estimatehits config option,

[idzebra-moved-to-github.git] / rset / rsbool.c

diff --git a/rset/rsbool.c b/rset/rsbool.c
index 66501d4..bc1a6dd 100644 (file)
--- a/rset/rsbool.c
+++ b/rset/rsbool.c
@@ -1,5 +1,5 @@
-/* $Id: rsbool.c,v 1.25 2004-02-12 15:15:54 heikki Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
+/* $Id: rsbool.c,v 1.34 2004-08-06 09:43:03 heikki Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
 This file is part of the Zebra server.
@@ -20,18 +20,18 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 02111-1307, USA.
 */
 
-
-
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <assert.h>
 
-#include <rsbool.h>
 #include <zebrautl.h>
+#include <rset.h>
+#include <rsbool.h>
 
-#include <../index/index.h> /* for log_keydump. Debugging only */
-
+#ifndef RSET_DEBUG
+#define RSET_DEBUG 0
+#endif
 
 static void *r_create(RSET ct, const struct rset_control *sel, void *parms);
 static RSFD r_open (RSET ct, int flag);
@@ -41,7 +41,7 @@ static void r_rewind (RSFD rfd);
 static int r_forward(RSET ct, RSFD rfd, void *buf, int *term_index,
                      int (*cmpfunc)(const void *p1, const void *p2),
                      const void *untilbuf);
-static int r_count (RSET ct);
+static void r_pos (RSFD rfd, zint *current, zint *total); 
 static int r_read_and (RSFD rfd, void *buf, int *term_index);
 static int r_read_or (RSFD rfd, void *buf, int *term_index);
 static int r_read_not (RSFD rfd, void *buf, int *term_index);
@@ -56,7 +56,7 @@ static const struct rset_control control_and =
     r_delete,
     r_rewind,
     r_forward, /* rset_default_forward, */
-    r_count,
+    r_pos,     /* rset_default_pos */
     r_read_and,
     r_write,
 };
@@ -69,12 +69,8 @@ static const struct rset_control control_or =
     r_close,
     r_delete,
     r_rewind,
-#if 1
     r_forward, 
-#else
-    rset_default_forward,
-#endif
-    r_count,
+    r_pos,
     r_read_or,
     r_write,
 };
@@ -88,7 +84,7 @@ static const struct rset_control control_not =
     r_delete,
     r_rewind,
     r_forward, 
-    r_count,
+    r_pos,
     r_read_not,
     r_write,
 };
@@ -104,10 +100,12 @@ struct rset_bool_info {
     RSET rset_r;
     int term_index_s;
     int (*cmp)(const void *p1, const void *p2);
+    void (*log_item)(int logmask, const void *p, const char *txt);
     struct rset_bool_rfd *rfd_list;
 };
 
 struct rset_bool_rfd {
+    zint hits;
     RSFD rfd_l;
     RSFD rfd_r;
     int  more_l;
@@ -133,6 +131,7 @@ static void *r_create (RSET ct, const struct rset_control *sel, void *parms)
     if (rset_is_volatile(info->rset_l) || rset_is_volatile(info->rset_r))
         ct->flags |= RSET_FLAG_VOLATILE;
     info->cmp = bool_parms->cmp;
+    info->log_item = bool_parms->log_item;
     info->rfd_list = NULL;
     
     info->term_index_s = info->rset_l->no_rset_terms;
@@ -164,6 +163,7 @@ static RSFD r_open (RSET ct, int flag)
     rfd->next = info->rfd_list;
     info->rfd_list = rfd;
     rfd->info = info;
+    rfd->hits=0;
 
     rfd->buf_l = xmalloc (info->key_size);
     rfd->buf_r = xmalloc (info->key_size);
@@ -218,6 +218,7 @@ static void r_rewind (RSFD rfd)
     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_r = rset_read (info->rset_r, p->rfd_r, p->buf_r, &p->term_index_r);
+    p->hits=0;
 }
 
 static int r_forward (RSET ct, RSFD rfd, void *buf, int *term_index,
@@ -226,56 +227,34 @@ static int r_forward (RSET ct, RSFD rfd, void *buf, int *term_index,
 {
     struct rset_bool_info *info = ((struct rset_bool_rfd*)rfd)->info;
     struct rset_bool_rfd *p = (struct rset_bool_rfd *) rfd;
-    int cmp=0;
     int rc;
 
+#if RSET_DEBUG
     logf (LOG_DEBUG, "rsbool_forward (L) [%p] '%s' (ct=%p rfd=%p m=%d,%d)",
                       rfd, ct->control->desc, ct, rfd, p->more_l, p->more_r);
+#endif
     if ( p->more_l && ((cmpfunc)(untilbuf,p->buf_l)==2) )
         p->more_l = rset_forward(info->rset_l, p->rfd_l, p->buf_l,
                         &p->term_index_l, info->cmp, untilbuf);
-
+#if RSET_DEBUG
     logf (LOG_DEBUG, "rsbool_forward (R) [%p] '%s' (ct=%p rfd=%p m=%d,%d)",
                       rfd, ct->control->desc, ct, rfd, p->more_l, p->more_r);
+#endif
     if ( p->more_r && ((cmpfunc)(untilbuf,p->buf_r)==2))
         p->more_r = rset_forward(info->rset_r, p->rfd_r, p->buf_r,
                         &p->term_index_r, info->cmp, untilbuf);
-
+#if RSET_DEBUG
     logf (LOG_DEBUG, "rsbool_forward [%p] calling read, m=%d,%d t=%d", 
                        rfd, p->more_l, p->more_r, p->tail);
+#endif
     
     p->tail=0; 
     rc = rset_read(ct,rfd,buf,term_index); 
+#if RSET_DEBUG
     logf (LOG_DEBUG, "rsbool_forward returning [%p] %d m=%d,%d", 
                        rfd, rc, p->more_l, p->more_r);
+#endif
     return rc;
-
-    if (p->more_l && p->more_r)
-           cmp = (*info->cmp)(p->buf_l, p->buf_r);
-       else if (p->more_l)
-           cmp = -2;
-       else
-           cmp = 2;
-    if ( (cmp<0) && (p->more_l) )
-    {
-        memcpy (buf, p->buf_l, info->key_size);
-           *term_index = p->term_index_l;
-        logf (LOG_DEBUG, "rsbool_forward returning L (cmp=%d)",cmp);
-        return 1;
-    } else if ( (cmp>0) && (p->more_r) )
-    {
-        memcpy (buf, p->buf_r, info->key_size);
-           *term_index = p->term_index_r + info->term_index_s;
-        logf (LOG_DEBUG, "rsbool_forward returning R (cmp=%d)",cmp);
-        return 1;
-    }
-    /* return ( p->more_l || p->more_r); */
-    return 0;
-}
-
-static int r_count (RSET ct)
-{
-    return 0;
 }
 
 
@@ -299,6 +278,8 @@ static int r_read_and (RSFD rfd, void *buf, int *term_index)
     struct rset_bool_rfd *p = (struct rset_bool_rfd *) rfd;
     struct rset_bool_info *info = p->info;
 
+    { zint cur,tot; r_pos(rfd, &cur, &tot); } 
+
     while (p->more_l || p->more_r)
     {
         int cmp;
@@ -309,10 +290,12 @@ static int r_read_and (RSFD rfd, void *buf, int *term_index)
             cmp = -2;
         else
             cmp = 2;
+#if RSET_DEBUG
         logf (LOG_DEBUG, "r_read_and [%p] looping: m=%d/%d c=%d t=%d",
                         rfd, p->more_l, p->more_r, cmp, p->tail);
-        key_logdump(LOG_DEBUG,p->buf_l);
-        key_logdump(LOG_DEBUG,p->buf_r);
+        (*info->log_item)(LOG_DEBUG, p->buf_l, "left ");
+        (*info->log_item)(LOG_DEBUG, p->buf_r, "right ");
+#endif
         if (!cmp)
         {
             memcpy (buf, p->buf_l, info->key_size);
@@ -328,9 +311,13 @@ static int r_read_and (RSFD rfd, void *buf, int *term_index)
             p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
                                   &p->term_index_r);
             p->tail = 1;
+#if RSET_DEBUG
             logf (LOG_DEBUG, "r_read_and [%p] returning R m=%d/%d c=%d",
                     rfd, p->more_l, p->more_r, cmp);
             key_logdump(LOG_DEBUG,buf);
+           (*info->log_item)(LOG_DEBUG, buf, "");
+#endif
+            p->hits++;
             return 1;
         }
         else if (cmp == -1)
@@ -340,9 +327,12 @@ static int r_read_and (RSFD rfd, void *buf, int *term_index)
             p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
                                   &p->term_index_l);
             p->tail = 1;
+#if RSET_DEBUG
             logf (LOG_DEBUG, "r_read_and [%p] returning L m=%d/%d c=%d",
                     rfd, p->more_l, p->more_r, cmp);
-            key_logdump(LOG_DEBUG,buf);
+           (*info->log_item)(LOG_DEBUG, buf, "");
+#endif
+            p->hits++;
             return 1;
         }
         else if (cmp > 1)  /* cmp == 2 */
@@ -358,9 +348,12 @@ static int r_read_and (RSFD rfd, void *buf, int *term_index)
             {
                 if (!p->more_r || (*info->cmp)(p->buf_r, buf) > 1)
                     p->tail = 0;
+#if RSET_DEBUG
                 logf (LOG_DEBUG, "r_read_and returning C m=%d/%d c=%d",
                         p->more_l, p->more_r, cmp);
-                key_logdump(LOG_DEBUG,buf);
+               (*info->log_item)(LOG_DEBUG, buf, "");
+#endif
+                p->hits++;
                 return 1;
             }
 #else
@@ -373,14 +366,20 @@ static int r_read_and (RSFD rfd, void *buf, int *term_index)
                                    &p->term_index_r);
                 if (!p->more_r || (*info->cmp)(p->buf_r, buf) > 1)
                     p->tail = 0;
+#if RSET_DEBUG
                 logf (LOG_DEBUG, "r_read_and [%p] returning R tail m=%d/%d c=%d",
                         rfd, p->more_l, p->more_r, cmp);
-                key_logdump(LOG_DEBUG,buf);
+               (*info->log_item)(LOG_DEBUG, buf, "");
+#endif
+                p->hits++;
                 return 1;
-            } else
+            }
+           else
             {
+#if RSET_DEBUG
                 logf (LOG_DEBUG, "r_read_and [%p] about to forward R m=%d/%d c=%d",
                         rfd, p->more_l, p->more_r, cmp);
+#endif
                 if (p->more_r && p->more_l)
                     p->more_r = rset_forward( 
                                     info->rset_r, p->rfd_r, 
@@ -402,9 +401,12 @@ static int r_read_and (RSFD rfd, void *buf, int *term_index)
              {
                  if (!p->more_l || (*info->cmp)(p->buf_l, buf) > 1)
                      p->tail = 0;
+#if RSET_DEBUG
                  logf (LOG_DEBUG, "r_read_and [%p] returning R tail m=%d/%d c=%d",
                         rfd, p->more_l, p->more_r, cmp);
-                 key_logdump(LOG_DEBUG,buf);
+                (*info->log_item)(LOG_DEBUG, buf, "");
+#endif
+                 p->hits++;
                  return 1;
              }
 #else
@@ -416,15 +418,20 @@ static int r_read_and (RSFD rfd, void *buf, int *term_index)
                                   &p->term_index_l);
                 if (!p->more_l || (*info->cmp)(p->buf_l, buf) > 1)
                     p->tail = 0;
+#if RSET_DEBUG
                 logf (LOG_DEBUG, "r_read_and [%p] returning L tail m=%d/%d c=%d",
                         rfd, p->more_l, p->more_r, cmp);
-                key_logdump(LOG_DEBUG,buf);
+               (*info->log_item)(LOG_DEBUG, buf, "");
+#endif
+                p->hits++;
                 return 1;
             }
             else
             {
+#if RSET_DEBUG
                 logf (LOG_DEBUG, "r_read_and [%p] about to forward L m=%d/%d c=%d",
                         rfd, p->more_l, p->more_r, cmp);
+#endif
                 if (p->more_r && p->more_l)
                     p->more_l = rset_forward( 
                     /* p->more_l = rset_default_forward( */
@@ -437,7 +444,9 @@ static int r_read_and (RSFD rfd, void *buf, int *term_index)
 #endif
         }
     }
+#if RSET_DEBUG
     logf (LOG_DEBUG, "r_read_and [%p] reached its end",rfd);
+#endif
     return 0;
 }
 
@@ -446,6 +455,7 @@ static int r_read_or (RSFD rfd, void *buf, int *term_index)
     struct rset_bool_rfd *p = (struct rset_bool_rfd *) rfd;
     struct rset_bool_info *info = p->info;
 
+    { zint cur,tot; r_pos(rfd, &cur, &tot); }
     while (p->more_l || p->more_r)
     {
         int cmp;
@@ -464,9 +474,12 @@ static int r_read_or (RSFD rfd, void *buf, int *term_index)
                                   &p->term_index_l);
             p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
                                   &p->term_index_r);
+#if RSET_DEBUG
             logf (LOG_DEBUG, "r_read_or returning A m=%d/%d c=%d",
                     p->more_l, p->more_r, cmp);
-            key_logdump(LOG_DEBUG,buf);
+            (*info->log_item)(LOG_DEBUG, buf, "");
+#endif
+            p->hits++;
             return 1;
         }
         else if (cmp > 0)
@@ -475,9 +488,12 @@ static int r_read_or (RSFD rfd, void *buf, int *term_index)
                *term_index = p->term_index_r + info->term_index_s;
             p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
                                   &p->term_index_r);
+#if RSET_DEBUG
             logf (LOG_DEBUG, "r_read_or returning B m=%d/%d c=%d",
                     p->more_l, p->more_r, cmp);
-            key_logdump(LOG_DEBUG,buf);
+            (*info->log_item)(LOG_DEBUG, buf, "");
+#endif
+            p->hits++;
             return 1;
         }
         else
@@ -486,9 +502,12 @@ static int r_read_or (RSFD rfd, void *buf, int *term_index)
                *term_index = p->term_index_l;
             p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
                                   &p->term_index_l);
+#if RSET_DEBUG
             logf (LOG_DEBUG, "r_read_or returning C m=%d/%d c=%d",
                     p->more_l, p->more_r, cmp);
-            key_logdump(LOG_DEBUG,buf);
+            (*info->log_item)(LOG_DEBUG, buf, "");
+#endif
+            p->hits++;
             return 1;
         }
     }
@@ -500,6 +519,7 @@ static int r_read_not (RSFD rfd, void *buf, int *term_index)
     struct rset_bool_rfd *p = (struct rset_bool_rfd *) rfd;
     struct rset_bool_info *info = p->info;
 
+    { zint cur,tot; r_pos(rfd, &cur, &tot); }
     while (p->more_l || p->more_r)
     {
         int cmp;
@@ -516,18 +536,16 @@ static int r_read_not (RSFD rfd, void *buf, int *term_index)
                *term_index = p->term_index_l;
             p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
                                   &p->term_index_l);
+            p->hits++;
             return 1;
         }
         else if (cmp > 1)
-#if 0
-            p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
-                                  &p->term_index_r);
-#else
-            p->more_r = rset_forward( 
-                            info->rset_r, p->rfd_r, 
-                            p->buf_r, &p->term_index_r, 
-                            (info->cmp), p->buf_l);
-#endif
+        {
+               p->more_r = rset_forward( 
+                   info->rset_r, p->rfd_r, 
+                   p->buf_r, &p->term_index_r, 
+                   (info->cmp), p->buf_l);
+        }
         else
         {
             memcpy (buf, p->buf_l, info->key_size);
@@ -559,3 +577,33 @@ static int r_write (RSFD rfd, const void *buf)
     return -1;
 }
 
+static void r_pos (RSFD rfd, zint *current, zint *total)
+{
+    struct rset_bool_rfd *p = (struct rset_bool_rfd *) rfd;
+    struct rset_bool_info *info = p->info;
+    zint lcur,ltot;
+    zint rcur,rtot;
+    float r;
+    ltot=-1; rtot=-1;
+    rset_pos(info->rset_l, p->rfd_l,  &lcur, &ltot);
+    rset_pos(info->rset_r, p->rfd_r,  &rcur, &rtot);
+    if ( (rtot<0) && (ltot<0)) { /*no position */
+        *current=rcur;  /* return same as you got */
+        *total=rtot;    /* probably -1 for not available */
+    }
+    if ( rtot<0) { rtot=0; rcur=0;} /* if only one useful, use it */
+    if ( ltot<0) { ltot=0; lcur=0;}
+    if ( rtot+ltot == 0 ) { /* empty rset */
+        *current=0;
+        *total=0;
+        return;
+    }
+    r=1.0*(lcur+rcur)/(ltot+rtot); /* weighed average of l and r */
+    *current=p->hits;
+    *total=(zint)(0.5+*current/r); 
+#if RSET_DEBUG
+    yaz_log(LOG_DEBUG,"bool_pos: (%s/%s) "ZINT_FORMAT"/"ZINT_FORMAT"= %0.4f ",
+                    info->rset_l->control->desc, info->rset_r->control->desc,
+                    *current, *total, r);
+#endif
+}