bool InMemorySort(int start_index, int num_blocks, vector<string> field_names)
{// InMemorySort for 2-pass algorithm
//count of blocks to sort in main memory
vector<Tuple> mem_tuples = mem.getTuples(start_index, num_blocks);
sort(mem_tuples.begin(), mem_tuples.end(), Tuple_Comparison(field_names));
for(int index=start_index; index< num_blocks; index++)
{
mem.getBlock(index)->clear(); // clearing the blocks
}
bool success = mem.setTuples(start_index, mem_tuples);
if(!success)
cout<<"Error in InMemorySort"<<endl;
return success;
}
//Main Memory Sorting
//In memory sorting operation based on tuples
bool MMSorting(int startMemIndex, int noOfBlocks, vector<string> fieldNames)
{
//total no of blocks to sort
vector<Tuple> tuplesInMem = mem.getTuples(startMemIndex, noOfBlocks);
sort(tuplesInMem.begin(), tuplesInMem.end(), CompareTuple(fieldNames));
//clear out any data in the given range of memory block
for(int i=startMemIndex; i< noOfBlocks; i++)
{
mem.getBlock(i)->clear();
}
bool success = mem.setTuples(startMemIndex, tuplesInMem);
if(!success)
cout<<"Error in MMSorting"<<endl;
return success;
}
string crossJoin(vector<string> attributes, string tableName1, string tableName2, string whereCondition, bool multi) {
string small,big;
bool proj = false;
if(schemaManager.getRelation(tableName1)->getNumOfBlocks()<=schemaManager.getRelation(tableName2)->getNumOfBlocks()) {
small = tableName1;
big = tableName2;
}
else {
small=tableName2;
big=tableName1;
}
Schema schema1 = schemaManager.getSchema(small);
Schema schema2 = schemaManager.getSchema(big);
vector<string> fieldNames;
vector<enum FIELD_TYPE> fieldTypes;
if(!multi) {
for(int i=0;i<schema1.getNumOfFields();i++) {
fieldNames.push_back(small+"."+schema1.getFieldName(i));
fieldTypes.push_back(schema1.getFieldType(i));
}
}
else {
for(int i=0;i<schema1.getNumOfFields();i++) {
fieldNames.push_back(schema1.getFieldName(i));
fieldTypes.push_back(schema1.getFieldType(i));
}
}
for(int i=0;i<schema2.getNumOfFields();i++) {
fieldNames.push_back(big+"."+schema2.getFieldName(i));
fieldTypes.push_back(schema2.getFieldType(i));
}
Schema schema(fieldNames,fieldTypes);
Relation *relation = schemaManager.createRelation(big+"_join",schema);
Relation *relation1 = schemaManager.getRelation(small);
Relation *relation2 = schemaManager.getRelation(big);
int size1 = relation1->getNumOfBlocks(), size2 = relation2->getNumOfBlocks();
if(!((attributes.size()==1 && attributes[0]=="*") || multi)) {
vector<string> fieldNames1;
vector<enum FIELD_TYPE> fieldTypes1;
for(int i=0;i<attributes.size();i++) {
int temp = schema.getFieldOffset(attributes[i]);
fieldNames1.push_back(schema.getFieldName(temp));
fieldTypes1.push_back(schema.getFieldType(attributes[i]));
}
Schema schema1(fieldNames1, fieldTypes1);
proj = true;
Relation *relationp = schemaManager.createRelation(big+"_joinp", schema1);
}
if(size1<=10) {
relation1->getBlocks(0,0,size1);
vector<Tuple> tuples = mainMemory.getTuples(0,size1);
for(int x=0;x<tuples.size();x++) {
for(int i=0;i<size2;i++) {
relation2->getBlock(i,1);
Block *block = mainMemory.getBlock(1);
for(int j=0;j<block->getNumTuples();j++) {
Tuple tuple2 = block->getTuple(j);
join(tuples[x], tuple2, small, big, whereCondition, multi, attributes);
}
}
}
}
else {
for(int x=0;x<size1;x++) {
relation1->getBlock(x,0);
Block *block0 = mainMemory.getBlock(0);
for(int y=0;y<block0->getNumTuples();y++) {
Tuple tuple1 = block0->getTuple(y);
for(int i=0;i<size2;i++) {
relation2->getBlock(i,1);
Block *block = mainMemory.getBlock(1);
for(int j=0;j<block->getNumTuples();j++) {
Tuple tuple2 = block->getTuple(j);
join(tuple1, tuple2, small, big, whereCondition, multi, attributes);
}
}
}
}
}
string rt = big+"_join";
if(proj) {
schemaManager.deleteRelation(rt);
rt = big+"_joinp";
}
return rt;
}
string distinct(string tableName) {
Relation *relation = schemaManager.getRelation(tableName);
Schema schema = relation->getSchema();
int size = relation->getNumOfBlocks();
vector<Tuple> tuples;
bool flag = true;
//one-pass
if(size<=10) {
relation->getBlocks(0,0,size);
tuples = mainMemory.getTuples(0,size);
tuples = getDistinctTuples(tuples);
Relation *relation1 = schemaManager.createRelation(tableName+"_distinct",schema);
insertTuples(tableName+"_distinct",tuples);
}
//two pass
else {
int index = 0, loadSize=10;
while(size>0) {
relation->getBlocks(index,0,loadSize);
for(int i=0;i<loadSize;i++) {
Block *block = mainMemory.getBlock(i);
for(int j=0;j<block->getNumTuples();j++) {
tuples.push_back(block->getTuple(j));
}
}
tuples = getDistinctTuples(tuples);
//partition(tuples, 0, tuples.size()-1);
if(flag) {
Relation *relation2= schemaManager.createRelation(tableName+"_dis", schema);
flag = false;
}
insertTuples(tableName+"_dis", tuples);
Relation *relation2 = schemaManager.getRelation(tableName+"_dis");
tuples.clear();
index = index+10;
size = size-10;
if(size<10)
loadSize = size;
}
if(size<=100) {
Relation *relation2 = schemaManager.createRelation(tableName+"_distinct", schema);
relation = schemaManager.getRelation(tableName+"_dis");
int buckets = relation->getNumOfBlocks()/10;
vector<Tuple> tuples;
for(int i=0;i<10;i++) {
for(int j=0;j<buckets;j++) {
if(j*10+i > relation->getNumOfBlocks()) break;
relation->getBlock(i+10*j,j);
Block *block = mainMemory.getBlock(j);
for(int k=0;k<block->getNumTuples();k++) {
tuples.push_back(block->getTuple(k));
}
}
}
tuples = getDistinctTuples(tuples);
insertTuples(tableName+"_distinct", tuples);
tuples.clear();
schemaManager.deleteRelation(tableName+"_dis");
}
else
cerr<<"Table size exceeds the limit size(mainMemory)^2"<<endl;
}
return tableName+"_distinct";
}