database/doris
2
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity
Files
b241bc4e9dfe5893a4292fe7671cf33ff9629718
doris/be/src/exprs/agg_fn_evaluator.cpp
Zhengguo Yang 50864aca7d [refactor] fix warings when compile with clang (#8069)
2022-02-19 11:29:02 +08:00

964 lines
35 KiB
C++
Raw Blame History

// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing,
// software distributed under the License is distributed on an
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the License for the
// specific language governing permissions and limitations
// under the License.
#if defined(__clang__)
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Warray-bounds"
#elif defined(__GNUC__) || defined(__GNUG__)
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Warray-bounds"
#pragma GCC diagnostic ignored "-Wstringop-overflow="
#endif
#include "exprs/agg_fn_evaluator.h"
#include <sstream>
#include "common/logging.h"
#include "exprs/aggregate_functions.h"
#include "exprs/anyval_util.h"
#include "runtime/datetime_value.h"
#include "runtime/mem_tracker.h"
#include "runtime/raw_value.h"
#include "runtime/user_function_cache.h"
#include "thrift/protocol/TDebugProtocol.h"
#include "udf/udf_internal.h"
#include "util/debug_util.h"
namespace doris {
using doris_udf::FunctionContext;
using doris_udf::BooleanVal;
using doris_udf::TinyIntVal;
using doris_udf::SmallIntVal;
using doris_udf::IntVal;
using doris_udf::BigIntVal;
using doris_udf::LargeIntVal;
using doris_udf::FloatVal;
using doris_udf::DoubleVal;
using doris_udf::DecimalV2Val;
using doris_udf::DateTimeVal;
using doris_udf::StringVal;
using doris_udf::AnyVal;
// typedef for builtin aggregate functions. Unfortunately, these type defs don't
// really work since the actual builtin is implemented not in terms of the base
// AnyVal* type. Due to this, there are lots of casts when we use these typedefs.
// TODO: these typedefs exists as wrappers to go from (TupleRow, Tuple) to the
// types the aggregation functions need. This needs to be done with codegen instead.
typedef void (*InitFn)(FunctionContext*, AnyVal*);
typedef void (*UpdateFn0)(FunctionContext*, AnyVal*);
typedef void (*UpdateFn1)(FunctionContext*, const AnyVal&, AnyVal*);
typedef void (*UpdateFn2)(FunctionContext*, const AnyVal&, const AnyVal&, AnyVal*);
typedef void (*UpdateFn3)(FunctionContext*, const AnyVal&, const AnyVal&, const AnyVal&, AnyVal*);
typedef void (*UpdateFn4)(FunctionContext*, const AnyVal&, const AnyVal&, const AnyVal&,
const AnyVal&, AnyVal*);
typedef void (*UpdateFn5)(FunctionContext*, const AnyVal&, const AnyVal&, const AnyVal&,
const AnyVal&, const AnyVal&, AnyVal*);
typedef void (*UpdateFn6)(FunctionContext*, const AnyVal&, const AnyVal&, const AnyVal&,
const AnyVal&, const AnyVal&, const AnyVal&, AnyVal*);
typedef void (*UpdateFn7)(FunctionContext*, const AnyVal&, const AnyVal&, const AnyVal&,
const AnyVal&, const AnyVal&, const AnyVal&, const AnyVal&, AnyVal*);
typedef void (*UpdateFn8)(FunctionContext*, const AnyVal&, const AnyVal&, const AnyVal&,
const AnyVal&, const AnyVal&, const AnyVal&, const AnyVal&, const AnyVal&,
AnyVal*);
typedef StringVal (*SerializeFn)(FunctionContext*, const StringVal&);
typedef AnyVal (*GetValueFn)(FunctionContext*, const AnyVal&);
typedef AnyVal (*FinalizeFn)(FunctionContext*, const AnyVal&);
Status AggFnEvaluator::create(ObjectPool* pool, const TExpr& desc, AggFnEvaluator** result) {
return create(pool, desc, false, result);
}
Status AggFnEvaluator::create(ObjectPool* pool, const TExpr& desc, bool is_analytic_fn,
AggFnEvaluator** result) {
*result = pool->add(new AggFnEvaluator(desc.nodes[0], is_analytic_fn));
int node_idx = 0;
for (int i = 0; i < desc.nodes[0].num_children; ++i) {
++node_idx;
Expr* expr = nullptr;
ExprContext* ctx = nullptr;
RETURN_IF_ERROR(
Expr::create_tree_from_thrift(pool, desc.nodes, nullptr, &node_idx, &expr, &ctx));
(*result)->_input_exprs_ctxs.push_back(ctx);
}
return Status::OK();
}
AggFnEvaluator::AggFnEvaluator(const TExprNode& desc, bool is_analytic_fn)
: _fn(desc.fn),
_is_merge(desc.agg_expr.is_merge_agg),
_is_analytic_fn(is_analytic_fn),
_return_type(TypeDescriptor::from_thrift(desc.fn.ret_type)),
_intermediate_type(TypeDescriptor::from_thrift(desc.fn.aggregate_fn.intermediate_type)),
_function_type(desc.fn.binary_type),
_total_mem_consumption(0),
_accumulated_mem_consumption(0),
_intermediate_slot_desc(nullptr),
_output_slot_desc(nullptr),
_init_fn(nullptr),
_update_fn(nullptr),
_remove_fn(nullptr),
_merge_fn(nullptr),
_serialize_fn(nullptr),
_get_value_fn(nullptr),
_finalize_fn(nullptr) {
if (_fn.name.function_name == "count") {
_agg_op = COUNT;
} else if (_fn.name.function_name == "min") {
_agg_op = MIN;
} else if (_fn.name.function_name == "max") {
_agg_op = MAX;
} else if (_fn.name.function_name == "sum") {
_agg_op = SUM;
} else if (_fn.name.function_name == "avg") {
_agg_op = AVG;
} else if (_fn.name.function_name == "ndv" || _fn.name.function_name == "ndv_no_finalize") {
_agg_op = NDV;
} else if (_fn.name.function_name == "count_distinct" ||
_fn.name.function_name == "count_distinct") {
_agg_op = COUNT_DISTINCT;
} else if (_fn.name.function_name == "sum_distinct" ||
_fn.name.function_name == "sum_distinct") {
_agg_op = SUM_DISTINCT;
} else if (_fn.name.function_name == "hll_union_agg") {
_agg_op = HLL_UNION_AGG;
} else {
_agg_op = OTHER;
}
}
Status AggFnEvaluator::prepare(RuntimeState* state, const RowDescriptor& desc, MemPool* pool,
const SlotDescriptor* intermediate_slot_desc,
const SlotDescriptor* output_slot_desc,
const std::shared_ptr<MemTracker>& mem_tracker,
FunctionContext** agg_fn_ctx) {
DCHECK(pool != nullptr);
DCHECK(intermediate_slot_desc != nullptr);
DCHECK(_intermediate_slot_desc == nullptr);
_output_slot_desc = output_slot_desc;
//DCHECK(_intermediate_slot_desc == nullptr);
_intermediate_slot_desc = intermediate_slot_desc;
_string_buffer_len = 0;
_mem_tracker = mem_tracker;
Status status = Expr::prepare(_input_exprs_ctxs, state, desc, _mem_tracker);
RETURN_IF_ERROR(status);
ObjectPool* obj_pool = state->obj_pool();
for (int i = 0; i < _input_exprs_ctxs.size(); ++i) {
_staging_input_vals.push_back(
create_any_val(obj_pool, input_expr_ctxs()[i]->root()->type()));
}
// window has intermediate_slot_type
if (_intermediate_slot_desc != nullptr) {
_staging_intermediate_val = create_any_val(obj_pool, _intermediate_slot_desc->type());
_staging_merge_input_val = create_any_val(obj_pool, _intermediate_slot_desc->type());
}
//_staging_output_val = create_any_val(obj_pool, _output_slot_desc->type());
_is_multi_distinct = false;
if (_agg_op == AggregationOp::COUNT_DISTINCT) {
_hybrid_map.reset(new HybridMap(TYPE_VARCHAR));
_is_multi_distinct = true;
_string_buffer.reset(new char[1024]);
_string_buffer_len = 1024;
} else if (_agg_op == AggregationOp::SUM_DISTINCT) {
_hybrid_map.reset(new HybridMap(input_expr_ctxs()[0]->root()->type().type));
_is_multi_distinct = true;
}
// TODO: this should be made identical for the builtin and UDA case by
// putting all this logic in an improved opcode registry.
// Load the function pointers. Merge is not required if this is evaluating an
// analytic function.
if (_fn.aggregate_fn.init_fn_symbol.empty() || _fn.aggregate_fn.update_fn_symbol.empty() ||
(!_is_analytic_fn && _fn.aggregate_fn.merge_fn_symbol.empty())) {
// This path is only for partially implemented builtins.
DCHECK_EQ(_fn.binary_type, TFunctionBinaryType::BUILTIN);
std::stringstream ss;
ss << "Function " << _fn.name.function_name << " is not implemented.";
return Status::InternalError(ss.str());
}
// Load the function pointers.
RETURN_IF_ERROR(UserFunctionCache::instance()->get_function_ptr(
_fn.id, _fn.aggregate_fn.init_fn_symbol, _fn.hdfs_location, _fn.checksum, &_init_fn,
nullptr));
RETURN_IF_ERROR(UserFunctionCache::instance()->get_function_ptr(
_fn.id, _fn.aggregate_fn.update_fn_symbol, _fn.hdfs_location, _fn.checksum, &_update_fn,
nullptr));
// Merge() is not loaded if evaluating the agg fn as an analytic function.
if (!_is_analytic_fn) {
RETURN_IF_ERROR(UserFunctionCache::instance()->get_function_ptr(
_fn.id, _fn.aggregate_fn.merge_fn_symbol, _fn.hdfs_location, _fn.checksum,
&_merge_fn, nullptr));
}
// Serialize and Finalize are optional
if (!_fn.aggregate_fn.serialize_fn_symbol.empty()) {
RETURN_IF_ERROR(UserFunctionCache::instance()->get_function_ptr(
_fn.id, _fn.aggregate_fn.serialize_fn_symbol, _fn.hdfs_location, _fn.checksum,
&_serialize_fn, nullptr));
}
if (!_fn.aggregate_fn.finalize_fn_symbol.empty()) {
RETURN_IF_ERROR(UserFunctionCache::instance()->get_function_ptr(
_fn.id, _fn.aggregate_fn.finalize_fn_symbol, _fn.hdfs_location, _fn.checksum,
&_finalize_fn, nullptr));
}
if (!_fn.aggregate_fn.get_value_fn_symbol.empty()) {
RETURN_IF_ERROR(UserFunctionCache::instance()->get_function_ptr(
_fn.id, _fn.aggregate_fn.get_value_fn_symbol, _fn.hdfs_location, _fn.checksum,
&_get_value_fn, nullptr));
}
if (!_fn.aggregate_fn.remove_fn_symbol.empty()) {
RETURN_IF_ERROR(UserFunctionCache::instance()->get_function_ptr(
_fn.id, _fn.aggregate_fn.remove_fn_symbol, _fn.hdfs_location, _fn.checksum,
&_remove_fn, nullptr));
}
std::vector<FunctionContext::TypeDesc> arg_types;
for (int j = 0; j < _input_exprs_ctxs.size(); ++j) {
arg_types.push_back(
AnyValUtil::column_type_to_type_desc(_input_exprs_ctxs[j]->root()->type()));
}
FunctionContext::TypeDesc intermediate_type =
AnyValUtil::column_type_to_type_desc(_intermediate_type);
FunctionContext::TypeDesc output_type =
AnyValUtil::column_type_to_type_desc(_output_slot_desc->type());
*agg_fn_ctx = FunctionContextImpl::create_context(state, pool, intermediate_type, output_type,
arg_types, 0, false);
return Status::OK();
}
Status AggFnEvaluator::open(RuntimeState* state, FunctionContext* agg_fn_ctx) {
RETURN_IF_ERROR(Expr::open(_input_exprs_ctxs, state));
// Now that we have opened all our input exprs, it is safe to evaluate any constant
// values for the UDA's FunctionContext (we cannot evaluate exprs before calling Open()
// on them).
std::vector<AnyVal*> constant_args(_input_exprs_ctxs.size());
for (int i = 0; i < _input_exprs_ctxs.size(); ++i) {
constant_args[i] = _input_exprs_ctxs[i]->root()->get_const_val(_input_exprs_ctxs[i]);
}
agg_fn_ctx->impl()->set_constant_args(constant_args);
return Status::OK();
}
void AggFnEvaluator::close(RuntimeState* state) {
Expr::close(_input_exprs_ctxs, state);
if (UNLIKELY(_total_mem_consumption > 0)) {
_mem_tracker->Release(_total_mem_consumption);
}
}
// Utility to put val into an AnyVal struct
inline void AggFnEvaluator::set_any_val(const void* slot, const TypeDescriptor& type, AnyVal* dst) {
if (slot == nullptr) {
dst->is_null = true;
return;
}
dst->is_null = false;
switch (type.type) {
case TYPE_NULL:
return;
case TYPE_BOOLEAN:
reinterpret_cast<BooleanVal*>(dst)->val = *reinterpret_cast<const bool*>(slot);
return;
case TYPE_TINYINT:
reinterpret_cast<TinyIntVal*>(dst)->val = *reinterpret_cast<const int8_t*>(slot);
return;
case TYPE_SMALLINT:
reinterpret_cast<SmallIntVal*>(dst)->val = *reinterpret_cast<const int16_t*>(slot);
return;
case TYPE_INT:
reinterpret_cast<IntVal*>(dst)->val = *reinterpret_cast<const int32_t*>(slot);
return;
case TYPE_BIGINT:
reinterpret_cast<BigIntVal*>(dst)->val = *reinterpret_cast<const int64_t*>(slot);
return;
case TYPE_FLOAT:
reinterpret_cast<FloatVal*>(dst)->val = *reinterpret_cast<const float*>(slot);
return;
case TYPE_DOUBLE:
reinterpret_cast<DoubleVal*>(dst)->val = *reinterpret_cast<const double*>(slot);
return;
case TYPE_CHAR:
case TYPE_VARCHAR:
case TYPE_HLL:
case TYPE_OBJECT:
case TYPE_STRING:
reinterpret_cast<const StringValue*>(slot)->to_string_val(
reinterpret_cast<StringVal*>(dst));
return;
case TYPE_DATE:
case TYPE_DATETIME:
reinterpret_cast<const DateTimeValue*>(slot)->to_datetime_val(
reinterpret_cast<DateTimeVal*>(dst));
return;
case TYPE_DECIMALV2:
reinterpret_cast<DecimalV2Val*>(dst)->val =
reinterpret_cast<const PackedInt128*>(slot)->value;
return;
case TYPE_LARGEINT:
memcpy(&reinterpret_cast<LargeIntVal*>(dst)->val, slot, sizeof(__int128));
return;
default:
DCHECK(false) << "NYI";
}
}
inline void AggFnEvaluator::set_output_slot(const AnyVal* src, const SlotDescriptor* dst_slot_desc,
Tuple* dst) {
if (src->is_null) {
dst->set_null(dst_slot_desc->null_indicator_offset());
return;
}
dst->set_not_null(dst_slot_desc->null_indicator_offset());
void* slot = dst->get_slot(dst_slot_desc->tuple_offset());
switch (dst_slot_desc->type().type) {
case TYPE_NULL:
return;
case TYPE_BOOLEAN:
*reinterpret_cast<bool*>(slot) = reinterpret_cast<const BooleanVal*>(src)->val;
return;
case TYPE_TINYINT:
*reinterpret_cast<int8_t*>(slot) = reinterpret_cast<const TinyIntVal*>(src)->val;
return;
case TYPE_SMALLINT:
*reinterpret_cast<int16_t*>(slot) = reinterpret_cast<const SmallIntVal*>(src)->val;
return;
case TYPE_INT:
*reinterpret_cast<int32_t*>(slot) = reinterpret_cast<const IntVal*>(src)->val;
return;
case TYPE_BIGINT:
*reinterpret_cast<int64_t*>(slot) = reinterpret_cast<const BigIntVal*>(src)->val;
return;
case TYPE_FLOAT:
*reinterpret_cast<float*>(slot) = reinterpret_cast<const FloatVal*>(src)->val;
return;
case TYPE_DOUBLE:
*reinterpret_cast<double*>(slot) = reinterpret_cast<const DoubleVal*>(src)->val;
return;
case TYPE_CHAR:
case TYPE_VARCHAR:
case TYPE_HLL:
case TYPE_OBJECT:
case TYPE_STRING:
*reinterpret_cast<StringValue*>(slot) =
StringValue::from_string_val(*reinterpret_cast<const StringVal*>(src));
return;
case TYPE_DATE:
case TYPE_DATETIME:
*reinterpret_cast<DateTimeValue*>(slot) =
DateTimeValue::from_datetime_val(*reinterpret_cast<const DateTimeVal*>(src));
return;
case TYPE_DECIMALV2:
*reinterpret_cast<PackedInt128*>(slot) = reinterpret_cast<const DecimalV2Val*>(src)->val;
return;
case TYPE_LARGEINT: {
memcpy(slot, &reinterpret_cast<const LargeIntVal*>(src)->val, sizeof(__int128));
return;
}
default:
DCHECK(false) << "NYI";
}
}
bool AggFnEvaluator::is_in_hybridmap(void* input_val, Tuple* dst, bool* is_add_buckets) {
bool is_in_hashset = false;
HybridSetBase* _set_ptr = nullptr;
_set_ptr = _hybrid_map->find_or_insert_set(reinterpret_cast<uint64_t>(dst), is_add_buckets);
is_in_hashset = _set_ptr->find(input_val);
if (!is_in_hashset) {
_set_ptr->insert(input_val);
}
return is_in_hashset;
}
// This function would be replaced in codegen.
void AggFnEvaluator::init(FunctionContext* agg_fn_ctx, Tuple* dst) {
DCHECK(_init_fn != nullptr);
reinterpret_cast<InitFn>(_init_fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(_staging_intermediate_val, _intermediate_slot_desc, dst);
agg_fn_ctx->impl()->set_num_updates(0);
agg_fn_ctx->impl()->set_num_removes(0);
}
void AggFnEvaluator::update_mem_limlits(int len) {
_accumulated_mem_consumption += len;
// per 16M , update mem_tracker one time
if (UNLIKELY(_accumulated_mem_consumption > 16777216)) {
_mem_tracker->Consume(_accumulated_mem_consumption);
_total_mem_consumption += _accumulated_mem_consumption;
_accumulated_mem_consumption = 0;
}
}
AggFnEvaluator::~AggFnEvaluator() {}
inline void AggFnEvaluator::update_mem_trackers(bool is_filter, bool is_add_buckets, int len) {
if (!is_filter) {
int total_len = len;
if (is_add_buckets) {
total_len += BIGINT_SIZE; //map's key size
}
update_mem_limlits(total_len);
}
}
bool AggFnEvaluator::count_distinct_data_filter(TupleRow* row, Tuple* dst) {
std::vector<int32_t> vec_string_len;
int total_len = 0;
// 1. calculate the total_len of all input parameters
for (int i = 0; i < input_expr_ctxs().size(); ++i) {
void* src_slot = input_expr_ctxs()[i]->get_value(row);
set_any_val(src_slot, input_expr_ctxs()[i]->root()->type(), _staging_input_vals[i]);
if (_staging_input_vals[i]->is_null) {
// even though only one parameter is null, the row will be abandon
return true;
}
if (input_expr_ctxs()[i]->root()->type().is_string_type()) {
const int string_len = reinterpret_cast<const StringVal*>(_staging_input_vals[i])->len;
vec_string_len.push_back(string_len);
total_len += string_len;
}
total_len += get_real_byte_size(input_expr_ctxs()[i]->root()->type().type);
}
int32_t vec_size = vec_string_len.size();
int32_t int_size = INT_SIZE;
total_len += vec_size * int_size;
// 2. merge multi parameter into one parameter(StringVal)
if (_string_buffer_len < total_len) {
_string_buffer_len = ((total_len << 10) + 1) >> 10; // (len/1024+1)*1024
_string_buffer.reset(new char[_string_buffer_len]);
}
StringValue string_val(_string_buffer.get(), total_len);
// the content of StringVal:
// header: the STRING_VALUE's len
// body: all input parameters' content
char* begin = string_val.ptr;
for (int i = 0; i < vec_size; i++) {
memcpy(begin, &vec_string_len[0], int_size);
begin += int_size;
}
for (int i = 0; i < input_expr_ctxs().size(); ++i) {
switch (input_expr_ctxs()[i]->root()->type().type) {
case TYPE_NULL:
return true;
case TYPE_BOOLEAN: {
*begin = (uint8_t) reinterpret_cast<BooleanVal*>(_staging_input_vals[i])->val;
begin += TINYINT_SIZE;
break;
}
case TYPE_TINYINT: {
memcpy(begin, &reinterpret_cast<TinyIntVal*>(_staging_input_vals[i])->val,
TINYINT_SIZE);
begin += TINYINT_SIZE;
break;
}
case TYPE_SMALLINT: {
memcpy(begin, &reinterpret_cast<SmallIntVal*>(_staging_input_vals[i])->val,
SMALLINT_SIZE);
begin += SMALLINT_SIZE;
break;
}
case TYPE_INT: {
memcpy(begin, &reinterpret_cast<IntVal*>(_staging_input_vals[i])->val, INT_SIZE);
begin += INT_SIZE;
break;
}
case TYPE_BIGINT: {
memcpy(begin, &reinterpret_cast<BigIntVal*>(_staging_input_vals[i])->val, BIGINT_SIZE);
begin += BIGINT_SIZE;
break;
}
case TYPE_LARGEINT: {
LargeIntVal* value = reinterpret_cast<LargeIntVal*>(_staging_input_vals[i]);
memcpy(begin, &value->val, LARGEINT_SIZE);
begin += LARGEINT_SIZE;
break;
}
case TYPE_FLOAT: {
memcpy(begin, &reinterpret_cast<FloatVal*>(_staging_input_vals[i])->val, FLOAT_SIZE);
begin += FLOAT_SIZE;
break;
}
case TYPE_DOUBLE: {
memcpy(begin, &reinterpret_cast<DoubleVal*>(_staging_input_vals[i])->val, DOUBLE_SIZE);
begin += DOUBLE_SIZE;
break;
}
case TYPE_DECIMALV2: {
DecimalV2Val* value = reinterpret_cast<DecimalV2Val*>(_staging_input_vals[i]);
memcpy(begin, value, sizeof(DecimalV2Val));
begin += sizeof(DecimalV2Val);
break;
}
case TYPE_CHAR:
case TYPE_VARCHAR:
case TYPE_HLL:
case TYPE_OBJECT:
case TYPE_STRING: {
StringVal* value = reinterpret_cast<StringVal*>(_staging_input_vals[i]);
memcpy(begin, value->ptr, value->len);
begin += value->len;
break;
}
case TYPE_DATE:
case TYPE_DATETIME: {
DateTimeVal* value = reinterpret_cast<DateTimeVal*>(_staging_input_vals[i]);
memcpy(begin, &value->packed_time, DATETIME_SIZE);
begin += DATETIME_SIZE;
break;
}
default: {
DCHECK(0) << "FYI" << input_expr_ctxs()[i]->root()->type();
}
}
}
DCHECK(begin == string_val.ptr + string_val.len)
<< "COUNT_DISTINCT: StringVal's len doesn't match";
bool is_add_buckets = false;
bool is_filter = is_in_hybridmap(&string_val, dst, &is_add_buckets);
update_mem_trackers(is_filter, is_add_buckets, string_val.len);
return is_filter;
}
bool AggFnEvaluator::sum_distinct_data_filter(TupleRow* row, Tuple* dst) {
DCHECK(input_expr_ctxs().size() == 1);
void* src_slot = input_expr_ctxs()[0]->get_value(row);
set_any_val(src_slot, input_expr_ctxs()[0]->root()->type(), _staging_input_vals[0]);
if (_staging_input_vals[0]->is_null) {
// if the parameter is null, the row will be abandon
return true;
}
bool is_filter = false;
bool is_add_buckets = false;
switch (input_expr_ctxs()[0]->root()->type().type) {
case TYPE_NULL: {
return true;
}
case TYPE_BIGINT: {
const BigIntVal* value = reinterpret_cast<BigIntVal*>(_staging_input_vals[0]);
is_filter = is_in_hybridmap((void*)&(value->val), dst, &is_add_buckets);
update_mem_trackers(is_filter, is_add_buckets, BIGINT_SIZE);
return is_filter;
}
case TYPE_FLOAT: {
const FloatVal* value = reinterpret_cast<FloatVal*>(_staging_input_vals[0]);
is_filter = is_in_hybridmap((void*)&(value->val), dst, &is_add_buckets);
update_mem_trackers(is_filter, is_add_buckets, FLOAT_SIZE);
return is_filter;
}
case TYPE_DOUBLE: {
const DoubleVal* value = reinterpret_cast<DoubleVal*>(_staging_input_vals[0]);
is_filter = is_in_hybridmap((void*)&(value->val), dst, &is_add_buckets);
update_mem_trackers(is_filter, is_add_buckets, DOUBLE_SIZE);
return is_filter;
}
case TYPE_DECIMALV2: {
const DecimalV2Val* value = reinterpret_cast<DecimalV2Val*>(_staging_input_vals[0]);
DecimalV2Value temp_value = DecimalV2Value::from_decimal_val(*value);
is_filter = is_in_hybridmap((void*)&(temp_value), dst, &is_add_buckets);
update_mem_trackers(is_filter, is_add_buckets, DECIMALV2_SIZE);
return is_filter;
}
case TYPE_LARGEINT: {
const LargeIntVal* value = reinterpret_cast<LargeIntVal*>(_staging_input_vals[0]);
is_filter = is_in_hybridmap((void*)&(value->val), dst, &is_add_buckets);
update_mem_trackers(is_filter, is_add_buckets, LARGEINT_SIZE);
return is_filter;
}
default: {
DCHECK(0) << "FYI";
}
}
return false;
}
void AggFnEvaluator::update_or_merge(FunctionContext* agg_fn_ctx, TupleRow* row, Tuple* dst,
void* fn) {
if (fn == nullptr) {
return;
}
bool dst_null = dst->is_null(_intermediate_slot_desc->null_indicator_offset());
void* dst_slot = nullptr;
if (!dst_null) {
dst_slot = dst->get_slot(_intermediate_slot_desc->tuple_offset());
}
set_any_val(dst_slot, _intermediate_slot_desc->type(), _staging_intermediate_val);
if (_is_multi_distinct) {
if (_agg_op == COUNT_DISTINCT) {
bool is_need_filter = count_distinct_data_filter(row, dst);
if (is_need_filter) {
_staging_input_vals[0]->is_null = true;
}
} else if (_agg_op == SUM_DISTINCT) {
bool is_need_filter = sum_distinct_data_filter(row, dst);
if (is_need_filter) {
_staging_input_vals[0]->is_null = true;
}
} else {
DCHECK(0) << "we only support count_distinct and sum_distinct";
}
} else {
for (int i = 0; i < input_expr_ctxs().size(); ++i) {
void* src_slot = input_expr_ctxs()[i]->get_value(row);
set_any_val(src_slot, input_expr_ctxs()[i]->root()->type(), _staging_input_vals[i]);
}
}
// TODO: this part is not so good and not scalable. It can be replaced with
// codegen but we can also consider leaving it for the first few cases for
// debugging.
// if _agg_op is TAggregationOp::COUNT_DISTINCT, it has only one
// input parameter, we consider the first parameter as the only input parameter
if (_is_multi_distinct && _agg_op == AggregationOp::COUNT_DISTINCT) {
reinterpret_cast<UpdateFn1>(fn)(agg_fn_ctx, *_staging_input_vals[0],
_staging_intermediate_val);
} else {
switch (input_expr_ctxs().size()) {
case 0:
reinterpret_cast<UpdateFn0>(fn)(agg_fn_ctx, _staging_intermediate_val);
break;
case 1:
reinterpret_cast<UpdateFn1>(fn)(agg_fn_ctx, *_staging_input_vals[0],
_staging_intermediate_val);
break;
case 2:
reinterpret_cast<UpdateFn2>(fn)(agg_fn_ctx, *_staging_input_vals[0],
*_staging_input_vals[1], _staging_intermediate_val);
break;
case 3:
reinterpret_cast<UpdateFn3>(fn)(agg_fn_ctx, *_staging_input_vals[0],
*_staging_input_vals[1], *_staging_input_vals[2],
_staging_intermediate_val);
break;
case 4:
reinterpret_cast<UpdateFn4>(fn)(agg_fn_ctx, *_staging_input_vals[0],
*_staging_input_vals[1], *_staging_input_vals[2],
*_staging_input_vals[3], _staging_intermediate_val);
break;
case 5:
reinterpret_cast<UpdateFn5>(fn)(agg_fn_ctx, *_staging_input_vals[0],
*_staging_input_vals[1], *_staging_input_vals[2],
*_staging_input_vals[3], *_staging_input_vals[4],
_staging_intermediate_val);
break;
case 6:
reinterpret_cast<UpdateFn6>(fn)(agg_fn_ctx, *_staging_input_vals[0],
*_staging_input_vals[1], *_staging_input_vals[2],
*_staging_input_vals[3], *_staging_input_vals[4],
*_staging_input_vals[5], _staging_intermediate_val);
break;
case 7:
reinterpret_cast<UpdateFn7>(fn)(
agg_fn_ctx, *_staging_input_vals[0], *_staging_input_vals[1],
*_staging_input_vals[2], *_staging_input_vals[3], *_staging_input_vals[4],
*_staging_input_vals[5], *_staging_input_vals[6], _staging_intermediate_val);
break;
case 8:
reinterpret_cast<UpdateFn8>(fn)(agg_fn_ctx, *_staging_input_vals[0],
*_staging_input_vals[1], *_staging_input_vals[2],
*_staging_input_vals[3], *_staging_input_vals[4],
*_staging_input_vals[5], *_staging_input_vals[6],
*_staging_input_vals[7], _staging_intermediate_val);
break;
default:
DCHECK(false) << "NYI";
}
}
set_output_slot(_staging_intermediate_val, _intermediate_slot_desc, dst);
}
void AggFnEvaluator::update(FunctionContext* agg_fn_ctx, TupleRow* row, Tuple* dst, void* fn,
MemPool* pool) {
return update_or_merge(agg_fn_ctx, row, dst, fn);
}
void AggFnEvaluator::merge(FunctionContext* agg_fn_ctx, TupleRow* row, Tuple* dst, MemPool* pool) {
return update_or_merge(agg_fn_ctx, row, dst, _merge_fn);
}
static void set_any_val2(const SlotDescriptor* desc, Tuple* tuple, AnyVal* dst) {
bool is_null = tuple->is_null(desc->null_indicator_offset());
void* slot = nullptr;
if (!is_null) {
slot = tuple->get_slot(desc->tuple_offset());
}
AnyValUtil::set_any_val(slot, desc->type(), dst);
}
void AggFnEvaluator::merge(FunctionContext* agg_fn_ctx, Tuple* src, Tuple* dst) {
DCHECK(_merge_fn != nullptr);
set_any_val2(_intermediate_slot_desc, dst, _staging_intermediate_val);
set_any_val2(_intermediate_slot_desc, src, _staging_merge_input_val);
// The merge fn always takes one input argument.
reinterpret_cast<UpdateFn1>(_merge_fn)(agg_fn_ctx, *_staging_merge_input_val,
_staging_intermediate_val);
set_output_slot(_staging_intermediate_val, _intermediate_slot_desc, dst);
}
void AggFnEvaluator::choose_update_or_merge(FunctionContext* agg_fn_ctx, TupleRow* row,
Tuple* dst) {
if (_is_merge) {
return update_or_merge(agg_fn_ctx, row, dst, _merge_fn);
} else {
return update_or_merge(agg_fn_ctx, row, dst, _update_fn);
}
}
void AggFnEvaluator::serialize_or_finalize(FunctionContext* agg_fn_ctx, Tuple* src,
const SlotDescriptor* dst_slot_desc, Tuple* dst,
void* fn, bool add_null) {
// DCHECK_EQ(dst_slot_desc->type().type, _return_type.type);
if (src == nullptr) {
src = dst;
}
if (fn == nullptr && src == dst) {
return;
}
// same
bool src_slot_null = add_null || src->is_null(_intermediate_slot_desc->null_indicator_offset());
void* src_slot = nullptr;
if (!src_slot_null) {
src_slot = src->get_slot(_intermediate_slot_desc->tuple_offset());
}
// not same
// if (_is_analytic_fn) {
// No fn was given but the src and dst tuples are different (doing a finalize()).
// Just copy the src slot into the dst tuple.
if (fn == nullptr) {
DCHECK_EQ(_intermediate_slot_desc->type(), dst_slot_desc->type());
RawValue::write(src_slot, dst, dst_slot_desc, nullptr);
return;
}
// }
set_any_val(src_slot, _intermediate_slot_desc->type(), _staging_intermediate_val);
switch (dst_slot_desc->type().type) {
case TYPE_BOOLEAN: {
typedef BooleanVal (*Fn)(FunctionContext*, AnyVal*);
BooleanVal v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
case TYPE_TINYINT: {
typedef TinyIntVal (*Fn)(FunctionContext*, AnyVal*);
TinyIntVal v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
case TYPE_SMALLINT: {
typedef SmallIntVal (*Fn)(FunctionContext*, AnyVal*);
SmallIntVal v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
case TYPE_INT: {
typedef IntVal (*Fn)(FunctionContext*, AnyVal*);
IntVal v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
case TYPE_BIGINT: {
typedef BigIntVal (*Fn)(FunctionContext*, AnyVal*);
BigIntVal v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
case TYPE_FLOAT: {
typedef FloatVal (*Fn)(FunctionContext*, AnyVal*);
FloatVal v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
case TYPE_DOUBLE: {
typedef DoubleVal (*Fn)(FunctionContext*, AnyVal*);
DoubleVal v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
case TYPE_CHAR:
case TYPE_VARCHAR:
case TYPE_HLL:
case TYPE_OBJECT:
case TYPE_STRING: {
typedef StringVal (*Fn)(FunctionContext*, AnyVal*);
StringVal v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
case TYPE_DATE:
case TYPE_DATETIME: {
typedef DateTimeVal (*Fn)(FunctionContext*, AnyVal*);
DateTimeVal v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
case TYPE_DECIMALV2: {
typedef DecimalV2Val (*Fn)(FunctionContext*, AnyVal*);
DecimalV2Val v = reinterpret_cast<Fn>(fn)(agg_fn_ctx, _staging_intermediate_val);
set_output_slot(&v, dst_slot_desc, dst);
break;
}
default:
DCHECK(false) << "NYI";
}
}
void AggFnEvaluator::serialize(FunctionContext* agg_fn_ctx, Tuple* tuple) {
serialize_or_finalize(agg_fn_ctx, nullptr, _intermediate_slot_desc, tuple, _serialize_fn);
}
//void AggFnEvaluator::finalize(FunctionContext* agg_fn_ctx, Tuple* tuple) {
// serialize_or_finalize(agg_fn_ctx, nullptr, _output_slot_desc, tuple, _finalize_fn);
//}
std::string AggFnEvaluator::debug_string(const std::vector<AggFnEvaluator*>& exprs) {
std::stringstream out;
out << "[";
for (int i = 0; i < exprs.size(); ++i) {
out << (i == 0 ? "" : " ") << exprs[i]->debug_string();
}
out << "]";
return out.str();
}
std::string AggFnEvaluator::debug_string() const {
std::stringstream out;
out << "AggFnEvaluator(op=" << _agg_op;
out << ")";
return out.str();
}
} // namespace doris
#if defined(__clang__)
#pragma clang diagnostic pop
#elif defined(__GNUC__) || defined(__GNUG__)
#pragma GCC diagnostic pop
#endif
Reference in New Issue View Git Blame Copy Permalink