wesm · wesm · Jan 1, 2019 · Jan 1, 2019 · Jan 2, 2019 · Jan 2, 2019
diff --git a/cpp/CMakeLists.txt b/cpp/CMakeLists.txt
@@ -287,8 +287,12 @@ Note that this requires linking Boost statically"
 
     set(BROTLI_MSVC_STATIC_LIB_SUFFIX "-static" CACHE STRING
       "Brotli static lib suffix used on Windows with MSVC (default -static)")
+    set(PROTOBUF_MSVC_STATIC_LIB_SUFFIX "" CACHE STRING
+      "Protobuf static lib suffix used on Windows with MSVC (default is empty string)")
+    set(RE2_MSVC_STATIC_LIB_SUFFIX "_static" CACHE STRING
+      "re2 static lib suffix used on Windows with MSVC (default is _static)")
     set(SNAPPY_MSVC_STATIC_LIB_SUFFIX "_static" CACHE STRING
-      "Snappy static lib suffix used on Windows with MSVC (default is empty string)")
+      "Snappy static lib suffix used on Windows with MSVC (default is _static)")
     set(LZ4_MSVC_STATIC_LIB_SUFFIX "_static" CACHE STRING
       "Lz4 static lib suffix used on Windows with MSVC (default _static)")
     set(ZSTD_MSVC_STATIC_LIB_SUFFIX "_static" CACHE STRING

diff --git a/cpp/build-support/lint_cpp_cli.py b/cpp/build-support/lint_cpp_cli.py
@@ -42,17 +42,21 @@ def _strip_comments(line):
         return m.group(0)
 
 
+def _path(*tokens):
+    return os.path.sep.join(tokens)
+
+
 def lint_file(path):
     fail_rules = [
         # rule, error message, rule-specific exclusions list
         (lambda x: '<mutex>' in x, 'Uses <mutex>', []),
         (lambda x: re.match(_NULLPTR_REGEX, x), 'Uses nullptr', []),
         (lambda x: re.match(_RETURN_NOT_OK_REGEX, x),
          'Use ARROW_RETURN_NOT_OK in header files',
-         ['arrow/status.h',
+         [_path('arrow', 'status.h'),
           'test',
-          'arrow/util/hash.h',
-          'arrow/python/util'])
+          _path('arrow', 'util', 'hash.h'),
+          _path('arrow', 'python', 'util')])
     ]
 
     with open(path) as f:
@@ -69,14 +73,14 @@ def lint_file(path):
 
 
 EXCLUSIONS = [
-    'arrow/python/iterators.h',
-    'arrow/util/hashing.h',
-    'arrow/util/macros.h',
-    'arrow/util/parallel.h',
-    'arrow/vendored',
-    'arrow/visitor_inline.h',
-    'gandiva/cache.h',
-    'gandiva/jni',
+    _path('arrow', 'python', 'iterators.h'),
+    _path('arrow', 'util', 'hashing.h'),
+    _path('arrow', 'util', 'macros.h'),
+    _path('arrow', 'util', 'parallel.h'),
+    _path('arrow', 'vendored'),
+    _path('arrow', 'visitor_inline.h'),
+    _path('gandiva', 'cache.h'),
+    _path('gandiva', 'jni'),
     'test',
     'internal'
 ]

diff --git a/cpp/cmake_modules/FindProtobuf.cmake b/cpp/cmake_modules/FindProtobuf.cmake
@@ -44,12 +44,12 @@ if (EXISTS "${_protobuf_path}/lib/${CMAKE_LIBRARY_ARCHITECTURE}")
   set (lib_dirs "lib/${CMAKE_LIBRARY_ARCHITECTURE}" ${lib_dirs})
 endif ()
 
-find_library (PROTOBUF_LIBRARY NAMES protobuf PATHS
+find_library (PROTOBUF_LIBRARY NAMES protobuf libprotobuf PATHS
   ${_protobuf_path}
   NO_DEFAULT_PATH
   PATH_SUFFIXES ${lib_dirs})
 
-find_library (PROTOC_LIBRARY NAMES protoc PATHS
+find_library (PROTOC_LIBRARY NAMES protoc libprotoc PATHS
   ${_protobuf_path}
   NO_DEFAULT_PATH
   PATH_SUFFIXES ${lib_dirs})
@@ -66,7 +66,7 @@ if (PROTOBUF_INCLUDE_DIR AND PROTOBUF_LIBRARY AND PROTOC_LIBRARY AND PROTOBUF_EX
   get_filename_component (PROTOBUF_LIBS ${PROTOBUF_LIBRARY} PATH)
   set (PROTOBUF_LIB_NAME protobuf)
   set (PROTOC_LIB_NAME protoc)
-  set (PROTOBUF_STATIC_LIB ${PROTOBUF_LIBS}/${CMAKE_STATIC_LIBRARY_PREFIX}${PROTOBUF_LIB_NAME}${CMAKE_STATIC_LIBRARY_SUFFIX})
+  set (PROTOBUF_STATIC_LIB ${PROTOBUF_LIBS}/${CMAKE_STATIC_LIBRARY_PREFIX}${PROTOBUF_LIB_NAME}${PROTOBUF_MSVC_STATIC_LIB_SUFFIX}${CMAKE_STATIC_LIBRARY_SUFFIX})
   set (PROTOC_STATIC_LIB ${PROTOBUF_LIBS}/${CMAKE_STATIC_LIBRARY_PREFIX}${PROTOC_LIB_NAME}${CMAKE_STATIC_LIBRARY_SUFFIX})
 else ()
   set (PROTOBUF_FOUND FALSE)
@@ -81,7 +81,7 @@ if (PROTOBUF_FOUND)
   message (STATUS "Found the Protoc executable: ${PROTOBUF_EXECUTABLE}")
 else()
   if (_protobuf_path)
-    set (PROTOBUF_ERR_MSG "Could not find Protobuf. Looked in ${_protobuf_path}.")
+    set (PROTOBUF_ERR_MSG "Could not find Protobuf. Looked in ${_protobuf_path}")
   else ()
     set (PROTOBUF_ERR_MSG "Could not find Protobuf in system search paths.")
   endif()
@@ -100,4 +100,3 @@ mark_as_advanced (
   PROTOBUF_STATIC_LIB
   PROTOC_STATIC_LIB
 )
-
diff --git a/cpp/cmake_modules/FindRE2.cmake b/cpp/cmake_modules/FindRE2.cmake
@@ -45,14 +45,18 @@ if (EXISTS "${_re2_path}/lib/${CMAKE_LIBRARY_ARCHITECTURE}")
   set (lib_dirs "lib/${CMAKE_LIBRARY_ARCHITECTURE}" ${lib_dirs})
 endif ()
 
-find_library(RE2_STATIC_LIB NAMES libre2${CMAKE_STATIC_LIBRARY_SUFFIX}
+set(RE2_LIB_NAME re2)
+set(RE2_STATIC_LIB_NAME ${CMAKE_STATIC_LIBRARY_PREFIX}${RE2_LIB_NAME}${RE2_MSVC_STATIC_LIB_SUFFIX}${CMAKE_STATIC_LIBRARY_SUFFIX})
+set(RE2_SHARED_LIB_NAME ${CMAKE_SHARED_LIBRARY_PREFIX}${RE2_LIB_NAME}${CMAKE_SHARED_LIBRARY_SUFFIX})
+
+find_library(RE2_STATIC_LIB NAMES ${RE2_STATIC_LIB_NAME}
   PATHS ${_re2_path}
         NO_DEFAULT_PATH
   PATH_SUFFIXES ${lib_dirs}
   DOC   "Google's re2 regex static library"
 )
 
-find_library(RE2_SHARED_LIB NAMES libre2${CMAKE_SHARED_LIBRARY_SUFFIX}
+find_library(RE2_SHARED_LIB NAMES ${RE2_SHARED_LIB_NAME}
   PATHS ${_re2_path}
         NO_DEFAULT_PATH
   PATH_SUFFIXES ${lib_dirs}

diff --git a/cpp/src/arrow/util/bit-util-test.cc b/cpp/src/arrow/util/bit-util-test.cc
@@ -734,6 +734,30 @@ TEST(BitUtil, CountLeadingZeros) {
   EXPECT_EQ(BitUtil::CountLeadingZeros(U64(ULLONG_MAX)), 0);
 }
 
+TEST(BitUtil, CountTrailingZeros) {
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(0)), 32);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(1) << 31), 31);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(1) << 30), 30);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(1) << 29), 29);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(1) << 28), 28);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(8)), 3);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(4)), 2);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(2)), 1);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(1)), 0);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U32(ULONG_MAX)), 0);
+
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(0)), 64);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(1) << 63), 63);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(1) << 62), 62);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(1) << 61), 61);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(1) << 60), 60);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(8)), 3);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(4)), 2);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(2)), 1);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(1)), 0);
+  EXPECT_EQ(BitUtil::CountTrailingZeros(U64(ULLONG_MAX)), 0);
+}
+
 #undef U32
 #undef U64
 

diff --git a/cpp/src/arrow/util/bit-util.h b/cpp/src/arrow/util/bit-util.h
@@ -45,6 +45,7 @@
 #if defined(_MSC_VER)
 #include <intrin.h>
 #pragma intrinsic(_BitScanReverse)
+#pragma intrinsic(_BitScanForward)
 #define ARROW_BYTE_SWAP64 _byteswap_uint64
 #define ARROW_BYTE_SWAP32 _byteswap_ulong
 #else
@@ -183,6 +184,48 @@ static inline int CountLeadingZeros(uint64_t value) {
 #endif
 }
 
+static inline int CountTrailingZeros(uint32_t value) {
+#if defined(__clang__) || defined(__GNUC__)
+  if (value == 0) return 32;
+  return static_cast<int>(__builtin_ctzl(value));
+#elif defined(_MSC_VER)
+  unsigned long index;  // NOLINT
+  if (_BitScanForward(&index, value)) {
+    return static_cast<int>(index);
+  } else {
+    return 32;
+  }
+#else
+  int bitpos = 0;
+  while (value != 0) {
+    value >>= 1;
+    ++bitpos;
+  }
+  return bitpos;
+#endif
+}
+
+static inline int CountTrailingZeros(uint64_t value) {
+#if defined(__clang__) || defined(__GNUC__)
+  if (value == 0) return 64;
+  return static_cast<int>(__builtin_ctzll(value));
+#elif defined(_MSC_VER)
+  unsigned long index;  // NOLINT
+  if (_BitScanForward64(&index, value)) {
+    return static_cast<int>(index);
+  } else {
+    return 64;
+  }
+#else
+  int bitpos = 0;
+  while (value != 0) {
+    value >>= 1;
+    ++bitpos;
+  }
+  return bitpos;
+#endif
+}
+
 // Returns the minimum number of bits needed to represent an unsigned value
 static inline int NumRequiredBits(uint64_t x) { return 64 - CountLeadingZeros(x); }
 

diff --git a/cpp/src/arrow/util/visibility.h b/cpp/src/arrow/util/visibility.h
@@ -20,6 +20,7 @@
 
 #if defined(_WIN32) || defined(__CYGWIN__)
 #if defined(_MSC_VER)
+#pragma warning(push)
 #pragma warning(disable : 4251)
 #else
 #pragma GCC diagnostic ignored "-Wattributes"

diff --git a/cpp/src/gandiva/CMakeLists.txt b/cpp/src/gandiva/CMakeLists.txt
@@ -43,29 +43,29 @@ configure_file(bc_file_path.cc.in ${BC_FILE_PATH_CC})
 add_definitions(-DGANDIVA_BYTE_COMPILE_FILE_PATH="${GANDIVA_BC_OUTPUT_PATH}")
 
 set(SRC_FILES annotator.cc
-      bitmap_accumulator.cc
-      configuration.cc
-      context_helper.cc
-      engine.cc
-      date_utils.cc
-      expr_decomposer.cc
-      expr_validator.cc
-      expression.cc
-      expression_registry.cc
-      exported_funcs_registry.cc
-      filter.cc
-      function_registry.cc
-      function_signature.cc
-      gdv_function_stubs.cc
-      llvm_generator.cc
-      llvm_types.cc
-      like_holder.cc
-      projector.cc
-      regex_util.cc
-      selection_vector.cc
-      tree_expr_builder.cc
-      to_date_holder.cc
-      ${BC_FILE_PATH_CC})
+  bitmap_accumulator.cc
+  configuration.cc
+  context_helper.cc
+  engine.cc
+  date_utils.cc
+  expr_decomposer.cc
+  expr_validator.cc
+  expression.cc
+  expression_registry.cc
+  exported_funcs_registry.cc
+  filter.cc
+  function_registry.cc
+  function_signature.cc
+  gdv_function_stubs.cc
+  llvm_generator.cc
+  llvm_types.cc
+  like_holder.cc
+  projector.cc
+  regex_util.cc
+  selection_vector.cc
+  tree_expr_builder.cc
+  to_date_holder.cc
+  ${BC_FILE_PATH_CC})
 
 set(GANDIVA_SHARED_PRIVATE_LINK_LIBS
   arrow_shared
@@ -95,6 +95,15 @@ ADD_ARROW_LIB(gandiva
   SHARED_PRIVATE_LINK_LIBS ${GANDIVA_SHARED_PRIVATE_LINK_LIBS}
   STATIC_LINK_LIBS ${GANDIVA_STATIC_LINK_LIBS})
 
+foreach(LIB_TARGET ${GANDIVA_LIBRARIES})
+  target_compile_definitions(${LIB_TARGET}
+    PRIVATE GANDIVA_EXPORTING)
+endforeach()
+
+if (ARROW_BUILD_STATIC AND WIN32)
+  target_compile_definitions(gandiva_static PUBLIC GANDIVA_STATIC)
+endif()
+
 add_dependencies(gandiva ${GANDIVA_LIBRARIES})
 
 # install for gandiva

diff --git a/cpp/src/gandiva/annotator.h b/cpp/src/gandiva/annotator.h
@@ -27,12 +27,13 @@
 #include "gandiva/eval_batch.h"
 #include "gandiva/gandiva_aliases.h"
 #include "gandiva/logging.h"
+#include "gandiva/visibility.h"
 
 namespace gandiva {
 
 /// \brief annotate the arrow fields in an expression, and use that
 /// to convert the incoming arrow-format row batch to an EvalBatch.
-class Annotator {
+class GANDIVA_EXPORT Annotator {
  public:
   Annotator() : buffer_count_(0), local_bitmap_count_(0) {}
 

diff --git a/cpp/src/gandiva/bitmap_accumulator.h b/cpp/src/gandiva/bitmap_accumulator.h
@@ -24,12 +24,13 @@
 #include "gandiva/dex.h"
 #include "gandiva/dex_visitor.h"
 #include "gandiva/eval_batch.h"
+#include "gandiva/visibility.h"
 
 namespace gandiva {
 
 /// \brief Extract bitmap buffer from either the input/buffer vectors or the
 /// local validity bitmap, and accumultes them to do the final computation.
-class BitMapAccumulator : public DexDefaultVisitor {
+class GANDIVA_EXPORT BitMapAccumulator : public DexDefaultVisitor {
  public:
   explicit BitMapAccumulator(const EvalBatch& eval_batch)
       : eval_batch_(eval_batch), all_invalid_(false) {}

diff --git a/cpp/src/gandiva/bitmap_accumulator_test.cc b/cpp/src/gandiva/bitmap_accumulator_test.cc
@@ -21,24 +21,22 @@
 #include <vector>
 
 #include <gtest/gtest.h>
+
+#include "arrow/test-util.h"
+
 #include "gandiva/dex.h"
 
 namespace gandiva {
 
 class TestBitMapAccumulator : public ::testing::Test {
  protected:
-  void FillBitMap(uint8_t* bmap, int nrecords);
+  void FillBitMap(uint8_t* bmap, uint32_t seed, int nrecords);
   void ByteWiseIntersectBitMaps(uint8_t* dst, const std::vector<uint8_t*>& srcs,
                                 int nrecords);
 };
 
-void TestBitMapAccumulator::FillBitMap(uint8_t* bmap, int nbytes) {
-  unsigned int cur = 0;
-
-  for (int i = 0; i < nbytes; ++i) {
-    rand_r(&cur);
-    bmap[i] = static_cast<uint8_t>(cur % UINT8_MAX);
-  }
+void TestBitMapAccumulator::FillBitMap(uint8_t* bmap, uint32_t seed, int nbytes) {
+  ::arrow::random_bytes(nbytes, seed, bmap);
 }
 
 void TestBitMapAccumulator::ByteWiseIntersectBitMaps(uint8_t* dst,
@@ -61,7 +59,7 @@ TEST_F(TestBitMapAccumulator, TestIntersectBitMaps) {
   uint8_t expected_bitmap[length];
 
   for (int i = 0; i < 4; i++) {
-    FillBitMap(src_bitmaps[i], length);
+    FillBitMap(src_bitmaps[i], i, length);
   }
 
   for (int i = 0; i < 4; i++) {

diff --git a/cpp/src/gandiva/compiled_expr.h b/cpp/src/gandiva/compiled_expr.h
@@ -18,7 +18,7 @@
 #ifndef GANDIVA_COMPILED_EXPR_H
 #define GANDIVA_COMPILED_EXPR_H
 
-#include <llvm/IR/IRBuilder.h>
+#include "gandiva/llvm_includes.h"
 #include "gandiva/value_validity_pair.h"
 
 namespace gandiva {