_dict_item (ZebraMaps zm, const char *s, int count ) } *to = '\0'; /* yaz_log (LOG_LOG, "%s", keybuf); */ - printf("%10d %s\n",count, 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, @@ -58,13 +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) { @@ -79,6 +87,13 @@ static int inv_stat_handle (char *name, const char *info, int pos, 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; @@ -106,6 +121,13 @@ static int inv_stat_handle (char *name, const char *info, int pos, 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; @@ -116,8 +138,9 @@ 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)) @@ -125,23 +148,34 @@ static int inv_stat_handle (char *name, const char *info, int pos, 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) @@ -160,6 +194,13 @@ static int inv_stat_handle (char *name, const char *info, int pos, 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; @@ -169,12 +210,13 @@ static int inv_stat_handle (char *name, const char *info, int pos, 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); + print_dict_item(stat_info->zh->reg->zebra_maps, name, occur, + firstsys,firstseq, lastsys, lastseq); return 0; } @@ -341,7 +383,15 @@ void zebra_register_statistics (ZebraHandle zh, int dumpdict) /* * * $Log: invstat.c,v $ - * Revision 1.30 2002-07-11 13:03:01 heikki + * 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 *