/*
* Create an instance of an aggregator for the specified aggregate
* type, column type, and result type. The object is constructed in
* memory from the provided memrory pool.
*/
AbstractAttributeAggregator *GetAttributeAggregatorInstance(ExpressionType agg_type) {
AbstractAttributeAggregator *aggregator;
switch (agg_type) {
case ExpressionType::AGGREGATE_COUNT:
aggregator = new CountAggregator();
break;
case ExpressionType::AGGREGATE_COUNT_STAR:
aggregator = new CountStarAggregator();
break;
case ExpressionType::AGGREGATE_SUM:
aggregator = new SumAggregator();
break;
case ExpressionType::AGGREGATE_AVG:
aggregator = new AvgAggregator(false);
break;
case ExpressionType::AGGREGATE_MIN:
aggregator = new MinAggregator();
break;
case ExpressionType::AGGREGATE_MAX:
aggregator = new MaxAggregator();
break;
default: {
std::string message =
"Unknown aggregate type " + ExpressionTypeToString(agg_type);
throw UnknownTypeException(static_cast<int>(agg_type), message);
}
}
return aggregator;
}
AbstractExpression *GetMoreSpecialized(ExpressionType c, L *l, R *r) {
assert(l);
assert(r);
switch (c) {
case (EXPRESSION_TYPE_COMPARE_EQUAL):
return new InlinedComparisonExpression<CmpEq, L, R>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_NOTEQUAL):
return new InlinedComparisonExpression<CmpNe, L, R>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_LESSTHAN):
return new InlinedComparisonExpression<CmpLt, L, R>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_GREATERTHAN):
return new InlinedComparisonExpression<CmpGt, L, R>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_LESSTHANOREQUALTO):
return new InlinedComparisonExpression<CmpLte, L, R>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_GREATERTHANOREQUALTO):
return new InlinedComparisonExpression<CmpGte, L, R>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_LIKE):
return new InlinedComparisonExpression<CmpLike, L, R>(
c, l, r); // added by michael
case (EXPRESSION_TYPE_COMPARE_IN):
return new InlinedComparisonExpression<CmpIn, L, R>(
c, l,
r); // added by michael
default:
char message[256];
sprintf(message,
"Invalid ExpressionType '%s' called for"
" ComparisonExpression",
ExpressionTypeToString(c).c_str());
throw ExpressionException(message);
}
}
void AbstractExpression::DeduceExpressionName() {
// If alias exists, it will be used in TrafficCop
if (!alias.empty()) return;
for (auto &child : children_) child->DeduceExpressionName();
// Aggregate expression already has correct expr_name_
if (ExpressionUtil::IsAggregateExpression(exp_type_)) return;
auto op_str = ExpressionTypeToString(exp_type_, true);
auto children_size = children_.size();
if (exp_type_ == ExpressionType::FUNCTION) {
auto expr = (FunctionExpression *)this;
expr_name_ = expr->GetFuncName() + "(";
for (size_t i = 0; i < children_size; i++) {
if (i > 0) expr_name_.append(",");
expr_name_.append(GetChild(i)->expr_name_);
}
expr_name_.append(")");
} else {
PL_ASSERT(children_size <= 2);
if (children_size == 2) {
expr_name_ = GetChild(0)->expr_name_ + " " + op_str + " " +
GetChild(1)->expr_name_;
} else if (children_size == 1) {
expr_name_ = op_str + " " + GetChild(0)->expr_name_;
}
}
}
AbstractExpression *GetGeneral(ExpressionType c, AbstractExpression *l,
AbstractExpression *r) {
PL_ASSERT(l);
PL_ASSERT(r);
switch (c) {
case (EXPRESSION_TYPE_COMPARE_EQUAL):
return new ComparisonExpression<CmpEq>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_NOTEQUAL):
return new ComparisonExpression<CmpNe>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_LESSTHAN):
return new ComparisonExpression<CmpLt>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_GREATERTHAN):
return new ComparisonExpression<CmpGt>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_LESSTHANOREQUALTO):
return new ComparisonExpression<CmpLte>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_GREATERTHANOREQUALTO):
return new ComparisonExpression<CmpGte>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_LIKE):
return new ComparisonExpression<CmpLike>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_NOTLIKE):
return new ComparisonExpression<CmpNotLike>(c, l, r);
case (EXPRESSION_TYPE_COMPARE_IN):
return new ComparisonExpression<CmpIn>(c, l, r);
default:
char message[256];
snprintf(message, 256,
"Invalid ExpressionType '%s' called"
" for ComparisonExpression",
ExpressionTypeToString(c).c_str());
throw Exception(message);
}
}
std::string AbstractExpression::Debug() const {
if (this == nullptr) return "";
std::ostringstream buffer;
buffer << "Expression[" << ExpressionTypeToString(GetExpressionType()) << ", "
<< GetExpressionType() << "]";
return (buffer.str());
}
SeqScanPlan::SeqScanPlan(parser::SelectStatement *select_node) {
LOG_DEBUG("Creating a Sequential Scan Plan");
auto target_table = static_cast<storage::DataTable *>(
catalog::Catalog::GetInstance()->GetTableWithName(
select_node->from_table->GetDatabaseName(),
select_node->from_table->GetTableName()));
SetTargetTable(target_table);
ColumnIds().clear();
// Check if there is an aggregate function in query
bool function_found = false;
for (auto elem : *select_node->select_list) {
if (elem->GetExpressionType() == EXPRESSION_TYPE_FUNCTION_REF) {
function_found = true;
break;
}
}
// Pass all columns
// TODO: This isn't efficient. Needs to be fixed
if (function_found) {
for (auto column : GetTable()->GetSchema()->GetColumns()) {
oid_t col_id = SeqScanPlan::GetColumnID(column.column_name);
SetColumnId(col_id);
}
}
// Pass columns in select_list
else {
if (select_node->select_list->at(0)->GetExpressionType() !=
EXPRESSION_TYPE_STAR) {
for (auto col : *select_node->select_list) {
LOG_TRACE("ExpressionType: %s",
ExpressionTypeToString(col->GetExpressionType()).c_str());
auto col_name = col->GetName();
oid_t col_id = SeqScanPlan::GetColumnID(std::string(col_name));
SetColumnId(col_id);
}
} else {
auto allColumns = GetTable()->GetSchema()->GetColumns();
for (uint i = 0; i < allColumns.size(); i++) SetColumnId(i);
}
}
// Check for "For Update" flag
if(select_node->is_for_update == true){
SetForUpdateFlag(true);
}
// Keep a copy of the where clause to be binded to values
if (select_node->where_clause != NULL) {
auto predicate = select_node->where_clause->Copy();
// Replace COLUMN_REF expressions with TupleValue expressions
expression::ExpressionUtil::ReplaceColumnExpressions(GetTable()->GetSchema(), predicate);
predicate_with_params_ =
std::unique_ptr<expression::AbstractExpression>(predicate->Copy());
SetPredicate(predicate);
}
}
const std::string ComparisonExpression::GetInfo(int num_indent) const {
std::ostringstream os;
os << StringUtil::Indent(num_indent) << "Expression ::\n"
<< StringUtil::Indent(num_indent + 1) << "expression type = Comparison,\n"
<< StringUtil::Indent(num_indent + 1)
<< "comparison type = " << ExpressionTypeToString(exp_type_) << "\n";
for (const auto &child : children_) {
os << child.get()->GetInfo(num_indent + 2);
}
return os.str();
}
// Produce the value that is the result of codegening the expression
codegen::Value ConjunctionTranslator::DeriveValue(CodeGen &codegen,
RowBatch::Row &row) const {
const auto &conjunction =
GetExpressionAs<expression::ConjunctionExpression>();
codegen::Value left = row.DeriveValue(codegen, *conjunction.GetChild(0));
codegen::Value right = row.DeriveValue(codegen, *conjunction.GetChild(1));
switch (conjunction.GetExpressionType()) {
case ExpressionType::CONJUNCTION_AND:
return left.LogicalAnd(codegen, right);
case ExpressionType::CONJUNCTION_OR:
return left.LogicalOr(codegen, right);
default:
throw Exception{"Received a non-conjunction expression type: " +
ExpressionTypeToString(conjunction.GetExpressionType())};
}
}
//===--------------------------------------------------------------------===//
// Plain Aggregator
//===--------------------------------------------------------------------===//
PlainAggregator::PlainAggregator(const planner::AggregatePlan *node,
storage::DataTable *output_table,
executor::ExecutorContext *econtext)
: AbstractAggregator(node, output_table, econtext) {
// allocate aggregators
aggregates = new Agg *[node->GetUniqueAggTerms().size()]();
// initialize aggregators
for (oid_t aggno = 0; aggno < node->GetUniqueAggTerms().size(); aggno++) {
LOG_TRACE("Aggregate term type: %s", ExpressionTypeToString(
node->GetUniqueAggTerms()[aggno].aggtype).c_str());
aggregates[aggno] =
GetAggInstance(node->GetUniqueAggTerms()[aggno].aggtype);
bool distinct = node->GetUniqueAggTerms()[aggno].distinct;
aggregates[aggno]->SetDistinct(distinct);
}
}
TEST_F(BooleanValueTests, ComparisonTest) {
std::vector<ExpressionType> compares = {
ExpressionType::COMPARE_EQUAL,
ExpressionType::COMPARE_NOTEQUAL,
ExpressionType::COMPARE_LESSTHAN,
ExpressionType::COMPARE_LESSTHANOREQUALTO,
ExpressionType::COMPARE_GREATERTHAN,
ExpressionType::COMPARE_GREATERTHANOREQUALTO};
int values[] = {true, false, type::PELOTON_BOOLEAN_NULL};
CmpBool result;
type::Value val0;
type::Value val1;
for (int i = 0; i < 2; i++) {
for (int j = 0; j < 2; j++) {
bool expected_null = false;
// VALUE #0
if (values[i] == type::PELOTON_BOOLEAN_NULL) {
val0 = type::ValueFactory::GetNullValueByType(type::TypeId::BOOLEAN);
expected_null = true;
} else {
val0 =
type::ValueFactory::GetBooleanValue(static_cast<bool>(values[i]));
}
// VALUE #1
if (values[j] == type::PELOTON_BOOLEAN_NULL) {
val1 = type::ValueFactory::GetNullValueByType(type::TypeId::BOOLEAN);
expected_null = true;
} else {
val1 =
type::ValueFactory::GetBooleanValue(static_cast<bool>(values[j]));
}
for (auto etype : compares) {
bool expected;
switch (etype) {
case ExpressionType::COMPARE_EQUAL:
expected = values[i] == values[j];
result = val0.CompareEquals(val1);
break;
case ExpressionType::COMPARE_NOTEQUAL:
expected = values[i] != values[j];
result = val0.CompareNotEquals(val1);
break;
case ExpressionType::COMPARE_LESSTHAN:
expected = values[i] < values[j];
result = val0.CompareLessThan(val1);
break;
case ExpressionType::COMPARE_LESSTHANOREQUALTO:
expected = values[i] <= values[j];
result = val0.CompareLessThanEquals(val1);
break;
case ExpressionType::COMPARE_GREATERTHAN:
expected = values[i] > values[j];
result = val0.CompareGreaterThan(val1);
break;
case ExpressionType::COMPARE_GREATERTHANOREQUALTO:
expected = values[i] >= values[j];
result = val0.CompareGreaterThanEquals(val1);
break;
default:
throw Exception("Unexpected comparison");
} // SWITCH
LOG_TRACE("%s %s %s => %d | %d\n", val0.ToString().c_str(),
ExpressionTypeToString(etype).c_str(),
val1.ToString().c_str(),
static_cast<int>(expected),
static_cast<int>(result));
if (expected_null) expected = false;
EXPECT_EQ(expected, result == CmpBool::CmpTrue);
EXPECT_EQ(!expected, result == CmpBool::CmpFalse);
EXPECT_EQ(expected_null, result == CmpBool::NULL_);
}
}
}
}
const planner::AbstractPlan *PlanTransformer::TransformAgg(
const AggPlanState *plan_state) {
// Alias all I need
const Agg *agg = plan_state->agg_plan;
auto numphases = plan_state->numphases;
auto numaggs = plan_state->numaggs;
auto targetlist = plan_state->ps_targetlist;
auto qual = plan_state->ps_qual;
auto peragg = plan_state->peragg;
auto tupleDesc = plan_state->result_tupleDescriptor;
auto aggstrategy = plan_state->agg_plan->aggstrategy;
LOG_INFO("Number of Agg phases: %d \n", numphases);
// When we'll have >1 phases?
if (numphases != 1) return nullptr;
/* Get project info */
std::unique_ptr<const planner::ProjectInfo> proj_info(
BuildProjectInfoFromTLSkipJunk(targetlist));
LOG_INFO("proj_info : \n%s", proj_info->Debug().c_str());
/* Get predicate */
std::unique_ptr<const expression::AbstractExpression> predicate(
BuildPredicateFromQual(qual));
/* Get Aggregate terms */
std::vector<planner::AggregatePlan::AggTerm> unique_agg_terms;
LOG_INFO("Number of (unique) Agg nodes: %d \n", numaggs);
for (int aggno = 0; aggno < numaggs; aggno++) {
auto transfn_oid = peragg[aggno].transfn_oid;
auto itr = peloton::bridge::kPgTransitFuncMap.find(transfn_oid);
if (kPgFuncMap.end() == itr) {
LOG_ERROR("Unmapped Transit function Id : %u\n", transfn_oid);
return nullptr;
}
// We don't check whether the mapped exprtype is a valid aggregate type
// here.
PltFuncMetaInfo fn_meta = itr->second;
// We only take the first argument as input to aggregator because
// we don't have multi-argument aggregator in Peloton at the moment.
// WARNING: there can be no arguments (e.g., COUNT(*))
auto arguments = peragg[aggno].aggrefstate->args;
expression::AbstractExpression *agg_expr = nullptr;
if (arguments) {
GenericExprState *gstate =
(GenericExprState *)lfirst(list_head(arguments));
LOG_INFO("Creating Agg Expr");
agg_expr = ExprTransformer::TransformExpr(gstate->arg);
LOG_INFO("Done creating Agg Expr");
}
/*
* AggStatePerAggData.sortColIdx along with other related attributes
* are used to handle ORDER BY and DISTINCT *within* aggregation.
* E.g.,
* SELECT count(DISTINCT x) ...
* SELECT str_agg(y ORDER BY x) ...
* Currently, we only handle the agg(DISTINCT x) case by
* checking whether numDistinctCols > 0.
* Note that numDistinctCols > 0 may be a necessary but not sufficient
* condition for agg(DISTINCT x).
*/
bool distinct = (peragg[aggno].numDistinctCols > 0);
unique_agg_terms.emplace_back(fn_meta.exprtype, agg_expr, distinct);
LOG_INFO(
"Unique Agg # : %d , transfn_oid : %u\n , aggtype = %s \n expr = %s, "
"numDistinctCols = %d",
aggno, transfn_oid, ExpressionTypeToString(fn_meta.exprtype).c_str(),
agg_expr ? agg_expr->Debug().c_str() : "<NULL>",
peragg[aggno].numDistinctCols);
for (int i = 0; i < peragg[aggno].numDistinctCols; i++) {
LOG_INFO("sortColIdx[%d] : %d \n", i, peragg[aggno].sortColIdx[i]);
}
} // end loop aggno
/* Get Group by columns */
std::vector<oid_t> groupby_col_ids;
LOG_INFO("agg.numCols = %d", agg->numCols);
for (int i = 0; i < agg->numCols; i++) {
LOG_INFO("agg.grpColIdx[%d] = %d \n", i, agg->grpColIdx[i]);
auto attrno = agg->grpColIdx[i];
if (AttributeNumberIsValid(attrno) &&
AttrNumberIsForUserDefinedAttr(attrno)) {
groupby_col_ids.emplace_back(AttrNumberGetAttrOffset(attrno));
}
}
/* Get output schema */
std::unique_ptr<catalog::Schema> output_schema(
SchemaTransformer::GetSchemaFromTupleDesc(tupleDesc));
/* Map agg stragegy */
LOG_INFO("aggstrategy : %s\n", (AGG_HASHED == aggstrategy)
? "HASH"
: (AGG_SORTED ? "SORT" : "PLAIN"));
PelotonAggType agg_type = AGGREGATE_TYPE_INVALID;
switch (aggstrategy) {
case AGG_SORTED:
agg_type = AGGREGATE_TYPE_SORTED;
break;
case AGG_HASHED:
agg_type = AGGREGATE_TYPE_HASH;
break;
case AGG_PLAIN:
agg_type = AGGREGATE_TYPE_PLAIN;
break;
}
std::vector<oid_t> column_ids;
for (auto agg_term : unique_agg_terms) {
if (agg_term.expression) {
LOG_INFO("AGG TERM :: %s", agg_term.expression->Debug().c_str());
}
BuildColumnListFromExpr(column_ids, agg_term.expression);
}
auto retval = new planner::AggregatePlan(
proj_info.release(), predicate.release(), std::move(unique_agg_terms),
std::move(groupby_col_ids), output_schema.release(), agg_type);
((planner::AggregatePlan *)retval)->SetColumnIds(column_ids);
// Find children
auto lchild = TransformPlan(outerAbstractPlanState(plan_state));
retval->AddChild(lchild);
return retval;
}
TEST_F(TimestampValueTests, ComparisonTest) {
std::vector<ExpressionType> compares = {
ExpressionType::COMPARE_EQUAL,
ExpressionType::COMPARE_NOTEQUAL,
ExpressionType::COMPARE_LESSTHAN,
ExpressionType::COMPARE_LESSTHANOREQUALTO,
ExpressionType::COMPARE_GREATERTHAN,
ExpressionType::COMPARE_GREATERTHANOREQUALTO};
uint64_t values[] = {1000000000, 2000000000, type::PELOTON_TIMESTAMP_NULL};
type::CmpBool result;
type::Value val0;
type::Value val1;
for (int i = 0; i < 3; i++) {
for (int j = 0; j < 3; j++) {
bool expected_null = false;
// VALUE #0
if (values[i] == type::PELOTON_TIMESTAMP_NULL) {
val0 = type::ValueFactory::GetNullValueByType(type::TypeId::TIMESTAMP);
expected_null = true;
} else {
val0 = type::ValueFactory::GetTimestampValue(
static_cast<uint64_t>(values[i]));
}
// VALUE #1
if (values[j] == type::PELOTON_TIMESTAMP_NULL) {
val1 = type::ValueFactory::GetNullValueByType(type::TypeId::TIMESTAMP);
expected_null = true;
} else {
val1 = type::ValueFactory::GetTimestampValue(
static_cast<uint64_t>(values[j]));
}
bool temp = expected_null;
for (auto etype : compares) {
bool expected;
expected_null = temp;
switch (etype) {
case ExpressionType::COMPARE_EQUAL:
expected = values[i] == values[j];
result = val0.CompareEquals(val1);
break;
case ExpressionType::COMPARE_NOTEQUAL:
expected = values[i] != values[j];
result = val0.CompareNotEquals(val1);
if (!val1.IsNull() && expected_null) {
expected_null = false;
}
break;
case ExpressionType::COMPARE_LESSTHAN:
expected = values[i] < values[j];
result = val0.CompareLessThan(val1);
break;
case ExpressionType::COMPARE_LESSTHANOREQUALTO:
expected = values[i] <= values[j];
result = val0.CompareLessThanEquals(val1);
break;
case ExpressionType::COMPARE_GREATERTHAN:
expected = values[i] > values[j];
result = val0.CompareGreaterThan(val1);
break;
case ExpressionType::COMPARE_GREATERTHANOREQUALTO:
expected = values[i] >= values[j];
result = val0.CompareGreaterThanEquals(val1);
break;
default:
throw Exception("Unexpected comparison");
} // SWITCH
LOG_TRACE("%s %s %s => %d | %d\n", val0.ToString().c_str(),
ExpressionTypeToString(etype).c_str(),
val1.ToString().c_str(), expected, result);
if (expected_null) {
EXPECT_EQ(expected_null, result == type::CMP_NULL);
} else {
EXPECT_EQ(expected, result == type::CMP_TRUE);
EXPECT_EQ(!expected, result == type::CMP_FALSE);
}
}
}
}
}
