X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Finvstat.c;h=dd808dae03ed68aded73721c32e4bdbcdfb959c7;hb=cc9f94a61cbd9dcc0df0cf7d0c7c41d2cec88189;hp=f4b548c59f451b65a66548313c1c99db28ee7e09;hpb=3e72747f4b28856e5f684dbe8984f8fb28680e62;p=idzebra-moved-to-github.git

diff --git a/index/invstat.c b/index/invstat.c
index f4b548c..dd808da 100644
--- 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][5];
+    int isamb_sizes[10];
+    int isamb_blocks[10];
+    unsigned long cksum;
 };
 
 #define SINGLETON_TYPE 8 /* the type to use for singletons that */ 
@@ -49,7 +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);
@@ -62,6 +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)
@@ -74,7 +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);
@@ -92,6 +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 ",
@@ -116,11 +125,26 @@ static int inv_stat_handle (char *name, const char *info, int pos,
     {
         ISAMB_PP pp;
         struct it_key key;
+        int cat = isam_p & 3;
+        int level;
+        int size;
+        int blocks;
+        
+        pp = isamb_pp_open_x(stat_info->zh->reg->isamb, isam_p, &level);
 
-        pp = isamb_pp_open(stat_info->zh->reg->isamb, isam_p);
         while (isamb_pp_read(pp, &key))
+        {
+            stat_info->cksum = stat_info->cksum * 65509 + 
+                key.sysno + 11 * key.seqno;
             occur++;
-        isamb_pp_close (pp);
+        }
+        isamb_pp_close_x (pp, &size, &blocks);
+        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])
@@ -136,6 +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];
@@ -172,34 +197,45 @@ 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++)
+    {
+        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;
+    }
+
     dict_scan (zh->reg->dict, term_dict, &before, &after, &stat_info,
                inv_stat_handle);
 
     if (zh->reg->isamc)
     {
-	fprintf (stderr, "   Blocks    Occur  Size KB   Bytes/Entry\n");
+	fprintf (stdout, "   Blocks    Occur  Size KB   Bytes/Entry\n");
 	for (i = 0; isc_block_used (zh->reg->isamc, i) >= 0; i++)
 	{
-	    fprintf (stderr, " %8d %8d", isc_block_used (zh->reg->isamc, i),
+	    fprintf (stdout, " %8d %8d", isc_block_used (zh->reg->isamc, i),
 		     stat_info.no_isam_entries[i]);
 
 	    if (stat_info.no_isam_entries[i])
-		fprintf (stderr, " %8d   %f",
+		fprintf (stdout, " %8d   %f",
 			 (int) ((1023.0 + (double)
                                  isc_block_used(zh->reg->isamc, i) *
 				 isc_block_size(zh->reg->isamc,i))/1024),
 			 ((double) isc_block_used(zh->reg->isamc, i) *
 			  isc_block_size(zh->reg->isamc,i))/
 			 stat_info.no_isam_entries[i]);
-	    fprintf (stderr, "\n");
+	    fprintf (stdout, "\n");
 	}
     }
     if (zh->reg->isamd)
     {
-	fprintf (stderr, "   Blocks   Occur      KB Bytes/Entry\n");
+	fprintf (stdout, "   Blocks   Occur      KB Bytes/Entry\n");
 	if (zh->reg->isamd->method->debug >0) 
             logf(LOG_LOG,"   Blocks   Occur      KB Bytes/Entry");
 	for (i = 0; i<=SINGLETON_TYPE; i++)
@@ -211,7 +247,7 @@ void zebra_register_statistics (ZebraHandle zh)
 	        blocks=size=0;
 	    if (stat_info.no_isam_entries[i]) 
 	    {
-		fprintf (stderr, "%c %7d %7d %7d %5.2f\n",
+		fprintf (stdout, "%c %7d %7d %7d %5.2f\n",
     		         (i==SINGLETON_TYPE)?('z'):('A'+i),
     		         blocks,
     		         count,
@@ -228,18 +264,48 @@ void zebra_register_statistics (ZebraHandle zh)
 	} /* for */
     } /* isamd */
     if ( (zh->reg->isamd) && (zh->reg->isamd->method->debug>0))
-        fprintf (stderr, "\n%d words using %d bytes\n",
+        fprintf (stdout, "\n%d words using %d bytes\n",
              stat_info.no_dict_entries, stat_info.no_dict_bytes);
-    fprintf (stderr, "    Occurrences     Words\n");
+
+    if (zh->reg->isamb)
+    {
+        for (i = 0; i<4; i++)
+        {
+            int j;
+            int bsize = isamb_block_info(zh->reg->isamb, i);
+            if (bsize < 0)
+                break;
+            fprintf (stdout, "Category   %d\n", i);
+            fprintf (stdout, "Block size %d\n", bsize);
+            fprintf (stdout, "Blocks:    %d\n", stat_info.isamb_blocks[i]);
+            fprintf (stdout, "Size:      %d\n", stat_info.isamb_sizes[i]);
+            fprintf (stdout, "Entries:   %d\n", stat_info.no_isam_entries[i]);
+            fprintf (stdout, "Total      %d\n", stat_info.isamb_blocks[i]*
+                     bsize);
+            for (j = 0; j<5; j++)
+                if (stat_info.isamb_levels[i][j])
+                    fprintf (stdout, "Level%d     %d\n", j,
+                             stat_info.isamb_levels[i][j]);
+            fprintf (stdout, "\n");
+        }
+    }
+    fprintf (stdout, "Checksum       %08lX\n", stat_info.cksum);
+
+    fprintf (stdout, "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 (stdout, "Word pos       %d\n", occur);
+    fprintf (stdout, "    Occurrences     Words\n");
     prev = 1;
     for (i = 0; stat_info.isam_bounds[i]; i++)
     {
         int here = stat_info.isam_bounds[i];
-        fprintf (stderr, "%7d-%-7d %7d\n",
+        fprintf (stdout, "%7d-%-7d %7d\n",
                  prev, here, stat_info.isam_occurrences[i]);
         prev = here+1;
     }
-    fprintf (stderr, "%7d-        %7d\n",
+    fprintf (stdout, "%7d-        %7d\n",
              prev, stat_info.isam_occurrences[i]);
     xmalloc_trav("unfreed"); /*! while hunting memory leaks */    
     zebra_end_read (zh);
@@ -249,7 +315,19 @@ void zebra_register_statistics (ZebraHandle zh)
 /*
  *
  * $Log: invstat.c,v $
- * Revision 1.25  2002-04-26 08:44:47  adam
+ * Revision 1.29  2002-06-19 10:29:17  adam
+ * align block sizes for isam sys. Better plot for test
+ *
+ * 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
  * Index statistics working again
  *
  * Revision 1.24  2002/04/05 08:46:26  adam