bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross/*
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * Copyright 2015 Nexenta Systmes, Inc. All rights reserved.
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * Copyright 2013 Garrett D'Amore <garrett@damore.org>
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * Copyright (c) 1995 Alex Tatmanjants <alex@elvisti.kiev.ua>
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * at Electronni Visti IA, Kiev, Ukraine.
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * All rights reserved.
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross *
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * Redistribution and use in source and binary forms, with or without
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * modification, are permitted provided that the following conditions
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * are met:
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * 1. Redistributions of source code must retain the above copyright
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * notice, this list of conditions and the following disclaimer.
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * 2. Redistributions in binary form must reproduce the above copyright
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * notice, this list of conditions and the following disclaimer in the
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * documentation and/or other materials provided with the distribution.
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross *
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * SUCH DAMAGE.
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross */
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#ifndef _COLLATEFILE_H_
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define _COLLATEFILE_H_
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross/*
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * This file defines the format of collation data files.
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * These are the files loaded to support LC_COLLATE category
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * locale data. Please note that this must define the file
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * format in a way that allows localedef to build such files
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * without assuming that the build system has all the same
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * locale.h defines and structures, which means this should
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * remain independent of things like limits.h values.
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross */
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#include <sys/types.h>
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross/* NB: libc build ensure this is == COLL_WEIGHTS_MAX (from limits.h) */
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define COLLATE_WEIGHTS_MAX 10
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define COLLATE_STR_LEN 24 /* should be 64-bit multiple */
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define COLLATE_VERSION "IllumosCollate2\n"
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define COLLATE_MAX_PRIORITY (0x7fffffff) /* max signed value */
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define COLLATE_SUBST_PRIORITY (0x40000000) /* bit indicates subst table */
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define DIRECTIVE_UNDEF 0x00
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define DIRECTIVE_FORWARD 0x01
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define DIRECTIVE_BACKWARD 0x02
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define DIRECTIVE_POSITION 0x04
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define DIRECTIVE_UNDEFINED 0x08 /* special last weight for UNDEFINED */
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#define DIRECTIVE_DIRECTION_MASK (DIRECTIVE_FORWARD | DIRECTIVE_BACKWARD)
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross/*
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * The collate file format is as follows:
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross *
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * char version[COLLATE_STR_LEN]; // must be COLLATE_VERSION
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * collate_info_t info; // see below, includes padding
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * collate_char_pri_t char_data[256]; // 8 bit char values
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * collate_subst_t subst[*]; // 0 or more substitutions
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * collate_chain_pri_t chains[*]; // 0 or more chains
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * collate_large_pri_t large[*]; // extended char priorities
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross *
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * Note that all structures must be 32-bit aligned, as each structure
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * contains 32-bit member fields. The entire file is mmap'd, so its
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * critical that alignment be observed. It is not generally safe to
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross * use any 64-bit values in the structures.
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross */
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Rosstypedef struct collate_info {
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross uint8_t directive_count;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross uint8_t directive[COLLATE_WEIGHTS_MAX];
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t pri_count[COLLATE_WEIGHTS_MAX];
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t flags;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t chain_count;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t large_count;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t subst_count[COLLATE_WEIGHTS_MAX];
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t undef_pri[COLLATE_WEIGHTS_MAX];
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross} collate_info_t;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Rosstypedef struct collate_char {
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t pri[COLLATE_WEIGHTS_MAX];
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross} collate_char_t;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Rosstypedef struct collate_chain {
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross wchar_t str[COLLATE_STR_LEN];
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t pri[COLLATE_WEIGHTS_MAX];
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross} collate_chain_t;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Rosstypedef struct collate_large {
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t val;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross collate_char_t pri;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross} collate_large_t;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Rosstypedef struct collate_subst {
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t key;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross int32_t pri[COLLATE_STR_LEN];
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross} collate_subst_t;
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross
bc09504ff1ed70f84c9713b732281f14a9ef49b2Gordon Ross#endif /* !_COLLATEFILE_H_ */