void VObject::fromNativeEncoding()
{
bool is_30 = !wcscmp(getVersion(), TEXT("3.0"));
for (int index = propertiesCount() - 1; index >= 0; index--) {
VProperty *vprop = getProperty(index);
if (vprop->equalsEncoding(TEXT("QUOTED-PRINTABLE"))) {
// remove this, we cannot recreate it
vprop->removeParameter(TEXT("ENCODING"));
}
WCHAR *native = vprop->getValue();
// in the worst case every comma/linebreak is replaced with
// two characters and each \n with =0D=0A
WCHAR *foreign = new WCHAR[6 * wcslen(native) + 1];
WCHAR curr;
int in = 0, out = 0;
// line break is encoded with either one or two
// characters on different platforms
const int linebreaklen = wcslen(SYNC4J_LINEBREAK);
// use backslash for special characters,
// if necessary do quoted-printable encoding
bool doquoted = !is_30 &&
wcsstr(native, SYNC4J_LINEBREAK) != NULL;
while ((curr = native[in]) != 0) {
in++;
switch (curr) {
case ',':
if (!is_30) {
// normal character
foreign[out] = curr;
out++;
break;
}
// no break!
case ';':
case '\\':
foreign[out] = '\\';
out++;
foreign[out] = curr;
out++;
break;
case SEMICOLON_REPLACEMENT:
foreign[out] = ';';
out++;
break;
default:
if (doquoted &&
(curr == '=' || (unsigned char)curr >= 128)) {
// escape = and non-ASCII characters
swprintf(foreign + out, 4, TEXT("=%02X"), (unsigned int)(unsigned char)curr);
out += 3;
} else if (!wcsncmp(native + in - 1,
SYNC4J_LINEBREAK,
linebreaklen)) {
// line break
if (is_30) {
foreign[out] = '\\';
out++;
foreign[out] = 'n';
out++;
} else {
wcscpy(foreign + out, TEXT("=0D=0A"));
out += 6;
}
in += linebreaklen - 1;
} else {
foreign[out] = curr;
out++;
}
break;
}
}
foreign[out] = 0;
vprop->setValue(foreign);
delete [] foreign;
if (doquoted) {
// we have used quoted-printable encoding
vprop->addParameter(TEXT("ENCODING"), TEXT("QUOTED-PRINTABLE"));
}
}
}
void VObject::toNativeEncoding()
{
bool is_30 = !wcscmp(getVersion(), TEXT("3.0"));
// line break is encoded with either one or two
// characters on different platforms
const int linebreaklen = wcslen(SYNC4J_LINEBREAK);
for (int index = propertiesCount() - 1; index >= 0; index--) {
VProperty *vprop = getProperty(index);
WCHAR *foreign = vprop->getValue();
// the native encoding is always shorter than the foreign one
WCHAR *native = new WCHAR[wcslen(foreign) + 1];
if (vprop->equalsEncoding(TEXT("QUOTED-PRINTABLE"))) {
int in = 0, out = 0;
WCHAR curr;
// this is a very crude quoted-printable decoder,
// based on Wikipedia's explanation of quoted-printable
while ((curr = foreign[in]) != 0) {
in++;
if (curr == '=') {
WCHAR values[2];
values[0] = foreign[in];
in++;
if (!values[0]) {
// incomplete?!
break;
}
values[1] = foreign[in];
in++;
if (values[0] == '\r' && values[1] == '\n') {
// soft line break, ignore it
} else {
native[out] = (hex2int(values[0]) << 4) |
hex2int(values[1]);
out++;
// replace \r\n with \n?
if ( linebreaklen == 1 &&
out >= 2 &&
native[out - 2] == '\r' &&
native[out - 1] == '\n' ) {
native[out - 2] = SYNC4J_LINEBREAK[0];
out--;
}
// the conversion to wchar on Windows is
// probably missing here
}
} else {
native[out] = curr;
out++;
}
}
native[out] = 0;
out++;
} else {
wcscpy(native, foreign);
}
// decode escaped characters after backslash:
// \n is line break only in 3.0
WCHAR curr;
int in = 0, out = 0;
while ((curr = native[in]) != 0) {
in++;
switch (curr) {
case '\\':
curr = native[in];
in++;
switch (curr) {
case 'n':
if (is_30) {
// replace with line break
wcsncpy(native + out, SYNC4J_LINEBREAK, linebreaklen);
out += linebreaklen;
} else {
// normal escaped character
native[out] = curr;
out++;
}
break;
case 0:
// unexpected end of string
break;
default:
// just copy next character
native[out] = curr;
out++;
break;
}
break;
case ';':
// field separator - must replace with something special
// so that we can encode it again in fromNativeEncoding()
native[out] = SEMICOLON_REPLACEMENT;
out++;
break;
default:
native[out] = curr;
out++;
}
}
native[out] = 0;
out++;
// charset handling:
// - doesn't exist at the moment, vCards have to be in ASCII or UTF-8
// - an explicit CHARSET parameter is removed because its parameter
// value might differ between 2.1 and 3.0 (quotation marks allowed in
// 3.0 but not 2.1) and thus would require extra code to convert it;
// when charsets really get supported this needs to be addressed
WCHAR *charset = vprop->getParameterValue(TEXT("CHARSET"));
if (charset) {
// proper decoding of the value and the property value text
// would go here, for the time being we just remove the
// value
if (_wcsicmp(charset, TEXT("UTF-8")) &&
_wcsicmp(charset, TEXT("\"UTF-8\""))) {
LOG.error("ignoring unsupported charset");
}
vprop->removeParameter(TEXT("CHARSET"));
}
vprop->setValue(native);
delete [] native;
}
}