Added a routine to get an array of terms in a query, in preparation

for ranking

diff --git a/include/rset.h b/include/rset.h
index 0b529a7..2fa67a3 100644 (file)
--- a/include/rset.h
+++ b/include/rset.h
@@ -1,4 +1,4 @@
-/* $Id: rset.h,v 1.38 2004-10-20 14:32:28 heikki Exp $
+/* $Id: rset.h,v 1.39 2004-10-22 10:12:51 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
    Index Data Aps
 
@@ -78,11 +78,19 @@ struct rsfd {  /* the stuff common to all rsfd's. */
  */
 struct rset_control
 {
-    char *desc; /* text description of set type (for debugging) */
+    /** text description of set type (for debugging) */
+    char *desc; 
 /* RSET rs_something_create(const struct rset_control *sel, ...); */
     void (*f_delete)(RSET ct);
+
+    /** recursively fills the terms array with terms. call with curterm=0 */
+    /* always counts them all into cur, but of course won't touch the term */
+    /* array past max. You can use this to count, set max=0 */
+    void (*f_getterms)(RSET ct, TERMID *terms, int maxterms, int *curterm);
+
     RSFD (*f_open)(RSET ct, int wflag);
     void (*f_close)(RSFD rfd);
+    /** forward behaves like a read, but it skips some items first */
     int (*f_forward)(RSFD rfd, void *buf, TERMID *term, const void *untilbuf);
     void (*f_pos)(RSFD rfd, double *current, double *total);
        /* returns -1,-1 if pos function not implemented for this type */
@@ -94,6 +102,15 @@ struct rset_control
 int rset_default_forward(RSFD rfd, void *buf, TERMID *term,
                      const void *untilbuf);
 
+/** rset_get_no_terms is a getterms function for those that don't have any */
+void rset_get_no_terms(RSET ct, TERMID *terms, int maxterms, int *curterm);
+
+/** 
+ * rset_get_one_term is a getterms function for those rsets that have
+ * exactly one term, like all rsisamX types. 
+ */
+void rset_get_one_term(RSET ct,TERMID *terms,int maxterms,int *curterm);
+
 /**
  * key_control contains all there is to know about the keys stored in 
  * an isam, and therefore operated by the rsets. Other than this info,
@@ -173,6 +190,10 @@ RSET rset_dup (RSET rs);
 #define rset_forward(rfd, buf, term, untilbuf) \
     (*(rfd)->rset->control->f_forward)((rfd),(buf),(term),(untilbuf))
 
+/* void rset_getterms(RSET ct, TERMID *terms, int maxterms, int *curterm); */
+#define rset_getterms(ct, terms, maxterms, curterm) \
+    (*(ct)->control->f_getterms)((ct),(terms),(maxterms),(curterm))
+
 /* int rset_pos(RSFD fd, double *current, double *total); */
 #define rset_pos(rfd,cur,tot) \
     (*(rfd)->rset->control->f_pos)( (rfd),(cur),(tot))

diff --git a/index/zsets.c b/index/zsets.c
index e77e5fb..cabefa6 100644 (file)
--- a/index/zsets.c
+++ b/index/zsets.c
@@ -1,4 +1,4 @@
-/* $Id: zsets.c,v 1.65 2004-10-21 12:43:09 heikki Exp $
+/* $Id: zsets.c,v 1.66 2004-10-22 10:12:51 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -733,7 +733,8 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
     struct it_key key;
     RSFD rfd;
     TERMID termid;
-    int i;
+    TERMID *terms;
+    int i,n;
     ZebraRankClass rank_class;
     struct rank_control *rc;
     struct zset_sort_info *sort_info;
@@ -810,6 +811,20 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
         (*rc->end) (zh->reg, handle);
     }
     rset_close (rfd);
+    n=0;
+    logf(LOG_LOG,"FIXME: Counting terms"); 
+    rset_getterms(rset,0,0,&n);
+    logf(LOG_LOG,"FIXME: Got %d terms", n);
+    terms=xmalloc( sizeof(*terms)*n);
+    i=n;
+    n=0;
+    rset_getterms(rset,terms,i,&n);
+    logf(LOG_LOG,"FIXME: Collected %d terms", n);
+    for (i=0;i<n;i++)
+        logf(LOG_LOG,"FIXME: term[%d]='%s'",i,terms[i]->name);
+    xfree(terms);
+
+
 /*
     for (i = 0; i < rset->no_rset_terms; i++)
     {

diff --git a/rset/rsbetween.c b/rset/rsbetween.c
index 1335a24..d099701 100644 (file)
--- a/rset/rsbetween.c
+++ b/rset/rsbetween.c
@@ -1,4 +1,4 @@
-/* $Id: rsbetween.c,v 1.27 2004-10-15 10:07:34 heikki Exp $
+/* $Id: rsbetween.c,v 1.28 2004-10-22 10:12:51 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
    Index Data Aps
 
@@ -48,11 +48,13 @@ static int r_forward_between(RSFD rfd, void *buf,
 static int r_read_between (RSFD rfd, void *buf, TERMID *term );
 static int r_write_between (RSFD rfd, const void *buf);
 static void r_pos_between (RSFD rfd, double *current, double *total);
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm);
 
 static const struct rset_control control = 
 {
     "between",
     r_delete_between,
+    r_get_terms,
     r_open_between,
     r_close_between,
     r_forward_between, 
@@ -447,3 +449,10 @@ static void r_pos_between (RSFD rfd, double *current, double *total)
                     *current, *total, r);
 #endif
 }
+
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm)
+{
+    struct rset_between_info *info = (struct rset_between_info *) ct->priv;
+    rset_getterms(info->rset_m, terms, maxterms, curterm);
+}
+

diff --git a/rset/rsbool.c b/rset/rsbool.c
index e993ecd..3f50b08 100644 (file)
--- a/rset/rsbool.c
+++ b/rset/rsbool.c
@@ -1,4 +1,4 @@
-/* $Id: rsbool.c,v 1.49 2004-10-15 10:07:34 heikki Exp $
+/* $Id: rsbool.c,v 1.50 2004-10-22 10:12:51 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -41,11 +41,14 @@ static int r_read_and (RSFD rfd, void *buf, TERMID *term);
 static int r_read_or (RSFD rfd, void *buf, TERMID *term);
 static int r_read_not (RSFD rfd, void *buf, TERMID *term);
 static int r_write (RSFD rfd, const void *buf);
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm);
+
 
 static const struct rset_control control_and = 
 {
     "and",
     r_delete,
+    r_get_terms,
     r_open,
     r_close,
     r_forward, 
@@ -58,6 +61,7 @@ static const struct rset_control control_or =
 {
     "or",
     r_delete,
+    r_get_terms,
     r_open,
     r_close,
     r_forward, 
@@ -70,6 +74,7 @@ static const struct rset_control control_not =
 {
     "not",
     r_delete,
+    r_get_terms,
     r_open,
     r_close,
     r_forward, 
@@ -512,3 +517,11 @@ static void r_pos (RSFD rfd, double *current, double *total)
                     *current, *total, r);
 #endif
 }
+
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm)
+{
+    struct rset_bool_info *info = (struct rset_bool_info *) ct->priv;
+    rset_getterms(info->rset_l, terms, maxterms, curterm);
+    rset_getterms(info->rset_r, terms, maxterms, curterm);
+}
+

diff --git a/rset/rset.c b/rset/rset.c
index a7c2ffe..42c20f8 100644 (file)
--- a/rset/rset.c
+++ b/rset/rset.c
@@ -1,4 +1,4 @@
-/* $Id: rset.c,v 1.35 2004-10-20 14:32:29 heikki Exp $
+/* $Id: rset.c,v 1.36 2004-10-22 10:12:51 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -148,6 +148,26 @@ int rset_default_forward(RSFD rfd, void *buf, TERMID *term,
     return more;
 }
 
+/** rset_get_no_terms is a getterms function for those that don't have any */
+void rset_get_no_terms(RSET ct, TERMID *terms, int maxterms, int *curterm)
+{
+    return;
+}
+
+/* rset_get_one_term gets that one term from an rset. Used by rsisamX */
+void rset_get_one_term(RSET ct,TERMID *terms,int maxterms,int *curterm)
+{
+    yaz_log(LOG_LOG,"FIXME: get_one_term: max=%d cur=%d", maxterms, *curterm);
+    if (ct->term)
+    {
+        yaz_log(LOG_LOG,"FIXME: get_one_term: '%s'", ct->term->name);
+        if (*curterm < maxterms)
+            terms[*curterm]=ct->term;
+        (*curterm)++;
+    }
+}
+
+
 TERMID rset_term_create (const char *name, int length, const char *flags,
                                     int type, NMEM nmem)
 

diff --git a/rset/rsisamb.c b/rset/rsisamb.c
index e7fbe40..094ba85 100644 (file)
--- a/rset/rsisamb.c
+++ b/rset/rsisamb.c
@@ -1,4 +1,4 @@
-/* $Id: rsisamb.c,v 1.24 2004-10-15 10:07:34 heikki Exp $
+/* $Id: rsisamb.c,v 1.25 2004-10-22 10:12:52 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -42,6 +42,7 @@ static const struct rset_control control =
 {
     "isamb",
     r_delete,
+    rset_get_one_term,
     r_open,
     r_close,
     r_forward, 

diff --git a/rset/rsisamc.c b/rset/rsisamc.c
index 4e8b827..7b54d15 100644 (file)
--- a/rset/rsisamc.c
+++ b/rset/rsisamc.c
@@ -1,4 +1,4 @@
-/* $Id: rsisamc.c,v 1.27 2004-10-15 10:07:34 heikki Exp $
+/* $Id: rsisamc.c,v 1.28 2004-10-22 10:12:52 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -40,6 +40,7 @@ static const struct rset_control control =
 {
     "isamc",
     r_delete,
+    rset_get_one_term,
     r_open,
     r_close,
     rset_default_forward,
@@ -131,3 +132,5 @@ static void r_pos (RSFD rfd, double *current, double *total)
     *total=-1;
 }
 
+
+

diff --git a/rset/rsisams.c b/rset/rsisams.c
index cd096e0..0607c2b 100644 (file)
--- a/rset/rsisams.c
+++ b/rset/rsisams.c
@@ -1,4 +1,4 @@
-/* $Id: rsisams.c,v 1.15 2004-10-15 10:07:34 heikki Exp $
+/* $Id: rsisams.c,v 1.16 2004-10-22 10:12:52 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
    Index Data Aps
 
@@ -38,6 +38,7 @@ static const struct rset_control control =
 {
     "isams",
     r_delete,
+    rset_get_one_term,
     r_open,
     r_close,
     rset_default_forward,
@@ -130,3 +131,5 @@ static void r_pos (RSFD rfd, double *current, double *total)
     *current=-1;  /* sorry, not implemented yet */
     *total=-1;
 }
+
+

diff --git a/rset/rsmultiandor.c b/rset/rsmultiandor.c
index fceb772..4f4bdcd 100644 (file)
--- a/rset/rsmultiandor.c
+++ b/rset/rsmultiandor.c
@@ -1,4 +1,4 @@
-/* $Id: rsmultiandor.c,v 1.6 2004-10-15 10:07:34 heikki Exp $
+/* $Id: rsmultiandor.c,v 1.7 2004-10-22 10:12:52 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
    Index Data Aps
 
@@ -55,11 +55,13 @@ static int r_forward_and(RSFD rfd, void *buf, TERMID *term,
 static int r_forward_or(RSFD rfd, void *buf, TERMID *term,
                      const void *untilbuf);
 static void r_pos (RSFD rfd, double *current, double *total);
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm);
 
 static const struct rset_control control_or = 
 {
     "multi-or",
     r_delete,
+    r_get_terms,
     r_open_or,
     r_close,
     r_forward_or,
@@ -71,6 +73,7 @@ static const struct rset_control control_and =
 {
     "multi-and",
     r_delete,
+    r_get_terms,
     r_open_and,
     r_close,
     r_forward_and,
@@ -579,3 +582,26 @@ static int r_write (RSFD rfd, const void *buf)
     logf (LOG_FATAL, "multior set type is read-only");
     return -1;
 }
+
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm)
+    /* Special case: Some multi-ors have all terms pointing to the same */
+    /* term. We do not want to duplicate those. Other multiors (and ands) */
+    /* have different terms under them. Those we want. */
+{
+    struct rset_multiandor_info *info = 
+        (struct rset_multiandor_info *) ct->priv;
+    int firstterm= *curterm;
+    int i;
+    for (i=0;i<info->no_rsets;i++)
+    {
+        rset_getterms(info->rsets[i], terms, maxterms, curterm);
+        yaz_log(LOG_DEBUG,"rsmulti: getterms: i=%d *cur=%d",i,*curterm);
+        /* FIXME - remove this log once we know it works */
+        if ( ( (*curterm) > firstterm+1 ) &&
+             ( (*curterm) <= maxterms ) &&
+             ( terms[(*curterm)-1] == terms[firstterm] ) )
+            *curterm--; /* forget the term, seen that before */
+    }
+}
+
+

diff --git a/rset/rsnull.c b/rset/rsnull.c
index ca452c1..f3d2135 100644 (file)
--- a/rset/rsnull.c
+++ b/rset/rsnull.c
@@ -1,4 +1,4 @@
-/* $Id: rsnull.c,v 1.28 2004-10-15 10:07:34 heikki Exp $
+/* $Id: rsnull.c,v 1.29 2004-10-22 10:12:52 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
    Index Data Aps
 
@@ -39,6 +39,7 @@ static const struct rset_control control =
 {
     "null",
     r_delete,
+    rset_get_no_terms,
     r_open,
     r_close,
     rset_default_forward,

diff --git a/rset/rsprox.c b/rset/rsprox.c
index 1c1827d..d24d2ef 100644 (file)
--- a/rset/rsprox.c
+++ b/rset/rsprox.c
@@ -1,4 +1,4 @@
-/* $Id: rsprox.c,v 1.19 2004-10-20 14:32:29 heikki Exp $
+/* $Id: rsprox.c,v 1.20 2004-10-22 10:12:52 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -39,11 +39,13 @@ static int r_forward(RSFD rfd, void *buf, TERMID *term, const void *untilbuf);
 static int r_read (RSFD rfd, void *buf, TERMID *term);
 static int r_write (RSFD rfd, const void *buf);
 static void r_pos (RSFD rfd, double *current, double *total);
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm);
 
 static const struct rset_control control = 
 {
     "prox",
     r_delete,
+    r_get_terms,
     r_open,
     r_close,
     r_forward,
@@ -337,3 +339,15 @@ static void r_pos (RSFD rfd, double *current, double *total)
     logf(LOG_DEBUG,"prox_pos: [%d] %0.1f/%0.1f= %0.4f ",
                     i,*current, *total, r);
 }
+
+
+
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm)
+{
+    struct rset_prox_info *info =
+              (struct rset_prox_info *) ct->priv;
+    int i;
+    for (i=0;i<info->rset_no;i++)
+        rset_getterms(info->rset[i], terms, maxterms, curterm);
+}
+

diff --git a/rset/rstemp.c b/rset/rstemp.c
index e8d4a7d..621addf 100644 (file)
--- a/rset/rstemp.c
+++ b/rset/rstemp.c
@@ -1,4 +1,4 @@
-/* $Id: rstemp.c,v 1.51 2004-10-15 10:07:34 heikki Exp $
+/* $Id: rstemp.c,v 1.52 2004-10-22 10:12:52 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
    Index Data Aps
 
@@ -48,6 +48,7 @@ static const struct rset_control control =
 {
     "temp",
     r_delete,
+    rset_get_no_terms,
     r_open,
     r_close,
     rset_default_forward,