CS 536 Program 3: The Parser

Due date: Friday, November 5 (by midnight)
Not accepted after midnight on Monday, November 8

Overview | Requirements | Announcements | Handin

Overview

For this assignment you will use the parser-generator Java Cup to write a parser for the C-- language. The parser will find syntax errors, and for syntactically correct programs, it will build an abstract-syntax tree (AST) representation of the program. You will also write methods to unparse the AST built by your parser, and an input file to test your parser (a main program that calls the parser and then the unparser has been provided). You will be graded on the correctness of your parser and your unparse methods, and on how thoroughly your input file tests the parser.

Getting Started

Skeleton files on which you should build are in: ~cs536-1/public/prog3

The files are:

c.jlex: A JLex specification for the C-- language (a solution to program 2). Use this if there were problems with your JLex specification.
c.cup: A Java Cup specification for a very small subset of the C-- language (you will need to add to this file). Note: you will need to change the terminal declarations for ID and STRINGLITERAL if you use your own scanner specification, and you used names other than IdTokenVal and StrLitTokenVal for the corresponding subclasses of TokenVal.
c.grammar: A CFG for the C-- language. Use this to guide the enhancements you make to c.cup.
ast.java: Contains class definitions for the AST structure that the parser will build (you will need to add to this file).
P3.java: The main program that calls the parser, then, for a successful parse, calls the unparser (no changes needed).
Makefile: A Makefile for program 3 (no changes needed).
test.C: Input for the current version of the parser (you will need to change this file).
Errors.java: Same as for program 2 (no changes needed).
IO.java: Same as for program 1; used by P3.java (no changes needed).

Here is a link to the Java Cup reference manual.

Operator Precedences and Associativities

The C-- grammar in the file c.grammar is ambiguous; it does not uniquely define the precedences and associativities of the arithmetic, relational, equality, and logical operators. You will need to add appropriate precedence and associativity declarations to your Java Cup specification.

All of the binary operators are left associative, except the relational and equality operators (<, >, <=, >=, ==, and !=), which are non-associative (i.e., expressions like a < b < c are not allowed and should cause a syntax error).
The unary minus and not (!) operators have the highest precedence, then multiplication and division, then addition and subtraction, then the relational and equality operators, and finally the logical operators (&& and ||).

Note that the same token (MINUS) is used for both the unary and binary minus operator, and that they have different precedences; however, the C-- grammar has been written so that the unary minus operator has the correct (highest) precedence; therefore, you can declare MINUS to have the precedence appropriate for the binary minus operator.

Java Cup will print a message telling you how many conflicts it found in your grammar. If the number is not zero, it means that your grammar is still ambiguous, and the parser is unlikely to work correctly. Do not ignore this! Go back and fix your specification so that your grammar is not ambiguous.

Building an Abstract-Syntax Tree

To make your parser build an abstract-syntax tree, you must add new productions, declarations, and actions to c.cup. You will need to decide, for each nonterminal that you add, what type its associated value should have. Then you must add the appropriate nonterminal declaration to the specification. For most nonterminals, the value will either be some kind of tree node (a subclass of ASTnode), or a LinkedList of some kind of node (use the information in ast.java to guide your decision). You must also add actions to each new grammar production to assign an appropriate value to RESULT. Note that the parser will return a Symbol whose value field contains the value assigned to RESULT in the production for the root nonterminal (nonterminal program).

Unparsing

To test your parser, you must write the unparse methods for the subclasses of ASTnode. When the unparse method of the root node of the program's abstract-syntax tree is called, it should print a nicely formatted version of the program (this is called unparsing the abstract-syntax tree). The output produced by calling unparse should be the same as the input to the parser except that:

There will be no comments in the output.
The output will be "pretty printed" (newlines and indentation will be used to make the program readable); and
Expressions will be fully parenthesized to reflect the order of evaluation.

For example, if the input program includes:

if (b == -1) { x = 4+3*5-y; while (c) { y = y*2+x; } } else { x = 0; }

the output of unparse should be something like the following:

if ((b == (-1))) {
    x = ((4 + (3 * 5)) - y);
    while (c) {
        y = ((y * 2) + x);
    }
}
else {
    x = 0;
}

To make grading easier, put open curly braces on the same line as the preceding code, and put closing curly braces on a line with no other code (as in the example above). Put the first statement in the body of an if or while on the line following the open curly brace. Whitespace within a line is up to you (as long as it looks reasonable).

Note: Trying to unparse a tree will help you determine whether you have built the tree correctly in the first place. Besides looking at the output of your unparser, you should try using it as the input to your parser; if it doesn't parse, you've made a mistake either in how you built your abstract-syntax tree, or in how you've written your unparser.

Another good way to test your code is to try compiling the output of your unparser using the C++ compiler (g++). If your input program uses I/O (cin or cout), you will first need to add: #include <iostream.h> at the beginning of the file.

It is a good idea to work incrementally:

Add a few grammar productions to c.cup.
Write the corresponding unparse operations.
Write a test program that uses the new language constructs.
Create a parser and run it on your test program.

Modifying `ast.java`

We will test your program by using our unparse methods on your abstract-syntax trees, and by using your unparse methods on our abstract-syntax trees. To make this work, you will need to:

Modify ast.java only by filling in the bodies of the unparse methods (and you must fill in all of the method bodies).
Make sure that no LinkedList field is null (i.e., when you call the constructor of a class with a LinkedList argument, that argument should never be null). Note that it is OK to make the ExpNode field of a ReturnStmtNode null (when no value is returned).
Follow the convention that the mySize field of a VarDeclNode has the value VarDeclNode.NOT_ARRAY if the type of the declared variable is a non-array type (i.e., if the declaration did not include square brackets).

Testing

Part of your task will be to write an input file that thoroughly tests your parser and your unparser. Note that since you are to provide only one input file, it should contain no syntax errors (you should also test your parser on some bad inputs, but don't hand those in).

Name your test file test.C. Use comments in the file to explain what aspects of the parser are being tested.

Working in Pairs

This assignment involves three main tasks:

Writing the parser specification (c.cup).
Writing the unparse methods for the AST nodes (in ast.java).
Writing an input file to test your implementation.

To ensure that both partners understand all aspects of the assignment, it is a good idea to share responsibility for all tasks.

I suggest that you proceed as follows, testing your parser after each change:

Working together, start by making a very small change to c.cup. For example, add the rules and actions for:
```
type ::= BOOL
type ::= VOID
```
Also update the appropriate unparse method in ast.java. Make sure that you can create and run the parser after making this small change.
Next, add the rules needed to allow array declarations.
Next, add the rules needed to allow programs to include methods with no arguments and with empty statement lists only, and update the corresponding unparse methods.
Still working together, add the rules (and unparse methods) for the simplest kind of expressions -- just plain identifiers.
Now divide up the statement nonterminals into two parts, one part for each person.
Each person should extend their own copy of c.cup by adding rules for their half of the statements, and should extend their own copy of ast.java to define the unparse methods needed for those statements.
Write test inputs for your statements.
After each person makes sure that their parser and unparser work on their own statements, combine the two by cutting and pasting one person's grammar rules into the other person's c.cup (and similarly for ast.java).
Now divide up the expression nonterminals into two parts, and implement those using a similar approach. Note that you will also need to give the operators the right precedences and associativities during this step (see above).
Divide up any remaining productions that need to be added, and add them.
Talk about what needs to be tested, and decide together what your final version of test.C should include.
When working on your own, do not try to implement all of your nonterminals at once. Instead, add one new rule at a time to the Java Cup specification, make the corresponding changes to the unparse methods in ast.java, and test your work by writing a C-- program that includes the new construct you added, and make sure that it is parsed and unparsed correctly.

If you worked alone on the previous program and are now working with a partner, see programming assignment 2 for more suggestions on how to work in pairs.

Announcements

Includes: Additions, Revisions, and FAQs (Frequently Asked Questions).
Please check here frequently.

10/15/2004 Program released.

Handin

What to turn in

See the assignments page for information about how to submit your code. The late policy is also found on the assignments page.

Electronically submit all of the files that are needed to create and run P3.class as well as your Makefile and your test.C. Do not copy any ".class" files, and do not create any subdirectories in your handin directory.

If you are working with a partner only one of you should hand in files. Include a comment at the top of P3.java with the names of both partners.

General information on program grading criteria can be found on the Grading Criteria for Programs page.