Download the PHP package sanmai/hoa-compiler without Composer

Install with:

This version of the library should be input-output-wise backward-compatible with the original package. For example, you can use this version together with jms/serializer to avoid known problems some of the dependencies of the original package have with PHP 7.4.

BC breaking changes include:

• Hoa\Exception\Exception sub-classes are no longer thrown, please switch to Hoa\Compiler\Exception
• The package itself does not depend on development time only dependencies. If you need to call getAST(), you need to manually install hoa/regex and hoa/file.

Hoa\Compiler

This library allows to manipulate LL(1) and LL(k) compiler compilers. A dedicated grammar description language is provided for the last one: the PP language.

Learn more.

Testing

To run all the test suites:

For more information, please read the contributor guide.

Quick usage

As a quick overview, we will look at the PP language and the LL(k) compiler compiler.

The PP language

A grammar is constituted by tokens (the units of a word) and rules (please, see the documentation for an introduction to the language theory). The PP language declares tokens with the following construction:

The default namespace is default. The value of a token is represented by a PCRE. We can skip tokens with the %skip construction.

As an example, we will take the simplified grammar of the JSON language. The complete grammar is in the hoa://Library/Json/Grammar.pp file. Thus:

We can see the PP constructions:

• rule() to call a rule;
• <token> and ::token:: to declare a token;
• | for a disjunction;
• (…) to group multiple declarations;
• e? to say that e is optional;
• e+ to say that e can appear at least 1 time;
• e* to say that e can appear 0 or many times;
• e{x,y} to say that e can appear between x and y times;
• #node to create a node the AST (resulting tree);
• token[i] to unify tokens value between them.

Unification is very useful. For example, if we have a token that expresses a quote (simple or double), we could have:

So, the data "foo" and 'foo' will be valid, but also "foo' and 'foo"! To avoid this, we can add a new constraint on token value by unifying them, thus:

All quote[0] for the rule instance must have the same value. Another example is the unification of XML tags name.

LL(k) compiler compiler

The Hoa\Compiler\Llk\Llk class provide helpers to manipulate (load or save) a compiler. The following code will use the previous grammar to create a compiler, and we will parse a JSON string. If the parsing succeed, it will produce an AST (stands for Abstract Syntax Tree) we can visit, for example to dump the AST:

Pretty simple.

Compiler in CLI

This library proposes a script to parse and apply a visitor on a data with a specific grammar. Very useful. Moreover, we can use pipe (because Hoa\File\Read —please, see the Hoa\File library— supports 0 as stdin), thus:

You can apply any visitor classes.

Errors

Errors are well-presented:

Samplers

Some algorithms are available to generate data based on a grammar. We will give only one example with the coverage-based generation algorithm that will activate all branches and tokens in the grammar:

Research papers

• Grammar-Based Testing using Realistic Domains in PHP, presented at A-MOST 2012 (Montréal, Canada) (article, presentation, details).

License

Hoa is under the New BSD License (BSD-3-Clause). Please, see LICENSE for details.