159243Sobrien/*
259243Sobrien * sh.char.h: Table for spotting special characters quickly
359243Sobrien * 	      Makes for very obscure but efficient coding.
459243Sobrien */
559243Sobrien/*-
659243Sobrien * Copyright (c) 1980, 1991 The Regents of the University of California.
759243Sobrien * All rights reserved.
859243Sobrien *
959243Sobrien * Redistribution and use in source and binary forms, with or without
1059243Sobrien * modification, are permitted provided that the following conditions
1159243Sobrien * are met:
1259243Sobrien * 1. Redistributions of source code must retain the above copyright
1359243Sobrien *    notice, this list of conditions and the following disclaimer.
1459243Sobrien * 2. Redistributions in binary form must reproduce the above copyright
1559243Sobrien *    notice, this list of conditions and the following disclaimer in the
1659243Sobrien *    documentation and/or other materials provided with the distribution.
17100616Smp * 3. Neither the name of the University nor the names of its contributors
1859243Sobrien *    may be used to endorse or promote products derived from this software
1959243Sobrien *    without specific prior written permission.
2059243Sobrien *
2159243Sobrien * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
2259243Sobrien * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
2359243Sobrien * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
2459243Sobrien * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
2559243Sobrien * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
2659243Sobrien * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
2759243Sobrien * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
2859243Sobrien * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
2959243Sobrien * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
3059243Sobrien * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
3159243Sobrien * SUCH DAMAGE.
3259243Sobrien */
3359243Sobrien#ifndef _h_sh_char
3459243Sobrien#define _h_sh_char
3559243Sobrien#if defined(NeXT) && defined(NLS)
3659243Sobrien# include <appkit/NXCType.h>
3759243Sobrien#else
3859243Sobrien# include <ctype.h>
39167465Smp# ifdef WIDE_STRINGS
40145479Smp#  ifdef HAVE_WCTYPE_H
41145479Smp#   include <wctype.h>
42145479Smp#  else
43145479Smp#   include <wchar.h>
44145479Smp#  endif
45145479Smp# endif
4659243Sobrien#endif
4759243Sobrien
4859243Sobrientypedef unsigned char tcshuc;
4959243Sobrien#ifdef _MINIX
5059243Sobrien# undef _SP
5159243Sobrien#endif /* _MINIX */
5259243Sobrienextern unsigned short _cmap[];
5359243Sobrien#if defined(DSPMBYTE)
5459243Sobrien# define CHECK_MBYTEVAR	STRdspmbyte
5559243Sobrien#endif
5659243Sobrienextern unsigned short _cmap_c[];
5759243Sobrienextern unsigned short _cmap_mbyte[];
5859243Sobrienextern short _enable_mbdisp;
5959243Sobrienextern unsigned short _mbmap[];
6059243Sobrienextern unsigned short _mbmap_euc[];
6159243Sobrienextern unsigned short _mbmap_sjis[];
6283098Smpextern unsigned short _mbmap_big5[];
63100616Smpextern unsigned short _mbmap_utf8[];
6459243Sobrien/* VARIABLE Check str */
6559243Sobrien/* same compiler require #define even not define DSPMBYTE */
66195609Smp#undef	_MB1
6759243Sobrien#define _MB1	0x0001
68195609Smp#undef	_MB2
6959243Sobrien#define _MB2	0x0002
7059243Sobrien
7159243Sobrien#ifndef NLS
7259243Sobrienextern tcshuc _cmap_lower[], _cmap_upper[];
7359243Sobrien
7459243Sobrien#endif
7559243Sobrien
76195609Smp#ifndef __QNXNTO__
77195609Smp#undef	_QF
7859243Sobrien#define	_QF	0x0001		/* '" (Forward quotes) */
79195609Smp#undef	_QB
8059243Sobrien#define	_QB	0x0002		/* ` (Backquote) */
81195609Smp#undef	_SP
8259243Sobrien#define	_SP	0x0004		/* space and tab */
83195609Smp#else
84195609Smp#undef	_XD
85195609Smp#define	_XD	0x0001		/* As in <ctype.h> */
86195609Smp#undef	_UP
87195609Smp#define	_UP	0x0002		/* As in <ctype.h> */
88195609Smp#undef	_SP
89195609Smp#define	_SP	0x0004		/* As in <ctype.h> */
90195609Smp#endif
91195609Smp#undef	_NL
9259243Sobrien#define	_NL	0x0008		/* \n */
93195609Smp#undef	_META
9459243Sobrien#define	_META	0x0010		/* lex meta characters, sp #'`";&<>()|\t\n */
95195609Smp#undef	_GLOB
9659243Sobrien#define	_GLOB	0x0020		/* glob characters, *?{[` */
97195609Smp#undef	_ESC
9859243Sobrien#define	_ESC	0x0040		/* \ */
99195609Smp#undef	_DOL
10059243Sobrien#define	_DOL	0x0080		/* $ */
101195609Smp#undef	_DIG
10259243Sobrien#define	_DIG  	0x0100		/* 0-9 */
103195609Smp#undef	_LET
104145479Smp#define	_LET  	0x0200		/* a-z, A-Z, _, or locale-specific */
105195609Smp#ifndef __QNXNTO__
106195609Smp#undef	_UP
107145479Smp#define	_UP   	0x0400		/* A-Z, or locale-specific */
108195609Smp#else
109195609Smp#undef	_QF
110195609Smp#define	_QF	0x0400		/* '" (Forward quotes) */
111195609Smp#endif
112195609Smp#undef	_DOW
113145479Smp#define	_DOW  	0x0800		/* a-z, or locale-specific */
114195609Smp#ifndef __QNXNTO__
115195609Smp#undef	_XD
11659243Sobrien#define	_XD 	0x1000		/* 0-9, a-f, A-F */
117195609Smp#else
118195609Smp#undef	_QB
119195609Smp#define	_QB	0x1000		/* 0-9, a-f, A-F */
120195609Smp#endif
121195609Smp#undef	_CMD
12259243Sobrien#define	_CMD	0x2000		/* lex end of command chars, ;&(|` */
123195609Smp#undef	_CTR
12459243Sobrien#define _CTR	0x4000		/* control */
125195609Smp#undef	_PUN
12659243Sobrien#define _PUN	0x8000		/* punctuation */
12759243Sobrien
128145479Smp#ifdef IS_ASCII
129167465Smp# define ASC(ch) (ch)
130167465Smp# define CTL_ESC(ch) (ch)
131145479Smp#else
132145479Smp# ifdef _OSD_POSIX
13359243Sobrien/* "BS2000 OSD" is a POSIX on a main frame using a EBCDIC char set */
134145479Smp#   include <ascii_ebcdic.h>
135145479Smp# else
13669408Sache/* "OS/390 USS" is a POSIX on a main frame using an IBM1047 char set */
137145479Smp# endif
138145479Smp  extern unsigned short _toascii[256];
139145479Smp  extern unsigned short _toebcdic[256];
14059243Sobrien
14159243Sobrien/* mainly for comparisons if (ASC(ch)=='\177')... */
142167465Smp#  define ASC(ch)     _toascii[(tcshuc)(ch)]
14359243Sobrien
14459243Sobrien/* Literal escapes ('\010') must be mapped to EBCDIC,
14559243Sobrien * for C-Escapes   ('\b'), the compiler already does it.
14659243Sobrien */
147167465Smp#  define CTL_ESC(ch) _toebcdic[(tcshuc)(ch)]
148145479Smp#endif /*IS_ASCII*/
14959243Sobrien
150145479Smp#ifdef WIDE_STRINGS
151145479Smp# define cmap(c, bits)	\
152231990Smp    (((c) & QUOTE) || (unsigned int)(c) >= 0x0080 ? 0 : \
153231990Smp	(_cmap[(tcshuc)ASC(c)] & (bits)))
154145479Smp#elif defined(SHORT_STRINGS) && defined(KANJI)
15569408Sache#  define cmap(c, bits)	\
156231990Smp    ((((c) & QUOTE) || ((tcshuc)(ASC(c) & 0x80) && adrof(STRnokanji))) ? \
157145479Smp	0 : (_cmap[(tcshuc)ASC(c)] & (bits)))
158145479Smp#else /* SHORT_STRINGS && KANJI */
159145479Smp# define cmap(c, bits)	\
160231990Smp    (((c) & QUOTE) ? 0 : (_cmap[(tcshuc)ASC(c)] & (bits)))
16169408Sache#endif /* SHORT_STRINGS && KANJI */
16259243Sobrien
163167465Smp#define isglob(c)	cmap((c), _GLOB)
164167465Smp#define isspc(c)	cmap((c), _SP)
165167465Smp#define ismeta(c)	cmap((c), _META)
166167465Smp#define iscmdmeta(c)	cmap((c), _CMD)
167145479Smp#ifdef WIDE_STRINGS
168145479Smp#define letter(c)	(((c) & QUOTE) ? 0 :  \
169316957Sdchagin			 (iswalpha((c)) || (c) == '_'))
170145479Smp#define alnum(c)	(((c) & QUOTE) ? 0 :  \
171316957Sdchagin		         (iswalnum((c)) || (c) == '_'))
172145479Smp#else
173145479Smp#define letter(c)	(((Char)(c) & QUOTE) ? 0 :  \
174316957Sdchagin			 ((isalpha((c)) && !(cmap((c), _PUN))) \
175195609Smp			  || (c) == '_'))
176145479Smp#define alnum(c)	(((Char)(c) & QUOTE) ? 0 :  \
177316957Sdchagin		         ((isalnum((c)) && !(cmap((c), _PUN))) \
178195609Smp			  || (c) == '_'))
179195609Smp
180145479Smp#endif
18159243Sobrien
18259243Sobrien#if defined(DSPMBYTE)
18359243Sobrien# define IsmbyteU(c)	(Ismbyte1((Char)(c))||(Ismbyte2((Char)(c))&&((c)&0200)))
18459243Sobrien#endif
18559243Sobrien
18659243Sobrien#ifdef NLS
187145479Smp# ifdef WIDE_STRINGS
188145479Smp#  define Isspace(c)	(((c) & QUOTE) ? 0 : iswspace(c))
189145479Smp#  define Isdigit(c)	(((c) & QUOTE) ? 0 : iswdigit(c))
190145479Smp#  define Isalpha(c)	(((c) & QUOTE) ? 0 : iswalpha(c))
191145479Smp#  define Islower(c)	(((c) & QUOTE) ? 0 : iswlower(c))
192145479Smp#  define Isupper(c)	(((c) & QUOTE) ? 0 : iswupper(c))
193145479Smp#  define Tolower(c) 	(((c) & QUOTE) ? 0 : (wchar_t)towlower(c))
194145479Smp#  define Toupper(c) 	(((c) & QUOTE) ? 0 : (wchar_t)towupper(c))
195145479Smp#  define Isxdigit(c)	(((c) & QUOTE) ? 0 : iswxdigit(c))
196145479Smp#  define Isalnum(c)	(((c) & QUOTE) ? 0 : iswalnum(c))
197145479Smp#  define Iscntrl(c) 	(((c) & QUOTE) ? 0 : iswcntrl(c))
198145479Smp#  define Isprint(c) 	(((c) & QUOTE) ? 0 : iswprint(c))
199145479Smp#  define Ispunct(c) 	(((c) & QUOTE) ? 0 : iswpunct(c))
200145479Smp# elif defined (NeXT)
20159243Sobrien#  define Isspace(c)	(((Char)(c) & QUOTE) ? 0 : NXIsSpace((unsigned) (c)))
20259243Sobrien#  define Isdigit(c)	(((Char)(c) & QUOTE) ? 0 : NXIsDigit((unsigned) (c)))
20359243Sobrien#  define Isalpha(c)	(((Char)(c) & QUOTE) ? 0 : NXIsAlpha((unsigned) (c)))
20459243Sobrien#  define Islower(c)	(((Char)(c) & QUOTE) ? 0 : NXIsLower((unsigned) (c)))
20559243Sobrien#  define Isupper(c)	(((Char)(c) & QUOTE) ? 0 : NXIsUpper((unsigned) (c)))
20659243Sobrien#  define Tolower(c) 	(((Char)(c) & QUOTE) ? 0 : NXToLower((unsigned) (c)))
20759243Sobrien#  define Toupper(c) 	(((Char)(c) & QUOTE) ? 0 : NXToUpper((unsigned) (c)))
20859243Sobrien#  define Isxdigit(c)	(((Char)(c) & QUOTE) ? 0 : NXIsXDigit((unsigned) (c)))
20959243Sobrien#if defined(DSPMBYTE)
21059243Sobrien#  define IscntrlM(c) 	(((Char)(c) & QUOTE) ? 0 : NXIsCntrl((unsigned) (c)))
21159243Sobrien#  define Iscntrl(c)	( (IscntrlM(c)) && !(_enable_mbdisp&&(IsmbyteU((c)))) )
21259243Sobrien#  define IsprintM(c) 	(((Char)(c) & QUOTE) ? 0 : NXIsPrint((unsigned) (c)))
21359243Sobrien#  define Isprint(c)	( (IsprintM(c)) || (_enable_mbdisp&&(IsmbyteU((c)))) )
21459243Sobrien#else
21559243Sobrien#  define Isalnum(c)	(((Char)(c) & QUOTE) ? 0 : NXIsAlNum((unsigned) (c)))
21659243Sobrien#  define Iscntrl(c) 	(((Char)(c) & QUOTE) ? 0 : NXIsCntrl((unsigned) (c)))
21759243Sobrien#  define Isprint(c) 	(((Char)(c) & QUOTE) ? 0 : NXIsPrint((unsigned) (c)))
21859243Sobrien#endif /* !defined(DSPMBYTE) */
21959243Sobrien#  define Ispunct(c) 	(((Char)(c) & QUOTE) ? 0 : NXIsPunct((unsigned) (c)))
22059243Sobrien# else /* !NeXT */
22169408Sache#  ifndef WINNT_NATIVE
22259243Sobrien#   define Isspace(c)	(((Char)(c) & QUOTE) ? 0 : isspace((tcshuc) (c)))
22359243Sobrien#   define Isdigit(c)	(((Char)(c) & QUOTE) ? 0 : isdigit((tcshuc) (c)))
22459243Sobrien#   define Isalpha(c)	(((Char)(c) & QUOTE) ? 0 : isalpha((tcshuc) (c)))
22559243Sobrien#   define Islower(c)	(((Char)(c) & QUOTE) ? 0 : islower((tcshuc) (c)))
22659243Sobrien#   define Isupper(c)	(((Char)(c) & QUOTE) ? 0 : isupper((tcshuc) (c)))
22759243Sobrien#   define Tolower(c) 	(((Char)(c) & QUOTE) ? 0 : tolower((tcshuc) (c)))
22859243Sobrien#   define Toupper(c) 	(((Char)(c) & QUOTE) ? 0 : toupper((tcshuc) (c)))
22959243Sobrien#   define Isxdigit(c)	(((Char)(c) & QUOTE) ? 0 : isxdigit((tcshuc) (c)))
23059243Sobrien#   define Isalnum(c)	(((Char)(c) & QUOTE) ? 0 : isalnum((tcshuc) (c)))
23159243Sobrien#if defined(DSPMBYTE)
23259243Sobrien#   define IscntrlM(c) 	(((Char)(c) & QUOTE) ? 0 : iscntrl((tcshuc) (c)))
23359243Sobrien#   define Iscntrl(c)	( (IscntrlM(c)) && !(_enable_mbdisp&&(IsmbyteU((c)))) )
23459243Sobrien#else
23559243Sobrien#   define Iscntrl(c) 	(((Char)(c) & QUOTE) ? 0 : iscntrl((tcshuc) (c)))
23659243Sobrien#endif /* !defined(DSPMBYTE) */
23759243Sobrien#   if SOLARIS2 == 24
23859243Sobrien    /*
23959243Sobrien     * From <casper@fwi.uva.nl> Casper Dik:
24059243Sobrien     * In Solaris 2.4, isprint('\t') returns true after setlocal(LC_ALL,"").
24159243Sobrien     * This breaks commandline editing when you include tabs.
24259243Sobrien     * (This is in the en_US locale).
24359243Sobrien     */
24459243Sobrien#if defined(DSPMBYTE)
24559243Sobrien#    define IsprintM(c) 	(((Char)(c) & QUOTE) ? 0 : \
24659243Sobrien				(isprint((tcshuc) (c)) && (c) != '\t'))
24759243Sobrien#else
24859243Sobrien#    define Isprint(c) 	(((Char)(c) & QUOTE) ? 0 : \
24959243Sobrien				(isprint((tcshuc) (c)) && (c) != '\t'))
25059243Sobrien#endif /* !defined(DSPMBYTE) */
25159243Sobrien#   else
25259243Sobrien#if defined(DSPMBYTE)
25359243Sobrien#    define IsprintM(c) (((Char)(c) & QUOTE) ? 0 : isprint((tcshuc) (c)))
25459243Sobrien#else
25559243Sobrien#    define Isprint(c) 	(((Char)(c) & QUOTE) ? 0 : isprint((tcshuc) (c)))
25659243Sobrien#endif /* !defined(DSPMBYTE) */
25759243Sobrien#   endif /* SOLARIS2 == 24 */
25859243Sobrien#if defined(DSPMBYTE)
25959243Sobrien#   define Isprint(c)	( (IsprintM(c)) || (_enable_mbdisp&&(IsmbyteU((c)))) )
26059243Sobrien#endif /* !defined(DSPMBYTE) */
26159243Sobrien#    define Ispunct(c) 	(((Char)(c) & QUOTE) ? 0 : ispunct((tcshuc) (c)))
26269408Sache#  else /* WINNT_NATIVE */
26359243Sobrien#   define Isspace(c) (((Char)(c) & QUOTE) ? 0 : isspace( oem_it((tcshuc)(c))))
26459243Sobrien#   define Isdigit(c) (((Char)(c) & QUOTE) ? 0 : isdigit( oem_it((tcshuc)(c))))
26559243Sobrien#   define Isalpha(c) (((Char)(c) & QUOTE) ? 0 : isalpha( oem_it((tcshuc)(c))))
26659243Sobrien#   define Islower(c) (((Char)(c) & QUOTE) ? 0 : islower( oem_it((tcshuc)(c))))
26759243Sobrien#   define Isupper(c) (((Char)(c) & QUOTE) ? 0 : isupper( oem_it((tcshuc)(c))))
26859243Sobrien#   define Tolower(c) (((Char)(c) & QUOTE) ? 0 : tolower( oem_it((tcshuc)(c))))
26959243Sobrien#   define Toupper(c) (((Char)(c) & QUOTE) ? 0 : toupper( oem_it((tcshuc)(c))))
27059243Sobrien#   define Isxdigit(c)(((Char)(c) & QUOTE) ? 0 : isxdigit(oem_it((tcshuc)(c))))
27159243Sobrien#   define Isalnum(c) (((Char)(c) & QUOTE) ? 0 : isalnum( oem_it((tcshuc)(c))))
27259243Sobrien#   define Ispunct(c) (((Char)(c) & QUOTE) ? 0 : ispunct( oem_it((tcshuc)(c))))
27359243Sobrien#if defined(DSPMBYTE)
27459243Sobrien#   define IscntrlM(c) (((Char)(c) & QUOTE) ? 0 : iscntrl( oem_it((tcshuc)(c))))
27559243Sobrien#   define Iscntrl(c)	( (IscntrlM(c)) && !(_enable_mbdisp&&(IsmbyteU((c)))) )
27659243Sobrien#   define IsprintM(c) (((Char)(c) & QUOTE) ? 0 : isprint( oem_it((tcshuc)(c))))
27759243Sobrien#   define Isprint(c)	( (IsprintM(c)) || (_enable_mbdisp&&(IsmbyteU((c)))) )
27859243Sobrien#else
27959243Sobrien#   define Iscntrl(c) (((Char)(c) & QUOTE) ? 0 : iscntrl( oem_it((tcshuc)(c))))
28059243Sobrien#   define Isprint(c) (((Char)(c) & QUOTE) ? 0 : isprint( oem_it((tcshuc)(c))))
28159243Sobrien#endif /* !defined(DSPMBYTE) */
28269408Sache#  endif /* WINNT_NATIVE */
28359243Sobrien# endif /* !NeXT */
28459243Sobrien#else /* !NLS */
285167465Smp# define Isspace(c)	cmap((c), _SP|_NL)
286167465Smp# define Isdigit(c)	cmap((c), _DIG)
287167465Smp# define Isalpha(c)	(cmap((c),_LET) && !(((c) & META) && AsciiOnly))
288167465Smp# define Islower(c)	(cmap((c),_DOW) && !(((c) & META) && AsciiOnly))
289167465Smp# define Isupper(c)	(cmap((c), _UP) && !(((c) & META) && AsciiOnly))
290145479Smp# define Tolower(c)	(_cmap_lower[ASC(c)])
291145479Smp# define Toupper(c)	(_cmap_upper[ASC(c)])
292167465Smp# define Isxdigit(c)	cmap((c), _XD)
293167465Smp# define Isalnum(c)	(cmap((c), _DIG|_LET) && !(((Char)(c) & META) && AsciiOnly))
29459243Sobrien#if defined(DSPMBYTE)
295167465Smp# define IscntrlM(c)	(cmap((c),_CTR) && !(((c) & META) && AsciiOnly))
29659243Sobrien# define Iscntrl(c)	( (IscntrlM(c)) && !(_enable_mbdisp&&(IsmbyteU((c)))) )
297167465Smp# define IsprintM(c)	(!cmap((c),_CTR) && !(((c) & META) && AsciiOnly))
29859243Sobrien# define Isprint(c)	( (IsprintM(c)) || (_enable_mbdisp&&(IsmbyteU((c)))) )
29959243Sobrien#else
300167465Smp# define Iscntrl(c)	(cmap((c),_CTR) && !(((c) & META) && AsciiOnly))
301167465Smp# define Isprint(c)	(!cmap((c),_CTR) && !(((c) & META) && AsciiOnly))
30259243Sobrien#endif /* !defined(DSPMBYTE) */
303167465Smp# define Ispunct(c)	(cmap((c),_PUN) && !(((c) & META) && AsciiOnly))
30459243Sobrien
30559243Sobrien#endif /* !NLS */
30659243Sobrien
30759243Sobrien#if defined(DSPMBYTE)
30859243Sobrien# define Ismbyte1(c)	((_mbmap[(c) & 0377] & _MB1) ? 1 : 0)
30959243Sobrien# define Ismbyte2(c)	((_mbmap[(c) & 0377] & _MB2) ? 1 : 0)
31059243Sobrien#endif
31159243Sobrien
31259243Sobrien#endif /* _h_sh_char */
313