common/sys/kiconv.h

d14d7d31f2a70b1e0a9c933607d6627f5718980eis/*
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * CDDL HEADER START
d14d7d31f2a70b1e0a9c933607d6627f5718980eis *
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * The contents of this file are subject to the terms of the
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * Common Development and Distribution License (the "License").
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * You may not use this file except in compliance with the License.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis *
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * or http://www.opensolaris.org/os/licensing.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * See the License for the specific language governing permissions
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * and limitations under the License.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis *
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * When distributing Covered Code, include this CDDL HEADER in each
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * If applicable, add the following below this CDDL HEADER, with the
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * fields enclosed by brackets "[]" replaced with your own identifying
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * information: Portions Copyright [yyyy] [name of copyright owner]
d14d7d31f2a70b1e0a9c933607d6627f5718980eis *
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * CDDL HEADER END
d14d7d31f2a70b1e0a9c933607d6627f5718980eis */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/*
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * Use is subject to license terms.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#ifndef _SYS_KICONV_H
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define _SYS_KICONV_H
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#pragma ident   "%Z%%M% %I% %E% SMI"
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#ifdef __cplusplus
d14d7d31f2a70b1e0a9c933607d6627f5718980eisextern "C" {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#endif
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#include <sys/types.h>
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#ifdef  _KERNEL
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/*
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * Supported fromcode/tocode values are saved in the following component type
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * of (name, id) pair. The id values of fromcode and tocode are used to
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * find out the corresponding code conversions.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis */
d14d7d31f2a70b1e0a9c933607d6627f5718980eistypedef struct {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    char        *name;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    size_t      id;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis} kiconv_code_list_t;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/*
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * Each unique kiconv code conversion identified by tocode and fromcode ids
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * have corresponding module id and internal function pointers to open(),
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * kiconv(), close(), and kiconvstr().
d14d7d31f2a70b1e0a9c933607d6627f5718980eis */
d14d7d31f2a70b1e0a9c933607d6627f5718980eistypedef struct {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    uint16_t    tid;        /* tocode id. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    uint16_t    fid;        /* fromcode id. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    uint16_t    mid;        /* module id. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    void        *(*open)(void);
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    size_t      (*kiconv)(void *, char **, size_t *, char **, size_t *,
d14d7d31f2a70b1e0a9c933607d6627f5718980eis            int *);
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    int     (*close)(void *);
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    size_t      (*kiconvstr)(char *, size_t *, char *, size_t *, int,
d14d7d31f2a70b1e0a9c933607d6627f5718980eis            int *);
d14d7d31f2a70b1e0a9c933607d6627f5718980eis} kiconv_conv_list_t;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/*
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * Each module id has a corresponding module name that is used to load
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * the module as needed and a reference counter.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis */
d14d7d31f2a70b1e0a9c933607d6627f5718980eistypedef struct {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    char        *name;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    uint_t      refcount;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis} kiconv_mod_list_t;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/*
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * The following two data structures are being used to transfer information
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * on the supported kiconv code conversions from a module to the framework.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis *
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * Details can be found from kiconv_ops(9S) and kiconv_module_info(9S)
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * man pages at PSARC/2007/173.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis */
d14d7d31f2a70b1e0a9c933607d6627f5718980eistypedef struct {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    char        *tocode;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    char        *fromcode;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    void        *(*kiconv_open)(void);
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    size_t      (*kiconv)(void *, char **, size_t *, char **, size_t *,
d14d7d31f2a70b1e0a9c933607d6627f5718980eis            int *);
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    int     (*kiconv_close)(void *);
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    size_t      (*kiconvstr)(char *, size_t *, char *, size_t *, int,
d14d7d31f2a70b1e0a9c933607d6627f5718980eis            int *);
d14d7d31f2a70b1e0a9c933607d6627f5718980eis} kiconv_ops_t;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eistypedef struct kiconv_mod_info {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    char        *module_name;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    size_t      kiconv_num_convs;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    kiconv_ops_t    *kiconv_ops_tbl;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    size_t      kiconv_num_aliases;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    char        **aliases;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    char        **canonicals;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    int     nowait;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis} kiconv_module_info_t;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/* The kiconv code conversion descriptor data structure. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eistypedef struct {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    void        *handle;    /* Handle from the actual open(). */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    size_t      id;     /* Index to the conv_list[]. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis} kiconv_data_t, *kiconv_t;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/* Common conversion state data structure. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eistypedef struct {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    uint8_t     id;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    uint8_t     bom_processed;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis} kiconv_state_data_t, *kiconv_state_t;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/* Common component types for possible code conversion mapping tables. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eistypedef struct {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    uchar_t     u8[3];
d14d7d31f2a70b1e0a9c933607d6627f5718980eis} kiconv_to_utf8_tbl_comp_t;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eistypedef struct {
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    uint32_t    u8:24;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    uint32_t    sb:8;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis} kiconv_to_sb_tbl_comp_t;
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/*
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * The maximum name length for any given codeset or alias names; the following
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * should be plenty big enough.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_MAX_CODENAME_LEN     63
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/* The following characters do not exist in the normalized code names. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_SKIPPABLE_CHAR(c)    \
d14d7d31f2a70b1e0a9c933607d6627f5718980eis    ((c) == '-' || (c) == '_' || (c) == '.' || (c) == '@')
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/*
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * When we encounter non-identical characters, as like iconv(3C) we have,
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * map them into either one of the replacement characters based on what is
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * the current target tocde.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis *
d14d7d31f2a70b1e0a9c933607d6627f5718980eis * The 0xefbfdb in UTF-8 is U+FFFD in Unicode scalar value.
d14d7d31f2a70b1e0a9c933607d6627f5718980eis */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_ASCII_REPLACEMENT_CHAR   ('?')
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_UTF8_REPLACEMENT_CHAR    (0xefbfbd)
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/* Numeric ids for kiconv modules. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_EMBEDDED         (0)
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_MODULE_ID_JA     (1)
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_MODULE_ID_SC     (2)
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_MODULE_ID_KO     (3)
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_MODULE_ID_TC     (4)
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_MODULE_ID_EMEA       (5)
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#define KICONV_MAX_MODULE_ID        KICONV_MODULE_ID_EMEA
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis/* Functions used in kiconv conversion and module management. */
d14d7d31f2a70b1e0a9c933607d6627f5718980eisextern void kiconv_init();
d14d7d31f2a70b1e0a9c933607d6627f5718980eisextern int  kiconv_register_module(kiconv_module_info_t *);
d14d7d31f2a70b1e0a9c933607d6627f5718980eisextern int  kiconv_unregister_module(kiconv_module_info_t *);
d14d7d31f2a70b1e0a9c933607d6627f5718980eisextern size_t   kiconv_module_ref_count(size_t);
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#endif  /* _KERNEL */
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#ifdef __cplusplus
d14d7d31f2a70b1e0a9c933607d6627f5718980eis}
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#endif
d14d7d31f2a70b1e0a9c933607d6627f5718980eis
d14d7d31f2a70b1e0a9c933607d6627f5718980eis#endif /* _SYS_KICONV_H */