using pch

Signed-off-by: Yuan Zhou <[email protected]>

native-sql-engine/cpp/src/codegen/arrow_compute/ext/codegen_common.cc

@@ -39,12 +39,7 @@ namespace extra {
 
 std::string BaseCodes() {
   return R"(
-#include <arrow/compute/api.h>
-#include <arrow/record_batch.h>
 
-#include "codegen/arrow_compute/ext/code_generator_base.h"
-#include "precompile/array.h"
-using namespace sparkcolumnarplugin::codegen::arrowcompute::extra;
 )";
 }
 
@@ -601,6 +596,7 @@ arrow::Status CompileCodes(std::string codes, std::string signature) {
   mkdir(outpath.c_str(), S_IRWXU | S_IRWXG | S_IROTH | S_IXOTH);
   std::string prefix = "/spark-columnar-plugin-codegen-";
   std::string cppfile = outpath + prefix + signature + ".cc";
+  std::string objfile = outpath + prefix + signature + ".o";
   std::string libfile = outpath + prefix + signature + ".so";
   std::string jarfile = outpath + prefix + signature + ".jar";
   std::string logfile = outpath + prefix + signature + ".log";
@@ -626,13 +622,15 @@ arrow::Status CompileCodes(std::string codes, std::string signature) {
   }
   std::string env_gcc = std::string(env_gcc_);
 
+  std::string env_codegen_option = " -O3 -march=native ";
   char* env_codegen_option_ = std::getenv("CODEGEN_OPTION");
 
-  if (env_codegen_option_ == nullptr) {
-    env_codegen_option_ = " -O3 -march=native ";
+  if (env_codegen_option_ != nullptr) {
+    env_codegen_option = std::string(env_codegen_option_);
   }
-  std::string env_codegen_option = std::string(env_codegen_option_);
 
+  std::string libwscgfile = GetTempPath() + "/nativesql_include/precompile/wscgapi.hpp";
+  std::string libwscg_pch = GetTempPath() + "/nativesql_include/precompile/wscgapi.hpp.gch";
   const char* env_arrow_dir = std::getenv("LIBARROW_DIR");
   std::string arrow_header;
   std::string arrow_lib, arrow_lib2;
@@ -646,14 +644,35 @@ arrow::Status CompileCodes(std::string codes, std::string signature) {
     arrow_lib2 = " -L" + std::string(env_arrow_dir) + "/lib ";
   }
   // compile the code
-  std::string cmd = env_gcc + " -std=c++14 -Wno-deprecated-declarations " + arrow_header +
-                    arrow_lib + arrow_lib2 + nativesql_header + nativesql_header_2 +
-                    nativesql_lib + cppfile + " -o " + libfile + env_codegen_option +
-                    " -shared -fPIC -lspark_columnar_jni 2> " + logfile;
+  std::string base_dir = GetTempPath();
+  chdir(base_dir.c_str());
+  std::string cmd = "";
+  struct stat pch_stat;
+  auto ret = stat(libwscg_pch.c_str(), &pch_stat);
+  if (ret == -1) {
+    cmd += env_gcc + " -std=c++14 -Wno-deprecated-declarations " + arrow_header +
+                    arrow_lib + arrow_lib2 + nativesql_header + nativesql_header_2
+                    + " -c " +
+                     libwscgfile + env_codegen_option + " -fPIC && ";
+
+  }
+
+  cmd += env_gcc + " -std=c++14 -Wno-deprecated-declarations " + arrow_header +
+                    nativesql_header + nativesql_header_2 + " -c " +
+                     cppfile  + " -o "+ objfile + env_codegen_option + "-fPIC && ";
+  // linking
+  cmd += env_gcc + arrow_lib + arrow_lib2 + nativesql_lib +
+                     objfile + " -o " + libfile + " -lspark_columnar_jni -shared && ";
+
+  // package
+  cmd += "cd " + outpath + " && jar -cf spark-columnar-plugin-codegen-precompile-" +
+        signature + ".jar spark-columnar-plugin-codegen-" + signature + ".so 2>" + logfile;
+
+
 #ifdef DEBUG
   std::cout << cmd << std::endl;
 #endif
-  int ret;
+
   int elapse_time = 0;
   TIME_MICRO(elapse_time, ret, system(cmd.c_str()));
 #ifdef DEBUG
@@ -664,15 +683,6 @@ arrow::Status CompileCodes(std::string codes, std::string signature) {
     std::cout << cmd << std::endl;
     return arrow::Status::Invalid("compilation failed, see ", logfile);
   }
-  cmd = "cd " + outpath + "; jar -cf spark-columnar-plugin-codegen-precompile-" +
-        signature + ".jar spark-columnar-plugin-codegen-" + signature + ".so";
-#ifdef DEBUG
-  std::cout << cmd << std::endl;
-#endif
-  ret = system(cmd.c_str());
-  if (WEXITSTATUS(ret) != EXIT_SUCCESS) {
-    return arrow::Status::Invalid("package jar failed");
-  }
 
 #ifdef DEBUG
   struct stat tstat;

native-sql-engine/cpp/src/codegen/arrow_compute/ext/conditioned_merge_join_kernel.cc

@@ -108,8 +108,7 @@ class ConditionedMergeJoinKernel::Impl {
     auto codegen_ctx = std::make_shared<CodeGenContext>();
     bool use_relation_for_stream = input.empty();
 
-    codegen_ctx->header_codes.push_back(
-        R"(#include "codegen/arrow_compute/ext/array_item_index.h")");
+
 
     std::vector<std::string> prepare_list;
     bool cond_check = false;
@@ -119,7 +118,7 @@ class ConditionedMergeJoinKernel::Impl {
     std::stringstream sort_define_ss;
     std::vector<gandiva::FieldVector> field_list = {left_field_list_, right_field_list_};
 
-    codegen_ctx->header_codes.push_back(R"(#include "codegen/common/sort_relation.h")");
+
     int idx = 0;
     for (auto relation_id : relation_id_) {
       auto relation_list_name = "sort_relation_" + std::to_string(relation_id) + "_";

native-sql-engine/cpp/src/codegen/arrow_compute/ext/conditioned_probe_kernel.cc

@@ -173,8 +173,7 @@ class ConditionedProbeKernel::Impl {
       std::shared_ptr<CodeGenContext>* codegen_ctx_out, int* var_id) {
     auto codegen_ctx = std::make_shared<CodeGenContext>();
 
-    codegen_ctx->header_codes.push_back(
-        R"(#include "codegen/arrow_compute/ext/array_item_index.h")");
+
 
     std::vector<std::string> prepare_list;
     bool cond_check = false;
@@ -192,7 +191,7 @@ class ConditionedProbeKernel::Impl {
     hash_prepare_ss << "RETURN_NOT_OK(typed_dependent_iter_list_" << hash_relation_id_
                     << "->Next("
                     << "&" << relation_list_name << "));" << std::endl;
-    codegen_ctx->header_codes.push_back(R"(#include "codegen/common/hash_relation.h")");
+
 
     hash_define_ss << "std::shared_ptr<HashRelation> " << relation_list_name << ";"
                    << std::endl;

native-sql-engine/cpp/src/codegen/arrow_compute/ext/hash_aggregate_kernel.cc

@@ -215,10 +215,7 @@ class HashAggregateKernel::Impl {
       std::shared_ptr<CodeGenContext>* codegen_ctx_out, int* var_id) {
     auto codegen_ctx = std::make_shared<CodeGenContext>();
 
-    codegen_ctx->header_codes.push_back(
-        R"(#include "codegen/arrow_compute/ext/array_item_index.h")");
-    codegen_ctx->header_codes.push_back(
-        R"(#include "codegen/arrow_compute/ext/actions_impl.h")");
+
 
     std::vector<std::string> prepare_list;
     // 1.0 prepare aggregate input expressions
@@ -243,7 +240,7 @@ class HashAggregateKernel::Impl {
     // 1. Get action list and action_prepare_project_list
     if (key_node_list.size() > 0 &&
         key_node_list[0]->return_type()->id() == arrow::Type::DECIMAL128) {
-      codegen_ctx->header_codes.push_back(R"(#include "precompile/hash_map.h")");
+
       aggr_prepare_ss << "aggr_hash_table_" << level << " = std::make_shared<"
                       << GetTypeString(key_node_list[0]->return_type(), "")
                       << "HashMap>(ctx_->memory_pool());" << std::endl;
@@ -254,7 +251,7 @@ class HashAggregateKernel::Impl {
     } else if (key_node_list.size() > 1 ||
                (key_node_list.size() > 0 &&
                 key_node_list[0]->return_type()->id() == arrow::Type::STRING)) {
-      codegen_ctx->header_codes.push_back(R"(#include "precompile/hash_map.h")");
+
       aggr_prepare_ss << "aggr_hash_table_" << level << " = std::make_shared<"
                       << GetTypeString(arrow::utf8(), "")
                       << "HashMap>(ctx_->memory_pool());" << std::endl;
@@ -263,7 +260,7 @@ class HashAggregateKernel::Impl {
 
     } else if (key_node_list.size() > 0) {
       auto type = key_node_list[0]->return_type();
-      codegen_ctx->header_codes.push_back(R"(#include "precompile/sparse_hash_map.h")");
+
       aggr_prepare_ss << "aggr_hash_table_" << level << " = std::make_shared<"
                       << "SparseHashMap<" << GetCTypeString(type)
                       << ">>(ctx_->memory_pool());" << std::endl;
@@ -308,8 +305,7 @@ class HashAggregateKernel::Impl {
         prepare_ss << "auto " << unsafe_row_name_validity << " = "
                    << project_output_list[i].first.first << "_validity;" << std::endl;
       } else {
-        codegen_ctx->header_codes.push_back(
-            R"(#include "third_party/row_wise_memory/unsafe_row.h")");
+
         std::stringstream unsafe_row_define_ss;
         unsafe_row_define_ss << "std::shared_ptr<UnsafeRow> " << unsafe_row_name
                              << "_unsafe_row = std::make_shared<UnsafeRow>("
@@ -562,7 +558,7 @@ class HashAggregateKernel::Impl {
     if (!result_expr_list_.empty()) {
       codegen_ctx->gandiva_projector = std::make_shared<GandivaProjector>(
           ctx_, arrow::schema(result_field_list_), GetGandivaKernel(result_expr_list_));
-      codegen_ctx->header_codes.push_back(R"(#include "precompile/gandiva_projector.h")");
+
       finish_ss << "RETURN_NOT_OK(gandiva_projector_list_[gp_idx++]->Evaluate(&"
                    "do_hash_"
                    "aggr_finish_"

native-sql-engine/cpp/src/codegen/arrow_compute/ext/sort_kernel.cc

@@ -1987,18 +1987,9 @@ class SortArraysCodegenKernel : public SortArraysToIndicesKernel::Impl {
         GetCachedVariablesClear(key_typed_codegen_list);
 
     return BaseCodes() + R"(
-#include <arrow/buffer.h>
 
-#include <algorithm>
-#include <cmath>
+#include "precompile/wscgapi.hpp"
 
-#include "codegen/arrow_compute/ext/array_item_index.h"
-#include "codegen/common/sort_relation.h"
-#include "precompile/builder.h"
-#include "precompile/type.h"
-#include "third_party/ska_sort.hpp"
-#include "third_party/timsort.hpp"
-using namespace sparkcolumnarplugin::precompile;
 
 class TypedSorterImpl : public CodeGenBase {
  public:

native-sql-engine/cpp/src/codegen/arrow_compute/ext/whole_stage_codegen_kernel.cc

@@ -306,8 +306,7 @@ class WholeStageCodeGenKernel::Impl {
     std::string out_list;
     std::stringstream define_ss;
     codes_ss << BaseCodes() << std::endl;
-    codes_ss << R"(#include "precompile/builder.h")" << std::endl;
-    codes_ss << R"(#include "utils/macros.h")" << std::endl;
+    codes_ss << R"(#include "precompile/wscgapi.hpp")" << std::endl;
     std::vector<std::string> headers;
     for (auto codegen_ctx : codegen_ctx_list) {
       for (auto header : codegen_ctx->header_codes) {

native-sql-engine/cpp/src/precompile/wscgapi.hpp

@@ -0,0 +1,33 @@
+#pragma once
+
+#include <arrow/compute/api.h>
+#include <arrow/record_batch.h>
+
+#include "codegen/arrow_compute/ext/code_generator_base.h"
+#include "precompile/array.h"
+
+#include "precompile/builder.h"
+#include "utils/macros.h"
+#include "codegen/arrow_compute/ext/array_item_index.h"
+#include "codegen/common/hash_relation.h"
+#include "codegen/arrow_compute/ext/actions_impl.h"
+#include "precompile/hash_map.h"
+#include "precompile/sparse_hash_map.h"
+#include "codegen/common/sort_relation.h"
+#include "third_party/row_wise_memory/unsafe_row.h"
+
+#include "precompile/type.h"
+#include "third_party/ska_sort.hpp"
+#include "third_party/timsort.hpp"
+#include "precompile/gandiva.h"
+
+#include <arrow/buffer.h>
+
+#include <algorithm>
+#include <cmath>
+
+#include <tuple>
+#include <numeric>
+using namespace sparkcolumnarplugin::precompile;
+
+using namespace sparkcolumnarplugin::codegen::arrowcompute::extra;