Skip to content
Tiny and fast parser for Octave or MATLAB®
Branch: master
Clone or download

Latest commit

Fetching latest commit…
Cannot retrieve the latest commit at this time.


Type Name Latest commit message Commit time
Failed to load latest commit information.

This is a fast, extensible, yet simple parser for GNU Octave and MATLAB®. It is a single file solution with about 200 lines of code.


GNU Octave version minimum 4.0 or MATLAB about version 2008 (verified for 2013b).


There are no dependencies.

Related Work

Simple Expression Example

p = ExprParser();
[ast, parseError] = p.parse('1+2*3');
y = ExprEvaluator(ast).exec()   % y = 7


You must subclass Parser to define your own grammar. The following subclasses are provided as examples


The class ExprParser extends Parser to support numbers, variables, strings, the binary operations +, -, *, / and the prefix - and + operations.

p = ExprParser();
[ast, parseError] = p.parse('x+2*3');
y = ExprEvaluator(ast).exec(scope)  % y = 7


The class 'FuncExprParser' extends ExprParser to also support function calls.

p = FuncExprParser();
[ast, parseError] = p.parse('power(sin(x),2) + power(cos(x),2)');
scope = struct('x', 1.2345);
y = ExprEvaluator(ast).exec(scope)   % y = 1


The class 'QtyExprParser' in the sister repository extends to quantities.

p = QtyExprParser();
[ast, parseError] = p.parse('2kg + 1kg');
qty = QtyEvaluator(ast).exec()      % qty = 3 kilogram

AST Explained

The parser emits an Abstract Syntax Tree (AST). It's not really a tree, but a Directed Acyclic Graph. On this AST you can, for example,

  • evaluate (as an example see test/evalExpr.m)
  • validate
  • compile

The emitted AST is

ast = p.parse('1.2+3.14')
  [1] = struct
      type = numerical
      value =  1.2
  [2] = struct
      type = numerical
      value =  3.14
  [3] = struct
      type = +
      head =  1
      tail =  2

returns a cell array with three nodes. A reference is expressed as an integer pointing to the referenced index. We will denote the reference to index I by $I.

Binary operations

prettyPrintAST(p.parse('1.2 + 3.14'))
Index Rule Head Tail
$1 numerical value: 1.2
$2 numerical value: 3.14
$3 + head: $1 tail: $2

Binary identifiers

prettyPrintAST(p.parse('3.14 + foo'))
Index Rule Head Tail
$1 numerical value: 3.14
$2 identifier name: foo
$3 + head: $1 tail: $2

Function Calls

prettyPrintAST(p.parse('power(2, 3)'))
Index Rule Head Tail
$1 identifier name: power
$2 numerical value: 2
$3 numerical value: 3
$4 funccall head: $1 tail: $2, $3

Prefix Operations

Index Rule Head Tail
$1 numerical value: 3
$2 uminus value: 1

# Extension Patterns


# Implementation Design

The parser is of the top-down operator precedence variety based on ideas by Vaughan Pratt, see or

TODO: Explain Advantage, disadvantage...

Code must run on both Octave and MATLAB. The former does not support handles to nested functions at least up to version 4.0.
As a workaround we use anonymous functions instead.
The scanner/tokenizer instantiates tokens, which in turn implement parser logic.

# Limitations Octave (as of 4.0)

* Cannot create handle to nested function
* No closures (read/write access to scoped variables)
* Nested functions to methods do not share method scope, not even read only.
You can’t perform that action at this time.