/*
- * Copyright (C) 1994-1995, Index Data I/S
+ * Copyright (C) 1994-2000, Index Data
* All rights reserved.
* Sebastian Hammer, Adam Dickmeiss
*
* $Log: dicttest.c,v $
- * Revision 1.19 1996-02-02 13:43:50 adam
+ * Revision 1.25 2000-12-05 09:59:10 adam
+ * Work on dict_delete_subtree.
+ *
+ * Revision 1.24 2000/09/05 14:04:05 adam
+ * Updates for prefix 'yaz_' for YAZ log functions.
+ *
+ * Revision 1.23 2000/07/07 12:49:20 adam
+ * Optimized resultSetInsert{Rank,Sort}.
+ *
+ * Revision 1.22 1999/02/02 14:50:19 adam
+ * Updated WIN32 code specific sections. Changed header.
+ *
+ * Revision 1.21 1996/10/29 14:00:03 adam
+ * Page size given by DICT_DEFAULT_PAGESIZE in dict.h.
+ *
+ * Revision 1.20 1996/03/20 09:35:16 adam
+ * Function dict_lookup_grep got extra parameter, init_pos, which marks
+ * from which position in pattern approximate pattern matching should occur.
+ *
+ * Revision 1.19 1996/02/02 13:43:50 adam
* The public functions simply use char instead of Dict_char to represent
* search strings. Dict_char is used internally only.
*
#include <ctype.h>
#include <dict.h>
+#include <zebrautl.h>
char *prog;
static Dict dict;
static int look_hits;
-static int grep_handle (char *name, const char *info, void *client)
+static int grep_handler (char *name, const char *info, void *client)
{
look_hits++;
printf ("%s\n", name);
return 0;
}
+static int scan_handler (char *name, const char *info, int pos, void *client)
+{
+ printf ("%s\n", name);
+ return 0;
+}
+
int main (int argc, char **argv)
{
+ Res my_resource = 0;
+ BFiles bfs;
const char *name = NULL;
const char *inputfile = NULL;
- const char *base = NULL;
+ const char *config = NULL;
+ const char *delete_term = NULL;
+ int scan_the_thing = 0;
int do_delete = 0;
int range = -1;
+ int srange = 0;
int rw = 0;
int infosize = 4;
int cache = 10;
if (argc < 2)
{
fprintf (stderr, "usage:\n "
- " %s [-d] [-r n] [-u] [-g pat] [-s n] [-v n] [-i f] [-w]"
- " [-c n] base file\n\n",
+ " %s [-d] [-D t] [-S] [-r n] [-p n] [-u] [-g pat] [-s n] "
+ "[-v n] [-i f] [-w] [-c n] config file\n\n",
prog);
fprintf (stderr, " -d delete instead of insert\n");
+ fprintf (stderr, " -D t delete subtree instead of insert\n");
fprintf (stderr, " -r n set regular match range\n");
+ fprintf (stderr, " -p n set regular match start range\n");
fprintf (stderr, " -u report if keys change during insert\n");
fprintf (stderr, " -g p try pattern n (see -r)\n");
fprintf (stderr, " -s n set info size to n (instead of 4)\n");
fprintf (stderr, " -i f read file with words\n");
fprintf (stderr, " -w insert/delete instead of lookup\n");
fprintf (stderr, " -c n cache size (number of pages)\n");
+ fprintf (stderr, " -S scan the dictionary\n");
exit (1);
}
- while ((ret = options ("dr:ug:s:v:i:wc:", argv, argc, &arg)) != -2)
+ while ((ret = options ("D:Sdr:p:ug:s:v:i:wc:", argv, argc, &arg)) != -2)
{
if (ret == 0)
{
- if (!base)
- base = arg;
+ if (!config)
+ config = arg;
else if (!name)
name = arg;
else
exit (1);
}
}
+ else if (ret == 'D')
+ {
+ delete_term = arg;
+ }
else if (ret == 'd')
do_delete = 1;
else if (ret == 'g')
{
range = atoi (arg);
}
+ else if (ret == 'p')
+ {
+ srange = atoi (arg);
+ }
else if (ret == 'u')
{
unique = 1;
rw = 1;
else if (ret == 'i')
inputfile = arg;
+ else if (ret == 'S')
+ scan_the_thing = 1;
else if (ret == 's')
{
infosize = atoi(arg);
}
else if (ret == 'v')
{
- log_init (log_mask_str(arg), prog, NULL);
+ yaz_log_init (yaz_log_mask_str(arg), prog, NULL);
}
else
{
exit (1);
}
}
- if (!base || !name)
+ if (!config || !name)
+ {
+ logf (LOG_FATAL, "no config and/or dictionary specified");
+ exit (1);
+ }
+ my_resource = res_open (config);
+ if (!my_resource)
{
- logf (LOG_FATAL, "no base and/or dictionary specified");
+ logf (LOG_FATAL, "cannot open resource `%s'", config);
exit (1);
}
- common_resource = res_open (base);
- if (!common_resource)
+ bfs = bfs_create (res_get(my_resource, "register"));
+ if (!bfs)
{
- logf (LOG_FATAL, "cannot open resource `%s'", base);
+ logf (LOG_FATAL, "bfs_create fail");
exit (1);
}
- dict = dict_open (name, cache, rw);
+ dict = dict_open (bfs, name, cache, rw, 0);
if (!dict)
{
logf (LOG_FATAL, "dict_open fail of `%s'", name);
ipf_ptr[i++] = '\0';
if (rw)
{
- if (do_delete)
+ if (do_delete)
switch (dict_delete (dict, ipf_ptr))
{
case 0:
{
look_hits = 0;
dict_lookup_grep (dict, ipf_ptr, range, NULL,
- &max_pos, grep_handle);
+ &max_pos, srange, grep_handler);
if (look_hits)
no_of_hits++;
else
no_of_misses++;
}
++no_of_iterations;
+ if ((no_of_iterations % 10000) == 0)
+ {
+ printf ("."); fflush(stdout);
+ }
ipf_ptr += (i-1);
}
}
}
fclose (ipf);
}
+ if (rw && delete_term)
+ {
+ logf (LOG_LOG, "dict_delete_subtree %s", delete_term);
+ dict_delete_subtree (dict, delete_term, 0, 0);
+ }
if (grep_pattern)
{
if (range < 0)
range = 0;
logf (LOG_LOG, "Grepping '%s'", grep_pattern);
dict_lookup_grep (dict, grep_pattern, range, NULL, &max_pos,
- grep_handle);
+ srange, grep_handler);
}
if (rw)
{
logf (LOG_LOG, "No of hits.... %d", no_of_hits);
logf (LOG_LOG, "No of misses.. %d", no_of_misses);
}
+ if (scan_the_thing)
+ {
+ char term_dict[1024];
+
+ int before = 1000000;
+ int after = 1000000;
+ logf (LOG_LOG, "dict_scan");
+ term_dict[0] = 1;
+ term_dict[1] = 0;
+ dict_scan (dict, term_dict, &before, &after, 0, scan_handler);
+ }
dict_close (dict);
- res_close (common_resource);
+ bfs_destroy (bfs);
+ res_close (my_resource);
return 0;
}
projects / idzebra-moved-to-github.git / blobdiff