isamb delete; more statistics

diff --git a/index/invstat.c b/index/invstat.c
index ca44bda..8e5770a 100644 (file)
--- a/index/invstat.c
+++ b/index/invstat.c
@@ -23,6 +23,7 @@ struct inv_stat_info {
     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 */ 
@@ -52,7 +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);
@@ -65,6 +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)
@@ -77,7 +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);
@@ -95,6 +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 ",
@@ -127,13 +133,18 @@ static int inv_stat_handle (char *name, const char *info, int pos,
         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_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])
@@ -149,6 +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];
@@ -185,6 +197,8 @@ 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;
 
@@ -265,15 +279,23 @@ void zebra_register_statistics (ZebraHandle zh)
             fprintf (stderr, "Block size %d\n", bsize);
             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++)
@@ -293,7 +315,10 @@ void zebra_register_statistics (ZebraHandle zh)
 /*
  *
  * $Log: invstat.c,v $
- * Revision 1.27  2002-04-30 08:28:37  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

diff --git a/isamb/isamb.c b/isamb/isamb.c
index 1625774..3437fc1 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 1.12 2002-04-30 08:28:37 adam Exp $
+ *  $Id: isamb.c,v 1.13 2002-04-30 19:31:09 adam Exp $
  */
 #include <yaz/xmalloc.h>
 #include <yaz/log.h>
@@ -91,7 +91,7 @@ ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
         isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
         b_size = b_size * 4;
         isamb->file[i].head_dirty = 0;
-        sprintf (fname, "%s-%d", name, i);
+        sprintf (fname, "%s%c", name, i+'A');
         isamb->file[i].bf =
             bf_open (bfs, fname, isamb->file[i].head.block_size, writeflag);
     
@@ -186,13 +186,14 @@ void close_block (ISAMB b, struct ISAMB_block *p)
 }
 
 int insert_sub (ISAMB b, struct ISAMB_block **p,
-                void *new_item,
+                void *new_item, int *mode,
                 ISAMC_I stream,
                 struct ISAMB_block **sp,
                 void *sub_item, int *sub_size,
                 void *max_item);
 
 int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
+                int *mode,
                 ISAMC_I stream, struct ISAMB_block **sp,
                 void *split_item, int *split_size)
 {
@@ -218,7 +219,8 @@ int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
         {
             sub_p1 = open_block (b, pos);
             assert (sub_p1);
-            more = insert_sub (b, &sub_p1, lookahead_item, stream, &sub_p2, 
+            more = insert_sub (b, &sub_p1, lookahead_item, mode,
+                               stream, &sub_p2, 
                                sub_item, &sub_size, src);
             break;
         }
@@ -229,7 +231,7 @@ int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
     {
         sub_p1 = open_block (b, pos);
         assert (sub_p1);
-        more = insert_sub (b, &sub_p1, lookahead_item, stream, &sub_p2, 
+        more = insert_sub (b, &sub_p1, lookahead_item, mode, stream, &sub_p2, 
                            sub_item, &sub_size, 0);
     }
     if (sub_p2)
@@ -297,7 +299,7 @@ int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
 
 
 int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
-                 ISAMC_I stream, struct ISAMB_block **sp2,
+                 int *lookahead_mode, ISAMC_I stream, struct ISAMB_block **sp2,
                  void *sub_item, int *sub_size,
                  void *max_item)
 {
@@ -329,17 +331,23 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
             char *dst_item = 0;
             char *dst_0 = dst;
             char *lookahead_next;
-            int lookahead_mode;
             int d = -1;
             
             if (lookahead_item)
                 d = (*b->method->compare_item)(file_item_buf, lookahead_item);
             
-            if (d > 0)  
+            if (d > 0)
+            {
                 dst_item = lookahead_item;
+                assert (*lookahead_mode);
+            }
             else
                 dst_item = file_item_buf;
-            if (!half1 && dst > cut)   
+            if (!*lookahead_mode && d == 0)
+            {
+                p->dirty = 1;
+            }
+            else if (!half1 && dst > cut)
             {
                 char *dst_item_0 = dst_item;
                 half1 = dst; /* candidate for splitting */
@@ -365,7 +373,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
                     lookahead_next = lookahead_item;
                     if (!(*stream->read_item)(stream->clientData,
                                               &lookahead_next,
-                                              &lookahead_mode))
+                                              lookahead_mode))
                     {
                         lookahead_item = 0;
                         more = 0;
@@ -384,7 +392,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
             {
                 lookahead_next = lookahead_item;
                 if (!(*stream->read_item)(stream->clientData,
-                                          &lookahead_next, &lookahead_mode))
+                                          &lookahead_next, lookahead_mode))
                 {
                     lookahead_item = 0;
                     more = 0;
@@ -406,7 +414,6 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
     maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
     while (lookahead_item)
     {
-        int lookahead_mode;
         char *dst_item = lookahead_item;
         char *dst_0 = dst;
         
@@ -416,7 +423,12 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
            yaz_log (LOG_LOG, "max_item 2");
             break;
         }
-        if (!half1 && dst > cut)   
+        if (!*lookahead_mode)
+        {
+            yaz_log (LOG_WARN, "Inconsistent register (2)");
+            abort();
+        }
+        else if (!half1 && dst > cut)   
         {
             char *dst_item_0 = dst_item;
             half1 = dst; /* candidate for splitting */
@@ -440,7 +452,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
             p->dirty = 1;
         dst_item = lookahead_item;
         if (!(*stream->read_item)(stream->clientData, &dst_item,
-                                  &lookahead_mode))
+                                  lookahead_mode))
         {
             lookahead_item = 0;
             more = 0;
@@ -506,16 +518,18 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
 }
 
 int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
+                int *mode,
                 ISAMC_I stream,
                 struct ISAMB_block **sp,
                 void *sub_item, int *sub_size,
                 void *max_item)
 {
     if (!*p || (*p)->leaf)
-        return insert_leaf (b, p, new_item, stream, sp, sub_item, sub_size, 
-                            max_item);
+        return insert_leaf (b, p, new_item, mode, stream, sp, sub_item, 
+                            sub_size, max_item);
     else
-        return insert_int (b, *p, new_item, stream, sp, sub_item, sub_size);
+        return insert_int (b, *p, new_item, mode, stream, sp, sub_item,
+                           sub_size);
 }
 
 int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
@@ -535,7 +549,7 @@ int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
         
         if (pos)
             p = open_block (b, pos);
-        more = insert_sub (b, &p, item_buf, stream, &sp,
+        more = insert_sub (b, &p, item_buf, &i_mode, stream, &sp,
                             sub_item, &sub_size, 0);
         if (sp)
         {    /* increase level of tree by one */