xref: /freebsd-src/usr.bin/tr/cmap.c (revision 1d386b48a555f61cb7325543adbbb5c3f3407a66)
1ca99cfddSTim J. Robbins /*-
2*4d846d26SWarner Losh  * SPDX-License-Identifier: BSD-2-Clause
31de7b4b8SPedro F. Giffuni  *
4ca99cfddSTim J. Robbins  * Copyright (c) 2004 Tim J. Robbins.
5ca99cfddSTim J. Robbins  * All rights reserved.
6ca99cfddSTim J. Robbins  *
7ca99cfddSTim J. Robbins  * Redistribution and use in source and binary forms, with or without
8ca99cfddSTim J. Robbins  * modification, are permitted provided that the following conditions
9ca99cfddSTim J. Robbins  * are met:
10ca99cfddSTim J. Robbins  * 1. Redistributions of source code must retain the above copyright
11ca99cfddSTim J. Robbins  *    notice, this list of conditions and the following disclaimer.
12ca99cfddSTim J. Robbins  * 2. Redistributions in binary form must reproduce the above copyright
13ca99cfddSTim J. Robbins  *    notice, this list of conditions and the following disclaimer in the
14ca99cfddSTim J. Robbins  *    documentation and/or other materials provided with the distribution.
15ca99cfddSTim J. Robbins  *
16ca99cfddSTim J. Robbins  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17ca99cfddSTim J. Robbins  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18ca99cfddSTim J. Robbins  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19ca99cfddSTim J. Robbins  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20ca99cfddSTim J. Robbins  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21ca99cfddSTim J. Robbins  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22ca99cfddSTim J. Robbins  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23ca99cfddSTim J. Robbins  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24ca99cfddSTim J. Robbins  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25ca99cfddSTim J. Robbins  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26ca99cfddSTim J. Robbins  * SUCH DAMAGE.
27ca99cfddSTim J. Robbins  */
28ca99cfddSTim J. Robbins /*
29ca99cfddSTim J. Robbins  * "Character map" ADT. Stores mappings between pairs of characters in a
30ca99cfddSTim J. Robbins  * splay tree, with a lookup table cache to simplify looking up the first
31ca99cfddSTim J. Robbins  * bunch of characters (which are presumably more common than others).
32ca99cfddSTim J. Robbins  */
33ca99cfddSTim J. Robbins 
34ca99cfddSTim J. Robbins #include <sys/cdefs.h>
35ca99cfddSTim J. Robbins #include <assert.h>
36821df508SXin LI #include <limits.h>
37821df508SXin LI #include <stdbool.h>
38ca99cfddSTim J. Robbins #include <stdlib.h>
39821df508SXin LI #include <wchar.h>
40ca99cfddSTim J. Robbins #include "cmap.h"
41ca99cfddSTim J. Robbins 
42ca99cfddSTim J. Robbins static struct cmapnode *cmap_splay(struct cmapnode *, wint_t);
43ca99cfddSTim J. Robbins 
44ca99cfddSTim J. Robbins /*
45ca99cfddSTim J. Robbins  * cmap_alloc --
46ca99cfddSTim J. Robbins  *	Allocate a character map.
47ca99cfddSTim J. Robbins  */
48ca99cfddSTim J. Robbins struct cmap *
cmap_alloc(void)49ca99cfddSTim J. Robbins cmap_alloc(void)
50ca99cfddSTim J. Robbins {
51ca99cfddSTim J. Robbins 	struct cmap *cm;
52ca99cfddSTim J. Robbins 
53ca99cfddSTim J. Robbins 	cm = malloc(sizeof(*cm));
54ca99cfddSTim J. Robbins 	if (cm == NULL)
55ca99cfddSTim J. Robbins 		return (NULL);
56ca99cfddSTim J. Robbins 	cm->cm_root = NULL;
57ca99cfddSTim J. Robbins 	cm->cm_def = CM_DEF_SELF;
58ca99cfddSTim J. Robbins 	cm->cm_havecache = false;
59ca99cfddSTim J. Robbins 	cm->cm_min = cm->cm_max = 0;
60ca99cfddSTim J. Robbins 	return (cm);
61ca99cfddSTim J. Robbins }
62ca99cfddSTim J. Robbins 
63ca99cfddSTim J. Robbins /*
64ca99cfddSTim J. Robbins  * cmap_add --
65ca99cfddSTim J. Robbins  *	Add a mapping from "from" to "to" to the map.
66ca99cfddSTim J. Robbins  */
67ca99cfddSTim J. Robbins bool
cmap_add(struct cmap * cm,wint_t from,wint_t to)68ca99cfddSTim J. Robbins cmap_add(struct cmap *cm, wint_t from, wint_t to)
69ca99cfddSTim J. Robbins {
70ca99cfddSTim J. Robbins 	struct cmapnode *cmn, *ncmn;
71ca99cfddSTim J. Robbins 
72ca99cfddSTim J. Robbins 	cm->cm_havecache = false;
73ca99cfddSTim J. Robbins 
74ca99cfddSTim J. Robbins 	if (cm->cm_root == NULL) {
75ca99cfddSTim J. Robbins 		cmn = malloc(sizeof(*cmn));
76ca99cfddSTim J. Robbins 		if (cmn == NULL)
77ca99cfddSTim J. Robbins 			return (false);
78ca99cfddSTim J. Robbins 		cmn->cmn_from = from;
79ca99cfddSTim J. Robbins 		cmn->cmn_to = to;
80ca99cfddSTim J. Robbins 		cmn->cmn_left = cmn->cmn_right = NULL;
81ca99cfddSTim J. Robbins 		cm->cm_root = cmn;
82ca99cfddSTim J. Robbins 		cm->cm_min = cm->cm_max = from;
83ca99cfddSTim J. Robbins 		return (true);
84ca99cfddSTim J. Robbins 	}
85ca99cfddSTim J. Robbins 
86ca99cfddSTim J. Robbins 	cmn = cm->cm_root = cmap_splay(cm->cm_root, from);
87ca99cfddSTim J. Robbins 
88ca99cfddSTim J. Robbins 	if (cmn->cmn_from == from) {
89ca99cfddSTim J. Robbins 		cmn->cmn_to = to;
90ca99cfddSTim J. Robbins 		return (true);
91ca99cfddSTim J. Robbins 	}
92ca99cfddSTim J. Robbins 
93ca99cfddSTim J. Robbins 	ncmn = malloc(sizeof(*ncmn));
94ca99cfddSTim J. Robbins 	if (ncmn == NULL)
95ca99cfddSTim J. Robbins 		return (false);
96ca99cfddSTim J. Robbins 	ncmn->cmn_from = from;
97ca99cfddSTim J. Robbins 	ncmn->cmn_to = to;
98ca99cfddSTim J. Robbins 	if (from < cmn->cmn_from) {
99ca99cfddSTim J. Robbins 		ncmn->cmn_left = cmn->cmn_left;
100ca99cfddSTim J. Robbins 		ncmn->cmn_right = cmn;
101ca99cfddSTim J. Robbins 		cmn->cmn_left = NULL;
102ca99cfddSTim J. Robbins 	} else {
103ca99cfddSTim J. Robbins 		ncmn->cmn_right = cmn->cmn_right;
104ca99cfddSTim J. Robbins 		ncmn->cmn_left = cmn;
105ca99cfddSTim J. Robbins 		cmn->cmn_right = NULL;
106ca99cfddSTim J. Robbins 	}
107ca99cfddSTim J. Robbins 	if (from < cm->cm_min)
108ca99cfddSTim J. Robbins 		cm->cm_min = from;
109ca99cfddSTim J. Robbins 	if (from > cm->cm_max)
110ca99cfddSTim J. Robbins 		cm->cm_max = from;
111ca99cfddSTim J. Robbins         cm->cm_root = ncmn;
112ca99cfddSTim J. Robbins 
113ca99cfddSTim J. Robbins 	return (true);
114ca99cfddSTim J. Robbins }
115ca99cfddSTim J. Robbins 
116ca99cfddSTim J. Robbins /*
117ca99cfddSTim J. Robbins  * cmap_lookup_hard --
118c75b8431STim J. Robbins  *	Look up the mapping for a character without using the cache.
119ca99cfddSTim J. Robbins  */
120ca99cfddSTim J. Robbins wint_t
cmap_lookup_hard(struct cmap * cm,wint_t ch)121ca99cfddSTim J. Robbins cmap_lookup_hard(struct cmap *cm, wint_t ch)
122ca99cfddSTim J. Robbins {
123ca99cfddSTim J. Robbins 
124ca99cfddSTim J. Robbins 	if (cm->cm_root != NULL) {
125ca99cfddSTim J. Robbins 		cm->cm_root = cmap_splay(cm->cm_root, ch);
126ca99cfddSTim J. Robbins 		if (cm->cm_root->cmn_from == ch)
127ca99cfddSTim J. Robbins 			return (cm->cm_root->cmn_to);
128ca99cfddSTim J. Robbins 	}
129ca99cfddSTim J. Robbins 	return (cm->cm_def == CM_DEF_SELF ? ch : cm->cm_def);
130ca99cfddSTim J. Robbins }
131ca99cfddSTim J. Robbins 
132ca99cfddSTim J. Robbins /*
133ca99cfddSTim J. Robbins  * cmap_cache --
134ca99cfddSTim J. Robbins  *	Update the cache.
135ca99cfddSTim J. Robbins  */
136ca99cfddSTim J. Robbins void
cmap_cache(struct cmap * cm)137ca99cfddSTim J. Robbins cmap_cache(struct cmap *cm)
138ca99cfddSTim J. Robbins {
139ca99cfddSTim J. Robbins 	wint_t ch;
140ca99cfddSTim J. Robbins 
141ca99cfddSTim J. Robbins 	for (ch = 0; ch < CM_CACHE_SIZE; ch++)
142ca99cfddSTim J. Robbins 		cm->cm_cache[ch] = cmap_lookup_hard(cm, ch);
143ca99cfddSTim J. Robbins 
144ca99cfddSTim J. Robbins 	cm->cm_havecache = true;
145ca99cfddSTim J. Robbins }
146ca99cfddSTim J. Robbins 
147ca99cfddSTim J. Robbins /*
148ca99cfddSTim J. Robbins  * cmap_default --
149ca99cfddSTim J. Robbins  *	Change the value that characters without mappings map to, and
150ca99cfddSTim J. Robbins  *	return the old value. The special character value CM_MAP_SELF
151ca99cfddSTim J. Robbins  *	means characters map to themselves.
152ca99cfddSTim J. Robbins  */
153ca99cfddSTim J. Robbins wint_t
cmap_default(struct cmap * cm,wint_t def)154ca99cfddSTim J. Robbins cmap_default(struct cmap *cm, wint_t def)
155ca99cfddSTim J. Robbins {
156ca99cfddSTim J. Robbins 	wint_t old;
157ca99cfddSTim J. Robbins 
158ca99cfddSTim J. Robbins 	old = cm->cm_def;
159ca99cfddSTim J. Robbins 	cm->cm_def = def;
160ca99cfddSTim J. Robbins 	cm->cm_havecache = false;
161ca99cfddSTim J. Robbins 	return (old);
162ca99cfddSTim J. Robbins }
163ca99cfddSTim J. Robbins 
164ca99cfddSTim J. Robbins static struct cmapnode *
cmap_splay(struct cmapnode * t,wint_t ch)165ca99cfddSTim J. Robbins cmap_splay(struct cmapnode *t, wint_t ch)
166ca99cfddSTim J. Robbins {
167ca99cfddSTim J. Robbins 	struct cmapnode N, *l, *r, *y;
168ca99cfddSTim J. Robbins 
169ca99cfddSTim J. Robbins 	/*
170ca99cfddSTim J. Robbins 	 * Based on public domain code from Sleator.
171ca99cfddSTim J. Robbins 	 */
172ca99cfddSTim J. Robbins 
173ca99cfddSTim J. Robbins 	assert(t != NULL);
174ca99cfddSTim J. Robbins 
175ca99cfddSTim J. Robbins 	N.cmn_left = N.cmn_right = NULL;
176ca99cfddSTim J. Robbins 	l = r = &N;
177ca99cfddSTim J. Robbins 	for (;;) {
178ca99cfddSTim J. Robbins 		if (ch < t->cmn_from) {
179ca99cfddSTim J. Robbins 			if (t->cmn_left != NULL &&
180ca99cfddSTim J. Robbins 			    ch < t->cmn_left->cmn_from) {
181ca99cfddSTim J. Robbins 				y = t->cmn_left;
182ca99cfddSTim J. Robbins 				t->cmn_left = y->cmn_right;
183ca99cfddSTim J. Robbins 				y->cmn_right = t;
184ca99cfddSTim J. Robbins 				t = y;
185ca99cfddSTim J. Robbins 			}
186ca99cfddSTim J. Robbins 			if (t->cmn_left == NULL)
187ca99cfddSTim J. Robbins 				break;
188ca99cfddSTim J. Robbins 			r->cmn_left = t;
189ca99cfddSTim J. Robbins 			r = t;
190ca99cfddSTim J. Robbins 			t = t->cmn_left;
191ca99cfddSTim J. Robbins 		} else if (ch > t->cmn_from) {
192ca99cfddSTim J. Robbins 			if (t->cmn_right != NULL &&
193ca99cfddSTim J. Robbins 			    ch > t->cmn_right->cmn_from) {
194ca99cfddSTim J. Robbins 				y = t->cmn_right;
195ca99cfddSTim J. Robbins 				t->cmn_right = y->cmn_left;
196ca99cfddSTim J. Robbins 				y->cmn_left = t;
197ca99cfddSTim J. Robbins 				t = y;
198ca99cfddSTim J. Robbins 			}
199ca99cfddSTim J. Robbins 			if (t->cmn_right == NULL)
200ca99cfddSTim J. Robbins 				break;
201ca99cfddSTim J. Robbins 			l->cmn_right = t;
202ca99cfddSTim J. Robbins 			l = t;
203ca99cfddSTim J. Robbins 			t = t->cmn_right;
204ca99cfddSTim J. Robbins 		} else
205ca99cfddSTim J. Robbins 			break;
206ca99cfddSTim J. Robbins 	}
207ca99cfddSTim J. Robbins 	l->cmn_right = t->cmn_left;
208ca99cfddSTim J. Robbins 	r->cmn_left = t->cmn_right;
209ca99cfddSTim J. Robbins 	t->cmn_left = N.cmn_right;
210ca99cfddSTim J. Robbins 	t->cmn_right = N.cmn_left;
211ca99cfddSTim J. Robbins 	return (t);
212ca99cfddSTim J. Robbins }
213