int main (int argc, char * const argv[])
{
// insert code here...
Moses::ResetUserTime();
Moses::PrintUserTime("Starting");
if (argc != 8) {
std::cerr << "Usage: " << argv[0] << " numSourceFactors numTargetFactors numScores tableLimit sortScoreIndex inputPath outputPath" << std::endl;
return 1;
}
int numSourceFactors = Moses::Scan<int>(argv[1])
, numTargetFactors = Moses::Scan<int>(argv[2])
, numScores = Moses::Scan<int>(argv[3])
, tableLimit = Moses::Scan<int>(argv[4]);
TargetPhraseCollection::s_sortScoreInd = Moses::Scan<int>(argv[5]);
assert(TargetPhraseCollection::s_sortScoreInd < numScores);
const string filePath = argv[6]
,destPath = argv[7];
Moses::InputFileStream inStream(filePath);
OnDiskWrapper onDiskWrapper;
bool retDb = onDiskWrapper.BeginSave(destPath, numSourceFactors, numTargetFactors, numScores);
assert(retDb);
PhraseNode &rootNode = onDiskWrapper.GetRootSourceNode();
size_t lineNum = 0;
char line[100000];
//while(getline(inStream, line))
while(inStream.getline(line, 100000)) {
lineNum++;
if (lineNum%1000 == 0) cerr << "." << flush;
if (lineNum%10000 == 0) cerr << ":" << flush;
if (lineNum%100000 == 0) cerr << lineNum << flush;
//cerr << lineNum << " " << line << endl;
std::vector<float> misc(1);
SourcePhrase sourcePhrase;
TargetPhrase *targetPhrase = new TargetPhrase(numScores);
OnDiskPt::PhrasePtr spShort = Tokenize(sourcePhrase, *targetPhrase, line, onDiskWrapper, numScores, misc);
assert(misc.size() == onDiskWrapper.GetNumCounts());
rootNode.AddTargetPhrase(sourcePhrase, targetPhrase, onDiskWrapper, tableLimit, misc, spShort);
}
rootNode.Save(onDiskWrapper, 0, tableLimit);
onDiskWrapper.EndSave();
Moses::PrintUserTime("Finished");
//pause();
return 0;
} // main()
void PhraseNode::GetChild(Word &wordFound, UINT64 &childFilePos, size_t ind, OnDiskWrapper &onDiskWrapper) const
{
size_t wordSize = onDiskWrapper.GetSourceWordSize();
size_t childSize = wordSize + sizeof(UINT64);
char *currMem = m_memLoad
+ sizeof(UINT64) * 2 // size & file pos of target phrase coll
+ sizeof(float) * onDiskWrapper.GetNumCounts() // count info
+ childSize * ind;
size_t memRead = ReadChild(wordFound, childFilePos, currMem);
CHECK(memRead == childSize);
}
PhraseNode::PhraseNode(UINT64 filePos, OnDiskWrapper &onDiskWrapper)
:m_counts(onDiskWrapper.GetNumCounts())
{
// load saved node
m_filePos = filePos;
size_t countSize = onDiskWrapper.GetNumCounts();
std::fstream &file = onDiskWrapper.GetFileSource();
file.seekg(filePos);
CHECK(filePos == (UINT64)file.tellg());
file.read((char*) &m_numChildrenLoad, sizeof(UINT64));
size_t memAlloc = GetNodeSize(m_numChildrenLoad, onDiskWrapper.GetSourceWordSize(), countSize);
m_memLoad = (char*) malloc(memAlloc);
// go to start of node again
file.seekg(filePos);
CHECK(filePos == (UINT64)file.tellg());
// read everything into memory
file.read(m_memLoad, memAlloc);
CHECK(filePos + memAlloc == (UINT64)file.tellg());
// get value
m_value = ((UINT64*)m_memLoad)[1];
// get counts
float *memFloat = (float*) (m_memLoad + sizeof(UINT64) * 2);
CHECK(countSize == 1);
m_counts[0] = memFloat[0];
m_memLoadLast = m_memLoad + memAlloc;
}
void PhraseNode::Save(OnDiskWrapper &onDiskWrapper, size_t pos, size_t tableLimit)
{
CHECK(!m_saved);
// save this node
m_targetPhraseColl.Sort(tableLimit);
m_targetPhraseColl.Save(onDiskWrapper);
m_value = m_targetPhraseColl.GetFilePos();
size_t numCounts = onDiskWrapper.GetNumCounts();
size_t memAlloc = GetNodeSize(GetSize(), onDiskWrapper.GetSourceWordSize(), numCounts);
char *mem = (char*) malloc(memAlloc);
//memset(mem, 0xfe, memAlloc);
size_t memUsed = 0;
UINT64 *memArray = (UINT64*) mem;
memArray[0] = GetSize(); // num of children
memArray[1] = m_value; // file pos of corresponding target phrases
memUsed += 2 * sizeof(UINT64);
// count info
float *memFloat = (float*) (mem + memUsed);
CHECK(numCounts == 1);
memFloat[0] = (m_counts.size() == 0) ? DEFAULT_COUNT : m_counts[0]; // if count = 0, put in very large num to make sure its still used. HACK
memUsed += sizeof(float) * numCounts;
// recursively save chm_countsildren
ChildColl::iterator iter;
for (iter = m_children.begin(); iter != m_children.end(); ++iter) {
const Word &childWord = iter->first;
PhraseNode &childNode = iter->second;
// recursive
if (!childNode.Saved())
childNode.Save(onDiskWrapper, pos + 1, tableLimit);
char *currMem = mem + memUsed;
size_t wordMemUsed = childWord.WriteToMemory(currMem);
memUsed += wordMemUsed;
UINT64 *memArray = (UINT64*) (mem + memUsed);
memArray[0] = childNode.GetFilePos();
memUsed += sizeof(UINT64);
}
// save this node
//Moses::DebugMem(mem, memAlloc);
CHECK(memUsed == memAlloc);
std::fstream &file = onDiskWrapper.GetFileSource();
m_filePos = file.tellp();
file.seekp(0, ios::end);
file.write(mem, memUsed);
UINT64 endPos = file.tellp();
CHECK(m_filePos + memUsed == endPos);
free(mem);
m_children.clear();
m_saved = true;
}