Create C library wrapper over existing Java library using GraalVM

native-schema-registry/README.md

@@ -0,0 +1,16 @@
+# Native Schema Registry 
+
+This module provides a native shared library (.so, .dll) version of the AWS Glue Schema Registry SerDes. 
+It uses GraalVM to generate the shared library. 
+
+## Build 
+
+Requires GraalVM (21.0+) with native-image support.
+
+The C data types module needs to be built before building the Java module.
+
+```asm
+cd c && cmake -S. -Bbuild
+cd build && cmake --build . --target native_schema_registry_c_data_types
+cd ../../ && mvn package -P native-image
+```

native-schema-registry/c/.clang-format

@@ -0,0 +1,57 @@
+---
+Language:        Cpp
+# BasedOnStyle:  Mozilla
+AlignAfterOpenBracket: AlwaysBreak
+AlignConsecutiveAssignments: false
+AlignConsecutiveDeclarations: false
+AlignEscapedNewlines: Right
+AlignOperands:   true
+AlignTrailingComments: true
+AllowAllParametersOfDeclarationOnNextLine: false
+AllowShortBlocksOnASingleLine: false
+AllowShortCaseLabelsOnASingleLine: false
+AllowShortFunctionsOnASingleLine: Inline
+AllowShortIfStatementsOnASingleLine: false
+AllowShortLoopsOnASingleLine: false
+AlwaysBreakAfterReturnType: None
+AlwaysBreakBeforeMultilineStrings: false
+BinPackArguments: false
+BinPackParameters: false
+BreakBeforeBinaryOperators: None
+BreakBeforeBraces: Attach
+BreakBeforeTernaryOperators: true
+BreakStringLiterals: true
+ColumnLimit:     120
+ContinuationIndentWidth: 4
+DerivePointerAlignment: false
+IncludeBlocks:   Preserve
+IndentCaseLabels: true
+IndentPPDirectives: AfterHash
+IndentWidth:     4
+IndentWrappedFunctionNames: true
+KeepEmptyLinesAtTheStartOfBlocks: true
+MacroBlockBegin: ''
+MacroBlockEnd:   ''
+MaxEmptyLinesToKeep: 1
+PenaltyBreakAssignment: 2
+PenaltyBreakBeforeFirstCallParameter: 19
+PenaltyBreakComment: 300
+PenaltyBreakFirstLessLess: 120
+PenaltyBreakString: 1000
+PenaltyExcessCharacter: 1000000
+PenaltyReturnTypeOnItsOwnLine: 100000
+PointerAlignment: Right
+ReflowComments:  true
+SortIncludes:    true
+SpaceAfterCStyleCast: false
+SpaceBeforeAssignmentOperators: true
+SpaceBeforeParens: ControlStatements
+SpaceInEmptyParentheses: false
+SpacesInContainerLiterals: true
+SpacesInCStyleCastParentheses: false
+SpacesInParentheses: false
+SpacesInSquareBrackets: false
+Standard:        Cpp11
+TabWidth:        4
+UseTab:          Never
+...

native-schema-registry/c/.clang-tidy

@@ -0,0 +1,12 @@
+---
+Checks:          'clang-diagnostic-*,clang-analyzer-*,readability-*,modernize-*,bugprone-*,misc-*,google-runtime-int,llvm-header-guard,fuchsia-restrict-system-includes,-clang-analyzer-valist.Uninitialized,-clang-analyzer-security.insecureAPI.rand,-clang-analyzer-alpha.*,-readability-magic-numbers,-readability-non-const-parameter,-readability-avoid-const-params-in-decls,-readability-else-after-return,-readability-isolate-declaration,-readability-uppercase-literal-suffix,-bugprone-sizeof-expression'
+WarningsAsErrors: '*'
+HeaderFilterRegex: '(./c/src/.*\.c$)|(./c/include/.*\.h$)'
+FormatStyle: 'file'
+CheckOptions:
+  - key:             readability-braces-around-statements.ShortStatementLines
+    value:           '1'
+  - key:             google-runtime-int.TypeSuffix
+    value:           '_t'
+  - key:             fuchsia-restrict-system-includes.Includes
+    value:           '*,-stdint.h,-stdbool.h,-assert.h'

native-schema-registry/c/CMakeLists.txt

@@ -0,0 +1,36 @@
+cmake_minimum_required(VERSION 3.19)
+IF (APPLE)
+    #Defaults to Apple Compile. Explicitly install and set to gcc.
+    set(CMAKE_C_COMPILER gcc)
+ENDIF()
+project(native_schema_registry_c C)
+set(CMAKE_C_STANDARD 99)
+
+##Global variables
+#Path to GraalVM generated shared library
+set(LIB_NATIVE_SCHEMA_REGISTRY_PATH "${PROJECT_SOURCE_DIR}/../target")
+set(LIB_NATIVE_SCHEMA_REGISTRY_LIBRARY_NAME_PREFIX libnativeschemaregistry)
+##OS Specific variables
+IF (WIN32)
+    set(LIB_NATIVE_SCHEMA_REGISTRY_LIBRARY_NAME ${LIB_NATIVE_SCHEMA_REGISTRY_LIBRARY_NAME_PREFIX}.dll)
+    set(LIB_NATIVE_SCHEMA_REGISTRY_LIBRARY_OBJ_NAME ${LIB_NATIVE_SCHEMA_REGISTRY_LIBRARY_NAME_PREFIX}.lib)
+ELSEIF(APPLE)
+    set(LIB_NATIVE_SCHEMA_REGISTRY_LIBRARY_NAME ${LIB_NATIVE_SCHEMA_REGISTRY_LIBRARY_NAME_PREFIX}.dylib)
+ELSE()
+    set(LIB_NATIVE_SCHEMA_REGISTRY_LIBRARY_NAME ${LIB_NATIVE_SCHEMA_REGISTRY_LIBRARY_NAME_PREFIX}.so)
+ENDIF()
+
+set(DATA_TYPES_MODULE_NAME native_schema_registry_c_data_types)
+set(SERDE_MODULE_NAME native_schema_registry_c)
+set(AWS_COMMON_MEMALLOC aws_common_memalloc)
+set(NATIVE_SCHEMA_REGISTRY_MODULE_NAME libnativeschemaregistry)
+set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS ON)
+
+include_directories("include")
+include(cmake/FetchAwsCommon.cmake)
+include_directories(${LIB_NATIVE_SCHEMA_REGISTRY_PATH})
+
+add_subdirectory("src")
+include (CTest)
+enable_testing()
+add_subdirectory("test")

native-schema-registry/c/README.md

@@ -0,0 +1,46 @@
+# Native Schema Registry in C 
+
+This module provides a C language based API for the schema registry serializer / de-serializers.
+
+## Build
+We use CMake to build the targets in this module.
+
+### Compile
+```asm
+#Run in c directory
+
+cmake -S. -Bbuild 
+cd build 
+cmake --build .
+
+#### Clean
+cmake --build . --target clean
+```
+### Testing
+```asm
+ctest .
+#Re-run failed tests with verbose output
+ctest --rerun-failed --output-on-failure
+```
+
+### Code Analysis
+Code is statically analyzed using clang-tidy.
+
+### Coverage
+Code coverage checks using gcov and lcov and fail if the coverage is below threshold.
+
+#### Installation
+You might have to install these modules using your OS package manager.
+
+### Sanitizers
+We use address,leak sanitizers to detect memory leaks and any potential issues during build. As of now, they only work on Linux.
+
+### Platform Support
+
+TBD
+
+## License
+
+**Project License** [Apache License Version 2.0](https://github.com/awslabs/aws-glue-schema-registry/blob/master/LICENSE.txt)
+
+N.B.: Although this repository is released under the Apache-2.0 license, its build dependencies include the third party Swig project. The Swig project's licensing includes the GPL-3.0 license.

native-schema-registry/c/cmake/FetchAwsCommon.cmake

@@ -0,0 +1,12 @@
+include(FetchContent)
+
+set(AWS_C_COMMON aws-c-common)
+
+FetchContent_Declare(
+        AWS_C_COMMON
+        GIT_REPOSITORY https://github.com/awslabs/aws-c-common.git
+        GIT_TAG        v0.7.4
+        GIT_SHALLOW    1
+)
+
+FetchContent_MakeAvailable(AWS_C_COMMON)

native-schema-registry/c/format-check.sh

@@ -0,0 +1,24 @@
+#!/usr/bin/env bash
+
+if [[ -z $CLANG_FORMAT ]] ; then
+    CLANG_FORMAT=clang-format
+fi
+
+if NOT type $CLANG_FORMAT 2> /dev/null ; then
+    echo "No appropriate clang-format found."
+    exit 1
+fi
+
+FAIL=0
+SOURCE_FILES=`find src include -type f \( -name '*.h' -o -name '*.c' \)`
+for i in $SOURCE_FILES
+do
+    $CLANG_FORMAT -output-replacements-xml $i | grep -c "<replacement " > /dev/null
+    if [ $? -ne 1 ]
+    then
+        echo "$i failed clang-format check."
+        FAIL=1
+    fi
+done
+
+exit $FAIL

native-schema-registry/c/include/glue_schema_registry_deserializer.h

@@ -0,0 +1,35 @@
+#ifndef GLUE_SCHEMA_REGISTRY_DESERIALIZER_H
+#define GLUE_SCHEMA_REGISTRY_DESERIALIZER_H
+
+#include "glue_schema_registry_error.h"
+#include "glue_schema_registry_schema.h"
+#include "mutable_byte_array.h"
+#include "read_only_byte_array.h"
+#include <stdbool.h>
+
+typedef struct glue_schema_registry_deserializer {
+    /* This is used for storing the instance context. Currently, being used for
+     * managing GraalVM instance. */
+    void *instance_context;
+} glue_schema_registry_deserializer;
+
+glue_schema_registry_deserializer *new_glue_schema_registry_deserializer(glue_schema_registry_error **p_err);
+
+void delete_glue_schema_registry_deserializer(glue_schema_registry_deserializer *deserializer);
+
+mutable_byte_array *glue_schema_registry_deserializer_decode(
+    const glue_schema_registry_deserializer *deserializer,
+    const read_only_byte_array *array,
+    glue_schema_registry_error **p_err);
+
+glue_schema_registry_schema *glue_schema_registry_deserializer_decode_schema(
+    const glue_schema_registry_deserializer *deserializer,
+    const read_only_byte_array *array,
+    glue_schema_registry_error **p_err);
+
+bool glue_schema_registry_deserializer_can_decode(
+    const glue_schema_registry_deserializer *deserializer,
+    const read_only_byte_array *array,
+    glue_schema_registry_error **p_err);
+
+#endif /* GLUE_SCHEMA_REGISTRY_DESERIALIZER_H */

native-schema-registry/c/include/glue_schema_registry_error.h

@@ -0,0 +1,56 @@
+#ifndef GLUE_SCHEMA_REGISTRY_ERROR_H
+#define GLUE_SCHEMA_REGISTRY_ERROR_H
+
+#include <stdio.h>
+
+/* Error codes are arbitrarily listed from 5000. No specific reason. */
+enum aws_gsr_error_code {
+    AWS_GSR_ERR_CODE_INVALID_STATE = 5000,
+    AWS_GSR_ERR_CODE_NULL_PARAMETERS = 5001,
+    AWS_GSR_ERR_CODE_GRAALVM_INIT_EXCEPTION = 5002,
+    AWS_GSR_ERR_CODE_GRAALVM_TEARDOWN_EXCEPTION = 5003,
+    AWS_GSR_ERR_CODE_INVALID_PARAMETERS = 5004,
+    AWS_GSR_ERR_CODE_RUNTIME_ERROR = 5005
+};
+
+/* TODO: Improve error reporting to respect logging levels. */
+#define log_warn(msg, code) fprintf(stderr, "WARN: %s, Code: %d\n", msg, code)
+
+#define AWS_GSR_MAX_ERROR_MSG_LEN 10000
+
+/** Defines the glue_schema_registry_error structure for holding error messages
+ * and codes resulting from function executions.
+ */
+typedef struct glue_schema_registry_error {
+    char *msg;
+    int code;
+} glue_schema_registry_error;
+
+glue_schema_registry_error *new_glue_schema_registry_error(const char *err_msg, int err_code);
+
+void delete_glue_schema_registry_error(glue_schema_registry_error *error);
+
+/* Copies the given error's msg into dst array trimming the size as necessary. */
+void glue_schema_registry_error_get_msg(const glue_schema_registry_error *error, char *dst, size_t len);
+
+/**
+ * Creates an instance of glue_schema_registry_error and writes it to the given
+ * glue_schema_registry_error pointer holder (*p_err). It is expected that
+ * *p_err is initialized by caller.
+ * @param p_err Initialized glue_schema_registry_error pointer holder.
+ * @param msg Error message to write.
+ * @param code Non-zero error code.
+ */
+void throw_error(glue_schema_registry_error **p_err, const char *msg, int code);
+
+/**
+ * Creates a pointer to hold an instance of glue_schema_registry_error
+ */
+glue_schema_registry_error **new_glue_schema_registry_error_holder(void);
+
+/**
+ * Deletes the pointer holder of glue_schema_registry_error and it's content.
+ */
+void delete_glue_schema_registry_error_holder(glue_schema_registry_error **p_err);
+
+#endif /* GLUE_SCHEMA_REGISTRY_ERROR_H */

native-schema-registry/c/include/glue_schema_registry_memory_allocator.h

@@ -0,0 +1,16 @@
+#ifndef GLUE_SCHEMA_REGISTRY_MEMORY_ALLOCATOR_H
+#define GLUE_SCHEMA_REGISTRY_MEMORY_ALLOCATOR_H
+
+#include <stdlib.h>
+
+/*
+ * Wrapper over AWS SDK Common memory allocator.
+ */
+
+void *aws_common_malloc(size_t size);
+
+void *aws_common_calloc(size_t count, size_t size);
+
+void aws_common_free(void *ptr);
+
+#endif /* GLUE_SCHEMA_REGISTRY_MEMORY_ALLOCATOR_H */

native-schema-registry/c/include/glue_schema_registry_schema.h

@@ -0,0 +1,47 @@
+#ifndef GLUE_SCHEMA_REGISTRY_SCHEMA_H
+#define GLUE_SCHEMA_REGISTRY_SCHEMA_H
+
+#include "glue_schema_registry_error.h"
+
+/*
+ * Glue Schema Registry Schema structure that represents
+ * schema object required by Glue Schema Registry Serializers / De-serializers.
+ */
+typedef struct glue_schema_registry_schema {
+    /* String name of the schema */
+    char *schema_name;
+
+    /* Complete definition of the schema as String */
+    char *schema_def;
+
+    /* Data format name, JSON, AVRO, PROTOBUF as String */
+    char *data_format;
+
+} glue_schema_registry_schema;
+
+/*
+ * Creates a new instance of glue_schema_registry_schema
+ */
+glue_schema_registry_schema *new_glue_schema_registry_schema(
+    const char *schema_name,
+    const char *schema_def,
+    const char *data_format,
+    glue_schema_registry_error **p_err);
+
+/*
+ * Deletes the glue schema registry schema.
+ */
+void delete_glue_schema_registry_schema(glue_schema_registry_schema *schema);
+
+/*
+ * Gets different attributes from glue_schema_registry_schema instance.
+ * These getter methods are translated into "Getter" methods in target
+ * languages.
+ */
+const char *glue_schema_registry_schema_get_schema_name(const glue_schema_registry_schema *schema);
+
+const char *glue_schema_registry_schema_get_schema_def(const glue_schema_registry_schema *schema);
+
+const char *glue_schema_registry_schema_get_data_format(const glue_schema_registry_schema *schema);
+
+#endif /* GLUE_SCHEMA_REGISTRY_SCHEMA_H */

native-schema-registry/c/include/glue_schema_registry_serializer.h

@@ -0,0 +1,27 @@
+#ifndef GLUE_SCHEMA_REGISTRY_SERIALIZER_H
+#define GLUE_SCHEMA_REGISTRY_SERIALIZER_H
+
+#include "glue_schema_registry_error.h"
+#include "glue_schema_registry_schema.h"
+#include "mutable_byte_array.h"
+#include "read_only_byte_array.h"
+
+typedef struct glue_schema_registry_serializer {
+    /* This is used for storing the instance context. Currently being used for
+     * managing GraalVM instance. */
+    void *instance_context;
+} glue_schema_registry_serializer;
+
+glue_schema_registry_serializer *new_glue_schema_registry_serializer(glue_schema_registry_error **p_err);
+
+void delete_glue_schema_registry_serializer(glue_schema_registry_serializer *serializer);
+
+// Encodes the GSR Schema with a byte array.
+mutable_byte_array *glue_schema_registry_serializer_encode(
+    const glue_schema_registry_serializer *serializer,
+    const read_only_byte_array *array,
+    const char *transport_name,
+    const glue_schema_registry_schema *gsr_schema,
+    glue_schema_registry_error **p_err);
+
+#endif /* GLUE_SCHEMA_REGISTRY_SERIALIZER_H */