X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Finvstat.c;h=ce82cd69447a5dd9735a436a4702327f31b9f7b8;hb=df117f4652b6999cf689cf5e2e65f8d8541ab4b0;hp=39cf04ab36eedb3cac5050aeb06ba09aa9a1c96c;hpb=7c82553e8fbccbb18a70c6f5e39db75069bfc9d3;p=idzebra-moved-to-github.git

diff --git a/index/invstat.c b/index/invstat.c
index 39cf04a..ce82cd6 100644
--- a/index/invstat.c
+++ b/index/invstat.c
@@ -11,16 +11,18 @@
 
 #include "index.h"
 #include "recindex.h"
-#include "../isamc/isamh-p.h"
+#if ZMBOL
 #include "../isamc/isamd-p.h"
+#endif
 
 struct inv_stat_info {
+    ISAMS isams;
+#if ZMBOL
     ISAM isam;
     ISAMC isamc;
-    ISAMS isams;
-    ISAMH isamh;
     ISAMD isamd;
-    int no_isam_entries[8];
+#endif
+    int no_isam_entries[9];
     int no_dict_entries;
     int no_dict_bytes;
     int isam_bounds[20];
@@ -28,21 +30,41 @@ 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)
 {
     int occur = 0;
     int i = 0;
     struct inv_stat_info *stat_info = (struct inv_stat_info*) client;
-    ISAM_P isam_p;
+    ISAMS_P isam_p;
 
     stat_info->no_dict_entries++;
     stat_info->no_dict_bytes += strlen(name);
 
-    assert (*info == sizeof(ISAM_P));
-    memcpy (&isam_p, info+1, sizeof(ISAM_P));
+    assert (*info == sizeof(ISAMS_P));
+    memcpy (&isam_p, info+1, sizeof(ISAMS_P));
+
+    if (stat_info->isams)
+    {
+        ISAMS_PP pp;
+        int occurx = 0;
+	struct it_key key;
 
-    //printf ("---\n");
+        pp = isams_pp_open (stat_info->isams, isam_p);
+        occur = isams_pp_num (pp);
+        while (isams_pp_read(pp, &key))
+	{
+	    //printf ("sysno=%d seqno=%d\n", key.sysno, key.seqno);
+            occurx++;
+	}
+        assert (occurx == occur);
+	stat_info->no_isam_entries[0] += occur;
+        isams_pp_close (pp);
+    }
+#if ZMBOL
     if (stat_info->isam)
     {
         ISPT ispt;
@@ -68,32 +90,6 @@ static int inv_stat_handle (char *name, const char *info, int pos,
 	stat_info->no_isam_entries[isc_type(isam_p)] += occur;
         isc_pp_close (pp);
     }
-    if (stat_info->isamh)
-    {
-        ISAMH_PP pp;
-        int occurx = 0;
-	struct it_key key;
-
-        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++;
-	    //logf (LOG_LOG,"sysno=%d seqno=%d (%x/%x) oc=%d/%d ofs=%d ",
-	    //       key.sysno, key.seqno,
-	    //       key.sysno, key.seqno,
-	    //       occur,occurx, pp->offset);
-	}
-        if (occurx != occur) {
-          logf(LOG_LOG,"Count error!!! read %d, counted %d", occur, occurx);
-          //isamh_pp_dump(stat_info->isamh, isam_p);
-          }
-	stat_info->no_isam_entries[isamh_type(isam_p)] += occur;
-        isamh_pp_close (pp);
-    }
     if (stat_info->isamd)
     {
         ISAMD_PP pp;
@@ -119,27 +115,13 @@ 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)
-    {
-        ISAMS_PP pp;
-        int occurx = 0;
-	struct it_key key;
-
-        pp = isams_pp_open (stat_info->isams, isam_p);
-        occur = isams_pp_num (pp);
-        while (isams_pp_read(pp, &key))
-	{
-	    //printf ("sysno=%d seqno=%d\n", key.sysno, key.seqno);
-            occurx++;
-	}
-        assert (occurx == occur);
-	stat_info->no_isam_entries[isc_type(isam_p)] += occur;
-        isams_pp_close (pp);
-    }
-
+#endif
     while (occur > stat_info->isam_bounds[i] && stat_info->isam_bounds[i])
         i++;
     ++(stat_info->isam_occurrences[i]);
@@ -149,18 +131,22 @@ static int inv_stat_handle (char *name, const char *info, int pos,
 void inv_prstat (BFiles bfs)
 {
     Dict dict;
+    ISAMS isams = NULL;
+#if ZMBOL
     ISAM  isam  = NULL;
     ISAMC isamc = NULL;
-    ISAMS isams = NULL;
-    ISAMH isamh = NULL;
     ISAMD isamd = NULL;
+    int blocks;
+    int size;
+    int count;
+#endif
     Records records;
     int i, prev;
     int before = 0;
     int after = 1000000000;
     struct inv_stat_info stat_info;
     char term_dict[2*IT_MAX_WORD+2];
-
+        
     term_dict[0] = 1;
     term_dict[1] = 0;
 
@@ -170,17 +156,7 @@ void inv_prstat (BFiles bfs)
         logf (LOG_FATAL, "dict_open fail");
         exit (1);
     }
-    if (res_get_match (common_resource, "isam", "i", NULL))
-    {
-        isam = is_open (bfs, FNAME_ISAM, key_compare, 0,
-			sizeof(struct it_key), common_resource);
-        if (!isam)
-        {
-            logf (LOG_FATAL, "is_open fail");
-            exit (1);
-        }
-    }
-    else if (res_get_match (common_resource, "isam", "s", NULL))
+    if (res_get_match (common_resource, "isam", "s", ISAM_DEFAULT))
     {
 	struct ISAMS_M_s isams_m;
         isams = isams_open (bfs, FNAME_ISAMS, 0,
@@ -191,16 +167,18 @@ void inv_prstat (BFiles bfs)
             exit (1);
         }
     }
-    else if (res_get_match (common_resource, "isam", "h", NULL))
+#if ZMBOL
+    else if (res_get_match (common_resource, "isam", "i", ISAM_DEFAULT))
     {
-        isamh = isamh_open (bfs, FNAME_ISAMH, 0, key_isamh_m(common_resource));
-        if (!isamh)
+        isam = is_open (bfs, FNAME_ISAM, key_compare, 0,
+			sizeof(struct it_key), common_resource);
+        if (!isam)
         {
-            logf (LOG_FATAL, "isamh_open fail");
+            logf (LOG_FATAL, "is_open fail");
             exit (1);
         }
     }
-    else if (res_get_match (common_resource, "isam", "d", NULL))
+    else if (res_get_match (common_resource, "isam", "d", ISAM_DEFAULT))
     {
 	struct ISAMD_M_s isamd_m;
         isamd = isamd_open (bfs, FNAME_ISAMD, 0, 
@@ -211,7 +189,7 @@ void inv_prstat (BFiles bfs)
             exit (1);
         }
     }
-    else
+    else if (res_get_match (common_resource, "isam", "c", ISAM_DEFAULT))
     {
 	struct ISAMC_M_s isamc_m;
         isamc = isc_open (bfs, FNAME_ISAMC, 0,
@@ -222,17 +200,19 @@ void inv_prstat (BFiles bfs)
             exit (1);
         }
     }
+#endif
     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;
+    stat_info.isams = isams;
+#if ZMBOL
     stat_info.isam = isam;
     stat_info.isamc = isamc;
-    stat_info.isams = isams;
-    stat_info.isamh = isamh;
     stat_info.isamd = isamd;
+#endif
     stat_info.isam_bounds[0] = 1;
     stat_info.isam_bounds[1] = 2;
     stat_info.isam_bounds[2] = 3;
@@ -258,6 +238,7 @@ void inv_prstat (BFiles bfs)
 
     dict_scan (dict, term_dict, &before, &after, &stat_info, inv_stat_handle);
 
+#if ZMBOL
     if (isamc)
     {
 	fprintf (stderr, "   Blocks    Occur  Size KB   Bytes/Entry\n");
@@ -276,9 +257,40 @@ void inv_prstat (BFiles bfs)
 	    fprintf (stderr, "\n");
 	}
     }
-	
-    fprintf (stderr, "\n%d words using %d bytes\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);
+#endif
     fprintf (stderr, "    Occurrences     Words\n");
     prev = 1;
     for (i = 0; stat_info.isam_bounds[i]; i++)
@@ -290,20 +302,19 @@ void inv_prstat (BFiles bfs)
     }
     fprintf (stderr, "%7d-        %7d\n",
              prev, stat_info.isam_occurrences[i]);
-
     rec_close (&records);
     dict_close (dict);
 
+    if (isams)
+        isams_close (isams);
+#if ZMBOL
     if (isam)
         is_close (isam);
     if (isamc)
         isc_close (isamc);
-    if (isams)
-        isams_close (isams);
-    if (isamh)
-        isamh_close (isamh);
     if (isamd)
         isamd_close (isamd);
+#endif
 
     xmalloc_trav("unfreed"); /*! while hunting memory leaks */    
 }
@@ -312,7 +323,19 @@ void inv_prstat (BFiles bfs)
 /*
  *
  * $Log: invstat.c,v $
- * Revision 1.17  1999-08-20 08:28:37  heikki
+ * Revision 1.21  2000-07-13 10:14:20  heikki
+ * Removed compiler warnings when making zebra
+ *
+ * Revision 1.20  1999/12/01 13:30:30  adam
+ * Updated configure for Zmbol/Zebra dependent settings.
+ *
+ * Revision 1.19  1999/11/30 13:48:03  adam
+ * Improved installation. Updated for inclusion of YAZ header files.
+ *
+ * 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