More isamb statistics

diff --git a/include/isamb.h b/include/isamb.h
index e89e8cc..9fb5f71 100644 (file)
--- a/include/isamb.h
+++ b/include/isamb.h
@@ -1,7 +1,7 @@
 /*
  * Copyright (C) 2000-2002, Index Data
  * All rights reserved.
- * $Id: isamb.h,v 1.2 2002-04-16 22:31:42 adam Exp $
+ * $Id: isamb.h,v 1.3 2002-04-29 18:03:45 adam Exp $
  */
 
 #ifndef ISAMB_H
@@ -27,4 +27,9 @@ void isamb_pp_close (ISAMB_PP pp);
 
 int isamb_pp_num (ISAMB_PP pp);
 
+ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level);
+void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks);
+
+int isamb_block_info (ISAMB isamb, int cat);
+
 #endif

diff --git a/index/invstat.c b/index/invstat.c
index f4b548c..7310b29 100644 (file)
--- a/index/invstat.c
+++ b/index/invstat.c
@@ -20,6 +20,10 @@ struct inv_stat_info {
     int isam_bounds[20];
     int isam_occurrences[20];
     char tmp[128];
+    int isamb_levels[10];
+    int isamb_sizes[10];
+    int isamb_blocks[10];
+    int isamb_no[10];
 };
 
 #define SINGLETON_TYPE 8 /* the type to use for singletons that */ 
@@ -116,11 +120,19 @@ static int inv_stat_handle (char *name, const char *info, int pos,
     {
         ISAMB_PP pp;
         struct it_key key;
-
-        pp = isamb_pp_open(stat_info->zh->reg->isamb, isam_p);
+        int cat = isam_p & 3;
+        int level;
+        int size;
+        int blocks;
+        
+        pp = isamb_pp_open_x(stat_info->zh->reg->isamb, isam_p, &level);
         while (isamb_pp_read(pp, &key))
             occur++;
-        isamb_pp_close (pp);
+        isamb_pp_close_x (pp, &size, &blocks);
+        stat_info->isamb_no[cat]++;
+        stat_info->isamb_levels[cat] += level;
+        stat_info->isamb_blocks[cat] += blocks;
+        stat_info->isamb_sizes[cat] += size;
     }
 
     while (occur > stat_info->isam_bounds[i] && stat_info->isam_bounds[i])
@@ -175,6 +187,14 @@ void zebra_register_statistics (ZebraHandle zh)
     for (i = 0; i<20; i++)
         stat_info.isam_occurrences[i] = 0;
 
+    for (i = 0; i<10; i++)
+    {
+        stat_info.isamb_levels[i] = 0;
+        stat_info.isamb_sizes[i] = 0;
+        stat_info.isamb_blocks[i] = 0;
+        stat_info.isamb_no[i] = 0;
+    }
+
     dict_scan (zh->reg->dict, term_dict, &before, &after, &stat_info,
                inv_stat_handle);
 
@@ -230,6 +250,24 @@ void zebra_register_statistics (ZebraHandle zh)
     if ( (zh->reg->isamd) && (zh->reg->isamd->method->debug>0))
         fprintf (stderr, "\n%d words using %d bytes\n",
              stat_info.no_dict_entries, stat_info.no_dict_bytes);
+
+    if (zh->reg->isamb)
+    {
+        for (i = 0; i<4; i++)
+        {
+            int bsize = isamb_block_info(zh->reg->isamb, i);
+            if (bsize < 0)
+                break;
+            fprintf (stderr, "Category   %d\n", i);
+            fprintf (stderr, "Block size %d\n", bsize);
+            fprintf (stderr, "Lists:     %d\n", stat_info.isamb_no[i]);
+            fprintf (stderr, "Blocks:    %d\n", stat_info.isamb_blocks[i]);
+            fprintf (stderr, "Size:      %d\n", stat_info.isamb_sizes[i]);
+            fprintf (stderr, "Total      %d\n", stat_info.isamb_blocks[i]*
+                     bsize);
+        }
+    }
+
     fprintf (stderr, "    Occurrences     Words\n");
     prev = 1;
     for (i = 0; stat_info.isam_bounds[i]; i++)
@@ -249,7 +287,10 @@ void zebra_register_statistics (ZebraHandle zh)
 /*
  *
  * $Log: invstat.c,v $
- * Revision 1.25  2002-04-26 08:44:47  adam
+ * Revision 1.26  2002-04-29 18:03:46  adam
+ * More isamb statistics
+ *
+ * Revision 1.25  2002/04/26 08:44:47  adam
  * Index statistics working again
  *
  * Revision 1.24  2002/04/05 08:46:26  adam

diff --git a/isamb/isamb.c b/isamb/isamb.c
index 5627af4..8578f96 100644 (file)
--- a/isamb/isamb.c
+++ b/isamb/isamb.c
@@ -2,7 +2,7 @@
  *  Copyright (c) 2000-2002, Index Data.
  *  See the file LICENSE for details.
  *
- *  $Id: isamb.c,v 1.10 2002-04-26 08:44:47 adam Exp $
+ *  $Id: isamb.c,v 1.11 2002-04-29 18:03:46 adam Exp $
  */
 #include <yaz/xmalloc.h>
 #include <yaz/log.h>
@@ -53,6 +53,8 @@ struct ISAMB_block {
 struct ISAMB_PP_s {
     ISAMB isamb;
     int level;
+    int total_size;
+    int no_blocks;
     struct ISAMB_block **block;
 };
 
@@ -558,7 +560,7 @@ int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
     return pos;
 }
 
-ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
+ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
 {
     ISAMB_PP pp = xmalloc (sizeof(*pp));
 
@@ -566,12 +568,17 @@ ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
     pp->block = xmalloc (10 * sizeof(*pp->block));
 
     pp->level = 0;
+    pp->total_size = 0;
+    pp->no_blocks = 0;
     while (1)
     {
         struct ISAMB_block *p = open_block (isamb, pos);
         char *src = p->bytes + p->offset;
         pp->block[pp->level] = p;
 
+        pp->total_size += p->size;
+        pp->no_blocks++;
+
         if (p->bytes[0]) /* leaf */
             break;
 
@@ -580,20 +587,43 @@ ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
         pp->level++;
     }
     pp->block[pp->level+1] = 0;
+    if (level)
+        *level = pp->level;
     return pp;
 }
 
-void isamb_pp_close (ISAMB_PP pp)
+ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
+{
+    return isamb_pp_open_x (isamb, pos, 0);
+}
+
+void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks)
 {
     int i;
     if (!pp)
         return;
+    if (size)
+        *size = pp->total_size;
+    if (blocks)
+        *blocks = pp->no_blocks;
     for (i = 0; i <= pp->level; i++)
         close_block (pp->isamb, pp->block[i]);
     xfree (pp->block);
     xfree (pp);
 }
 
+int isamb_block_info (ISAMB isamb, int cat)
+{
+    if (cat >= 0 && cat < isamb->no_cat)
+        return isamb->file[cat].head.block_size;
+    return -1;
+}
+
+void isamb_pp_close (ISAMB_PP pp)
+{
+    return isamb_pp_close_x (pp, 0, 0);
+}
+
 int isamb_pp_read (ISAMB_PP pp, void *buf)
 {
     char *dst = buf;
@@ -628,6 +658,9 @@ int isamb_pp_read (ISAMB_PP pp, void *buf)
         while (1)
         {
             pp->block[pp->level] = p = open_block (pp->isamb, pos);
+
+            pp->total_size += p->size;
+            pp->no_blocks++;
             
             if (p->leaf) /* leaf */
             {