nls_euc-jp.c File Reference

#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/nls.h>
#include <linux/errno.h>

Include dependency graph for nls_euc-jp.c:

Go to the source code of this file.

Macros
#define	IS_SJIS_LOW_BYTE(l)   ((0x40 <= (l)) && ((l) <= 0xFC) && ((l) != 0x7F))
#define	IS_SJIS_JISX0208(h, l)
#define	IS_SJIS_JISX0201KANA(c)   ((0xA1 <= (c)) && ((c) <= 0xDF))
#define	IS_SJIS_UDC_LOW(h, l)
#define	IS_SJIS_UDC_HI(h, l)
#define	IS_SJIS_IBM(h, l)
#define	IS_SJIS_NECIBM(h, l)
#define	MAP_SJIS2EUC(sjis_hi, sjis_lo, sjis_p, euc_hi, euc_lo, euc_p)
#define	SS2   (0x8E) /* Single Shift 2 */
#define	SS3   (0x8F) /* Single Shift 3 */
#define	IS_EUC_BYTE(c)   ((0xA1 <= (c)) && ((c) <= 0xFE))
#define	IS_EUC_JISX0208(h, l)   (IS_EUC_BYTE(h) && IS_EUC_BYTE(l))
#define	IS_EUC_JISX0201KANA(h, l)   (((h) == SS2) && (0xA1 <= (l) && (l) <= 0xDF))
#define	IS_EUC_UDC_LOW(h, l)
#define	IS_EUC_UDC_HI(h, l)   IS_EUC_UDC_LOW(h, l) /* G3 block */
#define	MAP_EUC2SJIS(euc_hi, euc_lo, euc_p, sjis_hi, sjis_lo, sjis_p)
#define	IS_EUC_IBM2JISX0208(h, l)    (((h) == 0xA2 && (l) == 0xCC) || ((h) == 0xA2 && (l) == 0xE8))
#define	MAP_ELEMENT_OF(map)   (sizeof(map) / sizeof(map[0]))

Functions
static int	sjisibm2euc (unsigned char *euc, const unsigned char sjis_hi, const unsigned char sjis_lo)
static int	euc2sjisibm_jisx0212 (unsigned char *sjis, const unsigned char euc_hi, const unsigned char euc_lo)
static int	euc2sjisibm_g3upper (unsigned char *sjis, const unsigned char euc_hi, const unsigned char euc_lo)
static int	euc2sjisibm (unsigned char *sjis, const unsigned char euc_hi, const unsigned char euc_lo)
static int	sjisnec2sjisibm (unsigned char *sjisibm, const unsigned char sjisnec_hi, const unsigned char sjisnec_lo)
static int	uni2char (const wchar_t uni, unsigned char *out, int boundlen)
static int	char2uni (const unsigned char *rawstring, int boundlen, wchar_t *uni)
static int __init	init_nls_euc_jp (void)
static void __exit	exit_nls_euc_jp (void)

Variables
static struct nls_table *	p_nls
static unsigned char	sjisibm2euc_map [][2]
struct {
unsigned short   euc
unsigned char   sjis [2]
}	euc2sjisibm_jisx0212_map []
static unsigned char	euc2sjisibm_g3upper_map [][2]
static struct nls_table	table

Macro Definition Documentation

IS_EUC_BYTE

#define IS_EUC_BYTE

(

c

)

((0xA1 <= (c)) && ((c) <= 0xFE))

Definition at line 43 of file nls_euc-jp.c.

IS_EUC_IBM2JISX0208

#define IS_EUC_IBM2JISX0208	(		h,
			l
	)		(((h) == 0xA2 && (l) == 0xCC) || ((h) == 0xA2 && (l) == 0xE8))

Definition at line 141 of file nls_euc-jp.c.

IS_EUC_JISX0201KANA

#define IS_EUC_JISX0201KANA	(		h,
			l
	)		(((h) == SS2) && (0xA1 <= (l) && (l) <= 0xDF))

Definition at line 45 of file nls_euc-jp.c.

IS_EUC_JISX0208

#define IS_EUC_JISX0208	(		h,
			l
	)		(IS_EUC_BYTE(h) && IS_EUC_BYTE(l))

Definition at line 44 of file nls_euc-jp.c.

IS_EUC_UDC_HI

#define IS_EUC_UDC_HI	(		h,
			l
	)		IS_EUC_UDC_LOW(h, l) /* G3 block */

Definition at line 48 of file nls_euc-jp.c.

IS_EUC_UDC_LOW

#define IS_EUC_UDC_LOW	(		h,
			l
	)

Value:

                 (((0xF5 <= (h)) && ((h) <= 0xFE))  \
                 && IS_EUC_BYTE(l))

Definition at line 46 of file nls_euc-jp.c.

IS_SJIS_IBM

#define IS_SJIS_IBM	(		h,
			l
	)

Value:

                 (((0xFA <= (h)) && ((h) <= 0xFC))  \
                 && IS_SJIS_LOW_BYTE(l))

Definition at line 27 of file nls_euc-jp.c.

IS_SJIS_JISX0201KANA

#define IS_SJIS_JISX0201KANA

(

c

)

((0xA1 <= (c)) && ((c) <= 0xDF))

Definition at line 22 of file nls_euc-jp.c.

IS_SJIS_JISX0208

#define IS_SJIS_JISX0208	(		h,
			l
	)

Value:

                 ((((0x81 <= (h)) && ((h) <= 0x9F)) \
                 || ((0xE0 <= (h)) && ((h) <= 0xEA)))   \
                 && IS_SJIS_LOW_BYTE(l))

Definition at line 19 of file nls_euc-jp.c.

IS_SJIS_LOW_BYTE

#define IS_SJIS_LOW_BYTE

(

l

)

((0x40 <= (l)) && ((l) <= 0xFC) && ((l) != 0x7F))

Definition at line 17 of file nls_euc-jp.c.

IS_SJIS_NECIBM

#define IS_SJIS_NECIBM	(		h,
			l
	)

Value:

                 (((0xED <= (h)) && ((h) <= 0xEE))  \
                 && IS_SJIS_LOW_BYTE(l))

Definition at line 29 of file nls_euc-jp.c.

IS_SJIS_UDC_HI

#define IS_SJIS_UDC_HI	(		h,
			l
	)

Value:

                 (((0xF5 <= (h)) && ((h) <= 0xF9))  \
                 && IS_SJIS_LOW_BYTE(l))

Definition at line 25 of file nls_euc-jp.c.

IS_SJIS_UDC_LOW

#define IS_SJIS_UDC_LOW	(		h,
			l
	)

Value:

                 (((0xF0 <= (h)) && ((h) <= 0xF4))  \
                 && IS_SJIS_LOW_BYTE(l))

Definition at line 23 of file nls_euc-jp.c.

MAP_ELEMENT_OF

#define MAP_ELEMENT_OF

(

map

)

(sizeof(map) / sizeof(map[0]))

Definition at line 271 of file nls_euc-jp.c.

MAP_EUC2SJIS

#define MAP_EUC2SJIS	(		euc_hi,
			euc_lo,
			euc_p,
			sjis_hi,
			sjis_lo,
			sjis_p
	)

Value:

    {       \
    if ((euc_hi) & 1) {                         \
        (sjis_hi) = (euc_hi) / 2 + ((sjis_p) - (euc_p) / 2);        \
        (sjis_lo) = (euc_lo) - ((euc_lo) >= 0xE0 ? 0x60 : 0x61);    \
    } else {                                \
        (sjis_hi) = (euc_hi) / 2 + (((sjis_p) - (euc_p) / 2) - 1);  \
        (sjis_lo) = (euc_lo) - 2;                   \
    }                                   \
} while(0)

Definition at line 49 of file nls_euc-jp.c.

MAP_SJIS2EUC

#define MAP_SJIS2EUC	(		sjis_hi,
			sjis_lo,
			sjis_p,
			euc_hi,
			euc_lo,
			euc_p
	)

Value:

    {       \
    if ((sjis_lo) >= 0x9F) {                        \
        (euc_hi) = (sjis_hi) * 2 - (((sjis_p) * 2 - (euc_p)) - 1);  \
        (euc_lo) = (sjis_lo) + 2;                   \
    } else {                                \
        (euc_hi) = (sjis_hi) * 2 - ((sjis_p) * 2 - (euc_p));        \
        (euc_lo) = (sjis_lo) + ((sjis_lo) >= 0x7F ? 0x60 : 0x61);   \
    }                                   \
} while(0)

Definition at line 31 of file nls_euc-jp.c.

SS2

#define SS2   (0x8E) /* Single Shift 2 */

Definition at line 41 of file nls_euc-jp.c.

SS3

#define SS3   (0x8F) /* Single Shift 3 */

Definition at line 42 of file nls_euc-jp.c.

Function Documentation

char2uni()

static int char2uni ( const unsigned char *  rawstring, int  boundlen, wchar_t *  uni  )

static

Definition at line 483 of file nls_euc-jp.c.

{
    unsigned char sjis_temp[2];
    int euc_offset, n;
 
    if ( !p_nls )
        return -EINVAL;
    if (boundlen <= 0)
        return -ENAMETOOLONG;
 
    /* translate EUC-JP into SJIS */
    if (rawstring[0] > 0x7F) {
        if (rawstring[0] == SS3) {
            if (boundlen < 3)
                return -EINVAL;
            euc_offset = 3;
 
            if (IS_EUC_UDC_HI(rawstring[1], rawstring[2])) {
                /* User defined characters half high */
                MAP_EUC2SJIS(rawstring[1], rawstring[2], 0xF5,
                             sjis_temp[0], sjis_temp[1], 0xF5);
            } else if (euc2sjisibm(sjis_temp,rawstring[1],rawstring[2])) {
                /* IBM extended characters */
            } else {
                /* JIS X 0212 and Invalid characters*/
                return -EINVAL;
 
                /* 'GETA' with SJIS coding */
                /* sjis_temp[0] = 0x81; */
                /* sjis_temp[1] = 0xAC; */
            }
        } else {
            if (boundlen < 2)
                return -EINVAL;
            euc_offset = 2;
 
            if (IS_EUC_JISX0201KANA(rawstring[0], rawstring[1])) {
                /* JIS X 0201 KANA */
                sjis_temp[0] = rawstring[1];
                sjis_temp[1] = 0x00;
            } else if (IS_EUC_UDC_LOW(rawstring[0], rawstring[1])) {
                /* User defined characters half low */
                MAP_EUC2SJIS(rawstring[0], rawstring[1], 0xF5,
                             sjis_temp[0], sjis_temp[1], 0xF0);
            } else if (IS_EUC_JISX0208(rawstring[0], rawstring[1])) {
                /* JIS X 0208 (include NEC spesial characters) */
                sjis_temp[0] = ((rawstring[0]-0x5f)/2) ^ 0xA0;
                if (!(rawstring[0] & 1))
                    sjis_temp[1] = rawstring[1] - 0x02;
                else if (rawstring[1] < 0xE0)
                    sjis_temp[1] = rawstring[1] - 0x61;
                else
                    sjis_temp[1] = rawstring[1] - 0x60;
            } else {
                /* Invalid characters */
                return -EINVAL;
            }
        }
    } else {
        euc_offset = 1;
 
        /* JIS X 0201 ROMAJI */
        sjis_temp[0] = rawstring[0];
        sjis_temp[1] = 0x00;
    }
 
    if ( (n = p_nls->char2uni(sjis_temp, sizeof(sjis_temp), uni)) < 0)
        return n;
 
    return euc_offset;
}

euc2sjisibm()

static int euc2sjisibm ( unsigned char *  sjis, const unsigned char  euc_hi, const unsigned char  euc_lo  )

inlinestatic

Definition at line 352 of file nls_euc-jp.c.

{
    int n;
 
#if 0
    if ((euc_hi == 0xA2) && (euc_lo == 0xCC)) {
        sjis[0] = 0xFA;
        sjis[1] = 0x54;
        return 2;
    } else if ((euc_hi == 0xA2) && (euc_lo == 0xE8)) {
        sjis[0] = 0xFA;
        sjis[1] = 0x5B;
        return 2;
    }
#endif
    if ((n = euc2sjisibm_g3upper(sjis, euc_hi, euc_lo))) {
        return n;
    } else if ((n = euc2sjisibm_jisx0212(sjis, euc_hi, euc_lo))) {
        return n;
    }
 
    return 0;
}

Referenced by char2uni().

euc2sjisibm_g3upper()

static int euc2sjisibm_g3upper ( unsigned char *  sjis, const unsigned char  euc_hi, const unsigned char  euc_lo  )

inlinestatic

Definition at line 332 of file nls_euc-jp.c.

{
    int index;
 
    if (euc_hi == 0xF3)
        index = ((euc_hi << 8) | euc_lo) - 0xF3F3;
    else
        index = ((euc_hi << 8) | euc_lo) - 0xF4A1 + 12;
 
    if ((index < 0) || (index >= MAP_ELEMENT_OF(euc2sjisibm_g3upper_map)))
        return 0;
 
    sjis[0] = euc2sjisibm_g3upper_map[index][0];
    sjis[1] = euc2sjisibm_g3upper_map[index][1];
 
    return 3;
}

Referenced by euc2sjisibm().

euc2sjisibm_jisx0212()

static int euc2sjisibm_jisx0212 ( unsigned char *  sjis, const unsigned char  euc_hi, const unsigned char  euc_lo  )

inlinestatic

Definition at line 306 of file nls_euc-jp.c.

{
    int index, min_index, max_index;
    unsigned short euc;
 
    min_index = 0;
    max_index = MAP_ELEMENT_OF(euc2sjisibm_jisx0212_map) - 1;
    euc = (euc_hi << 8) | euc_lo;
 
    while (min_index <= max_index) {
        index = (min_index + max_index) / 2;
        if (euc < euc2sjisibm_jisx0212_map[index].euc)
            max_index = index - 1;
        else
            min_index = index + 1;
        if (euc == euc2sjisibm_jisx0212_map[index].euc) {
            sjis[0] = euc2sjisibm_jisx0212_map[index].sjis[0];
            sjis[1] = euc2sjisibm_jisx0212_map[index].sjis[1];
            return 3;
        }
    }
    return 0;
}

Referenced by euc2sjisibm().

exit_nls_euc_jp()

static void __exit exit_nls_euc_jp ( void  )

static

Definition at line 578 of file nls_euc-jp.c.

{
    unregister_nls(&table);
    unload_nls(p_nls);
}

Referenced by Ext2UnloadAllNls().

init_nls_euc_jp()

static int __init init_nls_euc_jp ( void  )

static

Definition at line 565 of file nls_euc-jp.c.

{
    p_nls = load_nls("cp932");
 
    if (p_nls) {
        table.charset2upper = p_nls->charset2upper;
        table.charset2lower = p_nls->charset2lower;
        return register_nls(&table);
    }
 
    return -EINVAL;
}

Referenced by Ext2LoadAllNls().

sjisibm2euc()

static int sjisibm2euc ( unsigned char *  euc, const unsigned char  sjis_hi, const unsigned char  sjis_lo  )

inlinestatic

Definition at line 286 of file nls_euc-jp.c.

{
    int index;
 
    index = ((sjis_hi - 0xFA) * (0xFD - 0x40)) + (sjis_lo - 0x40);
    if (IS_EUC_IBM2JISX0208(sjisibm2euc_map[index][0],
                            sjisibm2euc_map[index][1])) {
        euc[0] = sjisibm2euc_map[index][0];
        euc[1] = sjisibm2euc_map[index][1];
        return 2;
    } else {
        euc[0] = SS3;
        euc[1] = sjisibm2euc_map[index][0];
        euc[2] = sjisibm2euc_map[index][1];
        return 3;
    }
}

Referenced by uni2char().

sjisnec2sjisibm()

static int sjisnec2sjisibm ( unsigned char *  sjisibm, const unsigned char  sjisnec_hi, const unsigned char  sjisnec_lo  )

inlinestatic

Definition at line 378 of file nls_euc-jp.c.

{
    int count;
 
    if (! IS_SJIS_NECIBM(sjisnec_hi, sjisnec_lo))
        return 0;
 
    if ((sjisnec_hi == 0xEE) && (sjisnec_lo == 0xF9)) {
        sjisibm[0] = 0x81;
        sjisibm[1] = 0xCA;
        return 2;
    }
 
    if ((sjisnec_hi == 0xEE) && (sjisnec_lo >= 0xEF)) {
        count = (sjisnec_hi << 8 | sjisnec_lo)
                - (sjisnec_lo <= 0xF9 ? 0xEEEF : (0xEEEF - 10));
    } else {
        count = (sjisnec_hi - 0xED) * (0xFC - 0x40)
                + (sjisnec_lo - 0x40) + (0x5C - 0x40);
        if (sjisnec_lo >= 0x7F)
            count--;
    }
 
    sjisibm[0] = 0xFA + (count / (0xFC - 0x40));
    sjisibm[1] = 0x40 + (count % (0xFC - 0x40));
    if (sjisibm[1] >= 0x7F)
        sjisibm[1]++;
 
    return 2;
}

Referenced by uni2char().

uni2char()

static int uni2char ( const wchar_t  uni, unsigned char *  out, int  boundlen  )

static

Definition at line 411 of file nls_euc-jp.c.

{
    int n;
 
    if (!p_nls)
        return -EINVAL;
    if ((n = p_nls->uni2char(uni, out, boundlen)) < 0)
        return n;
 
    /* translate SJIS into EUC-JP */
    if (n == 1) {
        if (IS_SJIS_JISX0201KANA(out[0])) {
            /* JIS X 0201 KANA */
            if (boundlen < 2)
                return -ENAMETOOLONG;
 
            out[1] = out[0];
            out[0] = SS2;
            return 2;
        }
    } else if (n == 2) {
        /* NEC/IBM extended characters to IBM extended characters */
        sjisnec2sjisibm(out, out[0], out[1]);
 
        if (IS_SJIS_UDC_LOW(out[0], out[1])) {
            /* User defined characters half low */
            MAP_SJIS2EUC(out[0], out[1], 0xF0, out[0], out[1], 0xF5);
        } else if (IS_SJIS_UDC_HI(out[0], out[1])) {
            /* User defined characters half high */
            unsigned char ch, cl;
 
            if (boundlen < 3)
                return -ENAMETOOLONG;
 
            n = 3;
            ch = out[0];
            cl = out[1];
            out[0] = SS3;
            MAP_SJIS2EUC(ch, cl, 0xF5, out[1], out[2], 0xF5);
        } else if (IS_SJIS_IBM(out[0], out[1])) {
            /* IBM extended characters */
            unsigned char euc[3], i;
 
            n = sjisibm2euc(euc, out[0], out[1]);
            if (boundlen < n)
                return -ENAMETOOLONG;
            for (i = 0; i < n; i++)
                out[i] = euc[i];
        } else if (IS_SJIS_JISX0208(out[0], out[1])) {
            /* JIS X 0208 (include NEC special characters) */
            out[0] = (out[0]^0xA0)*2 + 0x5F;
            if (out[1] > 0x9E)
                out[0]++;
 
            if (out[1] < 0x7F)
                out[1] = out[1] + 0x61;
            else if (out[1] < 0x9F)
                out[1] = out[1] + 0x60;
            else
                out[1] = out[1] + 0x02;
        } else {
            /* Invalid characters */
            return -EINVAL;
        }
    }
    else
        return -EINVAL;
 
    return n;
}

Variable Documentation

euc

unsigned short euc

Definition at line 146 of file nls_euc-jp.c.

Referenced by euc2sjisibm_jisx0212(), sjisibm2euc(), and uni2char().

euc2sjisibm_g3upper_map

unsigned char euc2sjisibm_g3upper_map[][2]

static

Initial value:

= {
    {0xFA, 0x40}, {0xFA, 0x41}, {0xFA, 0x42}, {0xFA, 0x43}, {0xFA, 0x44},
    {0xFA, 0x45}, {0xFA, 0x46}, {0xFA, 0x47}, {0xFA, 0x48}, {0xFA, 0x49},
    {0xFA, 0x4A}, {0xFA, 0x4B}, {0xFA, 0x4C}, {0xFA, 0x4D}, {0xFA, 0x4E},
    {0xFA, 0x4F}, {0xFA, 0x50}, {0xFA, 0x51}, {0xFA, 0x52}, {0xFA, 0x53},
    {0xFA, 0x56}, {0xFA, 0x57}, {0xFA, 0x58}, {0xFA, 0x59}, {0xFA, 0x5A},
    {0xFA, 0x62}, {0xFA, 0x6A}, {0xFA, 0x7C}, {0xFA, 0x83}, {0xFA, 0x8A},
    {0xFA, 0x8B}, {0xFA, 0x90}, {0xFA, 0x92}, {0xFA, 0x96}, {0xFA, 0x9B},
    {0xFA, 0x9C}, {0xFA, 0x9D}, {0xFA, 0xAA}, {0xFA, 0xAE}, {0xFA, 0xB0},
    {0xFA, 0xB1}, {0xFA, 0xBA}, {0xFA, 0xBD}, {0xFA, 0xC1}, {0xFA, 0xCD},
    {0xFA, 0xD0}, {0xFA, 0xD5}, {0xFA, 0xD8}, {0xFA, 0xE0}, {0xFA, 0xE5},
    {0xFA, 0xE8}, {0xFA, 0xEA}, {0xFA, 0xEE}, {0xFA, 0xF2}, {0xFB, 0x43},
    {0xFB, 0x44}, {0xFB, 0x50}, {0xFB, 0x58}, {0xFB, 0x5E}, {0xFB, 0x6E},
    {0xFB, 0x70}, {0xFB, 0x72}, {0xFB, 0x75}, {0xFB, 0x7C}, {0xFB, 0x7D},
    {0xFB, 0x7E}, {0xFB, 0x80}, {0xFB, 0x82}, {0xFB, 0x85}, {0xFB, 0x86},
    {0xFB, 0x89}, {0xFB, 0x8D}, {0xFB, 0x8E}, {0xFB, 0x92}, {0xFB, 0x94},
    {0xFB, 0x9D}, {0xFB, 0x9E}, {0xFB, 0x9F}, {0xFB, 0xA0}, {0xFB, 0xA1},
    {0xFB, 0xA9}, {0xFB, 0xAC}, {0xFB, 0xAE}, {0xFB, 0xB0}, {0xFB, 0xB1},
    {0xFB, 0xB3}, {0xFB, 0xB4}, {0xFB, 0xB6}, {0xFB, 0xB7}, {0xFB, 0xB8},
    {0xFB, 0xD3}, {0xFB, 0xDA}, {0xFB, 0xE8}, {0xFB, 0xE9}, {0xFB, 0xEA},
    {0xFB, 0xEE}, {0xFB, 0xF0}, {0xFB, 0xF2}, {0xFB, 0xF6}, {0xFB, 0xF7},
    {0xFB, 0xF9}, {0xFB, 0xFA}, {0xFB, 0xFC}, {0xFC, 0x42}, {0xFC, 0x49},
    {0xFC, 0x4B},
}

Definition at line 246 of file nls_euc-jp.c.

Referenced by euc2sjisibm_g3upper().

struct { ... } euc2sjisibm_jisx0212_map[]

Referenced by euc2sjisibm_jisx0212().

p_nls

struct nls_table* p_nls

static

Definition at line 15 of file nls_euc-jp.c.

Referenced by char2uni(), exit_nls_euc_jp(), init_nls_euc_jp(), and uni2char().

sjis

unsigned char sjis[2]

Definition at line 147 of file nls_euc-jp.c.

Referenced by euc2sjisibm(), euc2sjisibm_g3upper(), and euc2sjisibm_jisx0212().

sjisibm2euc_map

unsigned char sjisibm2euc_map[][2]

static

Definition at line 60 of file nls_euc-jp.c.

Referenced by sjisibm2euc().

table

struct nls_table table

static

Initial value:

= {
     "euc-jp",
       0,
     uni2char,
     char2uni,
    0, 0,
     THIS_MODULE,
}

Definition at line 556 of file nls_euc-jp.c.