Download the PHP package flaviovs/phpast without Composer

PHPAST - PHP Abstract Syntax Tree Structure and Interpreter

The PHPAST package provides an Abstract Syntax Tree (AST) structure, plus an interpreter to run programs represented by those structures. PHPAST provides not only all the operators commonly found in programming languages, such as addition, subtraction, logical operators ("and", "or", "not", "xor), etc., but also flow control structures ("if"), loops ("while", "for"). You can use variables, define functions -- both statically, or by using AST calls, which allow dynamic routines to be created --, and call pre-defined PHP built-ins or callbacks. This makes PHPAST extremely versatile and powerful, and allows you to power your custom programming language and do virtually anything a full-fledged language can do.

Note: this package contains only the AST structure and interpreter, to be used as the intermediate representation of a programming language. To properly implement a language you will need a parser to convert lexical constructions into the ASTs.

IMPORTANT: PHPAST should be considered work in progress. The API is not frozen yet, and fundamental pieces may be made to work differently at any time. Please, create a new issue if you need to discuss something you want/like/don't like about this package.

Installation

The easiest way to install and run PHPAST is through Composer:

$ composer require flaviovs/phpast

Running without composer

To run PHP without composer you must load the core library files, and implement a PSR-4 style auto-loader, pointing the PHPAST namespace to the src/ directory:

Basic Usage

The core concept of PHPAST (and ASTs in general) is the evaluation node, which comprises of objects that, once evaluated against a symbol table, produces another node. ASTs are organized as trees (duh!), so nodes can have children nodes which are also evaluated according to the operation to be carried by the parent node. Ultimately, the root node (your main program) returns the node that represents the complete evaluation of the tree -- which is the output of your "program".

Let's look at a basic example:

At this point, $output contains the results of evaluating our program, contained in the $program variable. Since our "program" is comprised of just a node representing the literal number 1, this very same node is returned. In other words, after running the code above, $output is the same as $program.

Notice that to evaluate a node you must provide a symbol table object. Symbol tables are like PHP arrays, and are used by PHPAST to hold program variables. We will cover symbol tables later, so for now just accept that they are required, and work like PHP arrays. In the example above we used $table to hold our symbol table.

Now let's look at a more complex example:

Whoa there! Things are becoming a bit more complicated. The program above is composed of a MulOp node, which receives two Integer nodes -- representing 2 and 4. MulOp is the node operation (hence the Op) corresponding to the multiplication operation. What it does is to multiply its two nodes operands, so when you run the program above, $output will be equivalent to an PHPAST\Integer(8) object (8 being the result of 2 × 4).

Now an even more complex program:

Did I say that PHPASTs programs are hierarchical structures? In the example above, the nodes passed to MulOp are nodes of type AddOp and SubOp. As you may already figured out, these two represents arithmetic addition and subtraction. In this example, $output will be equivalent to an PHPAST\Integer(-12) object ((2 + 4) * (2 - 4)).

This is the beauty of ASTs: by combining small, self-contained operation-nodes, you can create very complex programs.

Symbol Tables

As described above, a symbol table is required to evaluate a node. Symbol tables are like PHP arrays which you can use to assign a value to an indexed element. The different is that on a symbol table you assign nodes to a symbol (which is just a fancy name for "indexed element").

In fact, symbol tables implements the standard PHP array access syntax:

The syntax above is just to illustrate the expected behaviour of a symbol table. Usually you will not be writing code to manipulate symbol tables -- they will be handled by PHPAST, and the symbol/values managed by your PHPAST program.

FlatSymbolTable is the simplest symbol table, which works exactly like an PHP array. PHPAST also provides a ChainedScopeSymbolTable class, which can be used to implement local/global variables. You are free to choose any symbol table best suits your language, or even define your own.

Variables

Variables are entries in a symbol table, indexed by an identifier, which ultimately is used as the index to the node on the symbol table. PHPAST provides several nodes to handle basic variable operations. Let's look at some examples:

String Representation

PHPAST nodes have human-friendly (err... programmer-friendly) string representation of its node operations. This can be helpful to understand and/or debug PHPAST programs. To print the string representation, just use the node object on a string context. For example:

This should print

Outln (3.14159 ** 2)

Note: the string representation output by PHPAST does not serve as -- and was never meant to be -- a full-featured programming language. It is just a naive representation of the inner operation to be carried by a node.

Complex Operations

PHPAST also provides nodes that perform complex operations, such as program flow control and functions. Below are the description of some of these structures. Please refer to the classes in the src/ directory for a complete list of PHPAST operators.

Prog

This primitive node is used to evaluate several child nodes in sequence. The result of evaluation of the last node is returned. The node implements ArrayAccess, so a Prog can be manipulated using ordinary PHP array syntax. Progs are the building blocks to evaluate sequences of nodes in function bodies, control structures, and programs.

Example:

This should print:

foo := 0
bar := 1
foo += 10
For (Nop; (foo > bar); foo -= 2)
        Outln "foo = ", foo

Func & ReturnOp

Func nodes represent functions. Each function has an ArgList object that defines the arguments the function expects.

Func is a subclass of Prog, so its "array" elements define the nodes that make up the body of the function.

Like Prog, Func return the evaluation of last node. You can use a ReturnOp to end function evaluation and return an specific node.

Def & CallOp

With Def you can define a function, which basically means assigning a function body to a symbol in the symbol table. To "call" the function, use CallOp with a Ref to the symbol that defines the function, and a SymbolTable object that defines the function arguments.

Examples

The examples/ directory contains some examples of how to use PHPAST.

Support

Home page: https://github.com/flaviovs/phpast