mexpr

2015-07-09

mexpr

Build Status Join the chat at https://gitter.im/tmccombs/mexpr

Macro for Common Lisp to allow infix syntax for mathematical expressions.

The mexpr package contains the infix macro which converts infix expressions into Lisp S-Expressions.

Installation

The easiest way to install mexpr is with quicklisp.

(ql:quickload :mexpr)

Examples

The following are examples of how the infix macro can be used:

CL-USER> (infix 3 + 4)
7
CL-USER> (infix [ 3 + 4 ] * 5)
35
CL-USER (infix 3 + 4 * 5)
23
CL-USER> (let ((x 0))
               (infix [ 4 + (1+ x) ] / (sqrt 4)))
2.5
CL-USER> (infix 4 + 4 < 7 or 2 > 6 / 16)
T
CL-USER> (infix 5 = 6)
NIL
CL-USER> (infix 5 = 5)
T
CL-USER> (infix 2 expt 5)
32

You can use defop to add new operators:

CL-USER> (defop union 10) ; use function name and precedence
CL-USER> (infix '(1 2) union '(2 3))
(2 1 3 4)
CL-USER> (defop ++ 100 append) ; use operator symbol, precedence, and defition symbol
CL-USER> (infix '(1 2) ++ '(3 4))
(1 2 3 4)
CL-USER> (defop p 110 (lambda (a b) (+ (* a a) (* b b)))) ; use lambda for definition
CL-USER> (infix 3 p 4)
25

You can use a reader macro to make it a little simpler:

CL-USER> (enable-infix-syntax) ; equivalent to (cl-syntax:use-syntax :mexpr)
CL-USER> #n(3 + 4 ** 2)
19

Notes:

  1. There always needs to be whitespace between operands and operators.
  2. Variables and other forms can be used as operands.
  3. Operators have an associated precedence. Higher precedence operators are performed first.
  4. Operators of equal precedence are evaluated left to right.
  5. [ and ] are used for grouping expressions (overriding precedence).

Usage

The mexpr (or more verbose bytecurry.mexpr) package contains two main macros.

The infix macro parses it's arguments as an infix expression and produces the corresponding s-expression. Each argument is evaluated as one of the following forms:

  • Grouping: The special forms [ and ] are used for grouping expressions. (Parentheses were already taken.)
  • Operator: An operator is a symbol that has been registered using the defop macro. It represents a binary operation.
  • Operand: An operand is any form which is not an operator. This means that normal prefix forms can be embedded in the infix expression.

The infix macro can detect some syntax errors, in which case it will create a syntax-error condition. The type of the syntax error can be obtained with syntax-error-type. Unfortunately, at the moment some invalid forms simply produce strange results, such as a transposition of a operator and operand.


The defop macro can be used to define new operators. It takes two parameters, the first is the unquoted symbol of the operator, the second is the desired precedence of the operator (see below for precedence table). The symbol should correspond to a function or macro which can accept exactly two arguments (although it may have more optional arguments).


The function infix-reader is macro dispatch function which is available to the user to use in any reader macro he/she desires. The package also registers the "#n" dispatch with cl-syntax, so you can enable syntax of the form #n(<expr>) with (cl-syntax:use-syntax :mexpr). Alternatively, enable-infix-syntax is a wrapper around cl-syntax:use-syntax.

Precedence

Unlike prefix and postfix notations, infix notation uses operator precedence to determine the order of evaluation. mexpr uses a numeric precedence system, where the precedence of an operator is a positive integer. A higher number corresponds to a higher precedence. The precedence of the default operators is given below:

Operator Precedence Translation of a <op> b
** 110 (expt a b)
expt 110 (expt a b)
* 100 (* a b)
/ 100 (/ a b)
% 100 (mod a b)
mod 100 (mod a b)
rem 100 (rem a b)
+ 90 (+ a b)
- 90 (- a b)
ash 80 (ash a b)
<< 80 (ash a b)
>> 80 (ash a (- b))
< 70 (< a b)
> 70 (> a b)
<= 70 (<= a b)
>= 70 (>= a b)
= 60 (= a b)
/= 60 (/= a b)
logand 50 (logand a b)
& 50 (logand a b)
logxor 40 (logxor a b)
^ 40 (logxor a b)
logior 30 (logior a b)
\| 30 (logior a b)
and 20 (and a b)
or 10 (or a b)
Author
Thayne McCombs <bytecurry.software@gmail.com>
License
LLPGL, LLGPL