This page does not represent the most current semester of this course; it is present merely as an archive.

1 Task

In this assignment you will implement a reverse polish notation calculator, also known as a postfix notation calculator. That article also gives pseudocode for two algorithms; the left-to-right is probably a better match for line-by-line input, though you are welcome to read the full input and then run right-to-left algorithm (or any other correct algorithm you might design) if you’d prefer.

Your program (written in a file named rpn.c) should read stdin until either (a) it encounters an unknown token or (b) there is nothing left. It should split the input into tokens on whitespace (as defined by isspace in <ctype.h>), recognizing the four operators +, -, *, and / as well as integer literals.

Your program should halt when it

  • encounters an unrecognized literal
  • reaches the end of the input
  • is given an operator with insufficient operands to evaluate it

Before exiting for any of the above reasons, your program must print the remaining values on the stack. These must be presented on a single line, separated by commas within square brackets, with the top of the stack on the right. Optionally, your program may also print the contents of the stack every time it changes.

2 Examples

The following is one possible run of the program, with the optional print-stack feature included
2 3
[ 2 ]
[ 2, 3 ]
4     - 5
[ 2, 3, 4 ]
[ 2, -1 ]
[ 2, -1, 5 ]
+ * /
[ 2, 4 ]
[ 8 ]

Note that the program stopped when it encountered / on a stack with just one argument.


The following is one possible run of the program, with the optional print-stack feature not included
2
  3 -4 + not_a_number 5 4
[ 2, -1 ]

Note that the program stopped when it encountered not_a_number and did not continue running the 5 and 4.


You should also verify that if you end input early (by redirecting input, or by pressing Ctrl+D when running interactively) the program prints the final stack and exits:

Without intermediate stacks With intermediate stacks
echo -n 2 3 4 + 5 | ./a.out
[ 2, 7, 5 ]
echo -n 2 3 4 + 5 | ./a.out
[ 2 ]
[ 2, 3 ]
[ 2, 3, 4 ]
[ 2, 7 ]
[ 2, 7, 5 ]

Make sure your code works with all numbers as input, including the 0 that most string-to-number converters give if they detect an error.
2 1 0 -1 -2 - - - -
[ 2 ]
[ 2, 1 ]
[ 2, 1, 0 ]
[ 2, 1, 0, -1 ]
[ 2, 1, 0, -1, -2 ]
[ 2, 1, 0, 1 ]
[ 2, 1, -1 ]
[ 2, 2 ]
[ 0 ]

3 Tips

When reading text, check the manual page for your read function (read, fgets, etc) to see how it reports an end-of-file EOF

You are welcome to make either a linked-list or array-based stack. We will not supply any files to support this, though, so include your implementation in your rpn.c.

The following will print an array-based stack:

The following will print a singly-linked-list stack (doubly-linked printing code was supplied with the previous PA):

If you use strsep to split input into tokens, you’ll need to handle it retuning empty strings if two whitespaces appear in a row. Alternatively, making your own tokenizer is not difficult. Either way, you almost certainly want to test your tokenizer on its own before you try to merge it with your postfix evaluator.

Never compare strings with ==, as that compares addresses not contents. Use strcmp instead.

Both atoi and strtol will work to convert strings into numbers, but strtol can also detect non-numbers because if it finds a non-number then it will leave *endptr == nptr. Note that although strtol says it “may” set errno to EINVAL if given a non-number, in practice most implementations do not do this so do not rely on it in your code.

3.1 Things we do not test

You do not need to correctly handle any of the following

  • half-number tokens like 5more.
  • fractions like 2.3.
  • tokens without whitespace between them 2+3.
  • bigger-than-int values like 123456789012345678901235901234567890
  • no-input inputs like echo '' | ./a.out

3.2 Things we do test

  • very very long lines
  • many short lines
  • lines where a multi-digit number will be split across a buffer-capacity boundary