diff options
author | Corinna Vinschen <corinna@vinschen.de> | 2009-09-23 15:31:00 +0400 |
---|---|---|
committer | Corinna Vinschen <corinna@vinschen.de> | 2009-09-23 15:31:00 +0400 |
commit | e664429d770cb56bc6c2a8d07eeb5114748c22eb (patch) | |
tree | ab353e3265fdebf6293019638d63e62f68971b7a /winsup/cygwin/strfuncs.cc | |
parent | 70ecec052a16f4c9504442b25d1ea39cfb8e6ac9 (diff) |
* strfuncs.cc (sys_cp_wcstombs): Convert lone surrogate pair
second halves to unambiguous ASCII SO sequence. When converting
chars invalid in current codepage to ASCII SO sequence, make
sure to check for surrogate pair second half only if ct least
one wide characters is left. Decrement nwc if valid second half has
been converted.
(sys_cp_mbstowcs): Improve ASCII SO handling. Never break from loop
if invalid character has been found. Recognize ASCII SO sequence
representing originally invalid mulitbyte char converted into a
lone surrogate pair second half. Convert accordingly.
Diffstat (limited to 'winsup/cygwin/strfuncs.cc')
-rw-r--r-- | winsup/cygwin/strfuncs.cc | 95 |
1 files changed, 60 insertions, 35 deletions
diff --git a/winsup/cygwin/strfuncs.cc b/winsup/cygwin/strfuncs.cc index 88bbf767c..d4dbdaf55 100644 --- a/winsup/cygwin/strfuncs.cc +++ b/winsup/cygwin/strfuncs.cc @@ -435,8 +435,10 @@ sys_cp_wcstombs (wctomb_p f_wctomb, char *charset, char *dst, size_t len, surrogate pair in the 0xDCxx range specifying an invalid byte value when converting from MB to WC. The comment in sys_cp_mbstowcs below explains it. */ - buf[0] = (char) (pw & 0xff); - bytes = 1; + buf[0] = 0x0e; /* ASCII SO */ + buf[1] = 0xff; + buf[2] = (char) (pw & 0xff); + bytes = 3; } else if (bytes == -1 && *charset != 'U'/*TF-8*/) { @@ -451,7 +453,7 @@ sys_cp_wcstombs (wctomb_p f_wctomb, char *charset, char *dst, size_t len, continue; } ++bytes; /* Add the ASCII SO to the byte count. */ - if (ps.__count == -4) /* First half of a surrogate pair. */ + if (ps.__count == -4 && nwc > 0) /* First half of a surrogate pair. */ { ++pwcs; if ((*pwcs & 0xfc00) != 0xdc00) /* Invalid second half. */ @@ -461,6 +463,7 @@ sys_cp_wcstombs (wctomb_p f_wctomb, char *charset, char *dst, size_t len, continue; } bytes += __utf8_wctomb (_REENT, buf + bytes, *pwcs, charset, &ps); + nwc--; } } if (n + bytes <= len) @@ -546,42 +549,64 @@ sys_cp_mbstowcs (mbtowc_p f_mbtowc, char *charset, wchar_t *dst, size_t dlen, len = (size_t)-1; while (len > 0 && nms > 0) { - /* ASCII SO. Sanity check: If this is a lead SO byte for a following - UTF-8 sequence, there must be at least two more bytes left, and the - next byte must be a valid UTF-8 start byte. If the charset isn't - UTF-8 anyway, try to convert the following bytes as UTF-8 sequence. */ - if (*pmbs == 0x0e && nms > 2 && pmbs[1] >= 0xc2 - && pmbs[1] <= 0xf4 && *charset != 'U'/*TF-8*/) + /* ASCII SO handling. */ + if (*pmbs == 0x0e) { - pmbs++; - --nms; - bytes = __utf8_mbtowc (_REENT, ptr, (const char *) pmbs, nms, - charset, &ps); - if (bytes < 0) + /* Sanity check: If this is a lead SO byte for a following UTF-8 + sequence, there must be at least two more bytes left, and the + next byte must be a valid UTF-8 start byte. If the charset + isn't UTF-8 anyway, try to convert the following bytes as UTF-8 + sequence. */ + if (nms > 2 && pmbs[1] >= 0xc2 && pmbs[1] <= 0xf4 && *charset != 'U'/*TF-8*/) { - /* Invalid UTF-8 sequence? Treat the ASCII SO character as - stand-alone ASCII SO char. */ - bytes = 1; + bytes = __utf8_mbtowc (_REENT, ptr, (const char *) pmbs + 1, + nms - 1, charset, &ps); + if (bytes < 0) + { + /* Invalid UTF-8 sequence? Treat the ASCII SO character as + stand-alone ASCII SO char. */ + bytes = 1; + if (dst) + *ptr = 0x0e; + memset (&ps, 0, sizeof ps); + } + else + { + ++bytes; /* Count SO byte */ + if (bytes > 1 && ps.__count == 4) + { + /* First half of a surrogate. */ + wchar_t *ptr2 = dst ? ptr + 1 : NULL; + int bytes2 = __utf8_mbtowc (_REENT, ptr2, + (const char *) pmbs + bytes, + nms - bytes, charset, &ps); + if (bytes2 < 0) + memset (&ps, 0, sizeof ps); + else + { + bytes += bytes2; + ++count; + ptr = dst ? ptr + 1 : NULL; + --len; + } + } + } + } + /* Sequence for an invalid byte originally created in the next outer + else branch below. This must be converted back to a 0xDCxx value + as well. */ + else if (nms > 2 && pmbs[1] == 0xff) + { + bytes = 3; if (dst) - *ptr = 0x0e; - memset (&ps, 0, sizeof ps); - break; + *ptr = L'\xdc80' | pmbs[2]; } - if (bytes == 0) - break; - if (ps.__count == 4) /* First half of a surrogate. */ + /* Otherwise it's just a simple ASCII SO. */ + else { - wchar_t *ptr2 = dst ? ptr + 1 : NULL; - int bytes2 = __utf8_mbtowc (_REENT, ptr2, - (const char *) pmbs + bytes, - nms - bytes, charset, &ps); - if (bytes2 < 0) - break; - pmbs += bytes2; - nms -= bytes2; - ++count; - ptr = dst ? ptr + 1 : NULL; - --len; + bytes = 1; + if (dst) + *ptr = 0x0e; } } else if ((bytes = f_mbtowc (_REENT, ptr, (const char *) pmbs, nms, @@ -598,10 +623,10 @@ sys_cp_mbstowcs (mbtowc_p f_mbtowc, char *charset, wchar_t *dst, size_t dlen, characters converted to this format. It does allow processing of src to continue, however, which, since there is no way to signal decoding errors, seems like the best we can do. */ + bytes = 1; if (dst) *ptr = L'\xdc80' | *pmbs; memset (&ps, 0, sizeof ps); - bytes = 1; } if (bytes > 0) |