void iterate(HDT *hdt, char *query, ostream &out, bool measure) {
TripleString tripleString;
tripleString.read(query);
const char *subj = tripleString.getSubject().c_str();
const char *pred = tripleString.getPredicate().c_str();
const char *obj = tripleString.getObject().c_str();
if(strcmp(subj, "?")==0) {
subj="";
}
if(strcmp(pred, "?")==0) {
pred="";
}
if(strcmp(obj, "?")==0) {
obj="";
}
#if 0
cout << "Subject: |" << subj <<"|"<< endl;
cout << "Predicate: |" << pred <<"|"<< endl;
cout << "Object: |" << obj << "|"<<endl;
#endif
try {
IteratorTripleString *it = hdt->search(subj, pred, obj);
StopWatch st;
unsigned int numTriples=0;
while(it->hasNext() && interruptSignal==0) {
TripleString *ts = it->next();
if(!measure)
out << *ts << endl;
numTriples++;
}
cout << numTriples << " results in " << st << endl;
delete it;
interruptSignal=0; // Interrupt caught, enable again.
} catch (char *e) {
cerr << e << endl;
}
}
int main(int argc, char *argv[]) {
int c;
string inputFile, outputFile, limit;
string type = "null";
string dir = "data/hdt/";
while ((c = getopt(argc, argv, "hi:t:l:o:d:")) != -1) {
switch (c) {
case 'h':
help();
break;
case 'd':
dir = optarg;
break;
case 'i':
inputFile = optarg;
break;
case 'o':
outputFile = optarg;
break;
case 't':
type = optarg;
break;
case 'l':
limit = optarg;
break;
default:
cout << "ERROR: Unknown option" << endl;
help();
return 1;
}
}
// Load HDT file
vector<HDT*> HDTversions;
int numVersions = 0;
if (limit.length() > 0) {
numVersions = atoi((char*) limit.c_str());
} else {
cerr << "[WARNING] limit not provided, trying to load 58 versions"
<< endl;
numVersions = 58;
}
ostream *out;
ofstream outF;
if (outputFile != "") {
outF.open(outputFile.c_str());
out = &outF;
} else {
out = &cout;
}
for (int i = 0; i < numVersions; i++) {
std::stringstream sstm;
sstm << dir << (i + 1) << ".hdt";
cout << "Loading " << sstm.str() << endl;
HDTversions.push_back(
HDTManager::mapIndexedHDT((char*) sstm.str().c_str()));
//cout<<"loaded 1.hdt! Press any key to load 2.hdt"<<endl;
//int c = getchar();
}
cout << "WARMUP... " << endl;
for (int i = 0; i < numVersions; i++) {
// Enumerate all different predicates
cout << "Dataset " << (i + 1) << " contains "
<< HDTversions[i]->getDictionary()->getNpredicates()
<< " predicates." << endl;
// Enumerate all triples matching a pattern ("" means any)
IteratorTripleString *it = HDTversions[i]->search("", "", "");
int count = 0;
while (it->hasNext() && count < 100) {
TripleString *triple = it->next();
//cout << "Result Warmup: " << triple->getSubject() << ", " << triple->getPredicate() << ", " << triple->getObject() << endl;
count++;
}
delete it; // Remember to delete iterator to avoid memory leaks!
/*IteratorUCharString *itPred = HDTversions[i]->getDictionary()->getPredicates();
while(itPred->hasNext()) {
unsigned char *str = itPred->next(); // Warning this pointer is only valid until next call to next();
cout << str << endl;
itPred->freeStr(str);
}
delete itPred; // Remember to delete iterator to avoid memory leaks!
*/
}
cout << "... WARMUP finished!" << endl;
if (type == "null") {
cerr << "[ERROR] Please provide a type of query (-t [s,p,o])" << endl;
help();
exit(0);
}
//read queries
cout << "opening file:" << inputFile << endl;
std::ifstream file((char*) inputFile.c_str());
cout << "opened! " << endl;
if (!file.good())
throw "unable to open filter file";
string linea = "";
vector<double> times(numVersions, 0);
int num_queries = 0;
while (!file.eof()) {
getline(file, linea);
cout << "Reading line:" << linea << endl;
if (linea.length() == 0)
continue;
size_t pos = linea.find(' ');
if (pos != std::string::npos) {
string query = linea.substr(0, pos);
string subject = "", predicate = "", object = "";
if (type == "s") {
subject = query;
} else if (type == "p") {
predicate = query;
} else if (type == "o") {
object = query;
} else {
vector<string> elements = split(linea, " ");
if (type == "sp") {
subject = elements[0];
predicate = elements[1];
} else if (type == "so") {
subject = elements[0];
object = elements[1];
} else if (type == "po") {
predicate = elements[0];
object = elements[1];
} else if (type == "spo") {
subject = elements[0];
predicate = elements[1];
object = elements[2];
}
}
for (int i = 0; i < numVersions; i++) {
StopWatch st;
IteratorTripleString *it = HDTversions[i]->search(
subject.c_str(), predicate.c_str(), object.c_str());
int numResults = 0;
while (it->hasNext()) {
TripleString *triple = it->next();
//cout << "Result: " << triple->getSubject() << ", " << triple->getPredicate() << ", " << triple->getObject() << endl;
numResults++;
}
delete it;
double time = st.toMillis();
cout << numResults << " Results in " << time << " ms" << endl;
times[i] = times[i] + time;
}
num_queries++;
}
}
//compute mean of queries
*out << "<version>,<mean_time>,<total>" << endl;
for (int i = 0; i < numVersions; i++) {
*out << (i + 1) << "," << times[i] / num_queries<<","<<times[i] << endl;
}
for (int i = 0; i < numVersions; i++) {
delete HDTversions[i]; // Remember to delete instance when no longer needed!
}
if (outputFile != "") {
outF.close();
}
}
int main(int argc, char **argv) {
int c;
string outputFile;
while( (c = getopt(argc,argv,"ho:"))!=-1) {
switch(c) {
case 'h':
help();
break;
case 'o':
outputFile = optarg;
break;
default:
cout << "ERROR: Unknown option" << endl;
help();
return 1;
}
}
if(argc-optind<1) {
cout << "ERROR: You must supply an HDT File" << endl << endl;
help();
return 1;
}
try {
#ifdef HAVE_LIBZ
igzstream *inGz=NULL;
#endif
ifstream *inF=NULL;
istream *in=NULL;
string inputFile = argv[optind];
std::string suffix = inputFile.substr(inputFile.find_last_of(".") + 1);
std::string pipeCommand;
if( suffix == "gz"){
#ifdef HAVE_LIBZ
in = inGz = new igzstream(inputFile.c_str());
#else
throw std::runtime_error("Support for GZIP was not compiled in this version. Please Decompress the file before importing it.");
#endif
} else {
in = inF = new ifstream(inputFile.c_str(), ios::binary);
}
if (!in->good())
{
cerr << "Error opening file " << inputFile << endl;
throw std::runtime_error("Error opening file for reading");
}
ControlInformation controlInformation;
// Load Global Control Information
controlInformation.load(*in);
// Load header
controlInformation.load(*in);
Header *header = HDTFactory::readHeader(controlInformation);
header->load(*in, controlInformation);
if( suffix == "gz") {
#ifdef HAVE_LIBZ
inGz->close();
#endif
} else {
inF->close();
}
// Save
IteratorTripleString *it = header->search("","","");
while(it->hasNext()) {
TripleString *ts = it->next();
cout << *ts << " ."<< endl;
}
if(outputFile!="") {
ofstream out(outputFile.c_str());
if(!out.good()){
throw std::runtime_error("Could not open output file.");
}
RDFSerializerNTriples serializer(out, NTRIPLES);
serializer.serialize(it);
out.close();
} else {
RDFSerializerNTriples serializer(cout, NTRIPLES);
serializer.serialize(it);
}
delete it;
delete header;
} catch (std::exception& e) {
cerr << "ERROR: " << e.what() << endl;
}
}