ouked / OakTree

Personal Project: An interpreter written in C.

OakTree 🌳

A postfix notation expression evaluator, using a stack to build an abstract syntax tree.

This personal project has taught me a lot about data structures and pointers in C, and has given me an initial insight to the inner workings of compilers.

Note: I am currently rewriting this project in C++ as a learning exercise: ouked/OakTreeCPP

Example input/output ⚙️

Input	Output	Comments
6	6	There are no operators, so the last number is returned.
1 1 2	2	The same as above: unused values are ignored.
1 2 +	3	+ operator adds the two top values in the stack.
1 2 3 + *	5	2 3 + gets evaluated to 5, which then get's multiplied by the 1 (1 5 *)
4 2 5 * + 1 3 2 * + /	2	See below

The final example is evaluated as follows:

Evaluation	Resulting expression
	4 2 5 * + 1 3 2 * + /
2 5 * → 10	4 10 + 1 3 2 * + /
4 10 + → 14	14 1 3 2 * + /
3 2 * → 6	14 1 6 + /
1 6 + → 7	14 7 /
14 7 / → 2	2

How does it work? 🔎

After an expression is entered (currently through a file), the program reads it from left to right, and behaves according to the character that it reads.

• If the character is a digit...

  • The digit is added to the buffer.

• If the character is a space...

  • The contents of the buffer is used to create a new number node.
  • The node is pushed to the stack.
  • The buffer is cleared.

• If the character is an operator...

  • The stack is checked to verify there are enough nodes (For example, an addition operator requires two nodes, but a negation operator requires only 1)
  • The stack is popped.
  • A new operator node is made, with the nodes from the stack as children.

• If the character is an EOF...

  • If the buffer is not empty, the values are pushed to the stack.

Now that we have built an abstract syntax tree, we can evaluate it.

This is done by starting at the root node and recursively querying the value of both the left and the right child, before performing the operation (if any).

Testing 🧪

To ensure new features/bug fixes don't break any existing functionality, I've written my own simple unit testing framework. If the tests are designed to be representative of all the functionality of the program, all of them passing means that nothing has been broken (probably).

Running tests...

Test 01 passed.
Test 02 passed.
Test 03 passed.
Test 04 passed.
Test 05 passed.
Test 06 passed.
Test 07 passed.
Test 08 passed.


Success! All 8 tests passed.

By writing these tests before new features were developed, this was an example of test driven development: I knew a feature was implemented properly (and hadn't broken existing features) when all tests passed.

I also learnt a lot from writing the test framework itself, including how testing frameworks I've used before work.

Next steps 📈

The next steps will be to continue to head towards developing a stack-based language, like Forth or PostScript. I also want to make OakTree turing complete.

Specific features I'm considering include:

• character support
• variables
• functions