1 /*
2 * linux/fs/nls_euc-jp.c
3 *
4 * Added `OSF/JVC Recommended Code Set Conversion Specification
5 * between Japanese EUC and Shift-JIS' support: <hirofumi@mail.parknet.co.jp>
6 * (http://www.opengroup.or.jp/jvc/cde/sjis-euc-e.html)
7 */
8
9 #include <linux/module.h>
10 #include <linux/kernel.h>
11 #include <linux/string.h>
12 #include <linux/nls.h>
13 #include <linux/errno.h>
14
15 static struct nls_table *p_nls;
16
17 #define IS_SJIS_LOW_BYTE(l) ((0x40 <= (l)) && ((l) <= 0xFC) && ((l) != 0x7F))
18 /* JIS X 0208 (include NEC spesial characters) */
19 #define IS_SJIS_JISX0208(h, l) ((((0x81 <= (h)) && ((h) <= 0x9F)) \
20 || ((0xE0 <= (h)) && ((h) <= 0xEA))) \
21 && IS_SJIS_LOW_BYTE(l))
22 #define IS_SJIS_JISX0201KANA(c) ((0xA1 <= (c)) && ((c) <= 0xDF))
23 #define IS_SJIS_UDC_LOW(h, l) (((0xF0 <= (h)) && ((h) <= 0xF4)) \
24 && IS_SJIS_LOW_BYTE(l))
25 #define IS_SJIS_UDC_HI(h, l) (((0xF5 <= (h)) && ((h) <= 0xF9)) \
26 && IS_SJIS_LOW_BYTE(l))
27 #define IS_SJIS_IBM(h, l) (((0xFA <= (h)) && ((h) <= 0xFC)) \
28 && IS_SJIS_LOW_BYTE(l))
29 #define IS_SJIS_NECIBM(h, l) (((0xED <= (h)) && ((h) <= 0xEE)) \
30 && IS_SJIS_LOW_BYTE(l))
31 #define MAP_SJIS2EUC(sjis_hi, sjis_lo, sjis_p, euc_hi, euc_lo, euc_p) { \
32 if ((sjis_lo) >= 0x9F) { \
33 (euc_hi) = (sjis_hi) * 2 - (((sjis_p) * 2 - (euc_p)) - 1); \
34 (euc_lo) = (sjis_lo) + 2; \
35 } else { \
36 (euc_hi) = (sjis_hi) * 2 - ((sjis_p) * 2 - (euc_p)); \
37 (euc_lo) = (sjis_lo) + ((sjis_lo) >= 0x7F ? 0x60 : 0x61); \
38 } \
39 } while(0)
40
41 #define SS2 (0x8E) /* Single Shift 2 */
42 #define SS3 (0x8F) /* Single Shift 3 */
43 #define IS_EUC_BYTE(c) ((0xA1 <= (c)) && ((c) <= 0xFE))
44 #define IS_EUC_JISX0208(h, l) (IS_EUC_BYTE(h) && IS_EUC_BYTE(l))
45 #define IS_EUC_JISX0201KANA(h, l) (((h) == SS2) && (0xA1 <= (l) && (l) <= 0xDF))
46 #define IS_EUC_UDC_LOW(h, l) (((0xF5 <= (h)) && ((h) <= 0xFE)) \
47 && IS_EUC_BYTE(l))
48 #define IS_EUC_UDC_HI(h, l) IS_EUC_UDC_LOW(h, l) /* G3 block */
49 #define MAP_EUC2SJIS(euc_hi, euc_lo, euc_p, sjis_hi, sjis_lo, sjis_p) { \
50 if ((euc_hi) & 1) { \
51 (sjis_hi) = (euc_hi) / 2 + ((sjis_p) - (euc_p) / 2); \
52 (sjis_lo) = (euc_lo) - ((euc_lo) >= 0xE0 ? 0x60 : 0x61); \
53 } else { \
54 (sjis_hi) = (euc_hi) / 2 + (((sjis_p) - (euc_p) / 2) - 1); \
55 (sjis_lo) = (euc_lo) - 2; \
56 } \
57 } while(0)
58
59 /* SJIS IBM extended characters to EUC map */
60 static unsigned char sjisibm2euc_map[][2] = {
61 {0xF3, 0xF3}, {0xF3, 0xF4}, {0xF3, 0xF5}, {0xF3, 0xF6}, {0xF3, 0xF7},
62 {0xF3, 0xF8}, {0xF3, 0xF9}, {0xF3, 0xFA}, {0xF3, 0xFB}, {0xF3, 0xFC},
63 {0xF3, 0xFD}, {0xF3, 0xFE}, {0xF4, 0xA1}, {0xF4, 0xA2}, {0xF4, 0xA3},
64 {0xF4, 0xA4}, {0xF4, 0xA5}, {0xF4, 0xA6}, {0xF4, 0xA7}, {0xF4, 0xA8},
65 {0xA2, 0xCC}, {0xA2, 0xC3}, {0xF4, 0xA9}, {0xF4, 0xAA}, {0xF4, 0xAB},
66 {0xF4, 0xAC}, {0xF4, 0xAD}, {0xA2, 0xE8}, {0xD4, 0xE3}, {0xDC, 0xDF},
67 {0xE4, 0xE9}, {0xE3, 0xF8}, {0xD9, 0xA1}, {0xB1, 0xBB}, {0xF4, 0xAE},
68 {0xC2, 0xAD}, {0xC3, 0xFC}, {0xE4, 0xD0}, {0xC2, 0xBF}, {0xBC, 0xF4},
69 {0xB0, 0xA9}, {0xB0, 0xC8}, {0xF4, 0xAF}, {0xB0, 0xD2}, {0xB0, 0xD4},
70 {0xB0, 0xE3}, {0xB0, 0xEE}, {0xB1, 0xA7}, {0xB1, 0xA3}, {0xB1, 0xAC},
71 {0xB1, 0xA9}, {0xB1, 0xBE}, {0xB1, 0xDF}, {0xB1, 0xD8}, {0xB1, 0xC8},
72 {0xB1, 0xD7}, {0xB1, 0xE3}, {0xB1, 0xF4}, {0xB1, 0xE1}, {0xB2, 0xA3},
73 {0xF4, 0xB0}, {0xB2, 0xBB}, {0xB2, 0xE6}, {0x00, 0x00}, {0xB2, 0xED},
74 {0xB2, 0xF5}, {0xB2, 0xFC}, {0xF4, 0xB1}, {0xB3, 0xB5}, {0xB3, 0xD8},
75 {0xB3, 0xDB}, {0xB3, 0xE5}, {0xB3, 0xEE}, {0xB3, 0xFB}, {0xF4, 0xB2},
76 {0xF4, 0xB3}, {0xB4, 0xC0}, {0xB4, 0xC7}, {0xB4, 0xD0}, {0xB4, 0xDE},
77 {0xF4, 0xB4}, {0xB5, 0xAA}, {0xF4, 0xB5}, {0xB5, 0xAF}, {0xB5, 0xC4},
78 {0xB5, 0xE8}, {0xF4, 0xB6}, {0xB7, 0xC2}, {0xB7, 0xE4}, {0xB7, 0xE8},
79 {0xB7, 0xE7}, {0xF4, 0xB7}, {0xF4, 0xB8}, {0xF4, 0xB9}, {0xB8, 0xCE},
80 {0xB8, 0xE1}, {0xB8, 0xF5}, {0xB8, 0xF7}, {0xB8, 0xF8}, {0xB8, 0xFC},
81 {0xB9, 0xAF}, {0xB9, 0xB7}, {0xBA, 0xBE}, {0xBA, 0xDB}, {0xCD, 0xAA},
82 {0xBA, 0xE1}, {0xF4, 0xBA}, {0xBA, 0xEB}, {0xBB, 0xB3}, {0xBB, 0xB8},
83 {0xF4, 0xBB}, {0xBB, 0xCA}, {0xF4, 0xBC}, {0xF4, 0xBD}, {0xBB, 0xD0},
84 {0xBB, 0xDE}, {0xBB, 0xF4}, {0xBB, 0xF5}, {0xBB, 0xF9}, {0xBC, 0xE4},
85 {0xBC, 0xED}, {0xBC, 0xFE}, {0xF4, 0xBE}, {0xBD, 0xC2}, {0xBD, 0xE7},
86 {0xF4, 0xBF}, {0xBD, 0xF0}, {0xBE, 0xB0}, {0xBE, 0xAC}, {0xF4, 0xC0},
87 {0xBE, 0xB3}, {0xBE, 0xBD}, {0xBE, 0xCD}, {0xBE, 0xC9}, {0xBE, 0xE4},
88 {0xBF, 0xA8}, {0xBF, 0xC9}, {0xC0, 0xC4}, {0xC0, 0xE4}, {0xC0, 0xF4},
89 {0xC1, 0xA6}, {0xF4, 0xC1}, {0xC1, 0xF5}, {0xC1, 0xFC}, {0xF4, 0xC2},
90 {0xC1, 0xF8}, {0xC2, 0xAB}, {0xC2, 0xA1}, {0xC2, 0xA5}, {0xF4, 0xC3},
91 {0xC2, 0xB8}, {0xC2, 0xBA}, {0xF4, 0xC4}, {0xC2, 0xC4}, {0xC2, 0xD2},
92 {0xC2, 0xD7}, {0xC2, 0xDB}, {0xC2, 0xDE}, {0xC2, 0xED}, {0xC2, 0xF0},
93 {0xF4, 0xC5}, {0xC3, 0xA1}, {0xC3, 0xB5}, {0xC3, 0xC9}, {0xC3, 0xB9},
94 {0xF4, 0xC6}, {0xC3, 0xD8}, {0xC3, 0xFE}, {0xF4, 0xC7}, {0xC4, 0xCC},
95 {0xF4, 0xC8}, {0xC4, 0xD9}, {0xC4, 0xEA}, {0xC4, 0xFD}, {0xF4, 0xC9},
96 {0xC5, 0xA7}, {0xC5, 0xB5}, {0xC5, 0xB6}, {0xF4, 0xCA}, {0xC5, 0xD5},
97 {0xC6, 0xB8}, {0xC6, 0xD7}, {0xC6, 0xE0}, {0xC6, 0xEA}, {0xC6, 0xE3},
98 {0xC7, 0xA1}, {0xC7, 0xAB}, {0xC7, 0xC7}, {0xC7, 0xC3}, {0xC7, 0xCB},
99 {0xC7, 0xCF}, {0xC7, 0xD9}, {0xF4, 0xCB}, {0xF4, 0xCC}, {0xC7, 0xE6},
100 {0xC7, 0xEE}, {0xC7, 0xFC}, {0xC7, 0xEB}, {0xC7, 0xF0}, {0xC8, 0xB1},
101 {0xC8, 0xE5}, {0xC8, 0xF8}, {0xC9, 0xA6}, {0xC9, 0xAB}, {0xC9, 0xAD},
102 {0xF4, 0xCD}, {0xC9, 0xCA}, {0xC9, 0xD3}, {0xC9, 0xE9}, {0xC9, 0xE3},
103 {0xC9, 0xFC}, {0xC9, 0xF4}, {0xC9, 0xF5}, {0xF4, 0xCE}, {0xCA, 0xB3},
104 {0xCA, 0xBD}, {0xCA, 0xEF}, {0xCA, 0xF1}, {0xCB, 0xAE}, {0xF4, 0xCF},
105 {0xCB, 0xCA}, {0xCB, 0xE6}, {0xCB, 0xEA}, {0xCB, 0xF0}, {0xCB, 0xF4},
106 {0xCB, 0xEE}, {0xCC, 0xA5}, {0xCB, 0xF9}, {0xCC, 0xAB}, {0xCC, 0xAE},
107 {0xCC, 0xAD}, {0xCC, 0xB2}, {0xCC, 0xC2}, {0xCC, 0xD0}, {0xCC, 0xD9},
108 {0xF4, 0xD0}, {0xCD, 0xBB}, {0xF4, 0xD1}, {0xCE, 0xBB}, {0xF4, 0xD2},
109 {0xCE, 0xBA}, {0xCE, 0xC3}, {0xF4, 0xD3}, {0xCE, 0xF2}, {0xB3, 0xDD},
110 {0xCF, 0xD5}, {0xCF, 0xE2}, {0xCF, 0xE9}, {0xCF, 0xED}, {0xF4, 0xD4},
111 {0xF4, 0xD5}, {0xF4, 0xD6}, {0x00, 0x00}, {0xF4, 0xD7}, {0xD0, 0xE5},
112 {0xF4, 0xD8}, {0xD0, 0xE9}, {0xD1, 0xE8}, {0xF4, 0xD9}, {0xF4, 0xDA},
113 {0xD1, 0xEC}, {0xD2, 0xBB}, {0xF4, 0xDB}, {0xD3, 0xE1}, {0xD3, 0xE8},
114 {0xD4, 0xA7}, {0xF4, 0xDC}, {0xF4, 0xDD}, {0xD4, 0xD4}, {0xD4, 0xF2},
115 {0xD5, 0xAE}, {0xF4, 0xDE}, {0xD7, 0xDE}, {0xF4, 0xDF}, {0xD8, 0xA2},
116 {0xD8, 0xB7}, {0xD8, 0xC1}, {0xD8, 0xD1}, {0xD8, 0xF4}, {0xD9, 0xC6},
117 {0xD9, 0xC8}, {0xD9, 0xD1}, {0xF4, 0xE0}, {0xF4, 0xE1}, {0xF4, 0xE2},
118 {0xF4, 0xE3}, {0xF4, 0xE4}, {0xDC, 0xD3}, {0xDD, 0xC8}, {0xDD, 0xD4},
119 {0xDD, 0xEA}, {0xDD, 0xFA}, {0xDE, 0xA4}, {0xDE, 0xB0}, {0xF4, 0xE5},
120 {0xDE, 0xB5}, {0xDE, 0xCB}, {0xF4, 0xE6}, {0xDF, 0xB9}, {0xF4, 0xE7},
121 {0xDF, 0xC3}, {0xF4, 0xE8}, {0xF4, 0xE9}, {0xE0, 0xD9}, {0xF4, 0xEA},
122 {0xF4, 0xEB}, {0xE1, 0xE2}, {0xF4, 0xEC}, {0xF4, 0xED}, {0xF4, 0xEE},
123 {0xE2, 0xC7}, {0xE3, 0xA8}, {0xE3, 0xA6}, {0xE3, 0xA9}, {0xE3, 0xAF},
124 {0xE3, 0xB0}, {0xE3, 0xAA}, {0xE3, 0xAB}, {0xE3, 0xBC}, {0xE3, 0xC1},
125 {0xE3, 0xBF}, {0xE3, 0xD5}, {0xE3, 0xD8}, {0xE3, 0xD6}, {0xE3, 0xDF},
126 {0xE3, 0xE3}, {0xE3, 0xE1}, {0xE3, 0xD4}, {0xE3, 0xE9}, {0xE4, 0xA6},
127 {0xE3, 0xF1}, {0xE3, 0xF2}, {0xE4, 0xCB}, {0xE4, 0xC1}, {0xE4, 0xC3},
128 {0xE4, 0xBE}, {0xF4, 0xEF}, {0xE4, 0xC0}, {0xE4, 0xC7}, {0xE4, 0xBF},
129 {0xE4, 0xE0}, {0xE4, 0xDE}, {0xE4, 0xD1}, {0xF4, 0xF0}, {0xE4, 0xDC},
130 {0xE4, 0xD2}, {0xE4, 0xDB}, {0xE4, 0xD4}, {0xE4, 0xFA}, {0xE4, 0xEF},
131 {0xE5, 0xB3}, {0xE5, 0xBF}, {0xE5, 0xC9}, {0xE5, 0xD0}, {0xE5, 0xE2},
132 {0xE5, 0xEA}, {0xE5, 0xEB}, {0xF4, 0xF1}, {0xF4, 0xF2}, {0xF4, 0xF3},
133 {0xE6, 0xE8}, {0xE6, 0xEF}, {0xE7, 0xAC}, {0xF4, 0xF4}, {0xE7, 0xAE},
134 {0xF4, 0xF5}, {0xE7, 0xB1}, {0xF4, 0xF6}, {0xE7, 0xB2}, {0xE8, 0xB1},
135 {0xE8, 0xB6}, {0xF4, 0xF7}, {0xF4, 0xF8}, {0xE8, 0xDD}, {0xF4, 0xF9},
136 {0xF4, 0xFA}, {0xE9, 0xD1}, {0xF4, 0xFB}, {0xE9, 0xED}, {0xEA, 0xCD},
137 {0xF4, 0xFC}, {0xEA, 0xDB}, {0xEA, 0xE6}, {0xEA, 0xEA}, {0xEB, 0xA5},
138 {0xEB, 0xFB}, {0xEB, 0xFA}, {0xF4, 0xFD}, {0xEC, 0xD6}, {0xF4, 0xFE},
139 };
140
141 #define IS_EUC_IBM2JISX0208(h, l) \
142 (((h) == 0xA2 && (l) == 0xCC) || ((h) == 0xA2 && (l) == 0xE8))
143
144 /* EUC to SJIS IBM extended characters map (G3 JIS X 0212 block) */
145 static struct {
146 unsigned short euc;
147 unsigned char sjis[2];
148 } euc2sjisibm_jisx0212_map[] = {
149 {0xA2C3, {0xFA, 0x55}}, {0xB0A9, {0xFA, 0x68}}, {0xB0C8, {0xFA, 0x69}},
150 {0xB0D2, {0xFA, 0x6B}}, {0xB0D4, {0xFA, 0x6C}}, {0xB0E3, {0xFA, 0x6D}},
151 {0xB0EE, {0xFA, 0x6E}}, {0xB1A3, {0xFA, 0x70}}, {0xB1A7, {0xFA, 0x6F}},
152 {0xB1A9, {0xFA, 0x72}}, {0xB1AC, {0xFA, 0x71}}, {0xB1BB, {0xFA, 0x61}},
153 {0xB1BE, {0xFA, 0x73}}, {0xB1C8, {0xFA, 0x76}}, {0xB1D7, {0xFA, 0x77}},
154 {0xB1D8, {0xFA, 0x75}}, {0xB1DF, {0xFA, 0x74}}, {0xB1E1, {0xFA, 0x7A}},
155 {0xB1E3, {0xFA, 0x78}}, {0xB1F4, {0xFA, 0x79}}, {0xB2A3, {0xFA, 0x7B}},
156 {0xB2BB, {0xFA, 0x7D}}, {0xB2E6, {0xFA, 0x7E}}, {0xB2ED, {0xFA, 0x80}},
157 {0xB2F5, {0xFA, 0x81}}, {0xB2FC, {0xFA, 0x82}}, {0xB3B5, {0xFA, 0x84}},
158 {0xB3D8, {0xFA, 0x85}}, {0xB3DB, {0xFA, 0x86}}, {0xB3DD, {0xFB, 0x77}},
159 {0xB3E5, {0xFA, 0x87}}, {0xB3EE, {0xFA, 0x88}}, {0xB3FB, {0xFA, 0x89}},
160 {0xB4C0, {0xFA, 0x8C}}, {0xB4C7, {0xFA, 0x8D}}, {0xB4D0, {0xFA, 0x8E}},
161 {0xB4DE, {0xFA, 0x8F}}, {0xB5AA, {0xFA, 0x91}}, {0xB5AF, {0xFA, 0x93}},
162 {0xB5C4, {0xFA, 0x94}}, {0xB5E8, {0xFA, 0x95}}, {0xB7C2, {0xFA, 0x97}},
163 {0xB7E4, {0xFA, 0x98}}, {0xB7E7, {0xFA, 0x9A}}, {0xB7E8, {0xFA, 0x99}},
164 {0xB8CE, {0xFA, 0x9E}}, {0xB8E1, {0xFA, 0x9F}}, {0xB8F5, {0xFA, 0xA0}},
165 {0xB8F7, {0xFA, 0xA1}}, {0xB8F8, {0xFA, 0xA2}}, {0xB8FC, {0xFA, 0xA3}},
166 {0xB9AF, {0xFA, 0xA4}}, {0xB9B7, {0xFA, 0xA5}}, {0xBABE, {0xFA, 0xA6}},
167 {0xBADB, {0xFA, 0xA7}}, {0xBAE1, {0xFA, 0xA9}}, {0xBAEB, {0xFA, 0xAB}},
168 {0xBBB3, {0xFA, 0xAC}}, {0xBBB8, {0xFA, 0xAD}}, {0xBBCA, {0xFA, 0xAF}},
169 {0xBBD0, {0xFA, 0xB2}}, {0xBBDE, {0xFA, 0xB3}}, {0xBBF4, {0xFA, 0xB4}},
170 {0xBBF5, {0xFA, 0xB5}}, {0xBBF9, {0xFA, 0xB6}}, {0xBCE4, {0xFA, 0xB7}},
171 {0xBCED, {0xFA, 0xB8}}, {0xBCF4, {0xFA, 0x67}}, {0xBCFE, {0xFA, 0xB9}},
172 {0xBDC2, {0xFA, 0xBB}}, {0xBDE7, {0xFA, 0xBC}}, {0xBDF0, {0xFA, 0xBE}},
173 {0xBEAC, {0xFA, 0xC0}}, {0xBEB0, {0xFA, 0xBF}}, {0xBEB3, {0xFA, 0xC2}},
174 {0xBEBD, {0xFA, 0xC3}}, {0xBEC9, {0xFA, 0xC5}}, {0xBECD, {0xFA, 0xC4}},
175 {0xBEE4, {0xFA, 0xC6}}, {0xBFA8, {0xFA, 0xC7}}, {0xBFC9, {0xFA, 0xC8}},
176 {0xC0C4, {0xFA, 0xC9}}, {0xC0E4, {0xFA, 0xCA}}, {0xC0F4, {0xFA, 0xCB}},
177 {0xC1A6, {0xFA, 0xCC}}, {0xC1F5, {0xFA, 0xCE}}, {0xC1F8, {0xFA, 0xD1}},
178 {0xC1FC, {0xFA, 0xCF}}, {0xC2A1, {0xFA, 0xD3}}, {0xC2A5, {0xFA, 0xD4}},
179 {0xC2AB, {0xFA, 0xD2}}, {0xC2AD, {0xFA, 0x63}}, {0xC2B8, {0xFA, 0xD6}},
180 {0xC2BA, {0xFA, 0xD7}}, {0xC2BF, {0xFA, 0x66}}, {0xC2C4, {0xFA, 0xD9}},
181 {0xC2D2, {0xFA, 0xDA}}, {0xC2D7, {0xFA, 0xDB}}, {0xC2DB, {0xFA, 0xDC}},
182 {0xC2DE, {0xFA, 0xDD}}, {0xC2ED, {0xFA, 0xDE}}, {0xC2F0, {0xFA, 0xDF}},
183 {0xC3A1, {0xFA, 0xE1}}, {0xC3B5, {0xFA, 0xE2}}, {0xC3B9, {0xFA, 0xE4}},
184 {0xC3C9, {0xFA, 0xE3}}, {0xC3D8, {0xFA, 0xE6}}, {0xC3FC, {0xFA, 0x64}},
185 {0xC3FE, {0xFA, 0xE7}}, {0xC4CC, {0xFA, 0xE9}}, {0xC4D9, {0xFA, 0xEB}},
186 {0xC4EA, {0xFA, 0xEC}}, {0xC4FD, {0xFA, 0xED}}, {0xC5A7, {0xFA, 0xEF}},
187 {0xC5B5, {0xFA, 0xF0}}, {0xC5B6, {0xFA, 0xF1}}, {0xC5D5, {0xFA, 0xF3}},
188 {0xC6B8, {0xFA, 0xF4}}, {0xC6D7, {0xFA, 0xF5}}, {0xC6E0, {0xFA, 0xF6}},
189 {0xC6E3, {0xFA, 0xF8}}, {0xC6EA, {0xFA, 0xF7}}, {0xC7A1, {0xFA, 0xF9}},
190 {0xC7AB, {0xFA, 0xFA}}, {0xC7C3, {0xFA, 0xFC}}, {0xC7C7, {0xFA, 0xFB}},
191 {0xC7CB, {0xFB, 0x40}}, {0xC7CF, {0xFB, 0x41}}, {0xC7D9, {0xFB, 0x42}},
192 {0xC7E6, {0xFB, 0x45}}, {0xC7EB, {0xFB, 0x48}}, {0xC7EE, {0xFB, 0x46}},
193 {0xC7F0, {0xFB, 0x49}}, {0xC7FC, {0xFB, 0x47}}, {0xC8B1, {0xFB, 0x4A}},
194 {0xC8E5, {0xFB, 0x4B}}, {0xC8F8, {0xFB, 0x4C}}, {0xC9A6, {0xFB, 0x4D}},
195 {0xC9AB, {0xFB, 0x4E}}, {0xC9AD, {0xFB, 0x4F}}, {0xC9CA, {0xFB, 0x51}},
196 {0xC9D3, {0xFB, 0x52}}, {0xC9E3, {0xFB, 0x54}}, {0xC9E9, {0xFB, 0x53}},
197 {0xC9F4, {0xFB, 0x56}}, {0xC9F5, {0xFB, 0x57}}, {0xC9FC, {0xFB, 0x55}},
198 {0xCAB3, {0xFB, 0x59}}, {0xCABD, {0xFB, 0x5A}}, {0xCAEF, {0xFB, 0x5B}},
199 {0xCAF1, {0xFB, 0x5C}}, {0xCBAE, {0xFB, 0x5D}}, {0xCBCA, {0xFB, 0x5F}},
200 {0xCBE6, {0xFB, 0x60}}, {0xCBEA, {0xFB, 0x61}}, {0xCBEE, {0xFB, 0x64}},
201 {0xCBF0, {0xFB, 0x62}}, {0xCBF4, {0xFB, 0x63}}, {0xCBF9, {0xFB, 0x66}},
202 {0xCCA5, {0xFB, 0x65}}, {0xCCAB, {0xFB, 0x67}}, {0xCCAD, {0xFB, 0x69}},
203 {0xCCAE, {0xFB, 0x68}}, {0xCCB2, {0xFB, 0x6A}}, {0xCCC2, {0xFB, 0x6B}},
204 {0xCCD0, {0xFB, 0x6C}}, {0xCCD9, {0xFB, 0x6D}}, {0xCDAA, {0xFA, 0xA8}},
205 {0xCDBB, {0xFB, 0x6F}}, {0xCEBA, {0xFB, 0x73}}, {0xCEBB, {0xFB, 0x71}},
206 {0xCEC3, {0xFB, 0x74}}, {0xCEF2, {0xFB, 0x76}}, {0xCFD5, {0xFB, 0x78}},
207 {0xCFE2, {0xFB, 0x79}}, {0xCFE9, {0xFB, 0x7A}}, {0xCFED, {0xFB, 0x7B}},
208 {0xD0E5, {0xFB, 0x81}}, {0xD0E9, {0xFB, 0x83}}, {0xD1E8, {0xFB, 0x84}},
209 {0xD1EC, {0xFB, 0x87}}, {0xD2BB, {0xFB, 0x88}}, {0xD3E1, {0xFB, 0x8A}},
210 {0xD3E8, {0xFB, 0x8B}}, {0xD4A7, {0xFB, 0x8C}}, {0xD4D4, {0xFB, 0x8F}},
211 {0xD4E3, {0xFA, 0x5C}}, {0xD4F2, {0xFB, 0x90}}, {0xD5AE, {0xFB, 0x91}},
212 {0xD7DE, {0xFB, 0x93}}, {0xD8A2, {0xFB, 0x95}}, {0xD8B7, {0xFB, 0x96}},
213 {0xD8C1, {0xFB, 0x97}}, {0xD8D1, {0xFB, 0x98}}, {0xD8F4, {0xFB, 0x99}},
214 {0xD9A1, {0xFA, 0x60}}, {0xD9C6, {0xFB, 0x9A}}, {0xD9C8, {0xFB, 0x9B}},
215 {0xD9D1, {0xFB, 0x9C}}, {0xDCD3, {0xFB, 0xA2}}, {0xDCDF, {0xFA, 0x5D}},
216 {0xDDC8, {0xFB, 0xA3}}, {0xDDD4, {0xFB, 0xA4}}, {0xDDEA, {0xFB, 0xA5}},
217 {0xDDFA, {0xFB, 0xA6}}, {0xDEA4, {0xFB, 0xA7}}, {0xDEB0, {0xFB, 0xA8}},
218 {0xDEB5, {0xFB, 0xAA}}, {0xDECB, {0xFB, 0xAB}}, {0xDFB9, {0xFB, 0xAD}},
219 {0xDFC3, {0xFB, 0xAF}}, {0xE0D9, {0xFB, 0xB2}}, {0xE1E2, {0xFB, 0xB5}},
220 {0xE2C7, {0xFB, 0xB9}}, {0xE3A6, {0xFB, 0xBB}}, {0xE3A8, {0xFB, 0xBA}},
221 {0xE3A9, {0xFB, 0xBC}}, {0xE3AA, {0xFB, 0xBF}}, {0xE3AB, {0xFB, 0xC0}},
222 {0xE3AF, {0xFB, 0xBD}}, {0xE3B0, {0xFB, 0xBE}}, {0xE3BC, {0xFB, 0xC1}},
223 {0xE3BF, {0xFB, 0xC3}}, {0xE3C1, {0xFB, 0xC2}}, {0xE3D4, {0xFB, 0xCA}},
224 {0xE3D5, {0xFB, 0xC4}}, {0xE3D6, {0xFB, 0xC6}}, {0xE3D8, {0xFB, 0xC5}},
225 {0xE3DF, {0xFB, 0xC7}}, {0xE3E1, {0xFB, 0xC9}}, {0xE3E3, {0xFB, 0xC8}},
226 {0xE3E9, {0xFB, 0xCB}}, {0xE3F1, {0xFB, 0xCD}}, {0xE3F2, {0xFB, 0xCE}},
227 {0xE3F8, {0xFA, 0x5F}}, {0xE4A6, {0xFB, 0xCC}}, {0xE4BE, {0xFB, 0xD2}},
228 {0xE4BF, {0xFB, 0xD6}}, {0xE4C0, {0xFB, 0xD4}}, {0xE4C1, {0xFB, 0xD0}},
229 {0xE4C3, {0xFB, 0xD1}}, {0xE4C7, {0xFB, 0xD5}}, {0xE4CB, {0xFB, 0xCF}},
230 {0xE4D0, {0xFA, 0x65}}, {0xE4D1, {0xFB, 0xD9}}, {0xE4D2, {0xFB, 0xDC}},
231 {0xE4D4, {0xFB, 0xDE}}, {0xE4DB, {0xFB, 0xDD}}, {0xE4DC, {0xFB, 0xDB}},
232 {0xE4DE, {0xFB, 0xD8}}, {0xE4E0, {0xFB, 0xD7}}, {0xE4E9, {0xFA, 0x5E}},
233 {0xE4EF, {0xFB, 0xE0}}, {0xE4FA, {0xFB, 0xDF}}, {0xE5B3, {0xFB, 0xE1}},
234 {0xE5BF, {0xFB, 0xE2}}, {0xE5C9, {0xFB, 0xE3}}, {0xE5D0, {0xFB, 0xE4}},
235 {0xE5E2, {0xFB, 0xE5}}, {0xE5EA, {0xFB, 0xE6}}, {0xE5EB, {0xFB, 0xE7}},
236 {0xE6E8, {0xFB, 0xEB}}, {0xE6EF, {0xFB, 0xEC}}, {0xE7AC, {0xFB, 0xED}},
237 {0xE7AE, {0xFB, 0xEF}}, {0xE7B1, {0xFB, 0xF1}}, {0xE7B2, {0xFB, 0xF3}},
238 {0xE8B1, {0xFB, 0xF4}}, {0xE8B6, {0xFB, 0xF5}}, {0xE8DD, {0xFB, 0xF8}},
239 {0xE9D1, {0xFB, 0xFB}}, {0xE9ED, {0xFC, 0x40}}, {0xEACD, {0xFC, 0x41}},
240 {0xEADB, {0xFC, 0x43}}, {0xEAE6, {0xFC, 0x44}}, {0xEAEA, {0xFC, 0x45}},
241 {0xEBA5, {0xFC, 0x46}}, {0xEBFA, {0xFC, 0x48}}, {0xEBFB, {0xFC, 0x47}},
242 {0xECD6, {0xFC, 0x4A}},
243 };
244
245 /* EUC to SJIS IBM extended characters map (G3 Upper block) */
246 static unsigned char euc2sjisibm_g3upper_map[][2] = {
247 {0xFA, 0x40}, {0xFA, 0x41}, {0xFA, 0x42}, {0xFA, 0x43}, {0xFA, 0x44},
248 {0xFA, 0x45}, {0xFA, 0x46}, {0xFA, 0x47}, {0xFA, 0x48}, {0xFA, 0x49},
249 {0xFA, 0x4A}, {0xFA, 0x4B}, {0xFA, 0x4C}, {0xFA, 0x4D}, {0xFA, 0x4E},
250 {0xFA, 0x4F}, {0xFA, 0x50}, {0xFA, 0x51}, {0xFA, 0x52}, {0xFA, 0x53},
251 {0xFA, 0x56}, {0xFA, 0x57}, {0xFA, 0x58}, {0xFA, 0x59}, {0xFA, 0x5A},
252 {0xFA, 0x62}, {0xFA, 0x6A}, {0xFA, 0x7C}, {0xFA, 0x83}, {0xFA, 0x8A},
253 {0xFA, 0x8B}, {0xFA, 0x90}, {0xFA, 0x92}, {0xFA, 0x96}, {0xFA, 0x9B},
254 {0xFA, 0x9C}, {0xFA, 0x9D}, {0xFA, 0xAA}, {0xFA, 0xAE}, {0xFA, 0xB0},
255 {0xFA, 0xB1}, {0xFA, 0xBA}, {0xFA, 0xBD}, {0xFA, 0xC1}, {0xFA, 0xCD},
256 {0xFA, 0xD0}, {0xFA, 0xD5}, {0xFA, 0xD8}, {0xFA, 0xE0}, {0xFA, 0xE5},
257 {0xFA, 0xE8}, {0xFA, 0xEA}, {0xFA, 0xEE}, {0xFA, 0xF2}, {0xFB, 0x43},
258 {0xFB, 0x44}, {0xFB, 0x50}, {0xFB, 0x58}, {0xFB, 0x5E}, {0xFB, 0x6E},
259 {0xFB, 0x70}, {0xFB, 0x72}, {0xFB, 0x75}, {0xFB, 0x7C}, {0xFB, 0x7D},
260 {0xFB, 0x7E}, {0xFB, 0x80}, {0xFB, 0x82}, {0xFB, 0x85}, {0xFB, 0x86},
261 {0xFB, 0x89}, {0xFB, 0x8D}, {0xFB, 0x8E}, {0xFB, 0x92}, {0xFB, 0x94},
262 {0xFB, 0x9D}, {0xFB, 0x9E}, {0xFB, 0x9F}, {0xFB, 0xA0}, {0xFB, 0xA1},
263 {0xFB, 0xA9}, {0xFB, 0xAC}, {0xFB, 0xAE}, {0xFB, 0xB0}, {0xFB, 0xB1},
264 {0xFB, 0xB3}, {0xFB, 0xB4}, {0xFB, 0xB6}, {0xFB, 0xB7}, {0xFB, 0xB8},
265 {0xFB, 0xD3}, {0xFB, 0xDA}, {0xFB, 0xE8}, {0xFB, 0xE9}, {0xFB, 0xEA},
266 {0xFB, 0xEE}, {0xFB, 0xF0}, {0xFB, 0xF2}, {0xFB, 0xF6}, {0xFB, 0xF7},
267 {0xFB, 0xF9}, {0xFB, 0xFA}, {0xFB, 0xFC}, {0xFC, 0x42}, {0xFC, 0x49},
268 {0xFC, 0x4B},
269 };
270
271 #define MAP_ELEMENT_OF(map) (sizeof(map) / sizeof(map[0]))
272
273 static inline int sjisibm2euc(unsigned char *euc, const unsigned char sjis_hi,
274 const unsigned char sjis_lo);
275 static inline int euc2sjisibm_jisx0212(unsigned char *sjis, const unsigned char euc_hi,
276 const unsigned char euc_lo);
277 static inline int euc2sjisibm_g3upper(unsigned char *sjis, const unsigned char euc_hi,
278 const unsigned char euc_lo);
279 static inline int euc2sjisibm(unsigned char *sjis, const unsigned char euc_hi,
280 const unsigned char euc_lo);
281 static inline int sjisnec2sjisibm(unsigned char *sjisibm,
282 const unsigned char sjisnec_hi,
283 const unsigned char sjisnec_lo);
284
285 /* SJIS IBM extended characters to EUC */
sjisibm2euc(unsigned char * euc,const unsigned char sjis_hi,const unsigned char sjis_lo)286 static inline int sjisibm2euc(unsigned char *euc, const unsigned char sjis_hi,
287 const unsigned char sjis_lo)
288 {
289 int index;
290
291 index = ((sjis_hi - 0xFA) * (0xFD - 0x40)) + (sjis_lo - 0x40);
292 if (IS_EUC_IBM2JISX0208(sjisibm2euc_map[index][0],
293 sjisibm2euc_map[index][1])) {
294 euc[0] = sjisibm2euc_map[index][0];
295 euc[1] = sjisibm2euc_map[index][1];
296 return 2;
297 } else {
298 euc[0] = SS3;
299 euc[1] = sjisibm2euc_map[index][0];
300 euc[2] = sjisibm2euc_map[index][1];
301 return 3;
302 }
303 }
304
305 /* EUC to SJIS IBM extended characters (G3 JIS X 0212 block) */
euc2sjisibm_jisx0212(unsigned char * sjis,const unsigned char euc_hi,const unsigned char euc_lo)306 static inline int euc2sjisibm_jisx0212(unsigned char *sjis, const unsigned char euc_hi,
307 const unsigned char euc_lo)
308 {
309 int index, min_index, max_index;
310 unsigned short euc;
311
312 min_index = 0;
313 max_index = MAP_ELEMENT_OF(euc2sjisibm_jisx0212_map) - 1;
314 euc = (euc_hi << 8) | euc_lo;
315
316 while (min_index <= max_index) {
317 index = (min_index + max_index) / 2;
318 if (euc < euc2sjisibm_jisx0212_map[index].euc)
319 max_index = index - 1;
320 else
321 min_index = index + 1;
322 if (euc == euc2sjisibm_jisx0212_map[index].euc) {
323 sjis[0] = euc2sjisibm_jisx0212_map[index].sjis[0];
324 sjis[1] = euc2sjisibm_jisx0212_map[index].sjis[1];
325 return 3;
326 }
327 }
328 return 0;
329 }
330
331 /* EUC to SJIS IBM extended characters (G3 Upper block) */
euc2sjisibm_g3upper(unsigned char * sjis,const unsigned char euc_hi,const unsigned char euc_lo)332 static inline int euc2sjisibm_g3upper(unsigned char *sjis, const unsigned char euc_hi,
333 const unsigned char euc_lo)
334 {
335 int index;
336
337 if (euc_hi == 0xF3)
338 index = ((euc_hi << 8) | euc_lo) - 0xF3F3;
339 else
340 index = ((euc_hi << 8) | euc_lo) - 0xF4A1 + 12;
341
342 if ((index < 0) || (index >= MAP_ELEMENT_OF(euc2sjisibm_g3upper_map)))
343 return 0;
344
345 sjis[0] = euc2sjisibm_g3upper_map[index][0];
346 sjis[1] = euc2sjisibm_g3upper_map[index][1];
347
348 return 3;
349 }
350
351 /* EUC to SJIS IBM extended characters (G3 block) */
euc2sjisibm(unsigned char * sjis,const unsigned char euc_hi,const unsigned char euc_lo)352 static inline int euc2sjisibm(unsigned char *sjis, const unsigned char euc_hi,
353 const unsigned char euc_lo)
354 {
355 int n;
356
357 #if 0
358 if ((euc_hi == 0xA2) && (euc_lo == 0xCC)) {
359 sjis[0] = 0xFA;
360 sjis[1] = 0x54;
361 return 2;
362 } else if ((euc_hi == 0xA2) && (euc_lo == 0xE8)) {
363 sjis[0] = 0xFA;
364 sjis[1] = 0x5B;
365 return 2;
366 }
367 #endif
368 if ((n = euc2sjisibm_g3upper(sjis, euc_hi, euc_lo))) {
369 return n;
370 } else if ((n = euc2sjisibm_jisx0212(sjis, euc_hi, euc_lo))) {
371 return n;
372 }
373
374 return 0;
375 }
376
377 /* NEC/IBM extended characters to IBM extended characters */
sjisnec2sjisibm(unsigned char * sjisibm,const unsigned char sjisnec_hi,const unsigned char sjisnec_lo)378 static inline int sjisnec2sjisibm(unsigned char *sjisibm,
379 const unsigned char sjisnec_hi,
380 const unsigned char sjisnec_lo)
381 {
382 int count;
383
384 if (! IS_SJIS_NECIBM(sjisnec_hi, sjisnec_lo))
385 return 0;
386
387 if ((sjisnec_hi == 0xEE) && (sjisnec_lo == 0xF9)) {
388 sjisibm[0] = 0x81;
389 sjisibm[1] = 0xCA;
390 return 2;
391 }
392
393 if ((sjisnec_hi == 0xEE) && (sjisnec_lo >= 0xEF)) {
394 count = (sjisnec_hi << 8 | sjisnec_lo)
395 - (sjisnec_lo <= 0xF9 ? 0xEEEF : (0xEEEF - 10));
396 } else {
397 count = (sjisnec_hi - 0xED) * (0xFC - 0x40)
398 + (sjisnec_lo - 0x40) + (0x5C - 0x40);
399 if (sjisnec_lo >= 0x7F)
400 count--;
401 }
402
403 sjisibm[0] = 0xFA + (count / (0xFC - 0x40));
404 sjisibm[1] = 0x40 + (count % (0xFC - 0x40));
405 if (sjisibm[1] >= 0x7F)
406 sjisibm[1]++;
407
408 return 2;
409 }
410
uni2char(const wchar_t uni,unsigned char * out,int boundlen)411 static int uni2char(const wchar_t uni,
412 unsigned char *out, int boundlen)
413 {
414 int n;
415
416 if (!p_nls)
417 return -EINVAL;
418 if ((n = p_nls->uni2char(uni, out, boundlen)) < 0)
419 return n;
420
421 /* translate SJIS into EUC-JP */
422 if (n == 1) {
423 if (IS_SJIS_JISX0201KANA(out[0])) {
424 /* JIS X 0201 KANA */
425 if (boundlen < 2)
426 return -ENAMETOOLONG;
427
428 out[1] = out[0];
429 out[0] = SS2;
430 return 2;
431 }
432 } else if (n == 2) {
433 /* NEC/IBM extended characters to IBM extended characters */
434 sjisnec2sjisibm(out, out[0], out[1]);
435
436 if (IS_SJIS_UDC_LOW(out[0], out[1])) {
437 /* User defined characters half low */
438 MAP_SJIS2EUC(out[0], out[1], 0xF0, out[0], out[1], 0xF5);
439 } else if (IS_SJIS_UDC_HI(out[0], out[1])) {
440 /* User defined characters half high */
441 unsigned char ch, cl;
442
443 if (boundlen < 3)
444 return -ENAMETOOLONG;
445
446 n = 3; ch = out[0]; cl = out[1];
447 out[0] = SS3;
448 MAP_SJIS2EUC(ch, cl, 0xF5, out[1], out[2], 0xF5);
449 } else if (IS_SJIS_IBM(out[0], out[1])) {
450 /* IBM extended characters */
451 unsigned char euc[3], i;
452
453 n = sjisibm2euc(euc, out[0], out[1]);
454 if (boundlen < n)
455 return -ENAMETOOLONG;
456 for (i = 0; i < n; i++)
457 out[i] = euc[i];
458 } else if (IS_SJIS_JISX0208(out[0], out[1])) {
459 /* JIS X 0208 (include NEC special characters) */
460 out[0] = (out[0]^0xA0)*2 + 0x5F;
461 if (out[1] > 0x9E)
462 out[0]++;
463
464 if (out[1] < 0x7F)
465 out[1] = out[1] + 0x61;
466 else if (out[1] < 0x9F)
467 out[1] = out[1] + 0x60;
468 else
469 out[1] = out[1] + 0x02;
470 } else {
471 /* Invalid characters */
472 return -EINVAL;
473 }
474 }
475 else
476 return -EINVAL;
477
478 return n;
479 }
480
char2uni(const unsigned char * rawstring,int boundlen,wchar_t * uni)481 static int char2uni(const unsigned char *rawstring, int boundlen,
482 wchar_t *uni)
483 {
484 unsigned char sjis_temp[2];
485 int euc_offset, n;
486
487 if ( !p_nls )
488 return -EINVAL;
489 if (boundlen <= 0)
490 return -ENAMETOOLONG;
491
492 /* translate EUC-JP into SJIS */
493 if (rawstring[0] > 0x7F) {
494 if (rawstring[0] == SS3) {
495 if (boundlen < 3)
496 return -EINVAL;
497 euc_offset = 3;
498
499 if (IS_EUC_UDC_HI(rawstring[1], rawstring[2])) {
500 /* User defined characters half high */
501 MAP_EUC2SJIS(rawstring[1], rawstring[2], 0xF5,
502 sjis_temp[0], sjis_temp[1], 0xF5);
503 } else if (euc2sjisibm(sjis_temp,rawstring[1],rawstring[2])) {
504 /* IBM extended characters */
505 } else {
506 /* JIS X 0212 and Invalid characters*/
507 return -EINVAL;
508
509 /* 'GETA' with SJIS coding */
510 /* sjis_temp[0] = 0x81; */
511 /* sjis_temp[1] = 0xAC; */
512 }
513 } else {
514 if (boundlen < 2)
515 return -EINVAL;
516 euc_offset = 2;
517
518 if (IS_EUC_JISX0201KANA(rawstring[0], rawstring[1])) {
519 /* JIS X 0201 KANA */
520 sjis_temp[0] = rawstring[1];
521 sjis_temp[1] = 0x00;
522 } else if (IS_EUC_UDC_LOW(rawstring[0], rawstring[1])) {
523 /* User defined characters half low */
524 MAP_EUC2SJIS(rawstring[0], rawstring[1], 0xF5,
525 sjis_temp[0], sjis_temp[1], 0xF0);
526 } else if (IS_EUC_JISX0208(rawstring[0], rawstring[1])) {
527 /* JIS X 0208 (include NEC spesial characters) */
528 sjis_temp[0] = ((rawstring[0]-0x5f)/2) ^ 0xA0;
529 if (!(rawstring[0] & 1))
530 sjis_temp[1] = rawstring[1] - 0x02;
531 else if (rawstring[1] < 0xE0)
532 sjis_temp[1] = rawstring[1] - 0x61;
533 else
534 sjis_temp[1] = rawstring[1] - 0x60;
535 } else {
536 /* Invalid characters */
537 return -EINVAL;
538 }
539 }
540 } else {
541 euc_offset = 1;
542
543 /* JIS X 0201 ROMAJI */
544 sjis_temp[0] = rawstring[0];
545 sjis_temp[1] = 0x00;
546 }
547
548 if ( (n = p_nls->char2uni(sjis_temp, sizeof(sjis_temp), uni)) < 0)
549 return n;
550
551 return euc_offset;
552 }
553
554 static struct nls_table table = {
555 "euc-jp",
556 uni2char,
557 char2uni,
558 NULL,
559 NULL,
560 THIS_MODULE,
561 };
562
init_nls_euc_jp(void)563 static int __init init_nls_euc_jp(void)
564 {
565 p_nls = load_nls("cp932");
566
567 if (p_nls) {
568 table.charset2upper = p_nls->charset2upper;
569 table.charset2lower = p_nls->charset2lower;
570 return register_nls(&table);
571 }
572
573 return -EINVAL;
574 }
575
exit_nls_euc_jp(void)576 static void __exit exit_nls_euc_jp(void)
577 {
578 unregister_nls(&table);
579 unload_nls(p_nls);
580 }
581
582 module_init(init_nls_euc_jp)
583 module_exit(exit_nls_euc_jp)
584 MODULE_LICENSE("Dual BSD/GPL");
585
586 /*
587 * Overrides for Emacs so that we follow Linus's tabbing style.
588 * Emacs will notice this stuff at the end of the file and automatically
589 * adjust the settings for this buffer only. This must remain at the end
590 * of the file.
591 *
592 ---------------------------------------------------------------------------
593 * Local variables:
594 * c-indent-level: 8
595 * c-brace-imaginary-offset: 0
596 * c-brace-offset: -8
597 * c-argdecl-indent: 8
598 * c-label-offset: -8
599 * c-continued-statement-offset: 8
600 * c-continued-brace-offset: 0
601 * End:
602 */
603