#include "wx/fontmap.h"
#ifdef __DARWIN__
-#include "wx/mac/corefoundation/private/strconv_cf.h"
+#include "wx/osx/core/private/strconv_cf.h"
#endif //def __DARWIN__
const char *src, size_t srcLen) const
{
// although new conversion classes are supposed to implement this function
- // directly, the existins ones only implement the old MB2WC() and so, to
+ // directly, the existing ones only implement the old MB2WC() and so, to
// avoid to have to rewrite all conversion classes at once, we provide a
// default (but not efficient) implementation of this one in terms of the
// old function by copying the input to ensure that it's NUL-terminated and
// then using MB2WC() to convert it
+ //
+ // moreover, some conversion classes simply can't implement ToWChar()
+ // directly, the primary example is wxConvLibc: mbstowcs() only handles
+ // NUL-terminated strings
// the number of chars [which would be] written to dst [if it were not NULL]
size_t dstWritten = 0;
srcEnd = NULL;
}
+ // the idea of this code is straightforward: it converts a NUL-terminated
+ // chunk of the string during each iteration and updates the output buffer
+ // with the result
+ //
+ // all the complication come from the fact that this function, for
+ // historical reasons, must behave in 2 subtly different ways when it's
+ // called with a fixed number of characters and when it's called for the
+ // entire NUL-terminated string: in the former case (srcEnd == NULL) we
+ // must count all characters we convert, NUL or not; but in the latter we
+ // do not count the trailing NUL -- but still count all the NULs inside the
+ // string
+ //
+ // so for the (simple) former case we just always count the trailing NUL,
+ // but for the latter we need to wait until we see if there is going to be
+ // another loop iteration and only count it then
for ( ;; )
{
// try to convert the current chunk
if ( lenChunk == wxCONV_FAILED )
return wxCONV_FAILED;
- lenChunk++; // for the L'\0' at the end of this chunk
-
dstWritten += lenChunk;
+ if ( !srcEnd )
+ dstWritten++;
- if ( lenChunk == 1 )
+ if ( !lenChunk )
{
// nothing left in the input string, conversion succeeded
break;
if ( dstWritten > dstLen )
return wxCONV_FAILED;
- if ( MB2WC(dst, src, lenChunk) == wxCONV_FAILED )
+ // +1 is for trailing NUL
+ if ( MB2WC(dst, src, lenChunk + 1) == wxCONV_FAILED )
return wxCONV_FAILED;
dst += lenChunk;
+ if ( !srcEnd )
+ dst++;
}
if ( !srcEnd )
// note that ">=" (and not just "==") is needed here as the terminator
// we skipped just above could be inside or just after the buffer
- // delimited by inEnd
+ // delimited by srcEnd
if ( src >= srcEnd )
break;
+
+ // if we got here then this wasn't the last chunk in this string and
+ // hence we must count an extra char for L'\0' even when converting a
+ // fixed number of characters
+ if ( srcEnd )
+ {
+ dstWritten++;
+ if ( dst )
+ dst++;
+ }
}
return dstWritten;
// the number of chars [which would be] written to dst [if it were not NULL]
size_t dstWritten = 0;
+ // if we don't know its length we have no choice but to assume that it is
+ // NUL-terminated (notice that it can still be NUL-terminated even if
+ // explicit length is given but it doesn't change our return value)
+ const bool isNulTerminated = srcLen == wxNO_LEN;
+
// make a copy of the input string unless it is already properly
// NUL-terminated
- //
- // if we don't know its length we have no choice but to assume that it is,
- // indeed, properly terminated
wxWCharBuffer bufTmp;
- if ( srcLen == wxNO_LEN )
+ if ( isNulTerminated )
{
srcLen = wxWcslen(src) + 1;
}
if ( lenChunk == wxCONV_FAILED )
return wxCONV_FAILED;
- lenChunk += lenNul;
dstWritten += lenChunk;
+ if ( isNulTerminated )
+ dstWritten += lenNul;
if ( dst )
{
if ( dstWritten > dstLen )
return wxCONV_FAILED;
- if ( WC2MB(dst, src, lenChunk) == wxCONV_FAILED )
+ if ( WC2MB(dst, src, lenChunk + lenNul) == wxCONV_FAILED )
return wxCONV_FAILED;
dst += lenChunk;
+ if ( isNulTerminated )
+ dst += lenNul;
}
}
// because we want the buffer to always be NUL-terminated, even if the
// input isn't (as otherwise the caller has no way to know its length)
wxWCharBuffer wbuf(dstLen);
- wbuf.data()[dstLen - 1] = L'\0';
+ wbuf.data()[dstLen] = L'\0';
if ( ToWChar(wbuf.data(), dstLen, inBuff, inLen) != wxCONV_FAILED )
{
if ( outLen )
{
*outLen = dstLen;
- if ( wbuf[dstLen - 1] == L'\0' )
+
+ // we also need to handle NUL-terminated input strings
+ // specially: for them the output is the length of the string
+ // excluding the trailing NUL, however if we're asked to
+ // convert a specific number of characters we return the length
+ // of the resulting output even if it's NUL-terminated
+ if ( inLen == wxNO_LEN )
(*outLen)--;
}
{
*outLen = dstLen;
- if ( dstLen >= nulLen &&
- !NotAllNULs(buf.data() + dstLen - nulLen, nulLen) )
+ if ( inLen == wxNO_LEN )
{
- // in this case the output is NUL-terminated and we're not
- // supposed to count NUL
+ // in this case both input and output are NUL-terminated
+ // and we're not supposed to count NUL
*outLen -= nulLen;
}
}
// ----------------------------------------------------------------------------
// Implementation (C) 2004 Fredrik Roubert
+//
+// Changes to work in streaming mode (C) 2008 Vadim Zeitlin
//
// BASE64 decoding table
0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
};
-size_t wxMBConvUTF7::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t wxMBConvUTF7::ToWChar(wchar_t *dst, size_t dstLen,
+ const char *src, size_t srcLen) const
{
+ DecoderState stateOrig,
+ *statePtr;
+ if ( srcLen == wxNO_LEN )
+ {
+ // convert the entire string, up to and including the trailing NUL
+ srcLen = strlen(src) + 1;
+
+ // when working on the entire strings we don't update nor use the shift
+ // state from the previous call
+ statePtr = &stateOrig;
+ }
+ else // when working with partial strings we do use the shift state
+ {
+ statePtr = wx_const_cast(DecoderState *, &m_stateDecoder);
+
+ // also save the old state to be able to rollback to it on error
+ stateOrig = m_stateDecoder;
+ }
+
+ // but to simplify the code below we use this variable in both cases
+ DecoderState& state = *statePtr;
+
+
+ // number of characters [which would have been] written to dst [if it were
+ // not NULL]
size_t len = 0;
- while ( *psz && (!buf || (len < n)) )
+ const char * const srcEnd = src + srcLen;
+
+ while ( (src < srcEnd) && (!dst || (len < dstLen)) )
{
- unsigned char cc = *psz++;
- if (cc != '+')
- {
- // plain ASCII char
- if (buf)
- *buf++ = cc;
- len++;
- }
- else if (*psz == '-')
- {
- // encoded plus sign
- if (buf)
- *buf++ = cc;
- len++;
- psz++;
- }
- else // start of BASE64 encoded string
+ const unsigned char cc = *src++;
+
+ if ( state.IsShifted() )
{
- bool lsb, ok;
- unsigned int d, l;
- for ( ok = lsb = false, d = 0, l = 0;
- (cc = utf7unb64[(unsigned char)*psz]) != 0xff;
- psz++ )
+ const unsigned char dc = utf7unb64[cc];
+ if ( dc == 0xff )
+ {
+ // end of encoded part, check that nothing was left: there can
+ // be up to 4 bits of 0 padding but nothing else (we also need
+ // to check isLSB as we count bits modulo 8 while a valid UTF-7
+ // encoded sequence must contain an integral number of UTF-16
+ // characters)
+ if ( state.isLSB || state.bit > 4 ||
+ (state.accum & ((1 << state.bit) - 1)) )
+ {
+ if ( !len )
+ state = stateOrig;
+
+ return wxCONV_FAILED;
+ }
+
+ state.ToDirect();
+
+ // re-parse this character normally below unless it's '-' which
+ // is consumed by the decoder
+ if ( cc == '-' )
+ continue;
+ }
+ else // valid encoded character
{
- d <<= 6;
- d += cc;
- for (l += 6; l >= 8; lsb = !lsb)
+ // mini base64 decoder: each character is 6 bits
+ state.bit += 6;
+ state.accum <<= 6;
+ state.accum += dc;
+
+ if ( state.bit >= 8 )
{
- unsigned char c = (unsigned char)((d >> (l -= 8)) % 256);
- if (lsb)
+ // got the full byte, consume it
+ state.bit -= 8;
+ unsigned char b = (state.accum >> state.bit) & 0x00ff;
+
+ if ( state.isLSB )
{
- if (buf)
- *buf++ |= c;
- len ++;
- ok = true;
+ // we've got the full word, output it
+ if ( dst )
+ *dst++ = (state.msb << 8) | b;
+ len++;
+ state.isLSB = false;
}
- else
+ else // MSB
{
- if (buf)
- *buf = (wchar_t)(c << 8);
+ // just store it while we wait for LSB
+ state.msb = b;
+ state.isLSB = true;
}
}
}
+ }
- if ( !ok )
+ if ( state.IsDirect() )
+ {
+ // start of an encoded segment?
+ if ( cc == '+' )
{
- // in valid UTF7 we should have valid characters after '+'
- return wxCONV_FAILED;
+ if ( *src == '-' )
+ {
+ // just the encoded plus sign, don't switch to shifted mode
+ if ( dst )
+ *dst++ = '+';
+ len++;
+ src++;
+ }
+ else if ( utf7unb64[(unsigned)*src] == 0xff )
+ {
+ // empty encoded chunks are not allowed
+ if ( !len )
+ state = stateOrig;
+
+ return wxCONV_FAILED;
+ }
+ else // base-64 encoded chunk follows
+ {
+ state.ToShifted();
+ }
}
+ else // not '+'
+ {
+ // only printable 7 bit ASCII characters (with the exception of
+ // NUL, TAB, CR and LF) can be used directly
+ if ( cc >= 0x7f || (cc < ' ' &&
+ !(cc == '\0' || cc == '\t' || cc == '\r' || cc == '\n')) )
+ return wxCONV_FAILED;
- if (*psz == '-')
- psz++;
+ if ( dst )
+ *dst++ = cc;
+ len++;
+ }
}
}
- if ( buf && (len < n) )
- *buf = '\0';
+ if ( !len )
+ {
+ // as we didn't read any characters we should be called with the same
+ // data (followed by some more new data) again later so don't save our
+ // state
+ state = stateOrig;
+
+ return wxCONV_FAILED;
+ }
return len;
}
//
static const unsigned char utf7encode[128] =
{
- 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3,
+ 0, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3,
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
2, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 3, 0, 0, 0, 3,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 3, 3
};
-size_t wxMBConvUTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const
+static inline bool wxIsUTF7Direct(wchar_t wc)
{
+ return wc < 0x80 && utf7encode[wc] < 1;
+}
+
+size_t wxMBConvUTF7::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
+{
+ EncoderState stateOrig,
+ *statePtr;
+ if ( srcLen == wxNO_LEN )
+ {
+ // we don't apply the stored state when operating on entire strings at
+ // once
+ statePtr = &stateOrig;
+
+ srcLen = wxWcslen(src) + 1;
+ }
+ else // do use the mode we left the output in previously
+ {
+ stateOrig = m_stateEncoder;
+ statePtr = wx_const_cast(EncoderState *, &m_stateEncoder);
+ }
+
+ EncoderState& state = *statePtr;
+
+
size_t len = 0;
- while (*psz && ((!buf) || (len < n)))
+ const wchar_t * const srcEnd = src + srcLen;
+ while ( src < srcEnd && (!dst || len < dstLen) )
{
- wchar_t cc = *psz++;
- if (cc < 0x80 && utf7encode[cc] < 1)
+ wchar_t cc = *src++;
+ if ( wxIsUTF7Direct(cc) )
{
- // plain ASCII char
- if (buf)
- *buf++ = (char)cc;
+ if ( state.IsShifted() )
+ {
+ // pad with zeros the last encoded block if necessary
+ if ( state.bit )
+ {
+ if ( dst )
+ *dst++ = utf7enb64[((state.accum % 16) << (6 - state.bit)) % 64];
+ len++;
+ }
+
+ state.ToDirect();
+ if ( dst )
+ *dst++ = '-';
+ len++;
+ }
+
+ if ( dst )
+ *dst++ = (char)cc;
len++;
}
+ else if ( cc == '+' && state.IsDirect() )
+ {
+ if ( dst )
+ {
+ *dst++ = '+';
+ *dst++ = '-';
+ }
+
+ len += 2;
+ }
#ifndef WC_UTF16
else if (((wxUint32)cc) > 0xffff)
{
#endif
else
{
- if (buf)
- *buf++ = '+';
+ if ( state.IsDirect() )
+ {
+ state.ToShifted();
- len++;
- if (cc != '+')
+ if ( dst )
+ *dst++ = '+';
+ len++;
+ }
+
+ // BASE64 encode string
+ for ( ;; )
{
- // BASE64 encode string
- unsigned int lsb, d, l;
- for (d = 0, l = 0; /*nothing*/; psz++)
+ for ( unsigned lsb = 0; lsb < 2; lsb++ )
{
- for (lsb = 0; lsb < 2; lsb ++)
- {
- d <<= 8;
- d += lsb ? cc & 0xff : (cc & 0xff00) >> 8;
+ state.accum <<= 8;
+ state.accum += lsb ? cc & 0xff : (cc & 0xff00) >> 8;
- for (l += 8; l >= 6; )
- {
- l -= 6;
- if (buf)
- *buf++ = utf7enb64[(d >> l) % 64];
- len++;
- }
+ for (state.bit += 8; state.bit >= 6; )
+ {
+ state.bit -= 6;
+ if ( dst )
+ *dst++ = utf7enb64[(state.accum >> state.bit) % 64];
+ len++;
}
-
- cc = *psz;
- if (!(cc) || (cc < 0x80 && utf7encode[cc] < 1))
- break;
}
- if (l != 0)
- {
- if (buf)
- *buf++ = utf7enb64[((d % 16) << (6 - l)) % 64];
+ if ( src == srcEnd || wxIsUTF7Direct(cc = *src) )
+ break;
- len++;
- }
+ src++;
}
-
- if (buf)
- *buf++ = '-';
- len++;
}
}
- if (buf && (len < n))
- *buf = 0;
+ // we need to restore the original encoder state if we were called just to
+ // calculate the amount of space needed as we will presumably be called
+ // again to really convert the data now
+ if ( !dst )
+ state = stateOrig;
return len;
}
for ( const wchar_t *wp = src; ; wp++ )
{
- if ( !(srcLen == wxNO_LEN ? *wp : srcLen--) )
+ if ( !(srcLen == wxNO_LEN ? *wp : srcLen) )
{
// all done successfully, just add the trailing NULL if we are not
// using explicit length
return written;
}
+ if ( srcLen != wxNO_LEN )
+ srcLen--;
wxUint32 code;
#ifdef WC_UTF16
else // no destination buffer
{
// convert using temp buffer to calculate the size of the buffer needed
- wchar_t tbuf[8];
+ wchar_t tbuf[256];
res = 0;
do
if (ms_wcNeedsSwap)
{
// need to copy to temp buffer to switch endianness
- // (doing WC_BSWAP twice on the original buffer won't help, as it
+ // (doing WC_BSWAP twice on the original buffer won't work, as it
// could be in read-only memory, or be accessed in some other thread)
- tmpbuf = (wchar_t *)malloc(inbuflen + SIZEOF_WCHAR_T);
+ tmpbuf = (wchar_t *)malloc(inbuflen);
for ( size_t i = 0; i < srcLen; i++ )
tmpbuf[i] = WC_BSWAP(src[i]);
- tmpbuf[srcLen] = L'\0';
src = tmpbuf;
}
else // no destination buffer
{
// convert using temp buffer to calculate the size of the buffer needed
- char tbuf[16];
+ char tbuf[256];
res = 0;
do
{
dst = tbuf;
- outbuflen = 16;
+ outbuflen = WXSIZEOF(tbuf);
cres = iconv(w2m, ICONV_CHAR_CAST(&inbuf), &inbuflen, &dst, &outbuflen);
- res += 16 - outbuflen;
+ res += WXSIZEOF(tbuf) - outbuflen;
}
while ((cres == (size_t)-1) && (errno == E2BIG));
}
#if wxUSE_FONTMAP
m_encoding = wxFontMapperBase::GetEncodingFromName(charset);
+ if ( m_encoding == wxFONTENCODING_MAX )
+ {
+ // set to unknown/invalid value
+ m_encoding = wxFONTENCODING_SYSTEM;
+ }
+ else if ( m_encoding == wxFONTENCODING_DEFAULT )
+ {
+ // wxFONTENCODING_DEFAULT is same as US-ASCII in this context
+ m_encoding = wxFONTENCODING_ISO8859_1;
+ }
#else
m_encoding = wxFONTENCODING_SYSTEM;
#endif
return m_convReal->ToWChar(dst, dstLen, src, srcLen);
// latin-1 (direct)
- return wxMBConv::ToWChar(dst, dstLen, src, srcLen);
-}
+ if ( srcLen == wxNO_LEN )
+ srcLen = strlen(src) + 1; // take trailing NUL too
-size_t wxCSConv::FromWChar(char *dst, size_t dstLen,
- const wchar_t *src, size_t srcLen) const
-{
- CreateConvIfNeeded();
+ if ( dst )
+ {
+ if ( dstLen < srcLen )
+ return wxCONV_FAILED;
- if (m_convReal)
- return m_convReal->FromWChar(dst, dstLen, src, srcLen);
+ for ( size_t n = 0; n < srcLen; n++ )
+ dst[n] = (unsigned char)(src[n]);
+ }
- // latin-1 (direct)
- return wxMBConv::FromWChar(dst, dstLen, src, srcLen);
+ return srcLen;
}
-size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
+size_t wxCSConv::FromWChar(char *dst, size_t dstLen,
+ const wchar_t *src, size_t srcLen) const
{
CreateConvIfNeeded();
if (m_convReal)
- return m_convReal->MB2WC(buf, psz, n);
+ return m_convReal->FromWChar(dst, dstLen, src, srcLen);
// latin-1 (direct)
- size_t len = strlen(psz);
+ if ( srcLen == wxNO_LEN )
+ srcLen = wxWcslen(src) + 1;
- if (buf)
+ if ( dst )
{
- for (size_t c = 0; c <= len; c++)
- buf[c] = (unsigned char)(psz[c]);
- }
-
- return len;
-}
-
-size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
-{
- CreateConvIfNeeded();
-
- if (m_convReal)
- return m_convReal->WC2MB(buf, psz, n);
+ if ( dstLen < srcLen )
+ return wxCONV_FAILED;
- // latin-1 (direct)
- const size_t len = wxWcslen(psz);
- if (buf)
- {
- for (size_t c = 0; c <= len; c++)
+ for ( size_t n = 0; n < srcLen; n++ )
{
- if (psz[c] > 0xFF)
+ if ( src[n] > 0xFF )
return wxCONV_FAILED;
- buf[c] = (char)psz[c];
+ dst[n] = (char)src[n];
}
+
}
- else
+ else // still need to check the input validity
{
- for (size_t c = 0; c <= len; c++)
+ for ( size_t n = 0; n < srcLen; n++ )
{
- if (psz[c] > 0xFF)
+ if ( src[n] > 0xFF )
return wxCONV_FAILED;
}
}
- return len;
+ return srcLen;
}
size_t wxCSConv::GetMBNulLen() const
projects / wxWidgets.git / blobdiff