//This function returns current running PID
int CurrentPID() {
long currentContext = CurrentContext();
struct Process_Control_Block *PCB = findPCBbyContextID(currentContext);
if (PCB != NULL) {
return PCB->ProcessID;
}
else {
return -1;
}
}
std::vector< std::vector< int > > NHPYLM::Generate(std::string Mode, int NumWordsOrCharacters, int SentEndWordId, std::vector<double> *GeneratedWordLengthDistribution_) const
{
if (Mode == "CHPYLM") {
/* build vector of character ids */
std::vector<int> CharacterIds(CharactersEnd - CharactersBegin, 0);
std::iota(CharacterIds.begin(), CharacterIds.end(), CharactersBegin);
CharacterIds.push_back(EOW);
CharacterIds.push_back(EOS);
CharacterIds.push_back(PHI);
/* initialize base probabilities */
std::vector<double> BaseProbabilites;
BaseProbabilites.reserve(CharacterIds.size());
for (std::vector<int>::iterator Character = CharacterIds.begin(); Character != CharacterIds.end(); ++Character) {
if (*Character != PHI) {
BaseProbabilites.push_back(CHPYLMBaseProbabilities.find(*Character)->second);
} else {
BaseProbabilites.push_back(0);
}
}
/* sample the sentences */
int WordLength = 0;
int NumWords = 0;
double MeanWordLength = 0;
std::vector<double> GeneratedWordLengthDistribution;
std::vector<double> *WordLengthsProbabilities;
if (GeneratedWordLengthDistribution_) {
WordLengthsProbabilities = GeneratedWordLengthDistribution_;
WordLengthsProbabilities->clear();
} else {
WordLengthsProbabilities = &GeneratedWordLengthDistribution;
}
std::vector<int> CurrentContext(CHPYLMOrder - 1, EOW);
std::vector<int> GeneratedSentence;
std::vector<std::vector<int> > GeneratedSentences;
while (NumWords < NumWordsOrCharacters) {
GeneratedSentence.push_back(CHPYLM.GenerateWord(CurrentContext, CharacterIds, BaseProbabilites, true));
WordLength++;
if ((CHPYLMOrder > 1) && (GeneratedSentence.back() != EOW) && (GeneratedSentence.back() != EOS)) {
CurrentContext.erase(CurrentContext.begin());
CurrentContext.push_back(GeneratedSentence.back());
} else {
if (GeneratedSentence.back() == EOW) {
NumWords++;
MeanWordLength += WordLength;
WordLengthsProbabilities->resize(std::max(static_cast<size_t>(WordLength + 1), WordLengthsProbabilities->size()), 0);
WordLengthsProbabilities->at(WordLength)++;
WordLength = 0;
}
if (GeneratedSentence.back() == EOS) {
GeneratedSentence.push_back(EOW);
GeneratedSentences.push_back(GeneratedSentence);
GeneratedSentence.clear();
}
CurrentContext.assign(CHPYLMOrder - 1, EOW);
}
}
for (unsigned int WordLength = 0; WordLength < WordLengthsProbabilities->size(); WordLength++) {
WordLengthsProbabilities->at(WordLength) /= NumWords;
// std::cout << std::setprecision(4) << std::fixed << "Word length: " << WordLength << ", word length probability: " << WordLengthsProbabilities->at(WordLength) << std::endl;
}
MeanWordLength /= NumWords;
std::cout << " Mean length of generated words by CHPYLM: " << MeanWordLength << ", number of words: " << NumWords << std::endl << std::endl;
return GeneratedSentences;
} else if (Mode == "WHPYLM") {
/* build vector of word ids */
std::vector<int> WordIds;
WordIds.reserve(GetId2Word().size());
for (Id2WordHashmap::const_iterator Id2Word = GetId2Word().begin(); Id2Word != GetId2Word().end(); ++Id2Word) {
WordIds.push_back(Id2Word->first);
}
WordIds.push_back(PHI);
/* initialize base probabilities */
std::vector<double> BaseProbabilites;
BaseProbabilites.reserve(WordIds.size());
for (std::vector<int>::iterator Word = WordIds.begin(); Word != WordIds.end(); ++Word) {
if (*Word != PHI) {
BaseProbabilites.push_back(exp(CHPYLM.WordSequenceLoglikelihood(GetWordVector(*Word), CHPYLMBaseProbabilities)));
} else {
BaseProbabilites.push_back(0);
}
}
/* sample the sentences */
std::vector<int> CurrentContext(WHPYLMOrder - 1, SentEndWordId);
std::vector<int> GeneratedSentence;
std::vector<std::vector<int> > GeneratedSentences;
for (int NumWords = 0; NumWords < NumWordsOrCharacters; NumWords++) {
GeneratedSentence.push_back(WHPYLM.GenerateWord(CurrentContext, WordIds, BaseProbabilites, false));
if ((WHPYLMOrder > 1) && (GeneratedSentence.back() != SentEndWordId)) {
CurrentContext.erase(CurrentContext.begin());
CurrentContext.push_back(GeneratedSentence.back());
} else {
if (GeneratedSentence.back() == SentEndWordId) {
CurrentContext.assign(WHPYLMOrder - 1, SentEndWordId);
GeneratedSentences.push_back(GeneratedSentence);
GeneratedSentence.clear();
}
}
}
return GeneratedSentences;
} else {
return std::vector<std::vector<int> >();
}
}
//This function returns current running PCB
struct Process_Control_Block *CurrentPCB() {
long currentContext = CurrentContext();
struct Process_Control_Block *PCB = findPCBbyContextID(currentContext);
return PCB;
}
void LMSetMemErr(SInt16 value)
{
CurrentContext()->globals.LM_memErr = value;
}
SInt16 LMGetMemErr(void)
{
return CurrentContext()->globals.LM_memErr;
}
static inline MemoryManagerPrivate *_MemGetPrivate(void)
{
return CurrentContext()->memoryMgrPrivate;
}
