Compiler HWK#5 Due date: 2:20pm, June 16, 2005

Write a simple compiler to generate C-- code
for a simple PASCAL-like expression language with block structures.

A sample program is as follows:

       PROGRAM program_name
       BEGIN
          VAR %% beginning of variable declarations, required even no var's
          x, y, z: INTEGER;
          ENDVAR
          READ(y);
          x := (y + 3 * 4) - 5;
            BEGIN
                VAR
                  w, x, z: INTEGER;
                ENDVAR
                x := 2;
                WRITE(x);
                WRITESP();
                WRITE(y);
                WRITELN();
            END
          WRITE(y);
          WRITESP();
          WRITE(x);
          WRITELN();
       END

The definition of your language is as follows:
1. This language is case-sensitive with resevered words.
2. assignment is ":=", the statement terminator is ";"
   each line has at most one statement
3. variable must be declared before its usage.
   variable/program names uses the one in HWK1.
   can have only the type of integer (4-byte signed)
   3.1 BEGIN and END starts and ends a block.
   3.2 VAR ... ENDVAR is a declaration of local variables which can be
       empty or completely missing.
   3.3 VAR ... ENDVAR must be before any statements and can have only
       one appearance.
4. can have only three different types of statements
   a. assignment of expressions, where expressions
      can have variables, integer-constant, (, ), +, -, *, / and "uminus".
   b. input statement
         READ(single variable)
   c. output statements
         WRITE(single variable) ---  output a variable
         WRITESP()              ---  output a single space
         WRITELN()              ---  write a new line
   There is no space before and in between "()".
5. Anything after %% are comments until to the end of the line.
   Watch out for white spaces as in HWK 1.
6. Must detect the following error and reports its line number:
    a. invalid ID names
    b. undeclared variables
    c. illegal syntax
    d. duplicatedly declared variables
7. If there is no error, genertae a C-- target code.
8. The length of a variable name is 1024. There can have lots
of variables. I expect you to use either hash table or
balanced-binary search tree for symbol tables.
9. Your keyboard input becomes the standard input of your program.

Note:     (1) To submit your homework, use the software submission
system designed by TA. Read TA's web site for details.
          (2) You need to have LAX and YACC programs.

This homework MUST execute on Linux or FreeBSD workstations in our department.
Submitting files include a lex file, a Makefile, and optional C codes. The
Makefile MUST supports "make clean" and "make all" for cleaning,
and building the homework. An example of the Makefile can be
obtained from the course web. Do not use the samples without proper
changes.
The name of your executable after "make all" must be "hw5".

To compile a program, type
         hw5 filename.p
The object code is "filename.c".
We will then execute your C-- code.