diff options
Diffstat (limited to 'lib/unistr/u8-mbtouc.c')
| -rw-r--r-- | lib/unistr/u8-mbtouc.c | 166 |
1 files changed, 124 insertions, 42 deletions
diff --git a/lib/unistr/u8-mbtouc.c b/lib/unistr/u8-mbtouc.c index 96cd5b74..dc4607f6 100644 --- a/lib/unistr/u8-mbtouc.c +++ b/lib/unistr/u8-mbtouc.c @@ -1,5 +1,5 @@ /* Look at first character in UTF-8 string. - Copyright (C) 1999-2002, 2006-2007, 2009-2010 Free Software Foundation, Inc. + Copyright (C) 1999-2002, 2006-2007, 2009-2015 Free Software Foundation, Inc. Written by Bruno Haible <bruno@clisp.org>, 2001. This program is free software: you can redistribute it and/or modify it @@ -55,21 +55,32 @@ u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n) { /* incomplete multibyte character */ *puc = 0xfffd; - return n; + return 1; } } else if (c < 0xf0) { if (n >= 3) { - if ((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40 - && (c >= 0xe1 || s[1] >= 0xa0) - && (c != 0xed || s[1] < 0xa0)) + if ((s[1] ^ 0x80) < 0x40) { - *puc = ((unsigned int) (c & 0x0f) << 12) - | ((unsigned int) (s[1] ^ 0x80) << 6) - | (unsigned int) (s[2] ^ 0x80); - return 3; + 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; + return 3; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 2; } /* invalid multibyte character */ } @@ -77,26 +88,45 @@ u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n) { /* incomplete multibyte character */ *puc = 0xfffd; - return n; + if (n == 1 || (s[1] ^ 0x80) >= 0x40) + return 1; + else + return 2; } } else if (c < 0xf8) { if (n >= 4) { - if ((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40 - && (s[3] ^ 0x80) < 0x40 - && (c >= 0xf1 || s[1] >= 0x90) + if ((s[1] ^ 0x80) < 0x40) + { + if ((s[2] ^ 0x80) < 0x40) + { + if ((s[3] ^ 0x80) < 0x40) + { + if ((c >= 0xf1 || s[1] >= 0x90) #if 1 - && (c < 0xf4 || (c == 0xf4 && s[1] < 0x90)) + && (c < 0xf4 || (c == 0xf4 && s[1] < 0x90)) #endif - ) - { - *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; + ) + { + *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; + return 4; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 3; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 2; } /* invalid multibyte character */ } @@ -104,7 +134,12 @@ u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n) { /* incomplete multibyte character */ *puc = 0xfffd; - return n; + if (n == 1 || (s[1] ^ 0x80) >= 0x40) + return 1; + else if (n == 2 || (s[2] ^ 0x80) >= 0x40) + return 2; + else + return 3; } } #if 0 @@ -112,16 +147,37 @@ u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n) { if (n >= 5) { - if ((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40 - && (s[3] ^ 0x80) < 0x40 && (s[4] ^ 0x80) < 0x40 - && (c >= 0xf9 || s[1] >= 0x88)) + if ((s[1] ^ 0x80) < 0x40) { - *puc = ((unsigned int) (c & 0x03) << 24) - | ((unsigned int) (s[1] ^ 0x80) << 18) - | ((unsigned int) (s[2] ^ 0x80) << 12) - | ((unsigned int) (s[3] ^ 0x80) << 6) - | (unsigned int) (s[4] ^ 0x80); - return 5; + if ((s[2] ^ 0x80) < 0x40) + { + if ((s[3] ^ 0x80) < 0x40) + { + if ((s[4] ^ 0x80) < 0x40) + { + if (c >= 0xf9 || s[1] >= 0x88) + { + *puc = ((unsigned int) (c & 0x03) << 24) + | ((unsigned int) (s[1] ^ 0x80) << 18) + | ((unsigned int) (s[2] ^ 0x80) << 12) + | ((unsigned int) (s[3] ^ 0x80) << 6) + | (unsigned int) (s[4] ^ 0x80); + return 5; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 5; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 4; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 3; + } + /* invalid multibyte character */ + return 2; } /* invalid multibyte character */ } @@ -136,18 +192,44 @@ u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n) { if (n >= 6) { - if ((s[1] ^ 0x80) < 0x40 && (s[2] ^ 0x80) < 0x40 - && (s[3] ^ 0x80) < 0x40 && (s[4] ^ 0x80) < 0x40 - && (s[5] ^ 0x80) < 0x40 - && (c >= 0xfd || s[1] >= 0x84)) + if ((s[1] ^ 0x80) < 0x40) { - *puc = ((unsigned int) (c & 0x01) << 30) - | ((unsigned int) (s[1] ^ 0x80) << 24) - | ((unsigned int) (s[2] ^ 0x80) << 18) - | ((unsigned int) (s[3] ^ 0x80) << 12) - | ((unsigned int) (s[4] ^ 0x80) << 6) - | (unsigned int) (s[5] ^ 0x80); - return 6; + if ((s[2] ^ 0x80) < 0x40) + { + if ((s[3] ^ 0x80) < 0x40) + { + if ((s[4] ^ 0x80) < 0x40) + { + if ((s[5] ^ 0x80) < 0x40) + { + if (c >= 0xfd || s[1] >= 0x84) + { + *puc = ((unsigned int) (c & 0x01) << 30) + | ((unsigned int) (s[1] ^ 0x80) << 24) + | ((unsigned int) (s[2] ^ 0x80) << 18) + | ((unsigned int) (s[3] ^ 0x80) << 12) + | ((unsigned int) (s[4] ^ 0x80) << 6) + | (unsigned int) (s[5] ^ 0x80); + return 6; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 6; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 5; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 4; + } + /* invalid multibyte character */ + *puc = 0xfffd; + return 3; + } + /* invalid multibyte character */ + return 2; } /* invalid multibyte character */ } |