Skip to content

Latest commit

 

History

History
195 lines (146 loc) · 5.82 KB

README.md

File metadata and controls

195 lines (146 loc) · 5.82 KB
Raw

Small LISP interpreter in Python

THIS IS NOT SCHEME

This is a simple and small implementation of a LISP like interpreter for lambda calculus pratice. It has no data structures, everything has to be build with lambdas.

Tutorial

The REPL can be started by executing the lis.pyfile like python lis.py, it accepts multiple line input and readline integration with history etc. You can also type the contents in a file and execute it like this python lis.py < file.lispy. The extension doesn't matter.

About the language, the basic construct is a lambda :

(lambda (x) (+ 1 x))

Since this is a strict language we have if too, so not function is defined like this :

(lambda (x) (if x false true))

As you can see we have booleans, the basic values are booleans, keywords (like :foo) and integers like 1234.

Keywords behave like constants, you can use then when you need to test something for equality. And we have nil to represent absence of values.

(= :foo :foo) ;; prints True in REPL

You can define global symbols with define

(define not (lambda (x) (if x false true)))

Also we have print, print returns nil.

(print :hello-world)

And the arithmetic, boolean and bitwise operators, in that order : +, *, /, %, =, !=, >, <, >= <=, &, |:

The even? function can be defined like this

(define even? (lambda (x) (= (% x 2) 0)))

We have prog that make the role of a block, it evaluates all its arguments and return the result of the last one.

A trace function, that print and return it's input can be defined like this

(define trace (lambda (x)
	(prog
		(print x)
		x)))

We have fix to call anonymous functions recursively, so we can call anonymous factorial like this :

(fix (lambda (x fact) (if (= x 0) 1 (* x (fact (- x 1) fact)))) 5)

The above example will print 120 on REPL. fix will receive a function and call it passing itself as the last argument. Note that normal recursion also works :

(define fact (lambda (x)
	(if (= x 0)
		1
		(* x (fact (- x 1))))))
(fact 5)

Comments are done with ; and span up to the end of the line.

; this is a comment

And we have assert which is basically calls assert in Python

(assert (= 1 1)) ;; print nothing, return nil
(assert false) ;; abort with AssertionError

Thats it, congratulations you learned lis.py! Check out test.lispy for more examples.

We have macros, here how to define a let macro

(define-macro let (x e1 e2) ((lambda (x) e2) e1))

(let x 1 (+ 1 x)) ;; prints 2

Constructs

The basic constructs of the language

  • (lambda (x) x) : lambda keyword creates a lambda
  • (if true 1 0) : if for if expressions
  • (fix (lambda (x k) (if (= x 0) x (k (- x 0)))) 2): fix for recursion, fix will take a lambda and call it passing it as the last argument. The kstand for continuation, is pretty common for use k for continuation in the literature.
  • (define inc (lambda (x) (+ x 1))): define defines a symbol
  • (define-macro let (x e1 e2) ((lambda (x) e2) e1)) : define-macro defines a macro. Macros works by substitution, for example, assuming that the let macro is defined as (define-macro let (x e1 e2) ((lambda (x) e2) e1)) the evaluation of (let y 1 (+ y y)) follows by substitutions x -> y e1 -> 1, e2 -> (+ y y), in ((lambda (x) (e2)) e1), after the subtitutions we have (lambda (y) (+ y y) 1) which is then evaluated.
  • env returns the current environment, this is used for debugging as the object returned is a dict and lispy has no means to work with Python dicts
  • ; comment to the end of line : use ; for comments

Constants and values

  • nilis None
  • trueis True and false is False
  • Integers are integers, no floats, sorry
  • Symbol starting with : are called keywords and they evaluate to themselves. You can think of it as constants
  • We have no strings

Global functions

  • print, prints its argument, returns nil
  • + - * / % are the arithemetic operations, pay attention that / is // in python or flordiv. Also, these functions that two arguments, calling with three or more will give you an error
  • = > < >= <= != are the boolean copmarisons
  • | & bitwise operators
  • (assert x) : calls assert x in Python.
  • (prog *args) : Evaluate all the arguments and return the last one. Note that ((foo x) (bar y)) means Execute (foo x), then apply it's result to (bar y). If you need Execute (foo x), ignore it's result and then execute (bar y) then you need to write (prog (foo x) (bar y)).

REPL

If you execute python3 lis.pyit will start the REPl, you can type the code and press enter to submit, if you have unbalanced parenthesis it will exibihit the prompt "...>". It uses readline library for better input experience, you should get history for free too, it creates the ~/.lispy_history file, it's safe to delete this file if you want.

Running files

If you redirect the input to lis.py it will interpret the whole thing as a single string and execute the code. It will not print intermediary resulst so you may want to use print to see the results.

Motivation

  1. I want something easy to tweak to pratice lambda calculus, so this is why we don't have anything other data type beside lambdas, you have to use lambdas to construct otherstuff :), this is functional programming in it's essence.
  2. I want to try something where every logic is semantic, in regarding parsing there is only parenthesis and words.

Common errors:

  • Dunno what to do with (k ... when executing a recursive function, you probally forgot to call it using fix.

Examples

You can check the test.lispy file for examples. You can the run example by python3 lis.py < test.lispy