diff options
Diffstat (limited to 'lib/unistr/u8-mbtouc-aux.c')
| -rw-r--r-- | lib/unistr/u8-mbtouc-aux.c | 108 |
1 files changed, 73 insertions, 35 deletions
diff --git a/lib/unistr/u8-mbtouc-aux.c b/lib/unistr/u8-mbtouc-aux.c index 018cf507..22ed807b 100644 --- a/lib/unistr/u8-mbtouc-aux.c +++ b/lib/unistr/u8-mbtouc-aux.c @@ -1,5 +1,5 @@ /* Conversion UTF-8 to UCS-4. - Copyright (C) 2001-2002, 2006-2007, 2009-2022 Free Software Foundation, Inc. + Copyright (C) 2001-2002, 2006-2007, 2009-2024 Free Software Foundation, Inc. Written by Bruno Haible <bruno@clisp.org>, 2001. This file is free software: you can redistribute it and/or modify @@ -52,20 +52,15 @@ u8_mbtouc_aux (ucs4_t *puc, const uint8_t *s, size_t n) { if (n >= 3) { - if ((s[1] ^ 0x80) < 0x40) + if ((s[1] ^ 0x80) < 0x40 + && (c >= 0xe1 || s[1] >= 0xa0) + && (c != 0xed || s[1] < 0xa0)) { if ((s[2] ^ 0x80) < 0x40) { - if ((c >= 0xe1 || s[1] >= 0xa0) - && (c != 0xed || s[1] < 0xa0)) - { - *puc = ((unsigned int) (c & 0x0f) << 12) - | ((unsigned int) (s[1] ^ 0x80) << 6) - | (unsigned int) (s[2] ^ 0x80); - return 3; - } - /* invalid multibyte character */ - *puc = 0xfffd; + *puc = ((unsigned int) (c & 0x0f) << 12) + | ((unsigned int) (s[1] ^ 0x80) << 6) + | (unsigned int) (s[2] ^ 0x80); return 3; } /* invalid multibyte character */ @@ -73,38 +68,50 @@ u8_mbtouc_aux (ucs4_t *puc, const uint8_t *s, size_t n) return 2; } /* invalid multibyte character */ + *puc = 0xfffd; + return 1; } else { - /* incomplete multibyte character */ *puc = 0xfffd; - if (n == 1 || (s[1] ^ 0x80) >= 0x40) - return 1; + if (n == 1) + { + /* incomplete multibyte character */ + return 1; + } else - return 2; + { + if ((s[1] ^ 0x80) < 0x40 + && (c >= 0xe1 || s[1] >= 0xa0) + && (c != 0xed || s[1] < 0xa0)) + { + /* incomplete multibyte character */ + return 2; + } + else + { + /* invalid multibyte character */ + return 1; + } + } } } - else if (c < 0xf8) + else if (c <= 0xf4) { if (n >= 4) { - if ((s[1] ^ 0x80) < 0x40) + if ((s[1] ^ 0x80) < 0x40 + && (c >= 0xf1 || s[1] >= 0x90) + && (c < 0xf4 || (/* c == 0xf4 && */ s[1] < 0x90))) { if ((s[2] ^ 0x80) < 0x40) { if ((s[3] ^ 0x80) < 0x40) { - if ((c >= 0xf1 || s[1] >= 0x90) - && (c < 0xf4 || (c == 0xf4 && s[1] < 0x90))) - { - *puc = ((unsigned int) (c & 0x07) << 18) - | ((unsigned int) (s[1] ^ 0x80) << 12) - | ((unsigned int) (s[2] ^ 0x80) << 6) - | (unsigned int) (s[3] ^ 0x80); - return 4; - } - /* invalid multibyte character */ - *puc = 0xfffd; + *puc = ((unsigned int) (c & 0x07) << 18) + | ((unsigned int) (s[1] ^ 0x80) << 12) + | ((unsigned int) (s[2] ^ 0x80) << 6) + | (unsigned int) (s[3] ^ 0x80); return 4; } /* invalid multibyte character */ @@ -116,17 +123,48 @@ u8_mbtouc_aux (ucs4_t *puc, const uint8_t *s, size_t n) return 2; } /* invalid multibyte character */ + *puc = 0xfffd; + return 1; } else { - /* incomplete multibyte character */ *puc = 0xfffd; - if (n == 1 || (s[1] ^ 0x80) >= 0x40) - return 1; - else if (n == 2 || (s[2] ^ 0x80) >= 0x40) - return 2; + if (n == 1) + { + /* incomplete multibyte character */ + return 1; + } else - return 3; + { + if ((s[1] ^ 0x80) < 0x40 + && (c >= 0xf1 || s[1] >= 0x90) + && (c < 0xf4 || (/* c == 0xf4 && */ s[1] < 0x90))) + { + if (n == 2) + { + /* incomplete multibyte character */ + return 2; + } + else + { + if ((s[2] ^ 0x80) < 0x40) + { + /* incomplete multibyte character */ + return 3; + } + else + { + /* invalid multibyte character */ + return 2; + } + } + } + else + { + /* invalid multibyte character */ + return 1; + } + } } } } |