X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Finvstat.c;h=8e5770af2ee62d95f8c1a6c98622f865a01c8741;hb=a20e59c5087fb92c419f2330c786367f9ce8ccd2;hp=7310b29b9c7f059cc88418a464e2c738a92ffa1f;hpb=349ddf2c6256dc5d0fbabb693b688883610ae5fa;p=idzebra-moved-to-github.git

diff --git a/index/invstat.c b/index/invstat.c
index 7310b29..8e5770a 100644
--- a/index/invstat.c
+++ b/index/invstat.c
@@ -20,10 +20,10 @@ 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];
+    unsigned long cksum;
 };
 
 #define SINGLETON_TYPE 8 /* the type to use for singletons that */ 
@@ -53,7 +53,8 @@ static int inv_stat_handle (char *name, const char *info, int pos,
         occur = isams_pp_num (pp);
         while (isams_pp_read(pp, &key))
 	{
-	    //printf ("sysno=%d seqno=%d\n", key.sysno, key.seqno);
+            stat_info->cksum = stat_info->cksum * 65509 + 
+                key.sysno + 11 * key.seqno;
             occurx++;
 	}
         assert (occurx == occur);
@@ -66,6 +67,7 @@ static int inv_stat_handle (char *name, const char *info, int pos,
 
         ispt = is_position (stat_info->zh->reg->isam, isam_p);
         occur = is_numkeys (ispt);
+	stat_info->no_isam_entries[is_type(isam_p)] += occur;
         is_pt_free (ispt);
     }
     if (stat_info->zh->reg->isamc)
@@ -78,7 +80,8 @@ static int inv_stat_handle (char *name, const char *info, int pos,
         occur = isc_pp_num (pp);
         while (isc_pp_read(pp, &key))
 	{
-	    //printf ("sysno=%d seqno=%d\n", key.sysno, key.seqno);
+            stat_info->cksum = stat_info->cksum * 65509 + 
+                key.sysno + 11 * key.seqno;
             occurx++;
 	}
         assert (occurx == occur);
@@ -96,6 +99,8 @@ static int inv_stat_handle (char *name, const char *info, int pos,
         occur = isamd_pp_num (pp);
         while (isamd_pp_read(pp, &key))
 	{
+            stat_info->cksum = stat_info->cksum * 65509 + 
+                key.sysno + 11 * key.seqno;
             occurx++;
             if ( pp->is->method->debug >8 )
 	       logf (LOG_LOG,"sysno=%d seqno=%d (%x/%x) oc=%d/%d ofs=%d ",
@@ -126,13 +131,20 @@ 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))
+        {
+            stat_info->cksum = stat_info->cksum * 65509 + 
+                key.sysno + 11 * key.seqno;
             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] ++;
+	stat_info->no_isam_entries[cat] += occur;
     }
 
     while (occur > stat_info->isam_bounds[i] && stat_info->isam_bounds[i])
@@ -148,6 +160,7 @@ void zebra_register_statistics (ZebraHandle zh)
     int count;
     int i, prev;
     int before = 0;
+    int occur;
     int after = 1000000000;
     struct inv_stat_info stat_info;
     char term_dict[2*IT_MAX_WORD+2];
@@ -184,15 +197,18 @@ void zebra_register_statistics (ZebraHandle zh)
     stat_info.isam_bounds[17] = 1000000;
     stat_info.isam_bounds[18] = 0;
 
+    stat_info.cksum = 0;
+
     for (i = 0; i<20; i++)
         stat_info.isam_occurrences[i] = 0;
 
     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,
@@ -255,19 +271,31 @@ 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, "Entries:   %d\n", stat_info.no_isam_entries[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]);
+            fprintf (stderr, "\n");
         }
     }
+    fprintf (stderr, "Checksum       %08lX\n", stat_info.cksum);
 
+    fprintf (stderr, "Distinct words %d\n", stat_info.no_dict_entries);
+    occur = 0;
+    for (i = 0; i<9; i++)
+        occur += stat_info.no_isam_entries[i];
+    fprintf (stderr, "Word pos       %d\n", occur);
     fprintf (stderr, "    Occurrences     Words\n");
     prev = 1;
     for (i = 0; stat_info.isam_bounds[i]; i++)
@@ -287,7 +315,13 @@ void zebra_register_statistics (ZebraHandle zh)
 /*
  *
  * $Log: invstat.c,v $
- * Revision 1.26  2002-04-29 18:03:46  adam
+ * Revision 1.28  2002-04-30 19:31:09  adam
+ * isamb delete; more statistics
+ *
+ * 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