void FourSectionDictionary::save(std::ostream & output, ControlInformation & controlInformation, ProgressListener *listener)
{
controlInformation.setFormat(HDTVocabulary::DICTIONARY_TYPE_FOUR);
controlInformation.setUint("mapping", this->mapping);
controlInformation.setUint("sizeStrings", this->sizeStrings);
controlInformation.save(output);
IntermediateListener iListener(listener);
iListener.setRange(0,10);
iListener.notifyProgress(0, "Dictionary save shared area.");
shared->save(output);
iListener.setRange(10,45);
iListener.notifyProgress(0, "Dictionary save subjects.");
subjects->save(output);
iListener.setRange(45,60);
iListener.notifyProgress(0, "Dictionary save predicates.");
predicates->save(output);
iListener.setRange(60,100);
iListener.notifyProgress(0, "Dictionary save objects.");
objects->save(output);
}
void BasicHDT::saveToHDT(std::ostream & output, ProgressListener *listener)
{
ControlInformation controlInformation;
IntermediateListener iListener(listener);
controlInformation.clear();
controlInformation.setType(GLOBAL);
controlInformation.setFormat(HDTVocabulary::HDT_CONTAINER);
controlInformation.save(output);
controlInformation.clear();
controlInformation.setType(HEADER);
iListener.setRange(0,5);
header->save(output, controlInformation, &iListener);
controlInformation.clear();
controlInformation.setType(DICTIONARY);
iListener.setRange(5,70);
dictionary->save(output, controlInformation, &iListener);
controlInformation.clear();
controlInformation.setType(TRIPLES);
iListener.setRange(70,100);
triples->save(output, controlInformation, &iListener);
}
void PlainDictionary::save(std::ostream &output, ControlInformation &controlInformation, ProgressListener *listener)
{
controlInformation.setFormat(HDTVocabulary::DICTIONARY_TYPE_PLAIN);
controlInformation.setUint("mapping", this->mapping);
controlInformation.setUint("sizeStrings", this->sizeStrings);
controlInformation.setUint("numEntries", this->getNumberOfElements());
controlInformation.save(output);
unsigned int i = 0;
unsigned int counter=0;
const char marker = '\1';
//shared subjects-objects from subjects
for (i = 0; i < shared.size(); i++) {
output << shared[i]->str;
output.put(marker); //character to split file
counter++;
NOTIFYCOND(listener, "PlainDictionary saving shared", counter, getNumberOfElements());
}
output.put(marker); //extra line to set the begining of next part of dictionary
//not shared subjects
for (i = 0; i < subjects.size(); i++) {
output << subjects[i]->str;
output.put(marker); //character to split file
counter++;
NOTIFYCOND(listener, "PlainDictionary saving subjects", counter, getNumberOfElements());
}
output.put(marker); //extra line to set the begining of next part of dictionary
//not shared objects
for (i = 0; i < objects.size(); i++) {
output << objects[i]->str;
output.put(marker); //character to split file
counter++;
NOTIFYCOND(listener, "PlainDictionary saving objects", counter, getNumberOfElements());
}
output.put(marker); //extra line to set the begining of next part of dictionary
//predicates
for (i = 0; i < predicates.size(); i++) {
output << predicates[i]->str;
output.put(marker); //character to split file
counter++;
NOTIFYCOND(listener, "PlainDictionary saving predicates", counter, getNumberOfElements());
}
output.put(marker);
}
void TripleListDisk::save(std::ostream & output, ControlInformation &controlInformation, ProgressListener *listener)
{
controlInformation.setFormat(getType());
controlInformation.save(output);
for(unsigned int i=0; i<numTotalTriples; i++) {
TripleID *tid = getTripleID(i);
if(tid->isValid()) {
//cout << "Write: " << tid << " " << *tid << endl;
output.write((char *)tid, sizeof(TripleID));
}
}
}
void CompactTriples::save(std::ostream & output, ControlInformation &controlInformation, ProgressListener *listener)
{
controlInformation.clear();
controlInformation.setUint("numTriples", getNumberOfElements());
controlInformation.setFormat(HDTVocabulary::TRIPLES_TYPE_COMPACT);
controlInformation.setUint("order", order);
controlInformation.save(output);
IntermediateListener iListener(listener);
iListener.setRange(0,30);
iListener.notifyProgress(0, "CompactTriples saving Stream Y");
streamY->save(output);
iListener.setRange(30,100);
iListener.notifyProgress(0, "CompactTriples saving Stream Z");
streamZ->save(output);
}
void PlainHeader::save(std::ostream & output, ControlInformation &controlInformation, ProgressListener *listener)
{
// TODO: Choose format from spec (NTRIPLES, RDFXML...) and implement.
// Dump header into a stringbuffer to know size.
stringstream strbuf(stringstream::out);
for(vector<TripleString>::iterator it = triples.begin(); it!=triples.end(); it++){
strbuf << *it << " ." << endl;
}
string str = strbuf.str();
// Dump header
controlInformation.setFormat(HDTVocabulary::HEADER_NTRIPLES);
controlInformation.setUint("length", str.length());
controlInformation.save(output);
// Dump data
output << str;
}
void PlainTriples::save(std::ostream & output, ControlInformation &controlInformation, ProgressListener *listener)
{
controlInformation.clear();
controlInformation.setUint("numTriples", getNumberOfElements());
controlInformation.setFormat(HDTVocabulary::TRIPLES_TYPE_PLAIN);
controlInformation.setUint("order", order);
controlInformation.save(output);
IntermediateListener iListener(listener);
iListener.setRange(0,33);
iListener.notifyProgress(0, "PlainTriples saving subjects");
streamX->save(output);
iListener.setRange(33, 66);
iListener.notifyProgress(0, "PlainTriples saving predicates");
streamY->save(output);
iListener.setRange(66, 100);
iListener.notifyProgress(0, "PlainTriples saving objects");
streamZ->save(output);
}
int main(int argc, char **argv) {
int c;
string query, inputFile, outputFile;
bool measure = false;
while( (c = getopt(argc,argv,"hq:o:m"))!=-1) {
switch(c) {
case 'h':
break;
case 'q':
query = optarg;
break;
case 'o':
outputFile = optarg;
break;
case 'm':
measure = true;
break;
default:
cout << "ERROR: Unknown option" << endl;
return 1;
}
}
if(argc-optind<2) {
cout << "ERROR: You must supply an input and HDT File" << endl << endl;
return 1;
}
inputFile = argv[optind];
outputFile = argv[optind+1];
ConvertProgress progress;
StopWatch st;
try {
// LOAD
HDT *hdt = HDTManager::mapHDT(inputFile.c_str());
// CONVERT triples to TripleList
TriplesList tlist;
Triples *triples = hdt->getTriples();
cout << "Old Triples -> TriplesList" << endl;
st.reset();
IteratorTripleID *it = triples->searchAll();
tlist.insert(it);
delete it;
cout << " Old Triples -> TriplesList time" << st << endl;
// Convert tlist to OPS
cout << "TriplesList sort OPS" << endl;
st.reset();
tlist.sort(OPS, &progress);
cout << " TriplesList sort OPS time: " << st << endl;
// Generate new OPS BitmapTriples
cout << "TriplesList to new BitmapTriples" << endl;
HDTSpecification spec;
spec.set("triplesOrder", "OPS");
BitmapTriples bt(spec);
st.reset();
bt.load(tlist, &progress);
cout << " TriplesList to new BitmapTriples time" << st << endl;
// Update Header
#if 1
cout << "Update Header" << endl;
string rootNode("_:triples");
TripleString ts (rootNode, "", "");
hdt->getHeader()->remove(ts);
bt.populateHeader(*hdt->getHeader(), "_:triples");
#endif
// SAVE
cout << "Save to " << outputFile << endl;
ofstream out(outputFile.c_str(), ios::binary | ios::out);
ControlInformation ci;
ci.clear();
ci.setType(GLOBAL);
ci.setFormat(HDTVocabulary::HDT_CONTAINER);
ci.save(out);
// HEADER
ci.clear();
ci.setType(HEADER);
hdt->getHeader()->save(out, ci, NULL);
// DICTIONARY
ci.clear();
ci.setType(DICTIONARY);
hdt->getDictionary()->save(out, ci, NULL);
// NEW TRIPLES
ci.clear();
ci.setType(TRIPLES);
bt.save(out, ci, NULL);
out.close();
delete hdt;
} catch (char *e) {
cout << "ERROR: " << e << endl;
} catch (const char *e) {
cout << "ERROR: " << e << endl;
}
}
int main(int argc, char **argv) {
int c;
string query, inputFile, outputFile;
bool measure = false;
while( (c = getopt(argc,argv,"hq:o:m"))!=-1) {
switch(c) {
case 'h':
break;
case 'q':
query = optarg;
break;
case 'o':
outputFile = optarg;
break;
case 'm':
measure = true;
break;
default:
cout << "ERROR: Unknown option" << endl;
return 1;
}
}
if(argc-optind<1) {
cout << "ERROR: You must supply an input and HDT File" << endl << endl;
return 1;
}
inputFile = argv[optind];
try {
// LOAD
HDT *hdt = HDTManager::mapHDT(inputFile.c_str());
// CONVERT
Dictionary *dict = hdt->getDictionary();
//LiteralDictionary litDict;
FourSectionDictionary litDict;
StdoutProgressListener progress;
litDict.import(dict, &progress);
// SAVE
ofstream out(outputFile.c_str(), ios::binary | ios::out);
ControlInformation ci;
// GLOBAL
ci.clear();
ci.setType(GLOBAL);
ci.setFormat(HDTVocabulary::HDT_CONTAINER);
ci.save(out);
// HEADER
ci.clear();
ci.setType(HEADER);
hdt->getHeader()->save(out, ci, NULL);
// NEW DICTIONARY
ci.clear();
ci.setType(DICTIONARY);
litDict.save(out, ci, NULL);
// TRIPLES
ci.clear();
ci.setType(TRIPLES);
hdt->getTriples()->save(out, ci, NULL);
out.close();
delete hdt;
} catch (std::exception& e) {
cout << "ERROR: " << e.what() << endl;
}
}
