2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.23 2000-07-07 12:49:20 adam
8 * Optimized resultSetInsert{Rank,Sort}.
10 * Revision 1.22 1999/02/02 14:50:19 adam
11 * Updated WIN32 code specific sections. Changed header.
13 * Revision 1.21 1996/10/29 14:00:03 adam
14 * Page size given by DICT_DEFAULT_PAGESIZE in dict.h.
16 * Revision 1.20 1996/03/20 09:35:16 adam
17 * Function dict_lookup_grep got extra parameter, init_pos, which marks
18 * from which position in pattern approximate pattern matching should occur.
20 * Revision 1.19 1996/02/02 13:43:50 adam
21 * The public functions simply use char instead of Dict_char to represent
22 * search strings. Dict_char is used internally only.
24 * Revision 1.18 1996/02/01 20:39:52 adam
25 * Bug fix: insert didn't work on 8-bit characters due to unsigned char
26 * compares in dict_strcmp (strcmp) and signed Dict_char. Dict_char is
29 * Revision 1.17 1995/12/06 17:48:30 adam
30 * Bug fix: delete didn't work.
32 * Revision 1.16 1995/10/09 16:18:31 adam
33 * Function dict_lookup_grep got extra client data parameter.
35 * Revision 1.15 1995/09/04 12:33:31 adam
36 * Various cleanup. YAZ util used instead.
38 * Revision 1.14 1994/10/04 17:46:55 adam
39 * Function options now returns arg with error option.
41 * Revision 1.13 1994/10/04 12:08:05 adam
42 * Some bug fixes and some optimizations.
44 * Revision 1.12 1994/10/03 17:23:03 adam
45 * First version of dictionary lookup with regular expressions and errors.
47 * Revision 1.11 1994/09/28 13:07:09 adam
48 * Use log_mask_str now.
50 * Revision 1.10 1994/09/26 10:17:24 adam
53 * Revision 1.9 1994/09/22 14:43:56 adam
54 * First functional version of lookup with error correction. A 'range'
55 * specified the maximum number of insertions+deletions+substitutions.
57 * Revision 1.8 1994/09/22 10:43:44 adam
58 * Two versions of depend. Type 1 is the tail-type compatible with
59 * all make programs. Type 2 is the GNU make with include facility.
60 * Type 2 is default. depend rule chooses current rule.
62 * Revision 1.7 1994/09/19 16:34:26 adam
63 * Depend rule change. Minor changes in dicttest.c
65 * Revision 1.6 1994/09/16 15:39:12 adam
66 * Initial code of lookup - not tested yet.
68 * Revision 1.5 1994/09/06 13:05:14 adam
69 * Further development of insertion. Some special cases are
70 * not properly handled yet! assert(0) are put here. The
71 * binary search in each page definitely reduce usr CPU.
73 * Revision 1.4 1994/09/01 17:49:37 adam
74 * Removed stupid line. Work on insertion in dictionary. Not finished yet.
76 * Revision 1.3 1994/09/01 17:44:06 adam
77 * depend include change.
79 * Revision 1.2 1994/08/18 12:40:54 adam
80 * Some development of dictionary. Not finished at all!
82 * Revision 1.1 1994/08/16 16:26:47 adam
100 static int grep_handle (char *name, const char *info, void *client)
103 printf ("%s\n", name);
107 int main (int argc, char **argv)
111 const char *name = NULL;
112 const char *inputfile = NULL;
113 const char *config = NULL;
122 char *grep_pattern = NULL;
124 int no_of_iterations = 0;
125 int no_of_new = 0, no_of_same = 0, no_of_change = 0;
126 int no_of_hits = 0, no_of_misses = 0, no_not_found = 0, no_of_deleted = 0;
132 fprintf (stderr, "usage:\n "
133 " %s [-d] [-r n] [-p n] [-u] [-g pat] [-s n] [-v n] [-i f]"
134 " [-w] [-c n] config file\n\n",
136 fprintf (stderr, " -d delete instead of insert\n");
137 fprintf (stderr, " -r n set regular match range\n");
138 fprintf (stderr, " -p n set regular match start range\n");
139 fprintf (stderr, " -u report if keys change during insert\n");
140 fprintf (stderr, " -g p try pattern n (see -r)\n");
141 fprintf (stderr, " -s n set info size to n (instead of 4)\n");
142 fprintf (stderr, " -v n set logging level\n");
143 fprintf (stderr, " -i f read file with words\n");
144 fprintf (stderr, " -w insert/delete instead of lookup\n");
145 fprintf (stderr, " -c n cache size (number of pages)\n");
148 while ((ret = options ("dr:p:ug:s:v:i:wc:", argv, argc, &arg)) != -2)
158 logf (LOG_FATAL, "too many files specified\n");
192 infosize = atoi(arg);
196 log_init (log_mask_str(arg), prog, NULL);
200 logf (LOG_FATAL, "Unknown option '-%s'", arg);
204 if (!config || !name)
206 logf (LOG_FATAL, "no config and/or dictionary specified");
209 my_resource = res_open (config);
212 logf (LOG_FATAL, "cannot open resource `%s'", config);
215 bfs = bfs_create (res_get(my_resource, "register"));
218 logf (LOG_FATAL, "bfs_create fail");
221 dict = dict_open (bfs, name, cache, rw, 0);
224 logf (LOG_FATAL, "dict_open fail of `%s'", name);
233 memset (infobytes, 0, 120);
235 if (!(ipf = fopen(inputfile, "r")))
237 logf (LOG_FATAL|LOG_ERRNO, "cannot open %s", inputfile);
241 while (fgets (ipf_buf, 1023, ipf))
243 char *ipf_ptr = ipf_buf;
244 sprintf (infobytes, "%d", line);
245 for (;*ipf_ptr && *ipf_ptr != '\n';ipf_ptr++)
247 if (isalpha(*ipf_ptr) || *ipf_ptr == '_')
250 while (ipf_ptr[i] && (isalnum(ipf_ptr[i]) ||
258 switch (dict_delete (dict, ipf_ptr))
267 switch(dict_insert (dict, ipf_ptr,
268 infosize, infobytes))
276 logf (LOG_LOG, "%s change\n", ipf_ptr);
280 logf (LOG_LOG, "%s duplicate\n", ipf_ptr);
289 cp = dict_lookup (dict, ipf_ptr);
298 dict_lookup_grep (dict, ipf_ptr, range, NULL,
299 &max_pos, srange, grep_handle);
306 if ((no_of_iterations % 10000) == 0)
308 printf ("."); fflush(stdout);
321 logf (LOG_LOG, "Grepping '%s'", grep_pattern);
322 dict_lookup_grep (dict, grep_pattern, range, NULL, &max_pos,
323 srange, grep_handle);
327 logf (LOG_LOG, "Iterations.... %d", no_of_iterations);
330 logf (LOG_LOG, "No of deleted. %d", no_of_deleted);
331 logf (LOG_LOG, "No not found.. %d", no_not_found);
335 logf (LOG_LOG, "No of new..... %d", no_of_new);
336 logf (LOG_LOG, "No of change.. %d", no_of_change);
341 logf (LOG_LOG, "Lookups....... %d", no_of_iterations);
342 logf (LOG_LOG, "No of hits.... %d", no_of_hits);
343 logf (LOG_LOG, "No of misses.. %d", no_of_misses);
347 res_close (my_resource);