bool GaddagFactory::pushWord(const Quackle::LetterString &word)
{
++m_encodableWords;
hashWord(word);
// FIXME: This hash will fail if duplicate words are passed in.
// But testing for duplicate words isn't so easy without keeping
// an entirely separate list.
for (unsigned i = 1; i <= word.length(); i++)
{
Quackle::LetterString newword;
for (int j = i - 1; j >= 0; j--)
newword.push_back(word[j]);
if (i < word.length())
{
newword.push_back(internalSeparatorRepresentation); // "^"
for (unsigned j = i; j < word.length(); j++)
newword.push_back(word[j]);
}
m_gaddagizedWords.push_back(newword);
}
return true;
}
void GaddagFactory::Node::pushWord(const Quackle::LetterString& word)
{
if (word.length() == 0)
{
t = true;
return;
}
Quackle::Letter first = Quackle::String::front(word);
Quackle::LetterString rest = Quackle::String::allButFront(word);
int index = -1;
for (size_t i = 0; i < children.size(); i++)
{
if (children[i].c == first)
{
index = i;
i = children.size();
}
}
if (index == -1)
{
Node n;
n.c = first;
n.t = false;
n.pointer = 0;
n.lastchild = false;
children.push_back(n);
index = children.size() - 1;
}
children[index].pushWord(rest);
}
void GaddagFactory::hashWord(const Quackle::LetterString &word)
{
QCryptographicHash wordhash(QCryptographicHash::Md5);
wordhash.addData(word.constData(), word.length());
QByteArray wordhashbytes = wordhash.result();
m_hash.int32ptr[0] ^= ((const int32_t*)wordhashbytes.constData())[0];
m_hash.int32ptr[1] ^= ((const int32_t*)wordhashbytes.constData())[1];
m_hash.int32ptr[2] ^= ((const int32_t*)wordhashbytes.constData())[2];
m_hash.int32ptr[3] ^= ((const int32_t*)wordhashbytes.constData())[3];
}
void Node::pushword(Quackle::LetterString word, bool inSmaller, int pb) {
if (word.length() == 0) {
t = true;
playability = pb;
insmallerdict = inSmaller;
}
else {
char first = word[0];
Quackle::LetterString rest = word.substr(1, word.length() - 1);
int index = -1;
// cout << "first: " << first << ", rest: " << rest << endl;
for (unsigned int i = 0; i < children.size(); i++) {
if (children[i].c == first) {
index = i;
i = children.size();
}
}
if (index == -1) {
Node n;
n.c = first;
n.t = false;
n.playability = 0;
n.insmallerdict = false;
n.pointer = 0;
n.oldpointer = 0;
n.lastchild = false;
children.push_back(n);
index = children.size() - 1;
}
children[index].pushword(rest, inSmaller, pb);
}
dexplored = false;
sexplored = false;
lexplored = false;
deleted = false;
written = false;
}
int main(int argc, char **argv) {
QCoreApplication a(argc, argv);
GetOpt opts;
QString alphabet;
opts.addOption('a', "alphabet", &alphabet);
if (!opts.parse())
return 1;
if (alphabet.isNull())
alphabet = "english";
Quackle::AlphabetParameters *alphas = 0;
QString alphabetFile = QString("../data/alphabets/%1.quackle_alphabet").arg(alphabet);
UVcout << "Using alphabet file: " << QuackleIO::Util::qstringToString(alphabetFile) << endl;
QuackleIO::FlexibleAlphabetParameters *flexure = new QuackleIO::FlexibleAlphabetParameters;
flexure->load(alphabetFile);
alphas = flexure;
QString leavesFilename = "superleaves.raw";
QFile file(leavesFilename);
if (!file.exists())
{
UVcout << "leaves file does not exist: " << QuackleIO::Util::qstringToString(leavesFilename) << endl;
return false;
}
if (!file.open(QIODevice::ReadOnly | QIODevice::Text))
{
UVcout << "Could not open " << QuackleIO::Util::qstringToString(leavesFilename) << endl;
return false;
}
QTextStream stream(&file);
stream.setCodec(QTextCodec::codecForName("UTF-8"));
ofstream out("encoded");
int encodableLeaves = 0;
int unencodableLeaves = 0;
while (!stream.atEnd()) {
QString leaveQString;
stream >> leaveQString;
double value;
stream >> value;
//UVcout << "value: " << value << endl;
UVString leaveString = QuackleIO::Util::qstringToString(leaveQString);
if (stream.atEnd())
break;
//UVcout << "read original string: " << originalString << endl;
UVString leftover;
Quackle::LetterString encodedLeave = alphas->encode(leaveString, &leftover);
if (leftover.empty())
{
unsigned char leavelength = encodedLeave.length();
out.write((char*)(&leavelength), 1);
out.write(encodedLeave.begin(), encodedLeave.length());
unsigned short int intvalue = (value + 128) * 256;
//UVcout << "intvalue: " << intvalue << endl;
out.write((char*)(&intvalue), 2);
++encodableLeaves;
}
else
{
//UVcout << "not encodable without leftover: " << originalString << endl;
++unencodableLeaves;
}
}
file.close();
delete alphas;
UVcout << "encodable leaves: " << encodableLeaves << ", unencodable leaves: " << unencodableLeaves << endl;
}
