CMSC 330, Fall 2015

Organization of Programming Languages

Project 4 - Scheme Parser & Interpreter

Due 11:59pm November 9th, 2015 (Monday)


In this project, you will write an interpreter for a tiny subset of the Scheme programming language. As part of the project, you will also write a parser that translates a plain text Scheme program into an abstract syntax tree (AST), and an evaluator that executes the code represented as an AST.

For a quick introduction to Scheme, check out these lecture notes. In order to test your project can try running small Scheme programs online at here. In order to test your project you can try running it on Grace using the Scheme interpreter guile. It behaves similarly to the OCaml top-level. You might try to get the logic of your functions right, first, by writing them in OCaml, and then translate them to Scheme.

If you want to find out more about Scheme, you can download the Revised6 Report on the Algorithmic Language Scheme. You may also try Teach Yourself Scheme in Fixnum Days.

For purposes of this project, we will only test your interpreter with valid input. Thus your code may do whatever you want on a bad input. We do, however, recommend adding reasonable error handling code to your project for cases of malformed or otherwise incorrect Scheme input, because it will make developing your project easier. As you are testing your program, you may inadvertently create incorrect input data; substantial time may be lost in trying to debug the program, only to find a few mistyped characters in your input data are the source of the problem.

Also, even though you need all of these pieces written to have a "complete" Scheme interpreter, we've structured the project so that you can work on any of the separate phases independently and test them on their own. So if you get stuck on an earlier part of the project, you can still move on to a later part.

Getting Started

Download the following archive file and extract its contents.

Along with files used to make direct submissions to the submit server (submit.jar, .submit, submit.rb), you will find the following project files:

Part 1: Simple Scheme Programming

Put your solution to this part in the file schemeTest.txt.

Implement the following functions in regular Scheme. All of these functions should operate over integers and/or lists of integers. The purpose of this part is just to make sure you understand Scheme before you try to start writing an interpreter for it. You will find that Scheme is very much the same as OCaml, but with different syntax, and with dynamic (rather than static), typing. You might find these lecture notes useful.

  1. Write a function double x that returns two times x.
  2. Write a function powof2 x that returns true (which is written #t in Scheme) if and only if x is a power of 2. (Hint: Use the functions / and mod.)
  3. Write a function sum l that returns the sum of the integer list l, using recursion. Hint: You can use the function pair? to determine whether a value is a cons cell or not; and in this problem, you can assume if something is not a cons cell, then it's '().
  4. Write a function applyToList f l that returns a new list containing all the elements of l, in the same order, but with f applied to them. Your implementation should use recursion. You may not use Scheme's built-in map or for-each functions in writing your map function. Note applyToList directly takes 2 arguments (i.e., it is not using currying).

Part 2: Parsing Scheme

Put your solution to this part in the top half of file

Your next task is to write a parser for Scheme expressions, which in this case will be a function that turns a string into a Scheme abstract syntax tree (AST). Your parser will take as input a sequence of tokens, produced by a scanner, which are the terminals of the Scheme grammar, and output the Scheme AST.

We've supplied you with a function tokenize : string -> token list that acts as a scanner/lexer, converting the string input into a list of tokens, represented by the following data type:

type token =
  | Tok_Id of string
  | Tok_Num of int
  | Tok_String of string
  | Tok_True
  | Tok_False
  | Tok_LParen
  | Tok_RParen
  | Tok_END
For example, when called as tokenize "(foo (bar 3) 4 \"baz\")", the return value is
[Tok_LParen; Tok_Id "foo"; Tok_LParen; Tok_Id "bar"; Tok_Num 3; Tok_RParen; Tok_Num 4; Tok_String "baz"; Tok_RParen]

What to do: You must write a function parse : token list -> ast that takes as input a list of tokens (returned from tokenize) and returns an AST. Once you have done this, you can run it using the code in the file

You should use the idea of a recursive descent parser, as we discussed in class. Thus we suggest you write two functions: parse_S, which parses the non-terminal S representing a single Scheme expression, and parse_L, which parses the non-terminal L representing a list of Scheme expressions. The context free grammar for Scheme -- and notably for nonterminals S and L -- is given next, followed by the OCaml type definition ast of Scheme abstract syntax trees.

Scheme Grammar

The grammar for Scheme expressions you need to support is particularly simple:

  • S -> id | n | str | #t | #f | ( L )
  • L -> S L | epsilon
  • id are Scheme identifiers. Identifiers may contain upper- and lower-case letters, digits, =, *, +, /, < >, !, ?, and -. For example, valid identifiers are Foo, set!, <three, and +=. Identifiers may not begin with a digit. These correspond to the Tok_Id token. Identifiers are distinguished from integers and strings by their first character.
  • n are Integers (made up of digits). These correspond to the Tok_Num token.
  • str are Strings (beginning and ending with quotes). For purposes of this project, only alphanumeric and whitespace characters can appear within a string. These correspond to the Tok_String token.
  • #t and #f are tokens representing true and false. These correspond to the tokens Tok_True and Tok_False, respectively.
  • ( and ) are tokens (marking the beginning and end of a list); these correspond to the tokens Tok_LParen and Tok_RParen, respectively.

Scheme AST

For this project, Scheme expressions are represented using an AST (abstract syntax tree) defined using the following OCaml data type:

type ast =
  | Id of string
  | Num of int
  | Bool of bool
  | String of string
  | List of ast list
For example, the Scheme s-expression (foo (bar 3) 4 "baz") is represented by the AST List [Id "foo"; (List [Id "bar"; Num 3]); Num 4; String "baz"]. Your AST nodes for strings should not include quotes. For example, parse ["\"a\""] should return String "a", i.e., a String constructed from a string with one character, a.

Part 3: Scheme Interpreter

For part 3 of this project, your task is to write a evaluator for Scheme abstract syntax trees. This evaluator will form the basis of your Scheme interpreter. Put the code for the Scheme interpreter is in the bottom part of

What you will do: You will write a function eval that, given an environment and an AST, evaluates the expression corresponding to that AST in the given environment, producing a final value. The type of eval is (string * value) list -> ast -> value, where the first argument is the environment, the second is the AST, and the result is a value. We present the OCaml definition (and meaning) of values and environments below.

Note that what you will implement for this part corresponds very closely to the operational semantics for OCaml-like programs give in lecture, so that may serve as a good reference (and this project may serve as a good way to understand that lecture better).

Scheme Values

Your Scheme interpreter should represent the value of Scheme expressions using the following value user-variant data type:

type value =
  | Val_Num of int
  | Val_Bool of bool
  | Val_String of string
  | Val_Null
  | Val_Cons of value * value
  | Val_Define of (string * value) list
  | Val_Closure of ...
where the part labeled ... is for you to fill in. However, you must not change the part of value we have given you, because our grading scripts will look for exactly those constructors, with exactly those arguments as given, to test your interpreter.

When doing this project, be sure to keep straight the difference between textual entities that the programmer has written down (like the text "3") with the resulting value that your interpreter produces (Val_Num 3).

Scheme Environment

Your interpreter will need to maintain an environment to store bindings of values to symbols. The interpreter will represent the environment as an associative list of type (string * value) list. How the associative list is organized is up to you. The interpreter will need to maintain a top-level environment of all definitions encountered so far in the input program.

For example, eval [] (Num 3) should return Val_Num 3, meaning that in an empty environment, an AST node containing the number 3 evaluates to the integer 3. Calling eval [("x", Val_Num 3)] (Id "x") should also return Val_Num 3, since x is bound to 3 in the environment used to evaluate x.

Scheme Language Features

Here we present the language features your Scheme interpreter needs to support.
  • Basic Expressions
    • Values. Your evaluation function should evaluate integers, booleans, and strings to the corresponding values.
    • Null. For this project, we will vary slightly from Scheme and make null the built-in keyword for the empty list. Thus null should evaluate to Val_null. In actual Scheme, null is written '().

  • Functions Calls
    • In Scheme function calls appear as lists, where the name of the function is the first element in the list, and the remaining list arguments are the arguments passed to the function. E.g., (+ 1 2 3) calls the built-in + function and passes it three arguments, 1, 2, and 3.
    • Your implementation must support call-by-value, so that arguments are evaluated before they are passed to a closure. For example, consider the following Scheme expression: (+ (- 1 2) 3). The arguments to the + function must be first evaluated before they are passed to +. In this case, it means the expression (- 1 2) must be evaluated first, and its resulting value passed as the first argument to +.

  • Built-in Functions
    • Your interpreter should support a number of built-in Scheme functions. Some of these functions can take more than one argument, but you don't have to implement them with currying. You can just make these special cases inside your evaluator.
    • You do not have to treat primitive functions as first-class values; they will only be invoked and applied to arguments (and not passed as arguments or used as return values).
    • boolean?, number?, string?, pair?, and null?. These built-in functions return true if their single argument is a boolean, integer, string, cons cell, or null, respectively, and false otherwise.
    • Integer operations. +, -, *, and = on integers.
      • + sums its arguments, and should accept one or more arguments.
      • - may take one or more of arguments. It subtracts its 2nd through last argument from its first argument; given only one argument, it compute unary negation. For example, (- 3) evaluates to Val_Num (-3), while (- 4 3) evaluates to Val_Num 1 and (- 4 3 1) evaluates to Val_Num 0.
      • * multiplies its arguments, and should accept one or more arguments.
      • = compares its two arguments and evaluates to either Val_Bool true or Val_Bool false. You may use the OCaml = operator.
    • If expressions. You should allow both (if cond tr fl), which evaluates to tr if cond is true and fl otherwise, and (if cond tr), which evaluates to Val_Null if cond is false.
    • display. Prints its single argument, which is either a number or a string. Do not add a trailing newline. If a string, do not include any quotes around the string when it is printed. Display should return the value null.

  • Lists
    • Your interpreter should support lists constructed from multiple calls to the cons function. The car and cdr functions may be used to extract the head and tail of a list.
      • cons - Here (cons x y) should evaluate to the cons cell Val_Cons(a, b), where a and b are whatever x and y evaluate to, respectively.
      • car - When applied to a cons cell Val_Cons(a, b), car returns a.
      • cdr - When applied to a cons cell Val_Cons(a, b), cdr returns b.

  • Top-level Definitions
    • Your evaluation function must maintain a top-level environment containing the values of variables that have been bound using the built-in define function. For example, (define x y) will bind the value of y to the symbol x.
    • The top-level environment is passed to the eval function used to evaluate Scheme expressions. Scheme expressions containing unbound (free) variables may look in the top-level environment for values.
    • You may assume define is used only at the top level, and not within the body of a function; we will not test your interpreter with any such examples. You also do not need to handle cases where we shadow primitive operators.
    • The expression (define x ...) when evaluated returns a new top-level environment that contains the binding for x, as well as all bindings in the current top-level environment.
    • Future calls to eval at the top level must be made with the new top-level environment. Your implementation of eval must use the most recent binding for x within the top-level environment, if x is a free variable.
    • Remember that define may be used to bind different values (possibly with different type) to the same identifier, causing shadowing. When the value of a shadowed identifier is looked up, we return its latest definition. For instance:
        (define x 1)
        (+ x 2)  <-- produces 3
        (define x 4)
        (+ x 2)  <-- produces 6
    • The actual top-level environment in Scheme supports dynamic scoping (where identifier bindings may be changed by define). For this project, your top-level environment will be like the OCaml top-level environment, where identifier bindings do not change, but may be shadowed by newer bindings to an identifier with the same name.
    • You may assume that identifiers bound to user functions (lambda or dynamic) will not be shadowed (i.e., will only be defined once).

  • Identifiers
    • When presented with an identifier, your interpreter needs to look it up. The first place it should look it up is in the "local" environment, consisting of parameters of enclosing user-defined functions.
    • If an identifier is not in the local environment, the interpreter should look it up in the top-level environment (the one that define uses).
    • If an identifier is in neither environment, then it is unbound, and trying to evaluate it should produce an error. Since we are testing your code with only valid Scheme expressions, there will not be any unbound variables in the submit server test cases.
    • Scheme expressions use static lexical scoping to determine variable bindings in the local environment. For this project, user functions defined with the "dynamic" keyword will use dynamic scoping instead to determine where variables are bound in the local environment.

  • User Defined Functions
    • You should support creating anonymous functions of one argument. We will only test user-defined functions called with a single argument (though functions may use currying). You need to support functions using both static (lexical) and dynamic scoping. Both types of functions produce a Val_Closure value that you must define.
    • Lambda
      • User-defined functions supporting static lexical scoping are defined using the lambda keyword. Function definitions must be of the form (lambda (x) body), where x is the formal parameter, and contains the code for the function. Note x must be in a list, (lambda x body) is invalid syntax. The body of the function is a single expression; it does not have to be a list. For instance, the user-defined function may be returning a simple expression (e.g., 1, x).
      • To support static lexical scoping, your value for lambda must support closures. I.e., the function will need to be able to access values bound to identifiers not in the scope of the lambda expression, even if those values are bound to parameters of enclosing functions that have already returned.
      • The following example shows how user-defined functions should evaluated:
          let t = List [Id "define"; Id "next";
            List [Id "lambda"; List [Id "x"]; List [Id "+"; Id "x"; Num 1]]];;
          let (Val_Define newEnv) = eval env t;;
          eval newEnv (List [Id "next"; Num 3]);;
        Here we define t to be the abstract syntax tree corresponding to the Scheme code (define next (lambda (x) (+ x 1))). In evaluating t, we first evaluate the lambda to produce a closure. Then we bind next to the closure. The last evaluation line corresponds to the Scheme code (next 3). Thus we look up next in the top-level environment to return the closure. We evaluate the argument 3, which produces Val_Num 3. Then we apply the closure to the argument to produce 4.
    • Dynamic
      • User-defined functions supporting dynamic scoping are defined using the dynamic keyword. The syntax is otherwise identical to that of lambda.
      • The value of a dynamic function is also stored as a Val_Closure for convenience, but it does not need to be able to access non-local variables. Instead, values for any free variables in the body of the dynamic expression should be looked up in the environment where the function is called, rather than where it is declared. (Hint, this behavior is similar to that of a closure with an empty environment).
      • The following example shows the difference between functions defined with lambda and dynamic:
          (define x 1)
          (define foo (lambda (y) (+ x y)))
          (define bar (dynamic (y) (+ x y)))
          (foo 2)  <-- produces 3
          (bar 2)  <-- produces 3
          (define x 4)
          (foo 2)  <-- produces 3
          (bar 2)  <-- produces 6

    • There is one case where a lambda user-defined function foo may use dynamic scoping. A free variable x in foo that is also not in the top-level environment will not be saved in the closure for foo. The function foo may still be valid, if a define expression is used to bind a value to x before foo is called. In this case, the binding for x behaves as if foo used dynamic scoping.

      This behavior is different from how OCaml's top-level environment works, and is a result of the requirement that a Scheme expression look up variable bindings in both the local and top-level environment (and that define can add bindings to the top-level environment). For OCaml, a free variable in foo that is not in the top-level environment will cause an Unbound Variable error immediately.

      The public tests include an example of this situation. In the body of the user-defined lambda function bound to fact is a recursive call to fact. Similar code in OCaml would fail unless a "let rec" was used instead of "let". This code works in Scheme by relying on the assumption fact will have been added to the top-level environment using define by the time the function is actually called.

Testing and Submission

We will test your project by calling your parsing and evaluation functions directly, so be sure to give those functions the types we expect, as given above. You can work on the interpreter and parser in any order, we will test each part independently.

You may assume that all input test cases are syntactically correct. If the input Scheme code is not legal you may perform any action (e.g., exit, throw an exception).

All your code should be in two files, schemeTest.txt & You can submit your project in two ways:

  • Submit your schemeTest.txt & files directly to the submit server.
    • You can submit multiple files by putting the files in a .zip archive first. On Windows you can select the two files, then right click to select the "Send to->Compressed (zipped) Folder" option to create a .zip archive. Once your files are in a single zip archive, bring up the upload dialog box by clicking on the submit link in the column "web submission". Select your archive file using the "Browse" button, then press the "Submit project!" button.

    • The submit server now allows multiple files (from the same directory) to be selected. Bring up the upload dialog box by clicking on the submit link in the column "web submission". Browse to the directory containing your project files, then click on both schemeTest.txt and Now press the "Submit project!" button.

  • You may also submit directly by executing a Java program on a computer with Java and network access. Use the submit.jar file from the archive, To submit, go to the directory containing your project, then either execute submit.rb by typing:

    ruby submit.rb

    or use the java jar directly using the following command:

    java -jar submit.jar

    You will be asked to enter your class account and password, then all files in the directory (and its subdirectories) will be put in a jar file and submitted to the submit server. If your submission is successful you will see the message:

    Successful submission # received for project 4

Academic Integrity

The Campus Senate has adopted a policy asking students to include the following statement on each assignment in every course: "I pledge on my honor that I have not given or received any unauthorized assistance on this assignment." Consequently your program is requested to contain this pledge in a comment near the top.

Please carefully read the academic honesty section of the course syllabus. Any evidence of impermissible cooperation on projects, use of disallowed materials or resources, or unauthorized use of computer accounts, will be submitted to the Student Honor Council, which could result in an XF for the course, or suspension or expulsion from the University. Be sure you understand what you are and what you are not permitted to do in regards to academic integrity when it comes to project assignments. These policies apply to all students, and the Student Honor Council does not consider lack of knowledge of the policies to be a defense for violating them. Full information is found in the course syllabus---please review it at this time. [an error occurred while processing this directive]

Copyright Notice

This course project is copyright of Dr. Michael Hicks. ©Michael Hicks [2015]. All rights reserved. Any redistribution or reproduction of part or all of the contents in any form is prohibited without the express consent of the author.