#rewriteInC

Frontend for C, concentrating on: generation from code; single-file analysis; modification; and emission (to C code).

Use-cases:

• Generate/update free calling int StructName_cleanup(struct StructName*) functions, e.g.:
  • struct Foo { struct Bar *bar;}; struct Bar { int a; }; will generate:
  • int Bar_cleanup(struct Bar*); &
  • int Foo_cleanup(struct Foo*); (which internally will call Bar_cleanup)
• With fmt of JSON, INI, YAML, TOML, &etc.:
  • Generate/update parsers from const char* to struct: const int StructName_<fmt>_parse(struct StructName*, const char*) function;
  • Generate/update emitters from struct to char*: const int StructName_<fmt>_emit(const struct StructName*, char*) function;
• And helper functions inspired by Haskell typeclasses & Rust traits like:
  • bool StructName_eq(struct StructName*, struct StructName*);
  • int StructName_default(struct StructName*);
  • int StructName_deepcopy(const struct StructName*, struct StructName*);
  • int StructName_display(struct StructName*, FILE*);
  • int StructName_debug(struct StructName*, FILE*);
• Generate the Google Cloud client library for C ( with Google Cloud API Discovery Service as input);
• Generate arbitrary C client libraries (with OpenAPI as input);
• Generating #pragma for every function to expose them for calling from, e.g., assembly;
• Prepare C SDK to be wrapped in higher-level language SDKs (SWIG style)

In 2025, IMHO, there are four reasons to write in C:

0. Maintain an existing codebase;
1. Write low-level code (e.g., kernels, drivers, microcontrollers, compilers, assembly helpers);
2. Write performant code (e.g., databases, filesystems, math libraries);
3. Write interoperable code.

The focus of this venture is the final point… interoperability:

graph TD
  C[C]

  C --> C++
  C --> C#
  C --> Objective-C
  C --> Java
  C --> Python
  C --> Go
  C --> Rust
  C --> JavaScript
  C --> Perl
  C --> PHP

  C --> Swift
  C --> Kotlin
  C --> Lua
  C --> D
  C --> Scala
  C --> Tcl
  C --> Ruby
  C --> Fortran
  C --> Ada

  C --> Haskell
  C --> Elixir
  C --> Julia
  C --> Groovy
  C --> Crystal
  C --> Nim
  C --> OCaml
  C --> V
  C --> Eiffel
  C --> F#

  C --> TypeScript
  C --> COBOL
  C --> SQL
  C --> MATLAB
  C --> Lisp
  C --> Prolog
  C --> Dart
  C --> Erlang

  C --> R
  C --> Zig
  C --> Forth

(40 popular programming languages which can easily call into C)

This C compiler has a very unusual design, the macro and C languages are treated as one. The foci are:

• location start/end of function, struct, and feature macros (e.g., #ifdef JSON_EMIT then #endif /* JSON_EMIT */);
• struct fields.

…which enables a number of use-cases to be simply developed, e.g., see the list above.

• Macros aren't evaluated, which means a simple #define LBRACE { will break cdd-c.
• Like m4, CMake, and other tools; cdd-c must be run before your code is built.

Install: CMake ; C compiler toolchain ; git. Then:

$ git clone "https://github.com/offscale/vcpkg" -b "project0"
# Windows:
$ vcpkg\bootstrap-vcpkg.bat
# Non-Windows:
$ ./vcpkg/bootstrap-vcpkg.sh
# Both Windows and non-Windows:
$ git clone "https://github.com/SamuelMarks/cdd-c" && cd "cdd-c"  # Or your fork of this repo
# Windows
$ cmake -DCMAKE_BUILD_TYPE="Debug" -DCMAKE_TOOLCHAIN_FILE="..\vcpkg\scripts\buildsystems\vcpkg.cmake" -S . -B "build"
# Non-Windows
$ cmake -DCMAKE_BUILD_TYPE='Debug' -DCMAKE_TOOLCHAIN_FILE='../vcpkg/scripts/buildsystems/vcpkg.cmake' -S . -B 'build'
# Both Windows and non-Windows:
$ cmake --build "build"
# Test
$ cd "build" && ctest -C "Debug" --verbose

code2schema <header.h> <schema.json>
generate_build_system <build_system> <output_directory> <basename> [test_file]
jsonschema2tests <schema.json> <header_to_test.h> <output-test.h>
schema2code <schema.json> <basename>
sync_code <header.h> <impl.c>

For example, you can:

$ mkdir 'build'
# Create build system file for CMake
$ bin/c_cdd_cli generate_build_system cmake 'build' 'simp'
# Create simp.h and simp.c dataclass from simple_json.schema.json
$ bin/c_cdd_cli schema2code 'c_cdd/tests/mocks/simple_json.schema.json' 'build/src/simp'
# Create tests
$ bin/c_cdd_cli jsonschema2tests 'c_cdd/tests/mocks/simple_json.schema.json' 'simp.h' 'build/src/test/test_simp.h'
# Observe what was generated
$ tree --charset=ascii
.
|-- CMakeLists.txt
`-- src
    |-- CMakeLists.txt
    |-- lib_export.h
    |-- simp.c
    |-- simp.h
    |-- test
    |   |-- CMakeLists.txt
    |   |-- test_main.c
    |   `-- test_simp.h
    `-- vcpkg.json

3 directories, 9 files
# Configure with cmake (replace toolchain path with yours)
$ cmake -S 'build' -B 'build/cmake_debug_build' -DCMAKE_BUILD_TYPE='Debug' -DCMAKE_TOOLCHAIN_FILE='vcpkg/scripts/buildsystems/vcpkg.cmake'
# Build with cmake
$ cmake --build 'build/cmake_debug_build'
# Test with cmake (ctest)
$ ctest -C 'Debug' --build-run-dir 'build/cmake_debug_build' --verbose

Usage: code2schema <header.h> <schema.json>

Generates JSON Schema from structs located in specified C/C++ header or source file.

Usage: generate_build_system <build_system> <output_directory> <basename> [test_file]

Basic helpers to write build files for CMake, Makefile, Meson, Bazel.

• build_system: cmake | make | meson | bazel
• output_directory: where to generate the build files to
• basename: base name of the generated .c and .h files
• test_file: optional test .c file path (e.g., greatest.h based tests)

Generates build files in the current directory.

Usage: jsonschema2tests <schema.json> <header_to_test.h> <output-test.h>

Generates greatest.h based tests for the generated code.

Usage: schema2code <schema.json> <basename>

Generates structs from JSON Schema to specified C/C++ basename (that will generate header [basename.h] and source files [basename.c] for).

Usage: sync_code <header.h> <impl.c>

Update existing source code based on [potentially] handwritten changes to the code, e.g., changing the struct fields by hand should update all the to_json, from_json, eq, to_str, and from_str functions.

Licensed under either of

• Apache License, Version 2.0 (LICENSE-APACHE or https://www.apache.org/licenses/LICENSE-2.0)
• MIT license (LICENSE-MIT or https://opensource.org/licenses/MIT)

at your option.

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.

This repository uses clang-format to maintain source code formatted in LLVM style. Before committing for the first time, please install pre-commit on your system and then execute the following command to install pre-commit hooks: {{{ pre-commit install }}}