| @@ -1,186 +0,0 @@ |
| 1 | #define _WIN32_WINNT 0x0500 | | |
| 2 | #include <windows.h> | | |
| 3 | | | |
| 4 | #include "u.h" | | |
| 5 | #include "lib.h" | | |
| 6 | #include "r16.h" | | |
| 7 | | | |
| 8 | #define Bit(i) (7-(i)) | | |
| 9 | /* N 0's preceded by i 1's, T(Bit(2)) is 1100 0000 */ | | |
| 10 | #define T(i) (((1 << (Bit(i)+1))-1) ^ 0xFF) | | |
| 11 | /* 0000 0000 0000 0111 1111 1111 */ | | |
| 12 | #define RuneX(i) ((1 << (Bit(i) + ((i)-1)*Bitx))-1) | | |
| 13 | | | |
| 14 | enum | | |
| 15 | { | | |
| 16 | Bitx = Bit(1), | | |
| 17 | | | |
| 18 | Tx = T(1), /* 1000 0000 */ | | |
| 19 | Rune1 = (1<<(Bit(0)+0*Bitx))-1, /* 0000 0000 0000 0000 0111 1111 */ | | |
| 20 | | | |
| 21 | Maskx = (1<<Bitx)-1, /* 0011 1111 */ | | |
| 22 | Testx = Maskx ^ 0xFF, /* 1100 0000 */ | | |
| 23 | | | |
| 24 | SurrogateMax = 0xDFFF, | | |
| 25 | HiSurrogate = 0xD800, | | |
| 26 | LoSurrogate = 0xDC00, | | |
| 27 | | | |
| 28 | Bad = Runeerror, | | |
| 29 | }; | | |
| 30 | | | |
| 31 | Rune16* | | |
| 32 | runes16dup(Rune16 *r) | | |
| 33 | { | | |
| 34 | int n; | | |
| 35 | Rune16 *s; | | |
| 36 | | | |
| 37 | n = runes16len(r) + 1; | | |
| 38 | s = calloc(n, sizeof(Rune16)); | | |
| 39 | memcpy(s, r, n * sizeof(Rune16)); | | |
| 40 | return s; | | |
| 41 | } | | |
| 42 | | | |
| 43 | int | | |
| 44 | runes16len(Rune16 *r) | | |
| 45 | { | | |
| 46 | int n; | | |
| 47 | | | |
| 48 | n = 0; | | |
| 49 | while(*r++ != 0) | | |
| 50 | n++; | | |
| 51 | return n; | | |
| 52 | } | | |
| 53 | | | |
| 54 | char* | | |
| 55 | runes16toutf(char *p, Rune16 *r, int nc) | | |
| 56 | { | | |
| 57 | char *op, *ep; | | |
| 58 | int n, c; | | |
| 59 | Rune rc; | | |
| 60 | | | |
| 61 | op = p; | | |
| 62 | ep = p + nc; | | |
| 63 | while(c = *r++){ | | |
| 64 | n = 1; | | |
| 65 | if(c >= Runeself) | | |
| 66 | n = runelen(c); | | |
| 67 | if(p + n >= ep) | | |
| 68 | break; | | |
| 69 | if(c < Runeself){ | | |
| 70 | *p++ = c; | | |
| 71 | continue; | | |
| 72 | } | | |
| 73 | rc = c; | | |
| 74 | if(c >= LoSurrogate && c <= SurrogateMax) | | |
| 75 | rc = Bad; | | |
| 76 | else if(c >= HiSurrogate && c <= 0xDBFF){ /* decode a surrogate pair properly */ | | |
| 77 | if(p + n+1 >= ep) | | |
| 78 | rc = Bad; | | |
| 79 | else if((c = *r) >= LoSurrogate && c <= SurrogateMax){ | | |
| 80 | rc = 0x10000 | (*(r-1) - HiSurrogate) << 10 | (c - LoSurrogate); | | |
| 81 | r++; | | |
| 82 | }else | | |
| 83 | rc = Bad; | | |
| 84 | } | | |
| 85 | p += runetochar(p, &rc); | | |
| 86 | } | | |
| 87 | *p = '\0'; | | |
| 88 | return op; | | |
| 89 | } | | |
| 90 | | | |
| 91 | int | | |
| 92 | rune16nlen(Rune16 *r, int nrune) | | |
| 93 | { | | |
| 94 | int nb, i; | | |
| 95 | Rune c; | | |
| 96 | | | |
| 97 | nb = 0; | | |
| 98 | while(nrune--) { | | |
| 99 | c = *r++; | | |
| 100 | if(c <= Rune1){ | | |
| 101 | nb++; | | |
| 102 | } else { | | |
| 103 | for(i = 2; i < UTFmax + 1; i++) | | |
| 104 | if(c <= RuneX(i) || i == UTFmax){ | | |
| 105 | nb += i; | | |
| 106 | break; | | |
| 107 | } | | |
| 108 | } | | |
| 109 | } | | |
| 110 | return nb; | | |
| 111 | } | | |
| 112 | | | |
| 113 | Rune16* | | |
| 114 | utftorunes16(Rune16 *r, char *p, int nc) | | |
| 115 | { | | |
| 116 | Rune16 *or, *er; | | |
| 117 | Rune rc; | | |
| 118 | | | |
| 119 | or = r; | | |
| 120 | er = r + nc; | | |
| 121 | while(*p != '\0' && r + 1 < er){ | | |
| 122 | p += chartorune(&rc, p); | | |
| 123 | if(rc >= 0x10000){ /* got to encode it in a surrogate pair */ | | |
| 124 | rc -= 0x10000; | | |
| 125 | *r++ = (rc >> 10)+HiSurrogate; | | |
| 126 | *r++ = (rc & 0x3FF)+LoSurrogate; | | |
| 127 | }else | | |
| 128 | *r++ = rc; | | |
| 129 | } | | |
| 130 | *r = '\0'; | | |
| 131 | return or; | | |
| 132 | } | | |
| 133 | | | |
| 134 | int | | |
| 135 | runes16cmp(Rune16 *s1, Rune16 *s2) | | |
| 136 | { | | |
| 137 | Rune16 r1, r2; | | |
| 138 | | | |
| 139 | for(;;) { | | |
| 140 | r1 = *s1++; | | |
| 141 | r2 = *s2++; | | |
| 142 | if(r1 != r2) { | | |
| 143 | if(r1 > r2) | | |
| 144 | return 1; | | |
| 145 | return -1; | | |
| 146 | } | | |
| 147 | if(r1 == 0) | | |
| 148 | return 0; | | |
| 149 | } | | |
| 150 | } | | |
| 151 | | | |
| 152 | wchar_t * | | |
| 153 | widen(char *s) | | |
| 154 | { | | |
| 155 | int n; | | |
| 156 | wchar_t *ws; | | |
| 157 | | | |
| 158 | n = utflen(s) + 1; | | |
| 159 | ws = calloc(n, sizeof(wchar_t)); | | |
| 160 | utftorunes16(ws, s, n); | | |
| 161 | return ws; | | |
| 162 | } | | |
| 163 | | | |
| 164 | | | |
| 165 | char * | | |
| 166 | narrowen(wchar_t *ws) | | |
| 167 | { | | |
| 168 | char *s; | | |
| 169 | int n; | | |
| 170 | | | |
| 171 | n = widebytes(ws); | | |
| 172 | s = malloc(n); | | |
| 173 | runes16toutf(s, ws, n); | | |
| 174 | return s; | | |
| 175 | } | | |
| 176 | | | |
| 177 | | | |
| 178 | int | | |
| 179 | widebytes(wchar_t *ws) | | |
| 180 | { | | |
| 181 | int n = 0; | | |
| 182 | | | |
| 183 | while(*ws) | | |
| 184 | n += runelen(*ws++); | | |
| 185 | return n+1; | | |
| 186 | } | | |