//reads a subset of database determined by memory capacity into a DoublyLinkedList
//subsequently sorts the list, iterates through it to merge entries with duplicate primary fields separated by delim
//and concatenate secondary fields using conn as the separator
//and exports it into a temp file
//returns the count of temporary files
int splitDatabase(string &database, const char delim, const char conn)
{
ifstream fin(database);
ofstream fout;
int tempFileCount = 0;
for (char buffer[LINE_BUFFER_SIZE]; fin.getline(buffer,LINE_BUFFER_SIZE,'\n');)
{
static int charsRead;
static string tempFileName = string("temp") + to_string(tempFileCount++) + ".db";
static string lineRead = "";
static DoublyLinkedList<string> lines;
lineRead += buffer;
charsRead += fin.gcount();
//if \n was not found by getline
if(fin.fail())
{
fin.clear();
}
else
{
lines.add(lineRead);
lineRead = "";
if(charsRead > CHARS_ALLOWED_IN_MEMORY || fin.eof())
{
fout.open(tempFileName);
lines.sort();
for (DoublyLinkedList<string>::iterator iter = lines.begin(), jter = iter.next();
!iter.isNull(); iter = jter, jter = iter.next())
{
//if jter is not a null node
//and the first field of iter (iter[0] -> iter.find(delim)) is equal to the first field of jter
//append the second field of jter to the second field of iter separated by conn
//repeat until the first fields differ or jter goes out of the bounds of the list
while(!jter.isNull() && (*iter).substr(0,(*iter).find(delim)) == (*jter).substr(0,(*jter).find(delim)))
{
*iter += conn + (*jter).substr((*jter).find(delim)+1);
++jter;
}
//store iter in the current temporary database file
fout << *iter << endl;
}
fout.close();
tempFileName = string("temp") + to_string(tempFileCount++) + ".db";
lines.clearList();
charsRead = 0;
}
}
}
fin.close();
return tempFileCount-1;
}
int main(int argc, char const *argv[])
{
DoublyLinkedList<FileInfo> d;
FileInfo test;
test.fileName[0] = '.';
test.fileName[1] = '_';
test.fileName[2] = '.';
test.fileName[3] = 'T';
test.fileName[4] = 'r';
test.fileName[5] = 'a';
test.fileName[6] = 's';
test.fileName[7] = 'h';
test.fileName[8] = 'e';
test.fileName[9] = 's';
test.fileName[10] = '\0';
test.index = 96;
d.add(test);
FileInfo test2;
test2.fileName[0] = '.';
test2.fileName[1] = 'T';
test2.fileName[2] = 'r';
test2.fileName[3] = 'a';
test2.fileName[4] = 's';
test2.fileName[5] = 'h';
test2.fileName[6] = 'e';
test2.fileName[7] = 's';
test2.fileName[8] = '\0';
test2.index = 192;
d.add(test2);
FileInfo test3;
test3.fileName[0] = '.';
test3.fileName[1] = 'S';
test3.fileName[2] = 'p';
test3.fileName[3] = 'o';
test3.fileName[4] = 't';
test3.fileName[5] = 'l';
test3.fileName[6] = 'i';
test3.fileName[7] = 'g';
test3.fileName[8] = 'h';
test3.fileName[9] = 't';
test3.fileName[10] = '-';
test3.fileName[11] = 'V';
test3.fileName[12] = '1';
test3.fileName[13] = '0';
test3.fileName[14] = '0';
test3.fileName[15] = '\0';
test3.index = 288;
d.add(test3);
FileInfo test4;
test4.fileName[0] = '.';
test4.fileName[1] = 'f';
test4.fileName[2] = 's';
test4.fileName[3] = 'e';
test4.fileName[4] = 'v';
test4.fileName[5] = 'e';
test4.fileName[6] = 'n';
test4.fileName[7] = 't';
test4.fileName[8] = 's';
test4.fileName[9] = 'd';
test4.fileName[10] = '\0';
test4.index = 352;
d.add(test4);
FileInfo test5;
test5.fileName[0] = 'L';
test5.fileName[1] = 'O';
test5.fileName[2] = 'S';
test5.fileName[3] = 'T';
test5.fileName[4] = '.';
test5.fileName[5] = 'D';
test5.fileName[6] = 'I';
test5.fileName[7] = 'R';
test5.fileName[8] = '\0';
test5.index = 384;
d.add(test5);
FileInfo test6;
test6.fileName[0] = '.';
test6.fileName[1] = 'a';
test6.fileName[2] = 'n';
test6.fileName[3] = 'd';
test6.fileName[4] = 'r';
test6.fileName[5] = 'o';
test6.fileName[6] = 'i';
test6.fileName[7] = 'd';
test6.fileName[8] = '_';
test6.fileName[9] = 's';
test6.fileName[10] = 'e';
test6.fileName[11] = 'c';
test6.fileName[12] = 'u';
test6.fileName[13] = 'r';
test6.fileName[14] = 'e';
test6.fileName[15] = '\0';
test6.index = 480;
d.add(test6);
FileInfo test7;
test7.fileName[0] = 'A';
test7.fileName[1] = 'n';
test7.fileName[2] = 'd';
test7.fileName[3] = 'r';
test7.fileName[4] = 'o';
test7.fileName[5] = 'i';
test7.fileName[6] = 'd';
test7.fileName[7] = '\0';
test7.index = 544;
d.add(test7);
FileInfo test8;
test8.fileName[0] = 'S';
test8.fileName[1] = 'i';
test8.fileName[2] = 'z';
test8.fileName[3] = 'e';
test8.fileName[4] = 'T';
test8.fileName[5] = 'e';
test8.fileName[6] = 's';
test8.fileName[7] = 't';
test8.fileName[8] = '.';
test8.fileName[9] = 't';
test8.fileName[10] = 'x';
test8.fileName[11] = 't';
test8.fileName[12] = '\0';
test8.index = 608;
d.add(test8);
FileInfo test9;
test9.fileName[0] = 'L';
test9.fileName[1] = 'G';
test9.fileName[2] = 'B';
test9.fileName[3] = 'a';
test9.fileName[4] = 'c';
test9.fileName[5] = 'k';
test9.fileName[6] = 'u';
test9.fileName[7] = 'p';
test9.fileName[8] = '\0';
test9.index = 672;
d.add(test9);
FileInfo test10;
test10.fileName[0] = 'M';
test10.fileName[1] = 'u';
test10.fileName[2] = 's';
test10.fileName[3] = 'i';
test10.fileName[4] = 'c';
test10.fileName[5] = '\0';
test10.index = 736;
d.add(test10);
// ._.Trashes - 96
// .Trashes - 192
// .Spotlight-V100 - 288
// .fseventsd - 352
// LOST.DIR - 384
// .android_secure - 480
// Android - 544
// SizeTest.txt - 608
// LGBackup - 672
// Music - 736
// d.printList();
for(int i=0; i<d.getSize(); i++)
{
std::cout << i << " - ";
std::cout << (d.getAt(i)->fileName);
std::cout << (" - ");
std::cout << (d.getAt(i)->index) << std::endl;
}
std::cout << std::endl;
d.sort();
d.printList();
for(int i=0; i<d.getSize(); i++)
{
std::cout << i << " - ";
std::cout << (d.getAt(i)->fileName);
std::cout << (" - ");
std::cout << (d.getAt(i)->index) << std::endl;
}
return 0;
}