// construction/destruction
// ----------------------------------------------------------------------------
+// TODO: Do we want to include the null char in the stream? If so then
+// just add +1 to m_len in the ctor
wxStringInputStream::wxStringInputStream(const wxString& s)
#if wxUSE_UNICODE
- : m_str(s), m_buf(wxMBConvUTF8().cWX2MB(s).release()), m_len(strlen(m_buf))
+ // FIXME-UTF8: use wxCharBufferWithLength if we have it
+ : m_str(s), m_buf(s.utf8_str()), m_len(strlen(m_buf))
#else
- : m_str(s), m_buf((char*)s.c_str()), m_len(s.length())
+ : m_str(s), m_buf(s.mb_str()), m_len(s.length())
#endif
{
#if wxUSE_UNICODE
- wxASSERT_MSG(m_buf != NULL, _T("Could not convert string to UTF8!"));
+ wxASSERT_MSG(m_buf.data() != NULL, _T("Could not convert string to UTF8!"));
#endif
m_pos = 0;
}
-wxStringInputStream::~wxStringInputStream()
-{
-#if wxUSE_UNICODE
- delete[] m_buf;
-#endif
-}
-
// ----------------------------------------------------------------------------
// getlength
// ----------------------------------------------------------------------------
return wxInvalidOffset;
}
- if ( ofs < 0 || wx_static_cast(size_t, ofs) >= m_len )
+ if ( ofs < 0 || ofs > wx_static_cast(wxFileOffset, m_len) )
return wxInvalidOffset;
- m_pos = wx_static_cast(size_t, ofs);
+ // FIXME: this can't be right
+ m_pos = wx_truncate_cast(size_t, ofs);
return ofs;
}
size = sizeMax;
}
- memcpy(buffer, m_buf + m_pos, size);
+ memcpy(buffer, m_buf.data() + m_pos, size);
m_pos += size;
return size;
{
const char *p = wx_static_cast(const char *, buffer);
- // append the input buffer (may not be null terminated - thus
- // the literal length
- m_str->Append(wxString(p, m_conv, size));
+#if wxUSE_UNICODE_WCHAR
+ // the part of the string we have here may be incomplete, i.e. it can stop
+ // in the middle of an UTF-8 character and so converting it would fail; if
+ // this is the case, accumulate the part which we failed to convert until
+ // we get the rest (and also take into account the part which we might have
+ // left unconverted before)
+ const char *src;
+ size_t srcLen;
+ if ( m_unconv.GetDataLen() )
+ {
+ // append the new data to the data remaining since the last time
+ m_unconv.AppendData(p, size);
+ src = m_unconv;
+ srcLen = m_unconv.GetDataLen();
+ }
+ else // no unconverted data left, avoid extra copy
+ {
+ src = p;
+ srcLen = size;
+ }
- // return number of bytes actually written
+ size_t wlen;
+ wxWCharBuffer wbuf(m_conv.cMB2WC(src, srcLen, &wlen));
+ if ( wbuf )
+ {
+ // conversion succeeded, clear the unconverted buffer
+ m_unconv = wxMemoryBuffer(0);
+
+ m_str->append(wbuf, wlen);
+ }
+ else // conversion failed
+ {
+ // remember unconverted data if there had been none before (otherwise
+ // we've already got it in the buffer)
+ if ( src == p )
+ m_unconv.AppendData(src, srcLen);
+
+ // pretend that we wrote the data anyhow, otherwise the caller would
+ // believe there was an error and this might not be the case, but do
+ // not update m_pos as m_str hasn't changed
+ return size;
+ }
+#else // !wxUSE_UNICODE_WCHAR
+ // no recoding necessary, the data is supposed to already be in UTF-8 (if
+ // supported) or ASCII otherwise
+ m_str->append(p, size);
+#endif // wxUSE_UNICODE_WCHAR/!wxUSE_UNICODE_WCHAR
+
+ // update position
m_pos += size;
+ // return number of bytes actually written
return size;
}