Fixing all lint errors

.gitignore

@@ -107,3 +107,6 @@ java/.mvn/.develocity/
 # rat
 filtered_rat.txt
 rat.txt
+
+python/sample.parquet
+Dockerfile

cpp/src/arrow/util/io_util.cc

@@ -2248,7 +2248,8 @@ Result<void*> LoadDynamicLibrary(const char* path) {
   return LoadDynamicLibrary(platform_path);
 #else
   constexpr int kFlags =
-      // All undefined symbols in the shared object are resolved before dlopen() returns.
+      // All undefined symbols in the shared object are resolved before dlopen()
+      // returns.
       RTLD_NOW
       // Symbols defined in this shared object are not made available to
       // resolve references in subsequently loaded shared objects.
@@ -2310,7 +2311,8 @@ Status CloseDynamicLibrary(void* handle) {
   if (dlclose(handle) == 0) {
     return Status::OK();
   }
-  // dlclose(3) man page: "On success, dlclose() returns 0; on error, it returns a nonzero value."
+  // dlclose(3) man page: "On success, dlclose() returns 0; on error, it returns a
+  // nonzero value."
   auto* error = dlerror();
   return Status::IOError("dlclose() failed: ", error ? error : "unknown error");
 #endif

cpp/src/arrow/util/io_util.h

@@ -453,7 +453,8 @@ ARROW_EXPORT Result<void*> GetSymbol(void* handle, const char* name);
 ///
 /// This wraps dlclose() except on Windows, where FreeLibrary() is called.
 ///
-/// \return Status::OK() if the library was closed successfully, otherwise an error is returned.
+/// \return Status::OK() if the library was closed successfully, otherwise an error
+///         is returned.
 ARROW_EXPORT Status CloseDynamicLibrary(void* handle);
 
 template <typename T>

cpp/src/parquet/CMakeLists.txt

@@ -239,64 +239,58 @@ endif()
 
 if(PARQUET_REQUIRE_ENCRYPTION)
   list(APPEND PARQUET_SHARED_PRIVATE_LINK_LIBS ${ARROW_OPENSSL_LIBS})
-  set(PARQUET_SRCS ${PARQUET_SRCS} encryption/aes_encryption.cc
-                   encryption/openssl_internal.cc
-                   encryption/encryption_utils.cc
-                   encryption/external_dbpa_encryption.cc
-                   encryption/external/dbpa_library_wrapper.cc
-                   encryption/encoding_properties.cc
-                   encryption/external/dbpa_enum_utils.cc
-                   encryption/external/dbpa_executor.cc
-                   encryption/external/loadable_encryptor_utils.cc
-                   )
+  set(PARQUET_SRCS
+      ${PARQUET_SRCS}
+      encryption/aes_encryption.cc
+      encryption/openssl_internal.cc
+      encryption/encryption_utils.cc
+      encryption/external_dbpa_encryption.cc
+      encryption/external/dbpa_library_wrapper.cc
+      encryption/encoding_properties.cc
+      encryption/external/dbpa_enum_utils.cc
+      encryption/external/dbpa_executor.cc
+      encryption/external/loadable_encryptor_utils.cc)
 
   # Header-only dependencies for external DBPA code
   include(FetchContent)
 
   # magic_enum
-  FetchContent_Declare(
-    magic_enum_upstream
-    GIT_REPOSITORY https://github.com/Neargye/magic_enum.git
-    GIT_TAG v0.9.7
-  )
-  FetchContent_MakeAvailable(magic_enum_upstream)
+  fetchcontent_declare(magic_enum_upstream
+                       GIT_REPOSITORY https://github.com/Neargye/magic_enum.git
+                       GIT_TAG v0.9.7)
+  fetchcontent_makeavailable(magic_enum_upstream)
   add_library(magic_enum_header_only INTERFACE)
-  target_include_directories(magic_enum_header_only INTERFACE ${magic_enum_upstream_SOURCE_DIR}/include)
+  target_include_directories(magic_enum_header_only
+                             INTERFACE ${magic_enum_upstream_SOURCE_DIR}/include)
 
   # tcb/span
-  FetchContent_Declare(
-    tcb_span_upstream
-    GIT_REPOSITORY https://github.com/tcbrindle/span.git
-    GIT_TAG master
-  )
+  fetchcontent_declare(tcb_span_upstream
+                       GIT_REPOSITORY https://github.com/tcbrindle/span.git
+                       GIT_TAG master)
 
   # Populate sources only; do not add subproject/tests
-  FetchContent_GetProperties(tcb_span_upstream)
+  fetchcontent_getproperties(tcb_span_upstream)
   if(NOT tcb_span_upstream_POPULATED)
-    FetchContent_Populate(tcb_span_upstream)
+    fetchcontent_populate(tcb_span_upstream)
   endif()
   add_library(tcb_span INTERFACE)
   target_include_directories(tcb_span INTERFACE ${tcb_span_upstream_SOURCE_DIR}/include)
 
   # DBPS interface (header-only)
-  FetchContent_Declare(
-    dbps_agent
-    GIT_REPOSITORY https://github.com/protegrity/DataBatchProtectionService.git
-
-    #TODO: Change to a specific tag/commit when we have one.
-    #https://github.com/protegrity/arrow/issues/179
-    GIT_TAG 4c808b2233ed0bc04529c3b0dbf7c214c4901043
-    GIT_SHALLOW FALSE
-  )
-
-  FetchContent_GetProperties(dbps_agent)
+  fetchcontent_declare(dbps_agent
+                       GIT_REPOSITORY https://github.com/protegrity/DataBatchProtectionService.git
+                       #TODO: Change to a specific tag/commit when we have one.
+                       #https://github.com/protegrity/arrow/issues/179
+                       GIT_TAG 4c808b2233ed0bc04529c3b0dbf7c214c4901043
+                       GIT_SHALLOW FALSE)
+
+  fetchcontent_getproperties(dbps_agent)
   if(NOT dbps_agent_POPULATED)
-    FetchContent_Populate(dbps_agent)
+    fetchcontent_populate(dbps_agent)
   endif()
   add_library(dbps_interface INTERFACE)
   # Expose dbpa_interface.h and friends from DBPS
-  target_include_directories(dbps_interface INTERFACE
-    ${dbps_agent_SOURCE_DIR}/src/common)
+  target_include_directories(dbps_interface INTERFACE ${dbps_agent_SOURCE_DIR}/src/common)
 
   # Allows to disable building DBPS shared libraries
   option(PARQUET_BUILD_DBPS_LIBS "Build DBPS external libraries" ON)
@@ -305,32 +299,36 @@ if(PARQUET_REQUIRE_ENCRYPTION)
     include(ExternalProject)
 
     # Allow callers to inject additional CMake args for DBPS (e.g., BOOST_ROOT)
-    set(PARQUET_DBPS_CMAKE_ARGS "${PARQUET_DBPS_CMAKE_ARGS}" CACHE STRING "Extra CMake args for DBPS ExternalProject")
+    set(PARQUET_DBPS_CMAKE_ARGS
+        "${PARQUET_DBPS_CMAKE_ARGS}"
+        CACHE STRING "Extra CMake args for DBPS ExternalProject")
 
     #builds DBPS shared libraries alongside Parquet
     #same OS, same architecture, same compiler, same build type, same CMake args
-    ExternalProject_Add(dbps_external
-      DOWNLOAD_COMMAND "" # disables download, assumes (correctly) that sources are already fetched
-      SOURCE_DIR ${dbps_agent_SOURCE_DIR}
-      BINARY_DIR ${dbps_agent_BINARY_DIR}
-      CMAKE_ARGS
-        -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}
-        -DBUILD_SHARED_LIBS=ON
-        -DBUILD_TESTING=OFF
-
-        # Place DBPS outputs next to Arrow artifacts
-        -DCMAKE_LIBRARY_OUTPUT_DIRECTORY=${BUILD_OUTPUT_ROOT_DIRECTORY}
-        -DCMAKE_RUNTIME_OUTPUT_DIRECTORY=${BUILD_OUTPUT_ROOT_DIRECTORY}
-        -DCMAKE_ARCHIVE_OUTPUT_DIRECTORY=${BUILD_OUTPUT_ROOT_DIRECTORY}
-        ${PARQUET_DBPS_CMAKE_ARGS}
-      INSTALL_COMMAND "" #disables install - artifacts already copied to BUILD_OUTPUT_ROOT_DIRECTORY
+    externalproject_add(dbps_external
+                        DOWNLOAD_COMMAND "" # disables download, assumes (correctly) that sources are already fetched
+                        SOURCE_DIR ${dbps_agent_SOURCE_DIR}
+                        BINARY_DIR ${dbps_agent_BINARY_DIR}
+                        CMAKE_ARGS -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}
+                                   -DBUILD_SHARED_LIBS=ON
+                                   -DBUILD_TESTING=OFF
+                                   # Place DBPS outputs next to Arrow artifacts
+                                   -DCMAKE_LIBRARY_OUTPUT_DIRECTORY=${BUILD_OUTPUT_ROOT_DIRECTORY}
+                                   -DCMAKE_RUNTIME_OUTPUT_DIRECTORY=${BUILD_OUTPUT_ROOT_DIRECTORY}
+                                   -DCMAKE_ARCHIVE_OUTPUT_DIRECTORY=${BUILD_OUTPUT_ROOT_DIRECTORY}
+                                   ${PARQUET_DBPS_CMAKE_ARGS}
+                        INSTALL_COMMAND "" #disables install - artifacts already copied to BUILD_OUTPUT_ROOT_DIRECTORY
     )
   endif()
 
   list(APPEND PARQUET_SHARED_PRIVATE_LINK_LIBS magic_enum_header_only tcb_span)
   list(APPEND PARQUET_STATIC_LINK_LIBS magic_enum_header_only tcb_span)
   if(ARROW_TESTING)
-    list(APPEND PARQUET_TEST_LINK_LIBS magic_enum_header_only tcb_span dbps_interface)
+    list(APPEND
+         PARQUET_TEST_LINK_LIBS
+         magic_enum_header_only
+         tcb_span
+         dbps_interface)
   endif()
   list(APPEND PARQUET_SHARED_PRIVATE_LINK_LIBS dbps_interface)
   list(APPEND PARQUET_STATIC_LINK_LIBS dbps_interface)
@@ -509,7 +507,8 @@ add_parquet_test(arrow-reader-writer-test
                  arrow/arrow_reader_writer_test.cc
                  arrow/arrow_statistics_test.cc
                  arrow/variant_test.cc
-                 $<$<BOOL:${PARQUET_REQUIRE_ENCRYPTION}>:encryption/external/test_utils.cc>)
+                 $<$<BOOL:${PARQUET_REQUIRE_ENCRYPTION}>:encryption/external/test_utils.cc>
+)
 
 add_parquet_test(arrow-internals-test SOURCES arrow/path_internal_test.cc
                  arrow/reconstruct_internal_test.cc)
@@ -543,9 +542,7 @@ endif()
 
 # Those tests need to use static linking as they access thrift-generated
 # symbols which are not exported by parquet.dll on Windows (PARQUET-1420).
-add_parquet_test(file_deserialize_test
-                 SOURCES
-                 file_deserialize_test.cc
+add_parquet_test(file_deserialize_test SOURCES file_deserialize_test.cc
                  encryption/external/test_utils.cc)
 add_parquet_test(schema_test)
 

cpp/src/parquet/arrow/arrow_reader_writer_test.cc

@@ -77,7 +77,7 @@
 #include "parquet/test_util.h"
 
 #ifdef PARQUET_REQUIRE_ENCRYPTION
-#include "parquet/encryption/external/test_utils.h"
+#  include "parquet/encryption/external/test_utils.h"
 #endif
 
 using arrow::Array;
@@ -1973,7 +1973,7 @@ TEST(ExternalDbpaConcurrencyTest, FailsWhenUseThreadsTrue) {
   ASSERT_OK(b.Finish(&arr));
   auto schema = ::arrow::schema({::arrow::field("f0", ::arrow::int32())});
   auto table = ::arrow::Table::Make(schema, {arr});
-  
+
   ::arrow::util::SecureString column_key(std::string("key1234567890123"));
   ::arrow::util::SecureString footer_key(std::string("footer_key123456"));
 
@@ -1989,31 +1989,30 @@ TEST(ExternalDbpaConcurrencyTest, FailsWhenUseThreadsTrue) {
 
   parquet::ExternalFileEncryptionProperties::Builder fep_builder(footer_key);
   fep_builder.footer_key_metadata("kf")
-            ->encrypted_columns(enc_cols)
-           ->algorithm(parquet::ParquetCipher::AES_GCM_V1)
-           ->configuration_properties({{parquet::ParquetCipher::EXTERNAL_DBPA_V1,
-                                   {{"agent_library_path", lib_path},
-                                    {"file_path", "/tmp/test"}}}});
+      ->encrypted_columns(enc_cols)
+      ->algorithm(parquet::ParquetCipher::AES_GCM_V1)
+      ->configuration_properties(
+          {{parquet::ParquetCipher::EXTERNAL_DBPA_V1,
+            {{"agent_library_path", lib_path}, {"file_path", "/tmp/test"}}}});
 
-  auto writer_props =
-      parquet::WriterProperties::Builder().encryption(fep_builder.build_external())->build();
+  auto writer_props = parquet::WriterProperties::Builder()
+                          .encryption(fep_builder.build_external())
+                          ->build();
 
   ASSERT_OK_AND_ASSIGN(
-    auto sink, ::arrow::io::BufferOutputStream::Create(1 << 16, default_memory_pool()));
-  ASSERT_OK(parquet::arrow::WriteTable(*table, default_memory_pool(), sink, /*chunk_size=*/1024,
-                                       writer_props));
+      auto sink, ::arrow::io::BufferOutputStream::Create(1 << 16, default_memory_pool()));
+  ASSERT_OK(parquet::arrow::WriteTable(*table, default_memory_pool(), sink,
+                                       /*chunk_size=*/1024, writer_props));
   ASSERT_OK_AND_ASSIGN(auto buffer, sink->Finish());
 
   auto kr = std::make_shared<parquet::StringKeyIdRetriever>();
   kr->PutKey("kf", footer_key);
   kr->PutKey("key1234567890123", column_key);
 
   parquet::ExternalFileDecryptionProperties::Builder dep_builder;
-  dep_builder.key_retriever(kr)
-           ->app_context("{}")
-           ->configuration_properties({{parquet::ParquetCipher::EXTERNAL_DBPA_V1,
-                                  {{"agent_library_path", lib_path},
-                                    {"file_path", "/tmp/test"}}}});
+  dep_builder.key_retriever(kr)->app_context("{}")->configuration_properties(
+      {{parquet::ParquetCipher::EXTERNAL_DBPA_V1,
+        {{"agent_library_path", lib_path}, {"file_path", "/tmp/test"}}}});
   parquet::ReaderProperties rp = parquet::default_reader_properties();
   rp.file_decryption_properties(dep_builder.build_external());
 

cpp/src/parquet/arrow/reader.cc

@@ -139,7 +139,8 @@ bool IsExternalDBPAEncryptionUsedInColumns(std::shared_ptr<FileMetaData> metadat
       if (column_metadata->crypto_metadata()) {
         auto crypto_metadata = column_metadata->crypto_metadata();
         if (crypto_metadata->is_encryption_algorithm_set() &&
-            crypto_metadata->encryption_algorithm().algorithm == ParquetCipher::EXTERNAL_DBPA_V1) {
+            crypto_metadata->encryption_algorithm().algorithm ==
+                ParquetCipher::EXTERNAL_DBPA_V1) {
           return true;
         }
       }
@@ -164,12 +165,13 @@ class FileReaderImpl : public FileReader {
         reader_properties_(std::move(properties)) {}
 
   Status Init() {
-    // If the file is encrypted using EXTERNAL_DBPA_V1 on any of its columns, then it is not safe
-    // to use multiple threads to read the file.
+    // If the file is encrypted using EXTERNAL_DBPA_V1 on any of its columns, then it
+    // is not safe to use multiple threads to read the file.
     if (reader_properties_.use_threads()) {
       auto metadata = reader_->metadata();
       if (IsExternalDBPAEncryptionUsedInColumns(metadata)) {
-        return Status::Invalid("EXTERNAL_DBPA_V1 encryption does not support multiple threads");
+        return Status::Invalid(
+            "EXTERNAL_DBPA_V1 encryption does not support multiple threads");
       }
     }
     return SchemaManifest::Make(reader_->metadata()->schema(),