xref: /illumos-gate/usr/src/cmd/spell/spellin.c (revision 2e8cdddb)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License, Version 1.0 only
6  * (the "License").  You may not use this file except in compliance
7  * with the License.
8  *
9  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10  * or http://www.opensolaris.org/os/licensing.
11  * See the License for the specific language governing permissions
12  * and limitations under the License.
13  *
14  * When distributing Covered Code, include this CDDL HEADER in each
15  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16  * If applicable, add the following below this CDDL HEADER, with the
17  * fields enclosed by brackets "[]" replaced with your own identifying
18  * information: Portions Copyright [yyyy] [name of copyright owner]
19  *
20  * CDDL HEADER END
21  */
22 /*
23  * Copyright 2005 Sun Microsystems, Inc.  All rights reserved.
24  * Use is subject to license terms.
25  */
26 
27 /*	Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T	*/
28 /*	  All Rights Reserved   */
29 
30 
31 #include <unistd.h>
32 #include <stdlib.h>
33 #include <stdio.h>
34 #include <locale.h>
35 #include "hash.h"
36 #include "huff.h"
37 
38 int	encode(long, long *);
39 
40 #define	S (BYTE * sizeof (long))
41 #define	B (BYTE * sizeof (unsigned))
42 unsigned *table;
43 int hindex[NI];
44 unsigned wp;		/* word pointer */
45 int bp = B;		/* bit pointer */
46 static int ignore;
47 static int extra;
48 
49 static int
append(register unsigned w1,register int i)50 append(register unsigned w1, register int i)
51 {
52 	while (wp < ND - 1) {
53 		table[wp] |= w1>>(B-bp);
54 		i -= bp;
55 		if (i < 0) {
56 			bp = -i;
57 			return (1);
58 		}
59 		w1 <<= bp;
60 		bp = B;
61 		wp++;
62 	}
63 	return (0);
64 }
65 
66 
67 /*
68  *	usage: hashin N
69  *	where N is number of words in dictionary
70  *	and standard input contains sorted, unique
71  *	hashed words in octal
72  */
73 
74 int
main(int argc,char ** argv)75 main(int argc, char **argv)
76 {
77 	long h, k, d;
78 	int  i;
79 	long count;
80 	long w1;
81 	long x;
82 	int t, u;
83 	double z;
84 
85 	/* Set locale environment variables local definitions */
86 	(void) setlocale(LC_ALL, "");
87 #if !defined(TEXT_DOMAIN)	/* Should be defined by cc -D */
88 #define	TEXT_DOMAIN "SYS_TEST"	/* Use this only if it wasn't */
89 #endif
90 	(void) textdomain(TEXT_DOMAIN);
91 
92 	k = 0;
93 	u = 0;
94 	if (argc != 2) {
95 		(void) fprintf(stderr, gettext("%s: arg count\n"), argv[0]);
96 		exit(1);
97 	}
98 	table = (unsigned *)calloc(ND, sizeof (*table));
99 	if (table == 0) {
100 		(void) fprintf(stderr, gettext("%s: no space for table\n"),
101 		    argv[0]);
102 		exit(1);
103 	}
104 	if ((atof(argv[1])) == 0.0) {
105 		(void) fprintf(stderr, gettext("%s: illegal count"), argv[0]);
106 		exit(1);
107 	}
108 
109 	z = huff((1L<<HASHWIDTH)/atof(argv[1]));
110 	(void) fprintf(stderr, gettext("%s: expected code widths = %f\n"),
111 	    argv[0], z);
112 	for (count = 0; scanf("%lo", (unsigned long *)&h) == 1; ++count) {
113 		if ((t = h >> (HASHWIDTH - INDEXWIDTH)) != u) {
114 			if (bp != B)
115 				wp++;
116 			bp = B;
117 			while (u < t)
118 				hindex[++u] = wp;
119 			k =  (long)t<<(HASHWIDTH-INDEXWIDTH);
120 		}
121 		d = h-k;
122 		k = h;
123 		for (;;) {
124 			for (x = d; ; x /= 2) {
125 				i = encode(x, &w1);
126 				if (i > 0)
127 					break;
128 			}
129 			if (i > B) {
130 				if (!(append((unsigned)(w1>>(long)(i-B)), B) &&
131 				    append((unsigned)(w1<<(long)(B+B-i)),
132 				    i-B)))
133 					ignore++;
134 			} else
135 				if (!append((unsigned)(w1<<(long)(B-i)), i))
136 					ignore++;
137 			d -= x;
138 			if (d > 0)
139 				extra++;
140 			else
141 				break;
142 		}
143 	}
144 	if (bp != B)
145 		wp++;
146 	while (++u < NI)
147 		hindex[u] = wp;
148 	whuff();
149 	(void) fwrite((char *)hindex, sizeof (*hindex), NI, stdout);
150 	(void) fwrite((char *)table, sizeof (*table), wp, stdout);
151 	(void) fprintf(stderr,
152 	    gettext("%s: %ld items, %d ignored, %d extra, %u words occupied\n"),
153 	    argv[0], count, ignore, extra, wp);
154 	count -= ignore;
155 	(void) fprintf(stderr, "%s: %f table bits/item, %f table+index bits\n",
156 	    argv[0], (((float)BYTE * wp) * sizeof (*table) / count),
157 	    (BYTE * ((float)wp * sizeof (*table) + sizeof (hindex)) / count));
158 	return (0);
159 }
160