1241675Suqs/*	$Id: man_hash.c,v 1.25 2011/07/24 18:15:14 kristaps Exp $ */
2241675Suqs/*
3241675Suqs * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4241675Suqs *
5241675Suqs * Permission to use, copy, modify, and distribute this software for any
6241675Suqs * purpose with or without fee is hereby granted, provided that the above
7241675Suqs * copyright notice and this permission notice appear in all copies.
8241675Suqs *
9241675Suqs * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10241675Suqs * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11241675Suqs * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12241675Suqs * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13241675Suqs * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14241675Suqs * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15241675Suqs * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16241675Suqs */
17241675Suqs#ifdef HAVE_CONFIG_H
18241675Suqs#include "config.h"
19241675Suqs#endif
20241675Suqs
21241675Suqs#include <sys/types.h>
22241675Suqs
23241675Suqs#include <assert.h>
24241675Suqs#include <ctype.h>
25241675Suqs#include <limits.h>
26241675Suqs#include <stdlib.h>
27241675Suqs#include <string.h>
28241675Suqs
29241675Suqs#include "man.h"
30241675Suqs#include "mandoc.h"
31241675Suqs#include "libman.h"
32241675Suqs
33241675Suqs#define	HASH_DEPTH	 6
34241675Suqs
35241675Suqs#define	HASH_ROW(x) do { \
36241675Suqs		if (isupper((unsigned char)(x))) \
37241675Suqs			(x) -= 65; \
38241675Suqs		else \
39241675Suqs			(x) -= 97; \
40241675Suqs		(x) *= HASH_DEPTH; \
41241675Suqs	} while (/* CONSTCOND */ 0)
42241675Suqs
43241675Suqs/*
44241675Suqs * Lookup table is indexed first by lower-case first letter (plus one
45241675Suqs * for the period, which is stored in the last row), then by lower or
46241675Suqs * uppercase second letter.  Buckets correspond to the index of the
47241675Suqs * macro (the integer value of the enum stored as a char to save a bit
48241675Suqs * of space).
49241675Suqs */
50241675Suqsstatic	unsigned char	 table[26 * HASH_DEPTH];
51241675Suqs
52241675Suqs/*
53241675Suqs * XXX - this hash has global scope, so if intended for use as a library
54241675Suqs * with multiple callers, it will need re-invocation protection.
55241675Suqs */
56241675Suqsvoid
57241675Suqsman_hash_init(void)
58241675Suqs{
59241675Suqs	int		 i, j, x;
60241675Suqs
61241675Suqs	memset(table, UCHAR_MAX, sizeof(table));
62241675Suqs
63241675Suqs	assert(/* LINTED */
64241675Suqs			MAN_MAX < UCHAR_MAX);
65241675Suqs
66241675Suqs	for (i = 0; i < (int)MAN_MAX; i++) {
67241675Suqs		x = man_macronames[i][0];
68241675Suqs
69241675Suqs		assert(isalpha((unsigned char)x));
70241675Suqs
71241675Suqs		HASH_ROW(x);
72241675Suqs
73241675Suqs		for (j = 0; j < HASH_DEPTH; j++)
74241675Suqs			if (UCHAR_MAX == table[x + j]) {
75241675Suqs				table[x + j] = (unsigned char)i;
76241675Suqs				break;
77241675Suqs			}
78241675Suqs
79241675Suqs		assert(j < HASH_DEPTH);
80241675Suqs	}
81241675Suqs}
82241675Suqs
83241675Suqs
84241675Suqsenum mant
85241675Suqsman_hash_find(const char *tmp)
86241675Suqs{
87241675Suqs	int		 x, y, i;
88241675Suqs	enum mant	 tok;
89241675Suqs
90241675Suqs	if ('\0' == (x = tmp[0]))
91241675Suqs		return(MAN_MAX);
92241675Suqs	if ( ! (isalpha((unsigned char)x)))
93241675Suqs		return(MAN_MAX);
94241675Suqs
95241675Suqs	HASH_ROW(x);
96241675Suqs
97241675Suqs	for (i = 0; i < HASH_DEPTH; i++) {
98241675Suqs		if (UCHAR_MAX == (y = table[x + i]))
99241675Suqs			return(MAN_MAX);
100241675Suqs
101241675Suqs		tok = (enum mant)y;
102241675Suqs		if (0 == strcmp(tmp, man_macronames[tok]))
103241675Suqs			return(tok);
104241675Suqs	}
105241675Suqs
106241675Suqs	return(MAN_MAX);
107241675Suqs}
108