isamb fixes for pp_read. Statistics

diff --git a/index/invstat.c b/index/invstat.c
index 7310b29..ca44bda 100644 (file)
--- a/index/invstat.c
+++ b/index/invstat.c
@@ -20,10 +20,9 @@ struct inv_stat_info {
     int isam_bounds[20];
     int isam_occurrences[20];
     char tmp[128];
-    int isamb_levels[10];
+    int isamb_levels[10][5];
     int isamb_sizes[10];
     int isamb_blocks[10];
-    int isamb_no[10];
 };
 
 #define SINGLETON_TYPE 8 /* the type to use for singletons that */ 
@@ -126,13 +125,15 @@ static int inv_stat_handle (char *name, const char *info, int pos,
         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_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;
+        if (level > 4)
+            level = 4;
+        stat_info->isamb_levels[cat][level] ++;
     }
 
     while (occur > stat_info->isam_bounds[i] && stat_info->isam_bounds[i])
@@ -189,10 +190,11 @@ void zebra_register_statistics (ZebraHandle zh)
 
     for (i = 0; i<10; i++)
     {
-        stat_info.isamb_levels[i] = 0;
+        int j;
+        for (j = 0; j<5; j++)
+            stat_info.isamb_levels[i][j] = 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,
@@ -255,16 +257,20 @@ void zebra_register_statistics (ZebraHandle zh)
     {
         for (i = 0; i<4; i++)
         {
+            int j;
             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);
+            for (j = 0; j<5; j++)
+                if (stat_info.isamb_levels[i][j])
+                    fprintf (stderr, "Level%d     %d\n", j,
+                             stat_info.isamb_levels[i][j]);
         }
     }
 
@@ -287,7 +293,10 @@ void zebra_register_statistics (ZebraHandle zh)
 /*
  *
  * $Log: invstat.c,v $
- * Revision 1.26  2002-04-29 18:03:46  adam
+ * Revision 1.27  2002-04-30 08:28:37  adam
+ * isamb fixes for pp_read. Statistics
+ *
+ * Revision 1.26  2002/04/29 18:03:46  adam
  * More isamb statistics
  *
  * Revision 1.25  2002/04/26 08:44:47  adam

diff --git a/isamb/isamb.c b/isamb/isamb.c
index 8578f96..1625774 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 1.11 2002-04-29 18:03:46 adam Exp $
+ *  $Id: isamb.c,v 1.12 2002-04-30 08:28:37 adam Exp $
  */
 #include <yaz/xmalloc.h>
 #include <yaz/log.h>
@@ -39,7 +39,7 @@ struct ISAMB_s {
 };
 
 struct ISAMB_block {
-    int pos;
+    ISAMB_P pos;
     int cat;
     int size;
     int leaf;
@@ -52,6 +52,7 @@ struct ISAMB_block {
 
 struct ISAMB_PP_s {
     ISAMB isamb;
+    ISAMB_P pos;
     int level;
     int total_size;
     int no_blocks;
@@ -78,7 +79,7 @@ ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
     isamb->bfs = bfs;
     isamb->method = (ISAMC_M) xmalloc (sizeof(*method));
     memcpy (isamb->method, method, sizeof(*method));
-    isamb->no_cat = 4;
+    isamb->no_cat = 3;
 
     isamb->file = xmalloc (sizeof(*isamb->file) * isamb->no_cat);
     for (i = 0; i<isamb->no_cat; i++)
@@ -567,6 +568,7 @@ ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
     pp->isamb = isamb;
     pp->block = xmalloc (10 * sizeof(*pp->block));
 
+    pp->pos = pos;
     pp->level = 0;
     pp->total_size = 0;
     pp->no_blocks = 0;
@@ -579,7 +581,7 @@ ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
         pp->total_size += p->size;
         pp->no_blocks++;
 
-        if (p->bytes[0]) /* leaf */
+        if (p->leaf)
             break;
 
         decode_ptr (&src, &pos);
@@ -643,7 +645,7 @@ int isamb_pp_read (ISAMB_PP pp, void *buf)
             pp->block[pp->level] = 0;
             (pp->level)--;
             p = pp->block[pp->level];
-            assert (p->bytes[0] == 0);  /* must be int */
+            assert (!p->leaf);  /* must be int */
         }
         src = p->bytes + p->offset;
         
@@ -673,7 +675,7 @@ int isamb_pp_read (ISAMB_PP pp, void *buf)
         }
     }
     assert (p->offset < p->size);
-    assert (p->bytes[0]);
+    assert (p->leaf);
     src = p->bytes + p->offset;
     (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
                                     &dst, &src);

diff --git a/test/gils/zebra.cfg b/test/gils/zebra.cfg
index 33a5872..3241173 100644 (file)
--- a/test/gils/zebra.cfg
+++ b/test/gils/zebra.cfg
@@ -1,5 +1,5 @@
 # Simple Zebra configuration file
-# $Id: zebra.cfg,v 1.14 2002-04-17 08:17:13 adam Exp $
+# $Id: zebra.cfg,v 1.15 2002-04-30 08:28:37 adam Exp $
 #
 # Where the schema files, attribute files, etc are located.
 profilePath: .:../../tab:../../../yaz/tab
@@ -14,4 +14,4 @@ recordtype: grs.sgml
 #storekeys: 1
 #storedata: 1
 #recordId: (bib1,identifier-standard)
-isam: b
+isam: c