For this assignment you will write Haskell interpreter for a tiny programming language, let's call it TL (Tiny Language). The language contains just three types of statements:
let variableName = expression if expression goto label print expression1, expression2, ...
input variableName
Each statement may contain a preceding label. A label is an alphanumeric string ending with a colon (":").
A number of simplifying assumptions have been made about the syntax of the language.
• White space (blanks) are important and must be used to separate each token including around the arithmetic operators.
• There can be only one statement per line.
• The expressions are limited to constant numbers, constant strings, variable names, and binary expressions involving just one of the following operators: "+", "-", "*", "/", "<", "", "<=", "=", "==", or "!=", with their conventional meanings. Note again that the operators must be surrounded by spaces, which makes for easier parsing.
• The only types are strings and floating point numbers (Float in Haskell) and strings are only used in print statements. The result of Boolean operations is 0 if false and 1 if true. Furthermore any numeric expression can be used in an ifstatement and as with the C language, 0 is false and everything else is true.
• Blank lines are ignored.
let variableName = expression computes the value of expression then binds that value to the name variableName.
if expression goto label computes the value of expression, if the value is 0 execution continues with the next statement. If the value is non-zero then execution continues with the statement labeled label. If no such statement exists, the program terminates with the message: "tli: Illegal goto label at line x." where x is the actual line number of the illegal goto statement.
print expression1, expression2, ... evaluates each expression, then prints their values, all on one line, separated by spaces, terminating the line with a newline character.
input variableName attempts to read a number from the standard input. If successful that value is bound to the name variableName. If the read fails the program may simply terminate with the default message: "tli: Prelude.read: no parse".
If at any point an attempt is made to evaluate an expression that references a variable variableName for which there is no binding, the program terminates with the message: "tli: Undefined variable variableName at line x." where x is the actual line number of the failed expression.
Your program tli (tiny language interpreter) will take one command line argument, the name of the source file. It will compile the program into an internal form and then execute the compiled program. If the input program contains any syntax errors, the program should exit printing a message of the form "tli: Syntax error on line x." for the first syntax error detected. Although your program is allowed to continue and report additional syntax errors or provide additional error information, that is not required and will not affect your score.
tli should make just one pass over the program source building an internal representation that is a list of statements and a symbol table that maps labels into line numbers. You may use this same symbol table to store variable bindings during execution of the program. Each statement should be represented by a value of an appropriately extended version of these data types
data Expr = Constant Float | Var String | Plus Expr Expr deriving (Show)
data Stmt = Let String Expr | Print Expr deriving (Show)
Here is a tiny language program that prints out a sequence of numbers.
input start input end let x = start repeat: print x let x = x + 1 if x < end goto repeat
print "done", x
Assuming the above program is stored in "prog1.txt", when executed the command "tli prog1.txt" and the user entering 1 and 5, tli should produce the output
1.0
2.0
3.0
4.0 done 5.0
You should turn in one file, tli.hs. To help you get started, I have provided the file nano.hs that parses and executes a nano-subset of the tiny language. You can find that file in Canvas. I will spend time in class explaining nano.hs.
Scoring will be done according to the following with the points in ()s.
(1) parseExpr() works for all legal expressions
(2) parseTest() works for all legal statement lists including labels but the SymTable produced is incorrect
(2) parseTest() works for all legal statement lists including labels with a correct and complete SymTable of label to linenum mappings