neopb/pb2c.rb

#!/usr/bin/ruby

class Tokenizer
    TOKEN_TYPES = [
        [:function, /\bfunction\b/i],
        [:sub, /\bsub\b/i],
        [:end, /\bend\b/i],
        [:as, /\bas\b/i],
        [:typename, /\blong\b/i],
        [:identifier, /\b[a-zA-Z]+\b/],
        [:integer, /\b[0-9]+\b/],
        [:string, /".*"/],
        [:oparen, /\(/],
        [:cparen, /\)/],
        [:comma, /,/],
        [:quote, /'/],
    ]
    def initialize(code)
        @code = code
    end

    def tokenize
        tokens = []
        begin
            until @code.empty?
                tokens << tokenize_one_token
                @code = @code.strip
            end
        rescue RuntimeError => e
            puts tokens.join("\n")
            raise
        end
        tokens
    end

    def tokenize_one_token
        TOKEN_TYPES.each do |type, re|
            re = /\A(#{re})/
            if @code =~ re
                value = $1
                @code = @code[value.length..-1]
                return Token.new(type, value)
            end
        end
        raise RuntimeError.new(
            "Couldn't match token on #{@code.inspect}")
    end
end

Token = Struct.new(:type, :value)

class Parser
    def initialize(tokens)
        @tokens = tokens
    end

    def parse
        parse_function
    end

    def parse_function
        consume(:function)
        name = consume(:identifier).value
        arg_names = parse_arg_names
        consume(:as)
        rtype = consume(:typename).value
        body = parse_expr
        consume(:end)
        consume(:function)
        FunctionNode.new(name, rtype, arg_names, body)
    end

    def parse_arg_names
        arg_names = []
        consume(:oparen)
        if peek(:identifier)
            arg_names << consume(:identifier).value
            while peek(:comma)
                consume(:comma)
                arg_names << consume(:identifier).value
            end
        end
        consume(:cparen)
        arg_names
    end

    def parse_expr
        if peek(:integer)
            parse_integer
        elsif peek(:string)
            parse_string
        elsif peek(:identifier) && peek(:oparen, 1)
            parse_call
        elsif peek(:identifier) && peek(:string, 1)
            parse_stmt
        else
            parse_var_ref
        end
    end

    def parse_stmt
        name = consume(:identifier).value
        arg_exprs = consume(:string).value
        CallNode.new(name, arg_exprs)
    end

    def peek(expected_type, offset=0)
        @tokens.fetch(offset).type == expected_type
    end

    def consume(expected_type)
        token = @tokens.shift
        if token.type == expected_type
            token
        else
            raise RuntimeError.new(
                "Expected token type #{expected_type.inspect} but got #{token.type.inspect}")
        end
    end
end

FunctionNode = Struct.new(:name, :type, :arg_names, :body)
StringNode = Struct.new(:value)
CallNode = Struct.new(:name, :arg_exprs)

class Generator
    def generate(node)
        case node
        when FunctionNode
            "%s %s(%s) { return %s ; }" % [
                node.type.downcase,
                node.name,
                node.arg_names.join(','),
                generate(node.body),
            ]
        when CallNode
            "%s(%s)" % [
                node.name,
                node.arg_exprs
            ]
        when StringNode
            node.value
        else
            raise RuntimeError.new("Unexpected node type: #{node.class}")
        end
    end
end

tokens = Tokenizer.new(File.read("hello.bas")).tokenize
#puts "Tokens:\n"
#puts tokens.join("\n")
tree = Parser.new(tokens).parse
#puts "\nAST:\n"
#puts tree
RUNTIME = "#include <stdio.h>\n#define PRINT(a) printf(a)\n"
CMAIN   = "int main(void) { PBMain(); return 0; }"
generated = Generator.new.generate(tree)
#puts "\nGenerated:\n"
#puts generated
#puts "\nGenerated with preamble/postamble:\n"
puts [RUNTIME, generated, CMAIN].join("\n")