void SSE::storefile(string filename, uint64_t docID, double& cryptoduration){
size_t filesize = readFileSize(filename);
byte* docBytes = new byte[filesize];
readFile(filename, docBytes, filesize);
fstore.put(boost::lexical_cast<string>(docID), docBytes, filesize, cryptoduration);
delete[] docBytes;
}
std::unique_ptr<File> Extractor::readFile() {
auto fileName = readFileName();
readFileTimestamp();
readFileOwnerId();
readFileGroupId();
readFileMode();
auto fileSize = readFileSize();
readUntilEndOfFileHeader();
auto fileContent = readFileContent(fileSize);
return std::make_unique<StringFile>(fileContent, fileName);
}
void Extractor::readLookupTable() {
// In the GNU format, the special file name '/' denotes a lookup table.
// However, we need to ensure that it is just a standalone '/' because "//"
// denotes the start of a filename table.
if (hasLookupTableAt(i)) {
// The lookup table has the same format as a file. However, as we do
// not need it, throw it away after reading (i.e. do not store its
// content).
++i;
readFileTimestamp();
readFileOwnerId();
readFileGroupId();
readFileMode();
auto fileSize = readFileSize();
readUntilEndOfFileHeader();
readFileContent(fileSize);
}
}
bool SSE::retrieveIndex0(string keyword, vector<docid_t>& docIDs, double& diskTime){
OnlineSession session;
byte* docIDsBytes;
size_t size = session.updateRead(keyword, docIDsBytes, 0, diskTime);
cout << "Index0: Number of bytes retrieved are " << size << endl;
if(size == 0)
return false;
cout << "Size is " << size << endl;
// byte* updatedDocIDsBytes = new byte[size];
// memset(updatedDocIDsBytes, 0, size);
long totalDataDownloaded = 0;
int32_t j = 0;
for(int32_t i = 0; i < size/sizeof(docid_t); i++){
docid_t docID = *(docid_t*)(&docIDsBytes[i*sizeof(docid_t)]);
uint64_t docIDonDisk = getDocNameHash(boost::lexical_cast<string>(docID));
CLEAR_BIT(docIDonDisk, 0);
clock_t start = clock();
bool filepresent = fstore.isFilePresent(boost::lexical_cast<string>(docIDonDisk));
totalDataDownloaded += readFileSize(FILESTORE_PATH + boost::lexical_cast<string>(docIDonDisk));
diskTime += (double)(clock()-start)/CLOCKS_PER_SEC;
if(filepresent){
docIDs.push_back(docID);
// memcpy(&updatedDocIDsBytes[j*sizeof(docid_t)], &docIDsBytes[i*sizeof(docid_t)], sizeof(docid_t));
j++;
}
}
cout << "Index0: Total data downloaded is " << totalDataDownloaded << endl;
session.updateWrite(keyword, docIDsBytes, size, diskTime);
delete[] docIDsBytes;
return true;
}
int main(){
string directoryPath = "datasets/email/enron_mail_20110402/maildir/mann-k/inbox/";
BStore* store = new BStore();
int numFiles = 12;
string filenumbers[] = {"1", "2", "3", "4", "5", "6", "7", "8", "9" , "10", "267", "207"};
for(int i = 0; i < numFiles; i++){
string filename = directoryPath + filenumbers[i] + ".";
size_t size = readFileSize(filename);
byte fileBytes[size];
readFile(filename, fileBytes, size);
printhex(fileBytes, size, "FILE BYTES");
store->add(filename, fileBytes, size);
}
double execTime;
store->finalize(execTime);
delete store;
return 0;
}
bool SSE::retrieveIndex1(string keyword, vector<docid_t>& docIDs, double& diskTime){
OnlineSession session;
session.resetDiskAccessTime();
byte* docIDsBytes;
size_t size = session.updateRead(keyword, docIDsBytes, 0, diskTime);
cout << "Index1: Number of bytes retrieved are " << size << endl;
long totalDataDownloaded = 0;
if(size == 0)
return false;
for(int32_t i = 0; i < size/sizeof(docid_t); i++){
docid_t docID = *(docid_t*)(&docIDsBytes[i*sizeof(docid_t)]);
uint64_t docIDonDisk = getDocNameHash(boost::lexical_cast<string>(docID));
docIDs.push_back(docID);
totalDataDownloaded += readFileSize(boost::lexical_cast<string>(docIDonDisk));
}
cout << "Index1: Total data downloaded is " << totalDataDownloaded << endl;
delete[] docIDsBytes;
return true;
}
