X-Git-Url: http://sru.miketaylor.org.uk/?a=blobdiff_plain;f=rset%2Frsbetween.c;h=28fa70032cab3f91f3dda10b591ceba306adb3bd;hb=b4470de12aa07f21f5394c19af0d21b196087225;hp=f631f333edba31575957d256e420b72d19b60a2b;hpb=10590cc29e076afa175471c1c4d54dd3b2545b51;p=idzebra-moved-to-github.git diff --git a/rset/rsbetween.c b/rset/rsbetween.c index f631f33..28fa700 100644 --- a/rset/rsbetween.c +++ b/rset/rsbetween.c @@ -1,4 +1,4 @@ -/* $Id: rsbetween.c,v 1.12 2004-02-12 15:15:54 heikki Exp $ +/* $Id: rsbetween.c,v 1.19 2004-08-23 12:38:53 heikki Exp $ Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002 Index Data Aps @@ -35,20 +35,22 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include #include -#include #include +#include + +#define RSBETWEEN_DEBUG 0 static void *r_create_between(RSET ct, const struct rset_control *sel, void *parms); static RSFD r_open_between (RSET ct, int flag); static void r_close_between (RSFD rfd); static void r_delete_between (RSET ct); static void r_rewind_between (RSFD rfd); -static int r_forward_between(RSET ct, RSFD rfd, void *buf, int *term_index, +static int r_forward_between(RSET ct, RSFD rfd, void *buf, int (*cmpfunc)(const void *p1, const void *p2), const void *untilbuf); -static int r_count_between (RSET ct); -static int r_read_between (RSFD rfd, void *buf, int *term_index); +static int r_read_between (RSFD rfd, void *buf); static int r_write_between (RSFD rfd, const void *buf); +static void r_pos_between (RSFD rfd, double *current, double *total); static const struct rset_control control_between = { @@ -59,7 +61,7 @@ static const struct rset_control control_between = r_delete_between, r_rewind_between, r_forward_between, /* rset_default_forward, */ - r_count_between, + r_pos_between, r_read_between, r_write_between, }; @@ -73,7 +75,6 @@ struct rset_between_info { RSET rset_m; RSET rset_r; RSET rset_attr; - int term_index_s; int (*cmp)(const void *p1, const void *p2); char *(*printer)(const void *p1, char *buf); struct rset_between_rfd *rfd_list; @@ -88,9 +89,6 @@ struct rset_between_rfd { int more_m; int more_r; int more_attr; - int term_index_l; - int term_index_m; - int term_index_r; void *buf_l; void *buf_m; void *buf_r; @@ -98,8 +96,10 @@ struct rset_between_rfd { int level; struct rset_between_rfd *next; struct rset_between_info *info; + zint hits; }; +#if RSBETWEEN_DEBUG static void log2 (struct rset_between_rfd *p, char *msg, int cmp_l, int cmp_r) { char buf_l[32]; @@ -112,6 +112,7 @@ static void log2 (struct rset_between_rfd *p, char *msg, int cmp_l, int cmp_r) (*p->info->printer)(p->buf_r, buf_r), p->more_r, cmp_r, p->level); } +#endif static void *r_create_between (RSET ct, const struct rset_control *sel, void *parms) @@ -125,47 +126,9 @@ static void *r_create_between (RSET ct, const struct rset_control *sel, info->rset_m = between_parms->rset_m; info->rset_r = between_parms->rset_r; info->rset_attr = between_parms->rset_attr; - if (rset_is_volatile(info->rset_l) || - rset_is_volatile(info->rset_m) || - rset_is_volatile(info->rset_r)) - ct->flags |= RSET_FLAG_VOLATILE; info->cmp = between_parms->cmp; info->printer = between_parms->printer; info->rfd_list = NULL; - - info->term_index_s = info->rset_l->no_rset_terms; - if (info->rset_m) - { - ct->no_rset_terms = - info->rset_l->no_rset_terms + - info->rset_m->no_rset_terms + - info->rset_r->no_rset_terms; - ct->rset_terms = (RSET_TERM *) - xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms); - memcpy (ct->rset_terms, info->rset_l->rset_terms, - info->rset_l->no_rset_terms * sizeof(*ct->rset_terms)); - memcpy (ct->rset_terms + info->rset_l->no_rset_terms, - info->rset_m->rset_terms, - info->rset_m->no_rset_terms * sizeof(*ct->rset_terms)); - memcpy (ct->rset_terms + info->rset_l->no_rset_terms + - info->rset_m->no_rset_terms, - info->rset_r->rset_terms, - info->rset_r->no_rset_terms * sizeof(*ct->rset_terms)); - } - else - { - ct->no_rset_terms = - info->rset_l->no_rset_terms + - info->rset_r->no_rset_terms; - ct->rset_terms = (RSET_TERM *) - xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms); - memcpy (ct->rset_terms, info->rset_l->rset_terms, - info->rset_l->no_rset_terms * sizeof(*ct->rset_terms)); - memcpy (ct->rset_terms + info->rset_l->no_rset_terms, - info->rset_r->rset_terms, - info->rset_r->no_rset_terms * sizeof(*ct->rset_terms)); - } - return info; } @@ -193,20 +156,17 @@ static RSFD r_open_between (RSET ct, int flag) rfd->rfd_m = rset_open (info->rset_m, RSETF_READ); rfd->rfd_r = rset_open (info->rset_r, RSETF_READ); - rfd->more_l = rset_read (info->rset_l, rfd->rfd_l, rfd->buf_l, - &rfd->term_index_l); - rfd->more_m = rset_read (info->rset_m, rfd->rfd_m, rfd->buf_m, - &rfd->term_index_m); - rfd->more_r = rset_read (info->rset_r, rfd->rfd_r, rfd->buf_r, - &rfd->term_index_r); + rfd->more_l = rset_read (info->rset_l, rfd->rfd_l, rfd->buf_l); + rfd->more_m = rset_read (info->rset_m, rfd->rfd_m, rfd->buf_m); + rfd->more_r = rset_read (info->rset_r, rfd->rfd_r, rfd->buf_r); if (info->rset_attr) { - int dummy; rfd->rfd_attr = rset_open (info->rset_attr, RSETF_READ); rfd->more_attr = rset_read (info->rset_attr, rfd->rfd_attr, - rfd->buf_attr, &dummy); + rfd->buf_attr); } rfd->level=0; + rfd->hits=0; return rfd; } @@ -241,7 +201,6 @@ static void r_delete_between (RSET ct) struct rset_between_info *info = (struct rset_between_info *) ct->buf; assert (info->rfd_list == NULL); - xfree (ct->rset_terms); rset_delete (info->rset_l); rset_delete (info->rset_m); rset_delete (info->rset_r); @@ -255,52 +214,55 @@ static void r_rewind_between (RSFD rfd) struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info; struct rset_between_rfd *p = (struct rset_between_rfd *) rfd; +#if RSBETWEEN_DEBUG logf (LOG_DEBUG, "rsbetween_rewind"); +#endif rset_rewind (info->rset_l, p->rfd_l); rset_rewind (info->rset_m, p->rfd_m); rset_rewind (info->rset_r, p->rfd_r); - p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l, &p->term_index_l); - p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m, &p->term_index_m); - p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r, &p->term_index_r); + p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l); + p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m); + p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r); if (info->rset_attr) { - int dummy; rset_rewind (info->rset_attr, p->rfd_attr); - p->more_attr = rset_read (info->rset_attr, p->rfd_attr, p->buf_attr, - &dummy); + p->more_attr = rset_read (info->rset_attr, p->rfd_attr, p->buf_attr); } p->level=0; + p->hits=0; } -static int r_forward_between(RSET ct, RSFD rfd, void *buf, int *term_index, +static int r_forward_between(RSET ct, RSFD rfd, void *buf, int (*cmpfunc)(const void *p1, const void *p2), const void *untilbuf) { struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info; struct rset_between_rfd *p = (struct rset_between_rfd *) rfd; int rc; +#if RSBETWEEN_DEBUG log2( p, "fwd: before forward", 0,0); +#endif /* It is enough to forward the m pointer here, the read will */ /* naturally forward the l, m, and attr pointers */ if (p->more_m) p->more_m=rset_forward(info->rset_m,p->rfd_m, p->buf_m, - &p->term_index_m, info->cmp,untilbuf); + info->cmp,untilbuf); +#if RSBETWEEN_DEBUG log2( p, "fwd: after forward M", 0,0); - rc = r_read_between(rfd, buf, term_index); +#endif + rc = r_read_between(rfd, buf); +#if RSBETWEEN_DEBUG log2( p, "fwd: after forward", 0,0); +#endif return rc; } -static int r_count_between (RSET ct) -{ - return 0; -} -static int r_read_between (RSFD rfd, void *buf, int *term_index) +static int r_read_between (RSFD rfd, void *buf) { struct rset_between_rfd *p = (struct rset_between_rfd *) rfd; struct rset_between_info *info = p->info; @@ -310,7 +272,9 @@ static int r_read_between (RSFD rfd, void *buf, int *term_index) while (p->more_m) { +#if RSBETWEEN_DEBUG log2( p, "start of loop", cmp_l, cmp_r); +#endif /* forward L until past m, count levels, note rec boundaries */ if (p->more_l) @@ -320,7 +284,9 @@ static int r_read_between (RSFD rfd, void *buf, int *term_index) p->level = 0; cmp_l=2; /* past this record */ } +#if RSBETWEEN_DEBUG log2( p, "after first L", cmp_l, cmp_r); +#endif while (cmp_l < 0) /* l before m */ { @@ -335,7 +301,6 @@ static int r_read_between (RSFD rfd, void *buf, int *term_index) else { int cmp_attr; - int dummy_term; attr_match = 0; while (p->more_attr) { @@ -348,8 +313,8 @@ static int r_read_between (RSFD rfd, void *buf, int *term_index) else if (cmp_attr > 0) break; else if (cmp_attr==-1) - p->more_attr = rset_read (info->rset_attr, p->rfd_attr, - p->buf_attr, &dummy_term); + p->more_attr = rset_read (info->rset_attr, + p->rfd_attr, p->buf_attr); /* if we had a forward that went all the way to * the seqno, we could use that. But fwd only goes * to the sysno */ @@ -357,9 +322,10 @@ static int r_read_between (RSFD rfd, void *buf, int *term_index) { p->more_attr = rset_forward( info->rset_attr, p->rfd_attr, - p->buf_attr, &dummy_term, - info->cmp, p->buf_l); + p->buf_attr, info->cmp, p->buf_l); +#if RSBETWEEN_DEBUG logf(LOG_DEBUG, "btw: after frowarding attr m=%d",p->more_attr); +#endif } } /* while more_attr */ } @@ -370,24 +336,22 @@ static int r_read_between (RSFD rfd, void *buf, int *term_index) { if (p->more_l) { - p->more_l=rset_forward( - info->rset_l, p->rfd_l, - p->buf_l, &p->term_index_l, - info->cmp, p->buf_m); + p->more_l=rset_forward( info->rset_l, p->rfd_l, + p->buf_l, info->cmp, p->buf_m); if (p->more_l) cmp_l= (*info->cmp)(p->buf_l, p->buf_m); else cmp_l=2; +#if RSBETWEEN_DEBUG log2( p, "after forwarding L", cmp_l, cmp_r); +#endif } } else { - p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l, - &p->term_index_l); + p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l); } #else - p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l, - &p->term_index_l); + p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l); #endif if (p->more_l) { @@ -395,17 +359,23 @@ static int r_read_between (RSFD rfd, void *buf, int *term_index) } else cmp_l=2; +#if RSBETWEEN_DEBUG log2( p, "end of L loop", cmp_l, cmp_r); +#endif } /* forward L */ /* forward R until past m, count levels */ +#if RSBETWEEN_DEBUG log2( p, "Before moving R", cmp_l, cmp_r); +#endif if (p->more_r) cmp_r= (*info->cmp)(p->buf_r, p->buf_m); else cmp_r=2; +#if RSBETWEEN_DEBUG log2( p, "after first R", cmp_l, cmp_r); +#endif while (cmp_r < 0) /* r before m */ { /* -2, earlier record, don't count level */ @@ -416,27 +386,25 @@ static int r_read_between (RSFD rfd, void *buf, int *term_index) #if NEWCODE if (cmp_r==-2) { - p->more_r=rset_forward( - info->rset_r, p->rfd_r, - p->buf_r, &p->term_index_r, - info->cmp, p->buf_m); + p->more_r=rset_forward( info->rset_r, p->rfd_r, + p->buf_r, info->cmp, p->buf_m); } else { - p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r, - &p->term_index_r); + p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r); } if (p->more_r) cmp_r= (*info->cmp)(p->buf_r, p->buf_m); #else - p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r, - &p->term_index_r); + p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r); cmp_r= (*info->cmp)(p->buf_r, p->buf_m); #endif } else cmp_r=2; +#if RSBETWEEN_DEBUG log2( p, "End of R loop", cmp_l, cmp_r); +#endif } /* forward R */ if ( ( p->level <= 0 ) && ! p->more_l) @@ -445,42 +413,45 @@ static int r_read_between (RSFD rfd, void *buf, int *term_index) if ( attr_match && p->level > 0) /* within a tag pair (or deeper) */ { memcpy (buf, p->buf_m, info->key_size); - *term_index = p->term_index_m; +#if RSBETWEEN_DEBUG log2( p, "Returning a hit (and forwarding m)", cmp_l, cmp_r); - p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m, - &p->term_index_m); +#endif + p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m); if (cmp_l == 2) p->level = 0; + p->hits++; return 1; } else if ( ! p->more_l ) /* not in data, no more starts */ { +#if RSBETWEEN_DEBUG log2( p, "no more starts, exiting without a hit", cmp_l, cmp_r); +#endif return 0; /* ergo, nothing can be found. stop scanning */ } #if NEWCODE if (cmp_l == 2) { p->level = 0; - p->more_m=rset_forward( - info->rset_m, p->rfd_m, - p->buf_m, &p->term_index_m, - info->cmp, p->buf_l); + p->more_m=rset_forward( info->rset_m, p->rfd_m, + p->buf_m, info->cmp, p->buf_l); } else { - p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m, - &p->term_index_m); + p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m); } #else if (cmp_l == 2) p->level = 0; - p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m, - &p->term_index_m); + p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m); #endif +#if RSBETWEEN_DEBUG log2( p, "End of M loop", cmp_l, cmp_r); +#endif } /* while more_m */ +#if RSBETWEEN_DEBUG log2( p, "Exiting, nothing more in m", cmp_l, cmp_r); +#endif return 0; /* no more data possible */ @@ -493,3 +464,37 @@ static int r_write_between (RSFD rfd, const void *buf) return -1; } + +static void r_pos_between (RSFD rfd, double *current, double *total) +{ + struct rset_between_rfd *p = (struct rset_between_rfd *) rfd; + struct rset_between_info *info = p->info; + double lcur,ltot; + double mcur,mtot; + double rcur,rtot; + double r; + ltot=-1; rtot=-1; + rset_pos(info->rset_l, p->rfd_l, &lcur, <ot); + rset_pos(info->rset_m, p->rfd_m, &mcur, &mtot); + rset_pos(info->rset_r, p->rfd_r, &rcur, &rtot); + if ( (ltot<0) && (mtot<0) && (rtot<0) ) { /*no position */ + *current=mcur; /* return same as you got */ + *total=mtot; /* probably -1 for not available */ + } + if ( ltot<0) { ltot=0; lcur=0;} /* if only one useful, use it */ + if ( mtot<0) { mtot=0; mcur=0;} + if ( rtot<0) { rtot=0; rcur=0;} + if ( ltot+mtot+rtot < 1 ) { /* empty rset */ + *current=0; + *total=0; + return; + } + r=1.0*(lcur+mcur+rcur)/(ltot+mtot+rtot); /* weighed average of l and r */ + *current=p->hits; + *total=*current/r ; +#if RSBETWEEN_DEBUG + yaz_log(LOG_DEBUG,"betw_pos: (%s/%s) %0.1f/%0.1f= %0.4f ", + info->rset_l->control->desc, info->rset_r->control->desc, + *current, *total, r); +#endif +}