The C piler cplr is a tool for executing C code from the shell and is meant to break the barrier between the shell and the core language of UNIX, which it does by making the compiler a tool like sed, grep and awk.
The practical purpose of cplr is to allow the compilation and execution of short C programs for testing, system information retrieval, live coding and interactive development of C small sequences and programs.
$ cplr 'puts("hello!")'
hello!
$ cplr 'printf("%d\n", getpagesize())'
4096
cplr itself is a simple compiler front-end that does not have any comprehension of the C language or even data types, but this is why it works well because it does not try to do something hard. Instead we follow the UNIX philosophy and do just one thing: act as the interactive equivalent of cc.
The current backend of cplr is based on TinyCC, which gives it a language level of C99 or better and the ability to interact with almost any C library on a normal Linux system including large and complicated ones. You can run Gtk or Python in this tool without having to expect any issues.
A basic use is to retrieve type information:
$ cplr -i linux/kvm.h 'printf("%zu\n", sizeof(struct kvm_regs))'
144
You can also use cplr to create ad-hoc executables:
$ cplr -o realpath 'puts(realpath(argv[1],NULL))'
$ ./realpath .
/home/user
Small C libraries can be used directly from source:
$ cplr -s linenoise.c -i linenoise.h 'printf("Line: %s\n", linenoise("% "))'
% hello
Line: hello
Thanks to pkg-config you can easily call any library:
$ cplr -P tinfo -i term.h 'setupterm(NULL,1,NULL)' 'printf("%d\n", tigetnum("colors"))'
256
We build using CMake. You need readline. We bring our own copy of TinyCC.
$ git clone https://github.com/promovicz/cplr.git
$ cd cplr
$ cmake .
$ make
$ make install
- C REPL (experimental form exists)
- Garbage collection (in progress)
- Crash handlers (easy)
- Backtrace support
- Bounds checking
- Instruction trace
- Call trace
- Internal disassembler (capstone)
- Support other compilers using dlopen
- Support launching into gdb/lldb
- Source code and assembly dumping
- Editor and pager integration
- Live process code injection (researching)
- Live kernel code injection (needs gcc)
- Support for C++ and ObjC (needs gcc/clang)