mproved statistics on isam-d

diff --git a/index/invstat.c b/index/invstat.c
index 39cf04a..7f7cca3 100644 (file)
--- a/index/invstat.c
+++ b/index/invstat.c
@@ -20,7 +20,7 @@ struct inv_stat_info {
     ISAMS isams;
     ISAMH isamh;
     ISAMD isamd;
-    int no_isam_entries[8];
+    int no_isam_entries[9];
     int no_dict_entries;
     int no_dict_bytes;
     int isam_bounds[20];
@@ -28,6 +28,9 @@ struct inv_stat_info {
     char tmp[128];
 };
 
+#define SINGLETON_TYPE 8 /* the type to use for singletons that */ 
+                         /* have no block and no block type */
+
 static int inv_stat_handle (char *name, const char *info, int pos,
                             void *client)
 {
@@ -77,8 +80,6 @@ static int inv_stat_handle (char *name, const char *info, int pos,
         pp = isamh_pp_open (stat_info->isamh, isam_p);
         
         occur = isamh_pp_num (pp);
-         //  printf ("  opening item %d=%d:%d \n",
-         //    isam_p, isamh_type(isam_p),isamh_block(isam_p));
         while (isamh_pp_read(pp, &key))
        {
             occurx++;
@@ -119,7 +120,10 @@ static int inv_stat_handle (char *name, const char *info, int pos,
         if (occurx != occur) 
           logf(LOG_LOG,"Count error!!! read %d, counted %d", occur, occurx);
         assert (occurx == occur);
-       stat_info->no_isam_entries[isamd_type(isam_p)] += occur;
+        if ( is_singleton(isam_p) )
+           stat_info->no_isam_entries[SINGLETON_TYPE] += occur;
+       else
+           stat_info->no_isam_entries[isamd_type(isam_p)] += occur;
         isamd_pp_close (pp);
     }
     if (stat_info->isams)
@@ -160,7 +164,10 @@ void inv_prstat (BFiles bfs)
     int after = 1000000000;
     struct inv_stat_info stat_info;
     char term_dict[2*IT_MAX_WORD+2];
-
+    int blocks;
+    int size;
+    int count;
+        
     term_dict[0] = 1;
     term_dict[1] = 0;
 
@@ -224,7 +231,7 @@ void inv_prstat (BFiles bfs)
     }
     records = rec_open (bfs, 0, 0);
 
-    for (i = 0; i<8; i++)
+    for (i = 0; i<=SINGLETON_TYPE; i++)
        stat_info.no_isam_entries[i] = 0;
     stat_info.no_dict_entries = 0;
     stat_info.no_dict_bytes = 0;
@@ -276,21 +283,67 @@ void inv_prstat (BFiles bfs)
            fprintf (stderr, "\n");
        }
     }
-       
+    if (isamd)
+    {
+       fprintf (stderr, "   Blocks   Occur      KB Bytes/Entry\n");
+       if (isamd->method->debug >0) 
+            logf(LOG_LOG,"   Blocks   Occur      KB Bytes/Entry");
+       for (i = 0; i<=SINGLETON_TYPE; i++)
+       {
+           blocks= isamd_block_used(isamd,i);
+           size= isamd_block_size(isamd,i);
+           count=stat_info.no_isam_entries[i];
+           if (i==SINGLETON_TYPE) 
+               blocks=size=0;
+           if (stat_info.no_isam_entries[i]) 
+           {
+               fprintf (stderr, "%c %7d %7d %7d %5.2f\n",
+                        (i==SINGLETON_TYPE)?('z'):('A'+i),
+                        blocks,
+                        count,
+                        (int) ((1023.0 + (double) blocks * size)/1024),
+                        ((double) blocks * size)/count);
+               if (isamd->method->debug >0) 
+                   logf(LOG_LOG, "%c %7d %7d %7d %5.2f",
+                        (i==SINGLETON_TYPE)?('z'):('A'+i),
+                        blocks,
+                        count,
+                        (int) ((1023.0 + (double) blocks * size)/1024),
+                        ((double) blocks * size)/count);
+           } /* entries */
+       } /* for */
+    } /* isamd */
+    if ( (isamd) && (isamd->method->debug>0))
     fprintf (stderr, "\n%d words using %d bytes\n",
              stat_info.no_dict_entries, stat_info.no_dict_bytes);
     fprintf (stderr, "    Occurrences     Words\n");
+    if ( (isamd) && (isamd->method->debug>0) )
+    {
+       logf(LOG_LOG, "%d words using %d bytes",
+             stat_info.no_dict_entries, stat_info.no_dict_bytes);
+        logf(LOG_LOG, "    Occurrences     Words");
+    }
     prev = 1;
     for (i = 0; stat_info.isam_bounds[i]; i++)
     {
         int here = stat_info.isam_bounds[i];
         fprintf (stderr, "%7d-%-7d %7d\n",
                  prev, here, stat_info.isam_occurrences[i]);
+        if ( (isamd) && (isamd->method->debug>0) &&
+             stat_info.isam_occurrences[i] )
+        {
+           logf(LOG_LOG,"%7d-%-7d %7d",
+                 prev, here, stat_info.isam_occurrences[i]);
+        }
         prev = here+1;
     }
     fprintf (stderr, "%7d-        %7d\n",
              prev, stat_info.isam_occurrences[i]);
-
+    if ( (isamd) && (isamd->method->debug>0) &&
+         ( stat_info.isam_occurrences[i] ||
+           stat_info.isam_occurrences[i+1])  )
+        logf(LOG_LOG,"%7d-        %7d",
+        prev, stat_info.isam_occurrences[i]);
     rec_close (&records);
     dict_close (dict);
 
@@ -312,7 +365,10 @@ void inv_prstat (BFiles bfs)
 /*
  *
  * $Log: invstat.c,v $
- * Revision 1.17  1999-08-20 08:28:37  heikki
+ * Revision 1.18  1999-10-06 11:46:36  heikki
+ * mproved statistics on isam-d
+ *
+ * Revision 1.17  1999/08/20 08:28:37  heikki
  * Log levels
  *
  * Revision 1.16  1999/08/18 08:38:22  heikki

diff --git a/isamc/isamd.c b/isamc/isamd.c
index 191a51c..4b16ca6 100644 (file)
--- a/isamc/isamd.c
+++ b/isamc/isamd.c
@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 1995-1998, Index Data.
  * See the file LICENSE for details.
- * $Id: isamd.c,v 1.16 1999-10-05 09:57:40 heikki Exp $ 
+ * $Id: isamd.c,v 1.17 1999-10-06 11:46:36 heikki Exp $ 
  *
  * Isamd - isam with diffs 
  * Programmed by: Heikki Levanto
@@ -179,6 +179,8 @@ ISAMD isamd_open (BFiles bfs, const char *name, int writeflag, ISAMD_M method)
 
 int isamd_block_used (ISAMD is, int type)
 {
+    if ( type==-1) /* singleton */
+      return 0; 
     if (type < 0 || type >= is->no_files)
        return -1;
     return is->files[type].head.lastblock-1;
@@ -187,6 +189,8 @@ int isamd_block_used (ISAMD is, int type)
 int isamd_block_size (ISAMD is, int type)
 {
     ISAMD_filecat filecat = is->method->filecat;
+    if ( type==-1) /* singleton */
+      return 0; /* no bytes used */ 
     if (type < 0 || type >= is->no_files)
        return -1;
     return filecat[type].bsize;
@@ -246,14 +250,14 @@ int isamd_close (ISAMD is)
                   is->files[i].no_op_main,
                   is->files[i].no_op_diffonly);
         }
-        logf(LOG_LOG,"single   %8d", is->no_op_single);
-        logf(LOG_LOG,"new      %8d", is->no_op_new);
+        logf(LOG_LOG,"open single  %8d", is->no_op_single);
+        logf(LOG_LOG,"open new     %8d", is->no_op_new);
 
         logf(LOG_LOG, "new build   %8d", is->no_fbuilds);
         logf(LOG_LOG, "append      %8d", is->no_appds);
         logf(LOG_LOG, "  merges    %8d", is->no_merges);
         logf(LOG_LOG, "  singles   %8d", is->no_singles);
-        logf(LOG_LOG, "  non       %8d", is->no_non);
+        logf(LOG_LOG, "  no-ops    %8d", is->no_non);
 
         logf(LOG_LOG, "read blocks %8d", is->no_read);
         logf(LOG_LOG, "read keys:  %8d %8.1f k/bl", 
@@ -845,7 +849,10 @@ void isamd_pp_dump (ISAMD is, ISAMD_P ipos)
 
 /*
  * $Log: isamd.c,v $
- * Revision 1.16  1999-10-05 09:57:40  heikki
+ * Revision 1.17  1999-10-06 11:46:36  heikki
+ * mproved statistics on isam-d
+ *
+ * Revision 1.16  1999/10/05 09:57:40  heikki
  * Tuning the isam-d (and fixed a small "detail")
  *
  * Revision 1.15  1999/09/27 14:36:36  heikki