MNCC

MNCC is a small C-to-assembly transpiler prototype.

Current behavior is:

1. Read one C source file.
2. Lex it into tokens.
3. Parse top-level functions into an AST.
4. Generate NASM x86-64 assembly into .github/a.asm.
5. Execute assemble.sh, which runs NASM + LD to produce .github/a.out.

What It Is (Today)

• Single-file frontend + codegen prototype.
• Targets Linux x86-64 (uses _start + syscall exit sequence).
• Produces and assembles a fixed output path (.github/a.asm, .github/a.out).
• Supports only a narrow subset of C-like syntax.

What Is Actually Supported End-to-End

• Function definitions returning int
• Function parameters typed as int (parsed as declarations)
• Assignments with arithmetic expressions: +, -, *, /
• Function call expressions in assignments/returns
• return; and return <expr>;

Example that works with current pipeline:

int add(int a, int b)
{
    return a + b;
}

int main()
{
    return add(15, 8);
}

Important Current Limits

• CLI takes exactly one argument: ./mncc <path>
• Output filename cannot be configured from CLI
• No control-flow parsing (if, else, while, for)
• No comparisons/logical operators
• No unary minus expression handling
• No parenthesized expression handling
• Declaration without initializer (example: int x;) is not accepted by parser
• Standalone call statement (example: foo();) is parsed but not generated
• Parser/codegen failures are printed, but the process currently may still exit with 0 due to missing error propagation in main.c

Documentation

• docs/ARCHITECTURE.md
• docs/GRAMMAR.md
• docs/TOKENS.md
• docs/USAGE.md

Error Codes Defined In Headers

• 0: SUCCESS
• 84: ERROR
• 100: ELEX
• 101: EPAR
• 102: EINP
• 103: EGEN

Note: these codes are used internally by modules, but the top-level executable currently does not always propagate parse/codegen failures to its final exit code.