Try to revert back to boost to benchmark

blue-yonder · Oct 11, 2023 · 9194ffc · 9194ffc
1 parent 7086bbc
commit 9194ffc
Show file tree

Hide file tree

Showing 10 changed files with 41 additions and 6 deletions.
diff --git a/cmake_scripts/default_build_setup.cmake b/cmake_scripts/default_build_setup.cmake
@@ -64,3 +64,11 @@ add_custom_target(refresh_cmake_configuration
 	cmake -E touch ${CMAKE_PARENT_LIST_FILE} # make cmake detect configuration is changed on NEXT build
 	COMMENT "Forcing refreshing of the CMake configuration. This allows to use globbing safely."
 )
+
+if(WIN32)
+    add_definitions("-std=c++17")
+    link_directories("$ENV{PYTHON}/libs")
+    set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS "TRUE")
+    set(Boost_USE_STATIC_RUNTIME "ON")
+    set(Boost_USE_STATIC_LIBS "ON")
+endif()
diff --git a/cpp/cpp_odbc/CMakeLists.txt b/cpp/cpp_odbc/CMakeLists.txt
@@ -1,3 +1,6 @@
+find_package(Boost REQUIRED COMPONENTS locale)
+include_directories(SYSTEM ${Boost_INCLUDE_DIRS})
+
 find_package(odbc REQUIRED)
 include_directories(SYSTEM ${Odbc_INCLUDE_DIR})
 

diff --git a/cpp/cpp_odbc/Library/src/level2/level1_connector.cpp b/cpp/cpp_odbc/Library/src/level2/level1_connector.cpp
@@ -20,6 +20,8 @@
 #include <locale>
 #include <sstream>
 
+#include <boost/locale.hpp>
+
 namespace impl {
 
 template <typename Output_Handle, typename Input_Handle>
@@ -430,7 +432,12 @@ column_description level1_connector::do_describe_column_wide(statement_handle co
 
     impl::throw_on_error(return_code, *this, handle);
 
-    std::string const utf8_name = std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t>{}.to_bytes(name);
+    // TODO-Simeon: throws
+    // throw std::invalid_argument("Simeon conversion in level1_connector");
+    // TODO-Simeon: throws
+
+    //std::string const utf8_name = std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t>{}.to_bytes(name);
+    auto const utf8_name = boost::locale::conv::utf_to_utf<char>(static_cast<std::u16string>(name));
     return {utf8_name, data_type, size, decimal_digits, allows_nullable};
 }
 

diff --git a/cpp/turbodbc/CMakeLists.txt b/cpp/turbodbc/CMakeLists.txt
@@ -1,4 +1,4 @@
-find_package(Boost REQUIRED COMPONENTS system date_time)
+find_package(Boost REQUIRED COMPONENTS system date_time locale)
 include_directories(SYSTEM ${Boost_INCLUDE_DIRS})
 
 find_package(odbc REQUIRED)

diff --git a/cpp/turbodbc/Library/src/connect.cpp b/cpp/turbodbc/Library/src/connect.cpp
@@ -1,6 +1,8 @@
 #include "turbodbc/connect.h"
 #include <cpp_odbc/make_environment.h>
 
+using namespace std;
+
 turbodbc::connection turbodbc::connect(std::string const & connection_string, options options)
 {
 	auto environment = cpp_odbc::make_environment();

diff --git a/cpp/turbodbc/Library/src/cursor.cpp b/cpp/turbodbc/Library/src/cursor.cpp
@@ -14,12 +14,19 @@
 #include <sstream>
 #include <codecvt>
 
+#include <boost/locale.hpp>
+
 namespace {
 
     std::u16string as_utf16(std::string utf8_encoded) {
+        // TODO-Simeon: throws
+        // throw std::invalid_argument("Simeon conversion in cursor");
+        // TODO-Simeon: throws
+
         // not all compilers support the new C++11 conversion facets
-        static std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> converter;
-        return converter.from_bytes(utf8_encoded);
+        return boost::locale::conv::utf_to_utf<char16_t>(utf8_encoded);
+        //static std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> converter;
+        //return converter.from_bytes(utf8_encoded);
     }
 
 }

diff --git a/cpp/turbodbc_arrow/Library/src/arrow_result_set.cpp b/cpp/turbodbc_arrow/Library/src/arrow_result_set.cpp
@@ -15,6 +15,8 @@
 #include <codecvt>
 #include <vector>
 
+#include <boost/locale.hpp>
+
 using arrow::default_memory_pool;
 using arrow::AdaptiveIntBuilder;
 using arrow::ArrayBuilder;
@@ -127,7 +129,11 @@ Status AppendUnicodeStringsToBuilder(size_t rows_in_batch, BuilderType& builder,
         if (element.indicator == SQL_NULL_DATA) {
             ARROW_RETURN_NOT_OK(builder.AppendNullProxy());
         } else {
+            // TODO-Simeon: throws
+            // throw std::invalid_argument("Simeon conversion in arrow_result_set");
+            // TODO-Simeon: throws
             std::u16string str_u16(reinterpret_cast<const char16_t*>(element.data_pointer), element.indicator / 2);
+            //std::string u8string = std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t>{}.to_bytes(str_u16);
             std::string u8string = std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t>{}.to_bytes(str_u16);
             ARROW_RETURN_NOT_OK(builder.AppendProxy(u8string.data(), u8string.size()));
         }

diff --git a/cpp/turbodbc_arrow/Library/src/set_arrow_parameters.cpp b/cpp/turbodbc_arrow/Library/src/set_arrow_parameters.cpp
@@ -14,6 +14,8 @@
 #include <ciso646>
 #include <codecvt>
 
+#include <boost/locale.hpp>
+
 using arrow::BooleanArray;
 using arrow::BinaryArray;
 using arrow::ChunkedArray;

diff --git a/environment.yml b/environment.yml
@@ -4,7 +4,7 @@ channels:
 dependencies:
   - python>=3.9
   - pyarrow>=7.0,<14
-  - sel(unix): unixodbc
+  # - sel(unix): unixodbc
   - c-compiler
   - cxx-compiler
   - gmock=1.10.0

diff --git a/x_read.py b/x_read.py
@@ -36,7 +36,7 @@ def mssql_connect_turbodbc() -> turbodbc.connection.Connection:
     )
 
 
-smoke()
+# smoke()
 
 
 which = input("Which query: ") if len(sys.argv) < 2 else sys.argv[-1]