GaddagFactory::GaddagFactory(const UVString &alphabetFile)
: m_encodableWords(0), m_unencodableWords(0), m_alphas(NULL)
{
if (!alphabetFile.empty())
{
QuackleIO::FlexibleAlphabetParameters *flexure = new QuackleIO::FlexibleAlphabetParameters;
flexure->load(QuackleIO::Util::uvStringToQString(alphabetFile));
m_alphas = flexure;
}
// So the separator is sorted to last.
m_root.t = false;
m_root.c = QUACKLE_NULL_MARK; // "_"
m_root.pointer = 0;
m_root.lastchild = true;
m_hash.int32ptr[0] = m_hash.int32ptr[1] = m_hash.int32ptr[2] = m_hash.int32ptr[3] = 0;
}
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;
}
int MiniDawgMaker::executeFromArguments()
{
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;
root.t = false;
root.insmallerdict = false;
root.playability = 0;
root.c = QUACKLE_BLANK_MARK;
root.pointer = 0;
root.lastchild = true;
QString smallerDictFilename = "smaller.raw";
QFile smallerDict(smallerDictFilename);
if (!smallerDict.exists())
{
UVcout << "smaller dictionary does not exist: " << QuackleIO::Util::qstringToString(smallerDictFilename) << endl;
return false;
}
if (!smallerDict.open(QIODevice::ReadOnly | QIODevice::Text))
{
UVcout << "Could not open " << QuackleIO::Util::qstringToString(smallerDictFilename) << endl;
return false;
}
QTextStream smallerStream(&smallerDict);
smallerStream.setCodec(QTextCodec::codecForName("UTF-8"));
while (!smallerStream.atEnd())
{
QString originalQString;
smallerStream >> originalQString;
//UVcout << "this word is in the smaller dictionary: " << QuackleIO::Util::qstringToString(originalQString) << endl;
smallerMap[originalQString] = true;
}
QString playabilityFilename = "playabilities.raw";
QFile playability(playabilityFilename);
if (!playability.exists())
{
UVcout << "playability does not exist: " << QuackleIO::Util::qstringToString(playabilityFilename) << endl;
return false;
}
if (!playability.open(QIODevice::ReadOnly | QIODevice::Text))
{
UVcout << "Could not open " << QuackleIO::Util::qstringToString(playabilityFilename) << endl;
return false;
}
QTextStream playabilityStream(&playability);
playabilityStream.setCodec(QTextCodec::codecForName("UTF-8"));
while (!playabilityStream.atEnd())
{
int pb;
playabilityStream >> pb;
QString originalQString;
playabilityStream >> originalQString;
//UVcout << "playability: " << QuackleIO::Util::qstringToString(originalQString) << " " << pb << endl;
playabilityMap[originalQString] = pb;
}
QString dawgFilename = "dawginput.raw";
QFile file(dawgFilename);
if (!file.exists())
{
UVcout << "dawg does not exist: " << QuackleIO::Util::qstringToString(dawgFilename) << endl;
return false;
}
if (!file.open(QIODevice::ReadOnly | QIODevice::Text))
{
UVcout << "Could not open " << QuackleIO::Util::qstringToString(dawgFilename) << endl;
return false;
}
QTextStream stream(&file);
stream.setCodec(QTextCodec::codecForName("UTF-8"));
int encodableWords = 0;
int unencodableWords = 0;
while (!stream.atEnd())
{
QString originalQString;
stream >> originalQString;
bool inSmaller = smallerMap[originalQString];
int pb = playabilityMap[originalQString];
if (stream.atEnd())
break;
UVString originalString = QuackleIO::Util::qstringToString(originalQString);
//UVcout << "read original string: " << originalString;
//if (!inSmaller) UVcout << "#";
//UVcout << endl;
UVString leftover;
Quackle::LetterString encodedWord = alphas->encode(originalString, &leftover);
if (leftover.empty())
{
//for (Quackle::LetterString::iterator it = encodedWord.begin(); it != encodedWord.end(); ++it)
//UVcout << "got encoded letter: " << (int)(*it) << endl;
root.pushword(encodedWord, inSmaller, pb);
++encodableWords;
}
else
{
UVcout << "not encodable without leftover: " << originalString << endl;
++unencodableWords;
}
}
file.close();
delete alphas;
UVcout << "encodable words: " << encodableWords << ", unencodable words: " << unencodableWords << endl;
nodelist.push_back(&root);
root.print("");
UVcout << "nodelist.size(): " << nodelist.size() << endl;
minimize();
ofstream out("output.dawg", ios::out | ios::binary);
for (unsigned int i = 0; i < nodelist.size(); i++) {
//cout << nodelist[i]->c << " " << nodelist[i]->pointer << " " << nodelist[i]->t << " " << nodelist[i]->lastchild << endl;
Node* n = nodelist[i];
unsigned int p;
if (nodelist[i]->deleted)
{
p = (unsigned int)(nodelist[i]->cloneof->pointer);
// n = nodelist[i]->cloneof;
}
else
p = (unsigned int)(nodelist[i]->pointer);
char bytes[7];
unsigned char n1 = (p & 0x00FF0000) >> 16;
unsigned char n2 = (p & 0x0000FF00) >> 8;
unsigned char n3 = (p & 0x000000FF);
unsigned char n4 = n->c - QUACKLE_FIRST_LETTER;
unsigned int pb = n->playability;
unsigned char n5 = (pb & 0x00FF0000) >> 16;
unsigned char n6 = (pb & 0x0000FF00) >> 8;
unsigned char n7 = (pb & 0x000000FF);
if (n->t) {
n4 |= 32;
}
if (n->lastchild) {
n4 |= 64;
}
if (n->insmallerdict) {
n4 |= 128;
}
bytes[0] = n1; bytes[1] = n2; bytes[2] = n3; bytes[3] = n4;
bytes[4] = n5; bytes[5] = n6; bytes[6] = n7;
out.write(bytes, 7);
}
}