bool Parser::parse(IReader& reader, bool verbose) {
wchar_t entityBuf;
debug = verbose;
tagPage = 0;
attributePage = 0;
if (!reader.readByte(version))
return false;
if (version != 0x03)
return false;
if (debug) cout << " : Version";
if (!reader.readMultiByteInt(publicIdentifierId))
return false;
if (debug) cout << " : PublicIdentifierId";
unsigned int dtdIndex = 0;
if (publicIdentifierId == 0) {
if (!reader.readMultiByteInt(dtdIndex))
return false;
if (debug) cout << " : dtdIndex";
} else {
return false;
}
if (dtdIndex != 0)
return false;
if (!reader.readMultiByteInt(charSet))
return false;
if (charSet != 0x6a)
return false;
if (debug) cout << " : CharSet";
unsigned int len;
if (!reader.readMultiByteInt(len))
return false;
if (debug) cout << " : StringTable Len";
unsigned int offset = 0;
while (offset < len) {
wstring s;
unsigned short utflen;
if (!reader.readUTF(s, utflen))
return false;
unsigned char b;
if (!reader.readByte(b))
return false; // skip null termination
if (debug) cout << " : null termination";
reader.putStrT(offset, s);
offset += utflen + 2+ 1;
}
if (!contentHandler.startDocument())
return false;
if (publicIdentifierId == 0) {
wstring dtdPair;
if (!reader.getStrT(dtdIndex, dtdPair))
return false;
if (debug) cout << " : dtdPair";
size_t space = dtdPair.find(' ');
if (space == string::npos) {
cout << endl << "DTD specifier does not contain 'space'"<< endl;
exit(1);
}
wstring name = dtdPair.substr(0, space);
wstring dtd = dtdPair.substr(space+1);
istream* resolverStream = resolver.resolveEntity(name,L"" , dtd);
if (resolverStream != NULL) {
// FIXME, use the resolver stream
cout << endl << "Entity Resolver NOT Used." << endl;
}
}
unsigned char id;
while (reader.readByte(id)) {
switch (id) {
case WBXML::SWITCH_PAGE:
if (debug) cout << " : Switch Page";
if (!reader.readByte(tagPage)) return false;
if (debug) cout << " : TagPage: " << (unsigned int)tagPage;
break;
case WBXML::END: {
unsigned int tagID;
if (stack.empty()) return false;
tagID = stack.top();
stack.pop();
if (debug) cout << " : </" << tagHandler.getTag(tagID) << ">";
if (!contentHandler.endElement(tagID)) return false;
break;
}
case WBXML::ENTITY:
if (debug) cout << " : Entity";
if (!reader.readMultiByteInt((unsigned int&)entityBuf)) return false;
if (debug) cout << " : " << entityBuf;
if (!contentHandler.characters(&entityBuf, 0, 1)) return false;
break;
case WBXML::STR_I: {
if (debug) cout << " : STR_I";
wstring s;
if (!reader.readStrI(s)) return false;
if (!contentHandler.characters(s, 0, s.length())) return false;
break;
}
case WBXML::EXT_I_0:
case WBXML::EXT_I_1:
case WBXML::EXT_I_2:
case WBXML::EXT_T_0:
case WBXML::EXT_T_1:
case WBXML::EXT_T_2:
case WBXML::EXT_0:
case WBXML::EXT_1:
case WBXML::EXT_2:
case WBXML::OPAQUE: {
MutableAttributes result;
vector<wstring> value;
if (!handleExtensions(reader, id, stack.top(), -1, result, value)) return false;
break;
}
case WBXML::PI:
if (debug) cout << " : PI";
return false;
// throw new SAXException("PI Not Supported");
break;
case WBXML::STR_T: {
if (debug) cout << " : STR_T";
wstring s;
unsigned int pos;
if (!reader.readStrT(s, pos)) return false;
if (!contentHandler.characters(s, 0, s.size())) return false;
break;
}
default:
if (!readElement(reader, id)) return false;
break;
}
}
if (!stack.empty()) {
cout << endl << "Stack not empty " << stack.size();
return false;
// throw new SAXException("unclosed elements: " + stack);
}
return contentHandler.endDocument();
}
// FIXME, no handling of arrays yet
bool DefaultExtensionHandler::opaque(unsigned int len, IReader& reader,
unsigned int tagID, int attributeID, MutableAttributes& atts,
vector<wstring>& value) {
// read type
unsigned char type;
if (!reader.readByte(type))
return false;
if (debug) cout << " : Type: "<< atts.getTypeName((IAttributes::Types)type);
len--;
switch (type) {
case IAttributes::BOOLEAN:
if (len == 1) {
bool b;
if (!reader.readBoolean(b))
return false;
atts.set(attributeID, b);
len--;
} else {
// boolean[] array = new boolean[len];
// for (int i=0; i<array.length; i++) {
// array[i] = in.readByte() != 0;
// len--;
// }
// atts.set(attributeID, array);
}
break;
case IAttributes::BYTE:
if (len == 1) {
unsigned char b;
if (!reader.readByte(b))
return false;
atts.set(attributeID, b);
len--;
} else {
// byte[] array = new byte[len];
// for (int i=0; i<array.length; i++) {
// array[i] = in.readByte();
// len--;
// }
// atts.set(attributeID, array);
}
break;
case IAttributes::CHAR:
if (len == 2) {
wchar_t c;
if (!reader.readChar(c))
return false;
atts.set(attributeID, c);
len-=2;
} else {
// char[] array = new char[len/2];
// for (int i=0; i<array.length; i++) {
// array[i] = in.readChar();
// len-=2;
// }
// atts.set(attributeID, array);
}
break;
case IAttributes::DOUBLE:
if (len == 8) {
double d;
if (!reader.readDouble(d))
return false;
atts.set(attributeID, d);
len -= 8;
} else {
// double[] array = new double[len/8];
// for (int i=0; i<array.length; i++) {
// array[i] = in.readDouble();
// len-=8;
// }
// atts.set(attributeID, array);
}
break;
case IAttributes::FLOAT:
if (len == 4) {
float f;
if (!reader.readFloat(f))
return false;
atts.set(attributeID, f);
len -= 4;
} else {
// float[] array = new float[len/4];
// for (int i=0; i<array.length; i++) {
// array[i] = in.readFloat();
// len-=4;
// }
// atts.set(attributeID, array);
}
break;
case IAttributes::INT:
if (len == 4) {
int i;
if (!reader.readInt(i))
return false;
atts.set(attributeID, i);
len -= 4;
} else {
// int[] array = new int[len/4];
// for (int i=0; i<array.length; i++) {
// array[i] = in.readInt();
// len-=4;
// }
// atts.set(attributeID, array);
}
break;
case IAttributes::LONG:
if (len == 8) {
int64 l;
if (!reader.readLong(l))
return false;
atts.set(attributeID, l);
len -= 8;
} else {
// long[] array = new long[len/8];
// for (int i=0; i<array.length; i++) {
// array[i] = in.readLong();
// len-=8;
// }
// atts.set(attributeID, array);
}
break;
case IAttributes::SHORT:
if (len == 2) {
short s;
if (!reader.readShort(s))
return false;
atts.set(attributeID, s);
len -= 2;
} else {
// short[] array = new short[len/2];
// for (int i=0; i<array.length; i++) {
// array[i] = in.readShort();
// len-=2;
// }
// atts.set(attributeID, array);
}
break;
case IAttributes::STRING: {
vector<wstring> vs;
while (len != 0) {
wstring s;
unsigned short utflen;
if (!reader.readUTF(s, utflen))
return false;
unsigned char b;
if (!reader.readByte(b))
return false; // skip NULL termination
vs.push_back(s);
len -= utflen + 2+ 1;
}
if (vs.size() == 1) {
atts.set(attributeID, vs[0]);
} else {
// String[] array = new String[vs.size()];
// array = (String[])vs.toArray(array);
// atts.set(attributeID, array);
}
}
break;
default:
cerr << endl << "ERROR No opaque handler for attributeID: "<< attributeID
<< " part of tagID: "<< tagID << endl;
return false;
}
if (debug) cout << " : Value";
if (len != 0) {
cerr << endl << "Skipping "<< len << " unused OPAQUE bytes..."<< endl;
while (len != 0) {
unsigned char b;
if (!reader.readByte(b))
return false;
len--;
}
}
return true;
}
bool Parser::readAttr(IReader& reader, unsigned int tagID, MutableAttributes& atts) {
int intResult;
bool hasIntResult = false;
wchar_t charResult;
int hasCharResult = false;
unsigned char id;
if (!reader.readByte(id))
return false;
int attributeID = -1;
while (id != WBXML::END) {
// attribute start
while (id == WBXML::SWITCH_PAGE) {
if (debug) cout << " : Switch Page";
if (!reader.readByte(attributePage))
return false;
if (debug) cout << " : AttributePage: " << (unsigned int)attributePage;
if (!reader.readByte(id))
return false;
}
attributeID = getAttributeId(id);
if (debug) cout << " : AttributeID (0x" << hex << attributeID << "): " << tagHandler.getAttribute(attributeID);
vector<wstring> value;
// attribute value(s)
if (!reader.readByte(id))
return false;
while (id > 128|| id == WBXML::SWITCH_PAGE || id == WBXML::ENTITY || id
== WBXML::STR_I || id == WBXML::STR_T|| (id >= WBXML::EXT_I_0
&& id <= WBXML::EXT_I_2)|| (id >= WBXML::EXT_T_0 && id
<= WBXML::EXT_T_2)) {
switch (id) {
case WBXML::SWITCH_PAGE:
if (debug) cout << " : Switch Page";
if (!reader.readByte(attributePage))
return false;
if (debug) cout << " : AttributePage: " << (unsigned int)attributePage;
break;
case WBXML::ENTITY:
if (debug) cout << " : Entity";
if (!reader.readMultiByteInt((unsigned int&)charResult))
return false;
if (debug) cout << " : " << charResult;
hasCharResult = true;
value.push_back(&charResult);
break;
case WBXML::STR_I: {
if (debug) cout << " : STR_I";
wstring s;
if (!reader.readStrI(s))
return false;
value.push_back(s);
break;
}
case WBXML::EXT_I_0:
case WBXML::EXT_I_1:
case WBXML::EXT_I_2:
case WBXML::EXT_T_0:
case WBXML::EXT_T_1:
case WBXML::EXT_T_2:
case WBXML::EXT_0:
case WBXML::EXT_1:
case WBXML::EXT_2:
case WBXML::OPAQUE:
if (!handleExtensions(reader, id, tagID, attributeID, atts,
value))
return false;
break;
case WBXML::STR_T: {
if (debug) cout << " : STR_T";
wstring s;
unsigned int pos;
if (!reader.readStrT(s, pos))
return false;
value.push_back(s);
break;
}
default:
intResult = getAttributeId(id);
hasIntResult = true;
wstringstream s;
s << intResult;
value.push_back(s.str());
break;
}
if (!reader.readByte(id))
return false;
}
switch (value.size()) {
case 0:
// already handled
break;
case 1:
if (hasIntResult) {
atts.set(attributeID, intResult);
} else if (hasCharResult) {
atts.set(attributeID, charResult);
} else {
atts.set(attributeID, value[0]);
}
break;
default:
atts.set(attributeID, value);
break;
}
attributeID = -1;
}
if (debug) cout << " : AttributeEnd";
return true;
}