Work on relevance feedback.

diff --git a/rset/Makefile b/rset/Makefile
index 7769f03..570d1ac 100644 (file)
--- a/rset/Makefile
+++ b/rset/Makefile
@@ -1,7 +1,7 @@
 # Copyright (C) 1994, Index Data I/S 
 # All rights reserved.
 # Sebastian Hammer, Adam Dickmeiss
-# $Id: Makefile,v 1.6 1995-09-06 13:27:15 adam Exp $
+# $Id: Makefile,v 1.7 1995-09-08 14:52:41 adam Exp $
 
 SHELL=/bin/sh
 RANLIB=ranlib
@@ -11,7 +11,7 @@ INCLUDE=-I../include -I$(YAZ)/include
 DEFS=$(INCLUDE)
 LIB=../lib/rset.a
 PROG=
-PO=rset.o rstemp.o rsisam.o rsnull.o rsbool.o
+PO=rset.o rstemp.o rsisam.o rsnull.o rsbool.o rsrel.o
 CPP=cc -E
 
 all: $(LIB)

diff --git a/rset/rsbool.c b/rset/rsbool.c
index a20c452..f34a61f 100644 (file)
--- a/rset/rsbool.c
+++ b/rset/rsbool.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: rsbool.c,v $
- * Revision 1.4  1995-09-08 08:54:04  adam
+ * Revision 1.5  1995-09-08 14:52:41  adam
+ * Work on relevance feedback.
+ *
+ * Revision 1.4  1995/09/08  08:54:04  adam
  * More efficient and operation.
  *
  * Revision 1.3  1995/09/07  13:58:43  adam
@@ -109,7 +112,6 @@ static rset_control *r_create (const struct rset_control *sel, void *parms)
     rset_bool_parms *bool_parms = parms;
     struct rset_bool_info *info;
 
-    logf (LOG_DEBUG, "rsbool_create(%s)", sel->desc);
     newct = xmalloc(sizeof(*newct));
     memcpy (newct, sel, sizeof(*sel));
     newct->buf = xmalloc (sizeof(struct rset_bool_info));

diff --git a/rset/rset.c b/rset/rset.c
index e10f1b1..0776cb2 100644 (file)
--- a/rset/rset.c
+++ b/rset/rset.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: rset.c,v $
- * Revision 1.5  1995-09-07 13:58:43  adam
+ * Revision 1.6  1995-09-08 14:52:41  adam
+ * Work on relevance feedback.
+ *
+ * Revision 1.5  1995/09/07  13:58:43  adam
  * New parameter: result-set file descriptor (RSFD) to support multiple
  * positions within the same result-set.
  * Boolean operators: and, or, not implemented.
@@ -32,6 +35,7 @@ RSET rset_create(const rset_control *sel, void *parms)
 {
     RSET new;
 
+    logf (LOG_DEBUG, "rs_create(%s)", sel->desc);
     new = xmalloc(sizeof(*new));     /* make dynamic alloc scheme */
     if (!(new->control = (*sel->f_create)(sel, parms)))
        return 0;

diff --git a/rset/rsisam.c b/rset/rsisam.c
index 1eb1828..371b7c7 100644 (file)
--- a/rset/rsisam.c
+++ b/rset/rsisam.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: rsisam.c,v $
- * Revision 1.9  1995-09-07 13:58:43  adam
+ * Revision 1.10  1995-09-08 14:52:42  adam
+ * Work on relevance feedback.
+ *
+ * Revision 1.9  1995/09/07  13:58:43  adam
  * New parameter: result-set file descriptor (RSFD) to support multiple
  * positions within the same result-set.
  * Boolean operators: and, or, not implemented.
@@ -80,7 +83,6 @@ static rset_control *r_create(const struct rset_control *sel, void *parms)
     rset_isam_parms *pt = parms;
     struct rset_isam_info *info;
 
-    logf (LOG_DEBUG, "rsisam_create(%s)", sel->desc);
     newct = xmalloc(sizeof(*newct));
     memcpy(newct, sel, sizeof(*sel));
 

diff --git a/rset/rsnull.c b/rset/rsnull.c
index a278293..d46c9aa 100644 (file)
--- a/rset/rsnull.c
+++ b/rset/rsnull.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: rsnull.c,v $
- * Revision 1.2  1995-09-07 13:58:43  adam
+ * Revision 1.3  1995-09-08 14:52:42  adam
+ * Work on relevance feedback.
+ *
+ * Revision 1.2  1995/09/07  13:58:43  adam
  * New parameter: result-set file descriptor (RSFD) to support multiple
  * positions within the same result-set.
  * Boolean operators: and, or, not implemented.
@@ -47,7 +50,6 @@ static rset_control *r_create(const struct rset_control *sel, void *parms)
 {
     rset_control *newct;
 
-    logf (LOG_DEBUG, "rsnull_create(%s)", sel->desc);
     newct = xmalloc(sizeof(*newct));
     memcpy(newct, sel, sizeof(*sel));
     return newct;

diff --git a/rset/rsrel.c b/rset/rsrel.c
new file mode 100644 (file)
index 0000000..4faed5b
--- /dev/null
+++ b/rset/rsrel.c
@@ -0,0 +1,188 @@
+/*
+ * Copyright (C) 1994-1995, Index Data I/S 
+ * All rights reserved.
+ * Sebastian Hammer, Adam Dickmeiss
+ *
+ * $Log: rsrel.c,v $
+ * Revision 1.1  1995-09-08 14:52:42  adam
+ * Work on relevance feedback.
+ *
+ */
+
+#include <stdio.h>
+#include <assert.h>
+
+#include <isam.h>
+#include <rsrel.h>
+#include <alexutil.h>
+
+static rset_control *r_create(const struct rset_control *sel, void *parms);
+static RSFD r_open (rset_control *ct, int wflag);
+static void r_close (RSFD rfd);
+static void r_delete (rset_control *ct);
+static void r_rewind (RSFD rfd);
+static int r_count (rset_control *ct);
+static int r_read (RSFD rfd, void *buf);
+static int r_write (RSFD rfd, const void *buf);
+
+static const rset_control control = 
+{
+    "relevance set type",
+    0,
+    r_create,
+    r_open,
+    r_close,
+    r_delete,
+    r_rewind,
+    r_count,
+    r_read,
+    r_write
+};
+
+const rset_control *rset_kind_relevance = &control;
+
+struct rset_rel_info {
+    int     key_size;
+    int     max_rec;
+    int     no_rec;
+    int     (*cmp)(const void *p1, const void *p2);
+    void    *key_buf;
+    int     *score_buf;
+
+    struct rset_rel_rfd *rfd_list;
+};
+
+struct rset_rel_rfd {
+    int     position;
+    struct rset_rel_rfd *next;
+    struct rset_rel_info *info;
+};
+
+static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
+{
+    char *isam_buf;
+    int  *isam_r;
+    ISPT *isam_pt;
+    int i;
+
+    isam_buf = xmalloc (info->key_size * parms->no_isam_positions);
+    isam_r = xmalloc (sizeof (*isam_r) * parms->no_isam_positions);
+    isam_pt = xmalloc (sizeof (*isam_pt) * parms->no_isam_positions);
+
+    for (i = 0; i<parms->no_isam_positions; i++)
+    {
+        isam_pt[i] = is_position (parms->is, parms->isam_positions[i]);
+        isam_r[i] = is_readkey (isam_pt[i], isam_buf + i*info->key_size);
+    }
+
+    for (i = 0; i<parms->no_isam_positions; i++)
+        is_pt_free (isam_pt[i]);
+    xfree (isam_buf);
+    xfree (isam_r);
+    xfree (isam_pt);
+}
+
+static rset_control *r_create (const struct rset_control *sel, void *parms)
+{
+    rset_control *newct;
+    rset_relevance_parms *r_parms = parms;
+    struct rset_rel_info *info;
+
+    newct = xmalloc(sizeof(*newct));
+    memcpy(newct, sel, sizeof(*sel));
+    newct->buf = xmalloc (sizeof(struct rset_rel_info));
+
+    info = newct->buf;
+    info->key_size = r_parms->key_size;
+    assert (info->key_size > 1);
+    info->max_rec = r_parms->max_rec;
+    assert (info->max_rec > 1);
+    info->cmp = r_parms->cmp;
+
+    info->key_buf = xmalloc (info->key_size * info->max_rec);
+    info->score_buf = xmalloc (sizeof(*info->score_buf) * info->max_rec);
+    info->no_rec = 0;
+    info->rfd_list = NULL;
+
+    relevance (info, r_parms);
+    return newct;
+}
+
+static RSFD r_open (rset_control *ct, int wflag)
+{
+    struct rset_rel_rfd *rfd;
+    struct rset_rel_info *info = ct->buf;
+
+    if (wflag)
+    {
+       logf (LOG_FATAL, "relevance set type is read-only");
+       return NULL;
+    }
+    rfd = xmalloc (sizeof(*rfd));
+    rfd->next = info->rfd_list;
+    info->rfd_list = rfd;
+    rfd->position = 0;
+    rfd->info = info;
+    return rfd;
+}
+
+static void r_close (RSFD rfd)
+{
+    struct rset_rel_info *info = ((struct rset_rel_rfd*)rfd)->info;
+    struct rset_rel_rfd **rfdp;
+    
+    for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
+        if (*rfdp == rfd)
+        {
+            *rfdp = (*rfdp)->next;
+            free (rfd);
+            return;
+        }
+    logf (LOG_FATAL, "r_close but no rfd match!");
+    assert (0);
+}
+
+static void r_delete (rset_control *ct)
+{
+    struct rset_rel_info *info = ct->buf;
+
+    assert (info->rfd_list == NULL);
+    xfree (info->key_buf);
+    xfree (info->score_buf);
+    xfree (info);
+    xfree (ct);
+}
+
+static void r_rewind (RSFD rfd)
+{
+    ((struct rset_rel_rfd*) rfd)->position = 0;
+}
+
+static int r_count (rset_control *ct)
+{
+    struct rset_rel_info *info = ct->buf;
+
+    return info->no_rec;
+}
+
+static int r_read (RSFD rfd, void *buf)
+{
+    struct rset_rel_rfd *p = rfd;
+    struct rset_rel_info *info = p->info;
+
+    if (p->position >= info->max_rec)
+        return 0;
+    memcpy ((char*) buf + sizeof(*info->score_buf),
+            (char*) info->key_buf + info->key_size * p->position,
+            info->key_size);
+    memcpy ((char*) buf,
+            info->score_buf + p->position, sizeof(*info->score_buf));
+    ++(p->position);
+    return 1;
+}
+
+static int r_write (RSFD rfd, const void *buf)
+{
+    logf (LOG_FATAL, "relevance set type is read-only");
+    return -1;
+}

diff --git a/rset/rstemp.c b/rset/rstemp.c
index 0c835f9..e7a0a4c 100644 (file)
--- a/rset/rstemp.c
+++ b/rset/rstemp.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: rstemp.c,v $
- * Revision 1.7  1995-09-07 13:58:44  adam
+ * Revision 1.8  1995-09-08 14:52:42  adam
+ * Work on relevance feedback.
+ *
+ * Revision 1.7  1995/09/07  13:58:44  adam
  * New parameter: result-set file descriptor (RSFD) to support multiple
  * positions within the same result-set.
  * Boolean operators: and, or, not implemented.
@@ -88,7 +91,6 @@ static struct rset_control *r_create(const struct rset_control *sel,
     rset_temp_parms *temp_parms = parms;
     struct rset_temp_info *info;
     
-    logf (LOG_DEBUG, "ritemp_create(%s)", sel->desc);
     newct = xmalloc(sizeof(*newct));
     memcpy(newct, sel, sizeof(*sel));
     newct->buf = xmalloc (sizeof(struct rset_temp_info));