|
1 | 1 | /** |
2 | 2 | * @file utf32.h |
3 | 3 | * @brief UTF-32 |
4 | | - * @copyright Copyright (C) 1999-2001, 2008, 2011, 2016, 2024 Free Software Foundation, Inc. |
| 4 | + * @copyright Copyright (C) 1999-2024 Free Software Foundation, Inc. |
5 | 5 | * @copyright Copyright (C) 2024 The C++ Plus Project. |
6 | 6 | */ |
7 | 7 | /* |
|
38 | 38 | only for strings containing U+FEFF characters, which is quite rare.) |
39 | 39 | The default is big-endian. */ |
40 | 40 | /* The state is 0 if big-endian, 1 if little-endian. */ |
41 | | -static int |
42 | | -utf32_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, size_t n) |
| 41 | +static int utf32_mbtowc(conv_t conv, ucs4_t *pwc, const unsigned char *s, size_t n) |
43 | 42 | { |
44 | | - state_t state = conv->istate; |
45 | | - int count = 0; |
46 | | - for (; n >= 4 && count <= RET_COUNT_MAX && count <= INT_MAX-4;) { |
47 | | - ucs4_t wc = (state |
48 | | - ? (ucs4_t) s[0] |
49 | | - + ((ucs4_t) s[1] << 8) |
50 | | - + ((ucs4_t) s[2] << 16) |
51 | | - + ((ucs4_t) s[3] << 24) |
52 | | - : ((ucs4_t) s[0] << 24) |
53 | | - + ((ucs4_t) s[1] << 16) |
54 | | - + ((ucs4_t) s[2] << 8) |
55 | | - + (ucs4_t) s[3]); |
56 | | - if (wc == 0x0000feff) { |
57 | | - } else if (wc == 0xfffe0000u) { |
58 | | - state ^= 1; |
59 | | - } else { |
60 | | - if (wc < 0x110000 && !(wc >= 0xd800 && wc < 0xe000)) { |
61 | | - *pwc = wc; |
62 | | - conv->istate = state; |
63 | | - return count+4; |
64 | | - } else { |
65 | | - conv->istate = state; |
66 | | - return RET_SHIFT_ILSEQ(count); |
67 | | - } |
| 43 | + state_t state = conv->ibyteorder; |
| 44 | + int count = 0; |
| 45 | + for (; n >= 4 && count <= RET_COUNT_MAX && count <= INT_MAX - 4;) |
| 46 | + { |
| 47 | + ucs4_t wc = (state ? (ucs4_t)s[0] + ((ucs4_t)s[1] << 8) + ((ucs4_t)s[2] << 16) + ((ucs4_t)s[3] << 24) |
| 48 | + : ((ucs4_t)s[0] << 24) + ((ucs4_t)s[1] << 16) + ((ucs4_t)s[2] << 8) + (ucs4_t)s[3]); |
| 49 | + if (wc == 0x0000feff) |
| 50 | + { |
| 51 | + } |
| 52 | + else if (wc == 0xfffe0000u) |
| 53 | + { |
| 54 | + state ^= 1; |
| 55 | + } |
| 56 | + else |
| 57 | + { |
| 58 | + if (wc < 0x110000 && !(wc >= 0xd800 && wc < 0xe000)) |
| 59 | + { |
| 60 | + *pwc = wc; |
| 61 | + conv->ibyteorder = state; |
| 62 | + return count + 4; |
| 63 | + } |
| 64 | + else |
| 65 | + { |
| 66 | + conv->ibyteorder = state; |
| 67 | + return RET_SHIFT_ILSEQ(count); |
| 68 | + } |
| 69 | + } |
| 70 | + s += 4; |
| 71 | + n -= 4; |
| 72 | + count += 4; |
68 | 73 | } |
69 | | - s += 4; n -= 4; count += 4; |
70 | | - } |
71 | | - conv->istate = state; |
72 | | - return RET_TOOFEW(count); |
| 74 | + conv->ibyteorder = state; |
| 75 | + return RET_TOOFEW(count); |
73 | 76 | } |
74 | 77 |
|
75 | 78 | /* We output UTF-32 in big-endian order, with byte-order mark. */ |
76 | 79 | /* The state is 0 at the beginning, 1 after the BOM has been written. */ |
77 | | -static int |
78 | | -utf32_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, size_t n) |
| 80 | +static int utf32_wctomb(conv_t conv, unsigned char *r, ucs4_t wc, size_t n) |
79 | 81 | { |
80 | | - if (wc < 0x110000 && !(wc >= 0xd800 && wc < 0xe000)) { |
81 | | - int count = 0; |
82 | | - if (!conv->ostate) { |
83 | | - if (n >= 4) { |
84 | | - r[0] = 0x00; |
85 | | - r[1] = 0x00; |
86 | | - r[2] = 0xFE; |
87 | | - r[3] = 0xFF; |
88 | | - r += 4; n -= 4; count += 4; |
89 | | - } else |
90 | | - return RET_TOOSMALL; |
91 | | - } |
92 | | - if (wc < 0x110000) { |
93 | | - if (n >= 4) { |
94 | | - r[0] = 0; |
95 | | - r[1] = (unsigned char) (wc >> 16); |
96 | | - r[2] = (unsigned char) (wc >> 8); |
97 | | - r[3] = (unsigned char) wc; |
98 | | - conv->ostate = 1; |
99 | | - return count+4; |
100 | | - } else |
101 | | - return RET_TOOSMALL; |
| 82 | + if (wc < 0x110000 && !(wc >= 0xd800 && wc < 0xe000)) |
| 83 | + { |
| 84 | + int count = 0; |
| 85 | + if (!conv->ostate) |
| 86 | + { |
| 87 | + if (n >= 4) |
| 88 | + { |
| 89 | + r[0] = 0x00; |
| 90 | + r[1] = 0x00; |
| 91 | + r[2] = 0xFE; |
| 92 | + r[3] = 0xFF; |
| 93 | + r += 4; |
| 94 | + n -= 4; |
| 95 | + count += 4; |
| 96 | + } |
| 97 | + else |
| 98 | + return RET_TOOSMALL; |
| 99 | + } |
| 100 | + if (wc < 0x110000) |
| 101 | + { |
| 102 | + if (n >= 4) |
| 103 | + { |
| 104 | + r[0] = 0; |
| 105 | + r[1] = (unsigned char)(wc >> 16); |
| 106 | + r[2] = (unsigned char)(wc >> 8); |
| 107 | + r[3] = (unsigned char)wc; |
| 108 | + conv->ostate = 1; |
| 109 | + return count + 4; |
| 110 | + } |
| 111 | + else |
| 112 | + return RET_TOOSMALL; |
| 113 | + } |
102 | 114 | } |
103 | | - } |
104 | | - return RET_ILUNI; |
| 115 | + return RET_ILUNI; |
105 | 116 | } |
106 | 117 |
|
107 | 118 | #endif /* _UTF32_H_ */ |
0 commit comments