Added dumpdict command line option to dump the

dictionary before doing the usual stats

diff --git a/index/invstat.c b/index/invstat.c
index dd808da..f5bc6aa 100644 (file)
--- a/index/invstat.c
+++ b/index/invstat.c
@@ -24,11 +24,33 @@ struct inv_stat_info {
     int isamb_sizes[10];
     int isamb_blocks[10];
     unsigned long cksum;
+    int dumpwords;
 };
 
 #define SINGLETON_TYPE 8 /* the type to use for singletons that */ 
                          /* have no block and no block type */
 
+static void print_dict_item (ZebraMaps zm, const char *s, int count )
+{
+    int reg_type = s[1];
+    char keybuf[IT_MAX_WORD+1];
+    char *to = keybuf;
+    const char *from = s + 2;
+
+    while (*from)
+    {
+        const char *res = zebra_maps_output (zm, reg_type, &from);
+        if (!res)
+            *to++ = *from++;
+        else
+            while (*res)
+                *to++ = *res++;
+    }
+    *to = '\0';
+    /* yaz_log (LOG_LOG, "%s", keybuf); */
+    printf("%10d %s\n",count, keybuf);
+}
+
 static int inv_stat_handle (char *name, const char *info, int pos,
                             void *client)
 {
@@ -43,6 +65,7 @@ static int inv_stat_handle (char *name, const char *info, int pos,
     assert (*info == sizeof(ISAMS_P));
     memcpy (&isam_p, info+1, sizeof(ISAMS_P));
 
+
     if (stat_info->zh->reg->isams)
     {
         ISAMS_PP pp;
@@ -150,10 +173,12 @@ static int inv_stat_handle (char *name, const char *info, int pos,
     while (occur > stat_info->isam_bounds[i] && stat_info->isam_bounds[i])
         i++;
     ++(stat_info->isam_occurrences[i]);
+    if (stat_info->dumpwords)
+       print_dict_item(stat_info->zh->reg->zebra_maps, name, occur);
     return 0;
 }
 
-void zebra_register_statistics (ZebraHandle zh)
+void zebra_register_statistics (ZebraHandle zh, int dumpdict)
 {
     int blocks;
     int size;
@@ -169,6 +194,7 @@ void zebra_register_statistics (ZebraHandle zh)
        return;
 
     stat_info.zh = zh;
+    stat_info.dumpwords=dumpdict;
 
     term_dict[0] = 1;
     term_dict[1] = 0;
@@ -315,7 +341,11 @@ void zebra_register_statistics (ZebraHandle zh)
 /*
  *
  * $Log: invstat.c,v $
- * Revision 1.29  2002-06-19 10:29:17  adam
+ * Revision 1.30  2002-07-11 13:03:01  heikki
+ * Added dumpdict command line option to dump the
+ * dictionary before doing the usual stats
+ *
+ * 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

diff --git a/index/main.c b/index/main.c
index edf5212..c32c419 100644 (file)
--- a/index/main.c
+++ b/index/main.c
@@ -2,7 +2,7 @@
  * Copyright (C) 1994-2002, Index Data
  * All rights reserved.
  *
- * $Id: main.c,v 1.91 2002-06-19 13:19:43 adam Exp $
+ * $Id: main.c,v 1.92 2002-07-11 13:03:01 heikki Exp $
  */
 #include <stdio.h>
 #include <string.h>
@@ -151,7 +151,11 @@ int main (int argc, char **argv)
                 }
                 else if (!strcmp (arg, "stat") || !strcmp (arg, "status"))
                 {
-                    zebra_register_statistics (zh);
+                    zebra_register_statistics (zh,0);
+                }
+                else if (!strcmp (arg, "dump") || !strcmp (arg, "dumpdict"))
+                {
+                    zebra_register_statistics (zh,1);
                 }
                 else if (!strcmp (arg, "compact"))
                 {

diff --git a/index/zebraapi.h b/index/zebraapi.h
index 1b822b3..2551093 100644 (file)
--- a/index/zebraapi.h
+++ b/index/zebraapi.h
@@ -3,7 +3,7 @@
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
- * $Id: zebraapi.h,v 1.16 2002-04-26 08:44:47 adam Exp $
+ * $Id: zebraapi.h,v 1.17 2002-07-11 13:03:01 heikki Exp $
  */
 
 #ifndef ZEBRAAPI_H
@@ -153,7 +153,7 @@ YAZ_EXPORT
 void zebra_shadow_enable (ZebraHandle zh, int value);
 
 YAZ_EXPORT
-void zebra_register_statistics (ZebraHandle zh);
+void zebra_register_statistics (ZebraHandle zh, int dumpdict);
 
 YAZ_END_CDECL                                
 #endif