Revised since first distributed – please check for changes
Make sure you are working with the latest versions of the class code compatible with Julia 1.0.
This lab is designed to help you begin to understand how to design and implement an interpreter for a simple language. The concepts you will practice here will be foundation to further labs!
You will need to download and install Julia for this lab.
Note: if you are using Ubuntu, be careful using the
apt version of Julia (it may be too old!). You should use Julia 1.0 or later..
Note that the code required for this lab is available on LearningSuite, in the “Content” section, under the “Interpreters” subsection, on the page titled “Base Interpreter”. You will need the following files:
Lexer.jl- the lexical analyzer
Error.jl- defines an error exception we'll throw as appropriate
CI0.jl- the base interpreter built in class
You should not modify the
Error.jl, and these will stay the same for all of the interpreter assignments.
After playing with it to make sure it's working and you have everything set up correctly, make a copy of
CI0.jl and name it
RudInt.jl (short for “Rudimentary Interpreter”, or this assignment). You should then edit
RudInt.jl using the provided CI0 interpreter as a base. Remember to change the name of the module defined within the file to
RudInt as well.
For code development, you may wish to use any of the following:
The instructor and TAs are prepared to help you with the first two (Juno and Jupyter), but you're on your own if you choose an arbitrary editor or IDE that we aren't familiar with.
For this lab, you will create a simple interpreter for a language we call OWL (“Our Widdle Language”). You need to create a Julia module that implements all requisite functionality.
Your module should export
Remember that you will use multiple dispatch to implement different “versions” of
calc, based on the input type.
An important difference between the code that you will implement for this lab, and the code we went through in class, is that your code should properly abstract multiple binary AST nodes into a single class we'll call
Please name your module RudInt and submit just the one file.
Define a data structure to contain a mapping from operator symbols to the functions that actually implement that symbol. These functions can be either built-in ones or ones that you write, so long as you preserve the semantic meaning of the operation.
Dict(:+ => +)
Write the following function(s):
expr will always be the output of the lexer, and will consist of numbers, symbols, and lists of numbers and symbols.
expr into an OWL datastructure according to this grammar:
<AE> ::= number | (+ <AE> <AE>) | (- <AE> <AE>) | (* <AE> <AE>) | (/ <AE> <AE>) | (mod <AE> <AE>) | (collatz <AE>) | (- <AE>)
where number is a Julia real number literal.
You should have test cases for all legal and illegal types of expressions. For example, the expression
(+ 1 2 3) will pass the lexer just fine, but it is not accepted by our grammar – our grammar can only handle two arguments, not three!
If you are given invalid input, you must throw an error, such as
throw( LispError(“Whoa there! Unknown operation!”) ). This is defined in
Error.jl. Do not just print an error message, as this will not by caught by the autograder's
You must parse
expr into the following types (copy and paste this to the top of your code):
abstract type AE end struct NumNode <: AE n::Real end struct BinopNode <: AE op::Function lhs::AE rhs::AE end
Plus any other subtypes you deem necessary.
Write the following function:
Consumes an OWL abstract syntax tree representing an expression and computes the corresponding numerical result.
It should throw a Lisp error if division by zero or collatz of a negative number or zero is attempted.
This interpreter has another binary operation called
mod, similar to other binary operations. You should implement it in Julia using Julia's built-in
mod function (note that this is not the same as the
% in-line operator, at least for negative numbers!).
One of the fundamental “primitives” in our language is a function called
collatz. This function is commonly used as an example of a function that is simple to write, but which cannot be analyzed – illustrating the difficulty of the general problem of program analysis. It comes from the
Collatz conjecture. Our collatz function returns the number of times the function recurses. Note that for some numbers, such as
n=28, the function returns quite quickly – only 18 iterations – but for neighboring numbers, such as
n=27, the function takes 111 iterations!
The function is defined as:
function collatz( n::Real ) return collatz_helper( n, 0 ) end function collatz_helper( n::Real, num_iters::Int ) if n == 1 return num_iters end if mod(n,2)==0 return collatz_helper( n/2, num_iters+1 ) else return collatz_helper( 3*n+1, num_iters+1 ) end end
Note that the
- operator has two distinct usages: as a binary operation (the “minus” operation) and as a unary operation (“negation”). You should create a distinct class to handle unary operations.
If you change a module, it can sometimes be tricky to convince Julia to reload it and not use the previously compiled and cached version. I strongly recommend using the
Revise module and loading it before you load the rest of your code. If it's working as designed, it should automatically reload any subsequently loaded modules if the corresponding source code file changes. (If you have not already installed the
Revise package, use Julia's package manager to install it.)
using Revise # make sure to load before the others using Error using Lexer using RudInt
In order for your code to work with the solo-grader you will need to do following:
juliabinary is in your executable path (like you did for
~/.julia/config/startup.jlfile, then just make sure to run the autograder from the same directory in which you have RudInt, Lexer, and Error.
using Lexer. (These should already be in the base code we give you to start with.)
Changes since first given this semester: