In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT

is collected.

diff --git a/index/Makefile b/index/Makefile
index 63bccac..d0204ec 100644 (file)
--- a/index/Makefile
+++ b/index/Makefile
@@ -1,7 +1,7 @@
 # Copyright (C) 1995-1996, Index Data I/S 
 # All rights reserved.
 # Sebastian Hammer, Adam Dickmeiss
-# $Id: Makefile,v 1.36 1996-05-14 11:34:00 adam Exp $
+# $Id: Makefile,v 1.37 1996-05-14 14:04:33 adam Exp $
 
 SHELL=/bin/sh
 RANLIB=ranlib
@@ -20,7 +20,7 @@ TPROG3=zebrasrv
 DEFS=$(INCLUDE)
 O1 = main.o dir.o dirs.o trav.o extract.o kinput.o kcompare.o \
  symtab.o text.o recctrl.o structrec.o recindex.o regxread.o recstat.o \
- lockutil.o lockidx.o zinfo.o
+ lockutil.o lockidx.o zinfo.o invstat.o
 O2 = kdump.o
 O3 = zserver.o kcompare.o zrpn.o zsets.o text.o recctrl.o structrec.o  \
        attribute.o recindex.o zlogs.o regxread.o lockutil.o locksrv.o zinfo.o

diff --git a/index/index.h b/index/index.h
index 5491d85..694b36d 100644 (file)
--- a/index/index.h
+++ b/index/index.h
@@ -4,7 +4,11 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: index.h,v $
- * Revision 1.38  1996-04-12 07:02:23  adam
+ * Revision 1.39  1996-05-14 14:04:33  adam
+ * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
+ * is collected.
+ *
+ * Revision 1.38  1996/04/12  07:02:23  adam
  * File update of single files.
  *
  * Revision 1.37  1996/03/26 16:01:13  adam
@@ -203,6 +207,7 @@ void key_write (int cmd, struct it_key *k, const char *str);
 int key_compare (const void *p1, const void *p2);
 int key_qsort_compare (const void *p1, const void *p2);
 void key_logdump (int mask, const void *p);
+void inv_prstat (const char *dict_fname, const char *isam_fname);
 void key_input (const char *dict_fname, const char *isam_fname,
                  int nkeys, int cache);
 int merge_sort (char **buf, int from, int to);

diff --git a/index/invstat.c b/index/invstat.c
new file mode 100644 (file)
index 0000000..6cfeab9
--- /dev/null
+++ b/index/invstat.c
@@ -0,0 +1,130 @@
+/*
+ * Copyright (C) 1994-1996, Index Data I/S 
+ * All rights reserved.
+ * Sebastian Hammer, Adam Dickmeiss
+ *
+ * $Log: invstat.c,v $
+ * Revision 1.1  1996-05-14 14:04:34  adam
+ * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
+ * is collected.
+ *
+ */
+#include <stdio.h>
+#include <assert.h>
+#include <string.h>
+#include <ctype.h>
+
+#include "index.h"
+#include "recindex.h"
+
+struct inv_stat_info {
+    ISAM isam;
+    int no_dict_entries;
+    int no_dict_bytes;
+    int isam_bounds[20];
+    int isam_occurrences[20];
+    char tmp[128];
+};
+
+static int inv_stat_handle (char *name, const char *info, int pos,
+                            void *client)
+{
+    int occur;
+    int i = 0;
+    struct inv_stat_info *stat_info = (struct inv_stat_info*) client;
+    ISPT ispt;
+    ISAM_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));
+
+    ispt = is_position (stat_info->isam, isam_p);
+    
+    occur = is_numkeys (ispt);
+
+    is_pt_free (ispt);
+
+    while (occur > stat_info->isam_bounds[i] && stat_info->isam_bounds[i])
+        i++;
+    ++(stat_info->isam_occurrences[i]);
+
+    return 0;
+}
+
+void inv_prstat (const char *dict_fname, const char *isam_fname)
+{
+    Dict dict;
+    ISAM isam;
+    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;
+
+    dict = dict_open (dict_fname, 100, 0);
+    if (!dict)
+    {
+        logf (LOG_FATAL, "dict_open fail of `%s'", dict_fname);
+        exit (1);
+    }
+    isam = is_open (isam_fname, key_compare, 0, sizeof(struct it_key));
+    if (!isam)
+    {
+        logf (LOG_FATAL, "is_open fail of `%s'", isam_fname);
+        exit (1);
+    }
+    records = rec_open (0);
+
+    stat_info.no_dict_entries = 0;
+    stat_info.no_dict_bytes = 0;
+    stat_info.isam = isam;
+    stat_info.isam_bounds[0] = 1;
+    stat_info.isam_bounds[1] = 2;
+    stat_info.isam_bounds[2] = 3;
+    stat_info.isam_bounds[3] = 5;
+    stat_info.isam_bounds[4] = 10;
+    stat_info.isam_bounds[5] = 20;
+    stat_info.isam_bounds[6] = 30;
+    stat_info.isam_bounds[7] = 50;
+    stat_info.isam_bounds[8] = 100;
+    stat_info.isam_bounds[9] = 200;
+    stat_info.isam_bounds[10] = 5000;
+    stat_info.isam_bounds[11] = 10000;
+    stat_info.isam_bounds[12] = 20000;
+    stat_info.isam_bounds[13] = 50000;
+    stat_info.isam_bounds[14] = 100000;
+    stat_info.isam_bounds[15] = 200000;
+    stat_info.isam_bounds[16] = 500000;
+    stat_info.isam_bounds[17] = 1000000;
+    stat_info.isam_bounds[18] = 0;
+
+    for (i = 0; i<20; i++)
+        stat_info.isam_occurrences[i] = 0;
+
+    dict_scan (dict, term_dict, &before, &after, &stat_info, inv_stat_handle);
+
+    rec_close (&records);
+    dict_close (dict);
+    is_close (isam);
+
+    fprintf (stderr, "%d dictionary entries. %d bytes for strings\n",
+             stat_info.no_dict_entries, stat_info.no_dict_bytes);
+    fprintf (stderr, " size   occurrences\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",
+                 prev, here, stat_info.isam_occurrences[i]);
+        prev = here+1;
+    }
+    fprintf (stderr, "%7d-         %7d\n",
+             prev, stat_info.isam_occurrences[i]);
+}

diff --git a/index/main.c b/index/main.c
index bde872e..a3cbd58 100644 (file)
--- a/index/main.c
+++ b/index/main.c
@@ -1,10 +1,14 @@
 /*
- * Copyright (C) 1994-1995, Index Data I/S 
+ * Copyright (C) 1994-1996, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: main.c,v $
- * Revision 1.40  1996-04-26 10:00:23  adam
+ * Revision 1.41  1996-05-14 14:04:34  adam
+ * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
+ * is collected.
+ *
+ * Revision 1.40  1996/04/26  10:00:23  adam
  * Added option -V to zebraidx to display version information.
  * Removed stupid warnings from file update.
  *
@@ -290,6 +294,7 @@ int main (int argc, char **argv)
                         zebraIndexLockMsg ("r");
                     }
                     rec_prstat ();
+                    inv_prstat  (FNAME_WORD_DICT, FNAME_WORD_ISAM);
                 }
                 else
                 {

diff --git a/index/recstat.c b/index/recstat.c
index aa1ff65..c27ce8f 100644 (file)
--- a/index/recstat.c
+++ b/index/recstat.c
@@ -1,10 +1,14 @@
 /*
- * Copyright (C) 1994-1995, Index Data I/S 
+ * Copyright (C) 1994-1996, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: recstat.c,v $
- * Revision 1.1  1995-12-06 12:41:26  adam
+ * Revision 1.2  1996-05-14 14:04:34  adam
+ * In zebraidx, the 'stat' command is improved. Statistics about ISAM/DICT
+ * is collected.
+ *
+ * Revision 1.1  1995/12/06  12:41:26  adam
  * New command 'stat' for the index program.
  * Filenames can be read from stdin by specifying '-'.
  * Bug fix/enhancement of the transformation from terms to regular
@@ -15,7 +19,6 @@
 #include <assert.h>
 #include <string.h>
 #include <ctype.h>
-#include <assert.h>
 #include <fcntl.h>
 #include <unistd.h>