Merge 07ffc73 into c246da9

CMakeLists.txt

@@ -84,7 +84,7 @@ enable_testing()
 set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake_modules")
 set(BUILD_SUPPORT_DIR "${CMAKE_SOURCE_DIR}/build-support")
 
-set(CLANG_FORMAT_VERSION "5.0")
+set(CLANG_FORMAT_VERSION "6.0")
 find_package(ClangTools)
 if ("$ENV{CMAKE_EXPORT_COMPILE_COMMANDS}" STREQUAL "1" OR CLANG_TIDY_FOUND)
   # Generate a Clang compile_commands.json "compilation database" file for use

cmake_modules/FindClangTools.cmake

@@ -30,14 +30,20 @@
 #  CLANG_FORMAT_BIN, The path to the clang format binary
 #  CLANG_TIDY_FOUND, Whether clang format was found
 
+if (DEFINED ENV{HOMEBREW_PREFIX})
+  set(HOMEBREW_PREFIX "${ENV{HOMEBREW_PREFIX}")
+else()
+  set(HOMEBREW_PREFIX "/usr/local")
+endif()
+
 find_program(CLANG_TIDY_BIN
   NAMES clang-tidy-4.0
   clang-tidy-3.9
   clang-tidy-3.8
   clang-tidy-3.7
   clang-tidy-3.6
   clang-tidy
-  PATHS ${ClangTools_PATH} $ENV{CLANG_TOOLS_PATH} /usr/local/bin /usr/bin
+  PATHS ${ClangTools_PATH} $ENV{CLANG_TOOLS_PATH} /usr/local/bin /usr/bin "${HOMEBREW_PREFIX}/bin"
         NO_DEFAULT_PATH
 )
 
@@ -55,7 +61,7 @@ if (CLANG_FORMAT_VERSION)
       PATHS
             ${ClangTools_PATH}
             $ENV{CLANG_TOOLS_PATH}
-            /usr/local/bin /usr/bin
+            /usr/local/bin /usr/bin "${HOMEBREW_PREFIX}/bin"
             NO_DEFAULT_PATH
     )
 
@@ -67,16 +73,26 @@ if (CLANG_FORMAT_VERSION)
         if ("${CLANG_FORMAT_MINOR_VERSION}" STREQUAL "0")
             find_program(CLANG_FORMAT_BIN
               NAMES clang-format
-              PATHS /usr/local/opt/llvm@${CLANG_FORMAT_MAJOR_VERSION}/bin
+              PATHS "${HOMEBREW_PREFIX}/opt/llvm@${CLANG_FORMAT_MAJOR_VERSION}/bin"
                     NO_DEFAULT_PATH
             )
         else()
             find_program(CLANG_FORMAT_BIN
               NAMES clang-format
-              PATHS /usr/local/opt/llvm@${CLANG_FORMAT_VERSION}/bin
+              PATHS "${HOMEBREW_PREFIX}/opt/llvm@${CLANG_FORMAT_VERSION}/bin"
                     NO_DEFAULT_PATH
             )
         endif()
+
+        if ("${CLANG_FORMAT_BIN}" STREQUAL "CLANG_FORMAT_BIN-NOTFOUND")
+          # binary was still not found, look into Cellar
+          file(GLOB CLANG_FORMAT_PATH "${HOMEBREW_PREFIX}/Cellar/llvm/${CLANG_FORMAT_VERSION}.*")
+          find_program(CLANG_FORMAT_BIN
+            NAMES clang-format
+            PATHS "${CLANG_FORMAT_PATH}/bin"
+                  NO_DEFAULT_PATH
+          )
+        endif()
     endif()
 else()
     find_program(CLANG_FORMAT_BIN
@@ -86,7 +102,7 @@ else()
       clang-format-3.7
       clang-format-3.6
       clang-format
-      PATHS ${ClangTools_PATH} $ENV{CLANG_TOOLS_PATH} /usr/local/bin /usr/bin
+      PATHS ${ClangTools_PATH} $ENV{CLANG_TOOLS_PATH} /usr/local/bin /usr/bin "${HOMEBREW_PREFIX}/bin"
             NO_DEFAULT_PATH
     )
 endif()

src/parquet/arrow/arrow-reader-writer-test.cc

@@ -52,16 +52,16 @@ using arrow::Buffer;
 using arrow::ChunkedArray;
 using arrow::Column;
 using arrow::DataType;
+using arrow::default_memory_pool;
 using arrow::ListArray;
-using arrow::ResizableBuffer;
 using arrow::PrimitiveArray;
+using arrow::ResizableBuffer;
 using arrow::Status;
 using arrow::Table;
 using arrow::TimeUnit;
 using arrow::compute::Datum;
 using arrow::compute::DictionaryEncode;
 using arrow::compute::FunctionContext;
-using arrow::default_memory_pool;
 using arrow::io::BufferReader;
 
 using arrow::test::randint;
@@ -1453,13 +1453,13 @@ TEST(TestArrowReadWrite, ConvertedDateTimeTypes) {
 // Regression for ARROW-2802
 TEST(TestArrowReadWrite, CoerceTimestampsAndSupportDeprecatedInt96) {
   using ::arrow::Column;
+  using ::arrow::default_memory_pool;
   using ::arrow::Field;
   using ::arrow::Schema;
   using ::arrow::Table;
-  using ::arrow::TimeUnit;
   using ::arrow::TimestampBuilder;
   using ::arrow::TimestampType;
-  using ::arrow::default_memory_pool;
+  using ::arrow::TimeUnit;
 
   auto timestamp_type = std::make_shared<TimestampType>(TimeUnit::NANO);
 

src/parquet/arrow/arrow-schema-test.cc

@@ -62,8 +62,8 @@ class TestConvertParquetSchema : public ::testing::Test {
     for (int i = 0; i < expected_schema->num_fields(); ++i) {
       auto lhs = result_schema_->field(i);
       auto rhs = expected_schema->field(i);
-      EXPECT_TRUE(lhs->Equals(rhs)) << i << " " << lhs->ToString()
-                                    << " != " << rhs->ToString();
+      EXPECT_TRUE(lhs->Equals(rhs))
+          << i << " " << lhs->ToString() << " != " << rhs->ToString();
     }
   }
 
@@ -607,9 +607,10 @@ TEST_F(TestConvertParquetSchema, ParquetRepeatedNestedSchema) {
     auto inner_group_type = std::make_shared<::arrow::StructType>(inner_group_fields);
     auto outer_group_fields = {
         std::make_shared<Field>("leaf2", INT32, true),
-        std::make_shared<Field>("innerGroup", ::arrow::list(std::make_shared<Field>(
-                                                  "innerGroup", inner_group_type, false)),
-                                false)};
+        std::make_shared<Field>(
+            "innerGroup",
+            ::arrow::list(std::make_shared<Field>("innerGroup", inner_group_type, false)),
+            false)};
     auto outer_group_type = std::make_shared<::arrow::StructType>(outer_group_fields);
 
     arrow_fields.push_back(std::make_shared<Field>("leaf1", INT32, true));

src/parquet/arrow/reader.cc

@@ -301,9 +301,7 @@ class PARQUET_NO_EXPORT StructImpl : public ColumnReader::ColumnReaderImpl {
  public:
   explicit StructImpl(const std::vector<std::shared_ptr<ColumnReaderImpl>>& children,
                       int16_t struct_def_level, MemoryPool* pool, const Node* node)
-      : children_(children),
-        struct_def_level_(struct_def_level),
-        pool_(pool) {
+      : children_(children), struct_def_level_(struct_def_level), pool_(pool) {
     InitField(node, children);
   }
 

src/parquet/arrow/test-util.h

@@ -402,17 +402,16 @@ Status MakeEmptyListsArray(int64_t size, std::shared_ptr<Array>* out_array) {
                                         &offsets_buffer));
   memset(offsets_buffer->mutable_data(), 0, offsets_nbytes);
 
-  auto value_field = ::arrow::field("item", ::arrow::float64(),
-                                    false /* nullable_values */);
+  auto value_field =
+      ::arrow::field("item", ::arrow::float64(), false /* nullable_values */);
   auto list_type = ::arrow::list(value_field);
 
   std::vector<std::shared_ptr<Buffer>> child_buffers = {nullptr /* null bitmap */,
-                                                        nullptr /* values */ };
-  auto child_data = ::arrow::ArrayData::Make(value_field->type(), 0,
-                                             std::move(child_buffers));
+                                                        nullptr /* values */};
+  auto child_data =
+      ::arrow::ArrayData::Make(value_field->type(), 0, std::move(child_buffers));
 
-  std::vector<std::shared_ptr<Buffer>> buffers = {nullptr /* bitmap */,
-                                                  offsets_buffer };
+  std::vector<std::shared_ptr<Buffer>> buffers = {nullptr /* bitmap */, offsets_buffer};
   auto array_data = ::arrow::ArrayData::Make(list_type, size, std::move(buffers));
   array_data->child_data.push_back(child_data);
 

src/parquet/arrow/writer.cc

@@ -41,8 +41,8 @@ using arrow::Int16Builder;
 using arrow::ListArray;
 using arrow::MemoryPool;
 using arrow::NumericArray;
-using arrow::ResizableBuffer;
 using arrow::PrimitiveArray;
+using arrow::ResizableBuffer;
 using arrow::Status;
 using arrow::Table;
 using arrow::TimeUnit;
@@ -216,9 +216,11 @@ class LevelBuilder {
         if (level_null_count && level_valid_bitmap == nullptr) {
           // Special case: this is a null array (all elements are null)
           RETURN_NOT_OK(def_levels_.Append(static_cast<int16_t>(def_level + 1)));
-        } else if (nullable_level && ((level_null_count == 0) ||
-            BitUtil::GetBit(level_valid_bitmap,
-                            inner_offset + i + array_offsets_[recursion_level]))) {
+        } else if (nullable_level &&
+                   ((level_null_count == 0) ||
+                    BitUtil::GetBit(
+                        level_valid_bitmap,
+                        inner_offset + i + array_offsets_[recursion_level]))) {
           // Non-null element in a null level
           RETURN_NOT_OK(def_levels_.Append(static_cast<int16_t>(def_level + 2)));
         } else {

src/parquet/encoding-benchmark.cc

@@ -20,8 +20,8 @@
 #include "parquet/encoding-internal.h"
 #include "parquet/util/memory.h"
 
-using arrow::MemoryPool;
 using arrow::default_memory_pool;
+using arrow::MemoryPool;
 
 namespace parquet {
 

src/parquet/encoding-test.cc

@@ -30,8 +30,8 @@
 #include "parquet/util/memory.h"
 #include "parquet/util/test-common.h"
 
-using arrow::MemoryPool;
 using arrow::default_memory_pool;
+using arrow::MemoryPool;
 
 using std::string;
 using std::vector;

src/parquet/encoding.h

@@ -51,12 +51,12 @@ class Encoder {
   virtual void PutSpaced(const T* src, int num_values, const uint8_t* valid_bits,
                          int64_t valid_bits_offset) {
     std::shared_ptr<ResizableBuffer> buffer;
-    auto status = ::arrow::AllocateResizableBuffer(pool_, num_values * sizeof(T),
-                                                   &buffer);
+    auto status =
+        ::arrow::AllocateResizableBuffer(pool_, num_values * sizeof(T), &buffer);
     if (!status.ok()) {
       std::ostringstream ss;
-      ss << "AllocateResizableBuffer failed in Encoder.PutSpaced in "
-         << __FILE__ << ", on line " << __LINE__;
+      ss << "AllocateResizableBuffer failed in Encoder.PutSpaced in " << __FILE__
+         << ", on line " << __LINE__;
       throw ParquetException(ss.str());
     }
     int32_t num_valid_values = 0;

src/parquet/printer.cc

@@ -84,8 +84,8 @@ void ParquetFilePrinter::DebugPrint(std::ostream& stream, std::list<int> selecte
         std::string min = stats->EncodeMin(), max = stats->EncodeMax();
         stream << ", Null Values: " << stats->null_count()
                << ", Distinct Values: " << stats->distinct_count() << std::endl
-               << "  Max: " << FormatStatValue(descr->physical_type(), max.c_str())
-               << ", Min: " << FormatStatValue(descr->physical_type(), min.c_str());
+               << "  Max: " << FormatStatValue(descr->physical_type(), max)
+               << ", Min: " << FormatStatValue(descr->physical_type(), min);
       } else {
         stream << "  Statistics Not Set";
       }
@@ -207,9 +207,8 @@ void ParquetFilePrinter::JSONPrint(std::ostream& stream, std::list<int> selected
         std::string min = stats->EncodeMin(), max = stats->EncodeMax();
         stream << "\"NumNulls\": \"" << stats->null_count() << "\", "
                << "\"DistinctValues\": \"" << stats->distinct_count() << "\", "
-               << "\"Max\": \"" << FormatStatValue(descr->physical_type(), max.c_str())
-               << "\", "
-               << "\"Min\": \"" << FormatStatValue(descr->physical_type(), min.c_str())
+               << "\"Max\": \"" << FormatStatValue(descr->physical_type(), max) << "\", "
+               << "\"Min\": \"" << FormatStatValue(descr->physical_type(), min)
                << "\" },";
       } else {
         stream << "\"False\",";

src/parquet/statistics-test.cc

@@ -36,8 +36,8 @@
 #include "parquet/types.h"
 #include "parquet/util/memory.h"
 
-using arrow::MemoryPool;
 using arrow::default_memory_pool;
+using arrow::MemoryPool;
 
 namespace parquet {
 
@@ -194,8 +194,9 @@ bool* TestRowGroupStatistics<BooleanType>::GetValuesPointer(std::vector<bool>& v
 }
 
 template <typename TestType>
-typename std::vector<typename TestType::c_type> TestRowGroupStatistics<
-    TestType>::GetDeepCopy(const std::vector<typename TestType::c_type>& values) {
+typename std::vector<typename TestType::c_type>
+TestRowGroupStatistics<TestType>::GetDeepCopy(
+    const std::vector<typename TestType::c_type>& values) {
   return values;
 }
 

src/parquet/statistics.cc

@@ -24,8 +24,8 @@
 #include "parquet/statistics.h"
 #include "parquet/util/memory.h"
 
-using arrow::MemoryPool;
 using arrow::default_memory_pool;
+using arrow::MemoryPool;
 
 namespace parquet {
 

src/parquet/types-test.cc

@@ -62,54 +62,77 @@ TEST(TestLogicalTypeToString, LogicalTypes) {
 }
 
 TEST(TypePrinter, StatisticsTypes) {
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wdeprecated-declarations"
   std::string smin;
   std::string smax;
   int32_t int_min = 1024;
   int32_t int_max = 2048;
   smin = std::string(reinterpret_cast<char*>(&int_min), sizeof(int32_t));
   smax = std::string(reinterpret_cast<char*>(&int_max), sizeof(int32_t));
+  ASSERT_STREQ("1024", FormatStatValue(Type::INT32, smin).c_str());
   ASSERT_STREQ("1024", FormatStatValue(Type::INT32, smin.c_str()).c_str());
+  ASSERT_STREQ("2048", FormatStatValue(Type::INT32, smax).c_str());
   ASSERT_STREQ("2048", FormatStatValue(Type::INT32, smax.c_str()).c_str());
 
   int64_t int64_min = 10240000000000;
   int64_t int64_max = 20480000000000;
   smin = std::string(reinterpret_cast<char*>(&int64_min), sizeof(int64_t));
   smax = std::string(reinterpret_cast<char*>(&int64_max), sizeof(int64_t));
+  ASSERT_STREQ("10240000000000", FormatStatValue(Type::INT64, smin).c_str());
   ASSERT_STREQ("10240000000000", FormatStatValue(Type::INT64, smin.c_str()).c_str());
+  ASSERT_STREQ("20480000000000", FormatStatValue(Type::INT64, smax).c_str());
   ASSERT_STREQ("20480000000000", FormatStatValue(Type::INT64, smax.c_str()).c_str());
 
   float float_min = 1.024f;
   float float_max = 2.048f;
   smin = std::string(reinterpret_cast<char*>(&float_min), sizeof(float));
   smax = std::string(reinterpret_cast<char*>(&float_max), sizeof(float));
+  ASSERT_STREQ("1.024", FormatStatValue(Type::FLOAT, smin).c_str());
   ASSERT_STREQ("1.024", FormatStatValue(Type::FLOAT, smin.c_str()).c_str());
+  ASSERT_STREQ("2.048", FormatStatValue(Type::FLOAT, smax).c_str());
   ASSERT_STREQ("2.048", FormatStatValue(Type::FLOAT, smax.c_str()).c_str());
 
   double double_min = 1.0245;
   double double_max = 2.0489;
   smin = std::string(reinterpret_cast<char*>(&double_min), sizeof(double));
   smax = std::string(reinterpret_cast<char*>(&double_max), sizeof(double));
+  ASSERT_STREQ("1.0245", FormatStatValue(Type::DOUBLE, smin).c_str());
   ASSERT_STREQ("1.0245", FormatStatValue(Type::DOUBLE, smin.c_str()).c_str());
+  ASSERT_STREQ("2.0489", FormatStatValue(Type::DOUBLE, smax).c_str());
   ASSERT_STREQ("2.0489", FormatStatValue(Type::DOUBLE, smax.c_str()).c_str());
 
   Int96 Int96_min = {{1024, 2048, 4096}};
   Int96 Int96_max = {{2048, 4096, 8192}};
   smin = std::string(reinterpret_cast<char*>(&Int96_min), sizeof(Int96));
   smax = std::string(reinterpret_cast<char*>(&Int96_max), sizeof(Int96));
+  ASSERT_STREQ("1024 2048 4096", FormatStatValue(Type::INT96, smin).c_str());
   ASSERT_STREQ("1024 2048 4096", FormatStatValue(Type::INT96, smin.c_str()).c_str());
+  ASSERT_STREQ("2048 4096 8192", FormatStatValue(Type::INT96, smax).c_str());
   ASSERT_STREQ("2048 4096 8192", FormatStatValue(Type::INT96, smax.c_str()).c_str());
 
   smin = std::string("abcdef");
   smax = std::string("ijklmnop");
+  ASSERT_STREQ("abcdef", FormatStatValue(Type::BYTE_ARRAY, smin).c_str());
   ASSERT_STREQ("abcdef", FormatStatValue(Type::BYTE_ARRAY, smin.c_str()).c_str());
+  ASSERT_STREQ("ijklmnop", FormatStatValue(Type::BYTE_ARRAY, smax).c_str());
   ASSERT_STREQ("ijklmnop", FormatStatValue(Type::BYTE_ARRAY, smax.c_str()).c_str());
 
+  // PARQUET-1357: FormatStatValue truncates binary statistics on zero character
+  smax.push_back('\0');
+  ASSERT_EQ(smax, FormatStatValue(Type::BYTE_ARRAY, smax));
+  // This fails, thus the call to FormatStatValue(.., const char*) was deprecated.
+  // ASSERT_EQ(smax, FormatStatValue(Type::BYTE_ARRAY, smax.c_str()));
+
   smin = std::string("abcdefgh");
   smax = std::string("ijklmnop");
+  ASSERT_STREQ("abcdefgh", FormatStatValue(Type::FIXED_LEN_BYTE_ARRAY, smin).c_str());
   ASSERT_STREQ("abcdefgh",
                FormatStatValue(Type::FIXED_LEN_BYTE_ARRAY, smin.c_str()).c_str());
+  ASSERT_STREQ("ijklmnop", FormatStatValue(Type::FIXED_LEN_BYTE_ARRAY, smax).c_str());
   ASSERT_STREQ("ijklmnop",
                FormatStatValue(Type::FIXED_LEN_BYTE_ARRAY, smax.c_str()).c_str());
+#pragma GCC diagnostic pop
 }
 
 }  // namespace parquet