Kaleidoscope compiler using libfirm.
C
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
examples
src
Makefile
README.md

README.md

Kaleidoscope

Kaleidoscope is a very simple programming language, and the compiler itself.

The project is actually an example about using libfirm.

This work is based on (if not copied from) a tutorial from researchers in the Programming paradigms group at the Karlsruhe Institute of Technology. That tutorial included the parser itself, that I replaced by flex/bison because I thought it was not the goal of the tutorial to teach about parsing code.

Installation

The following tools are needed to build the project:

  • gcc
  • g++
  • bison
  • flex
  • make

and libfirm.

Then, just run

make

How it works

kaleidoscope.lex contains the token definitions for the lexical analyzer. That is, flex uses it to transform the Kaleidoscope code (as a plain character string) to a list of blocks which have a semantic value.

kaleidoscope.y is a language grammar for the parser generator. That is, bison uses it to transform the tokens into libfirm's graphs, finally generating an Assembly file.

During the parsing of an expression, an abstract syntax tree is built, because at this time the "owner" of the corresponding graph nodes is still unknown. These nodes are created on a function definition.

Language reference

def defines a new function. It has a prototype (a name followed by parenthesis that may enclose a list of variables), and an expression as the body (a value, an elementary operation, or a function call).

main is a special keyword defining the main function of the program. It takes no prototype (because it can't change: it has no parameter), just a body.

extern declares extern functions (with just a prototype, no body).

# introduces a comment: everything on the line after the sharp will be ignored by the compiler.

See the examples directory.

Complete syntax

Complete syntax in EBNF (does not describe precedence and comments):

input = {stat}

stat = "main" , expr  (* main function *)
     | "def" , prototype , expr  (* function definition *)
     | "extern" , prototype  (* function declaration *)

prototype = identifier , "(" , [ { identifier , "," } identifier ] , ")"

expr = double  (* constant value *)
     | identifier  (* variable *)
     | identifier , "(" , [ { expr , "," } , expr ] , ")"  (* function call *)
     | expr , binop , expr  (* binary operation *)

binop = "+" | "*"  (* hey! this is just an example *)

(* double and identifier are trivially defined *)

The syntax is split in the .lex and .y files, and is a little bit longer due to the recursive rules.