mmnewmedia / schematron-parser

1.0.0 2026-10-13 08:45 UTC

Requires

• php: ^8.4|^8.5
• ext-dom: *
• ext-mbstring: *
• symfony/console: ^7.3 || ^8.0

Requires (Dev)

• phpunit/phpunit: ^13.0
• vimeo/psalm: ^6.5

BSD-3-Clause

• Marcel Maaß

• xml
• parser
• validation
• Xpath
• compiler
• schematron
• EN16931

Published on Packagist.org on 2026-10-13 08:59 UTC

README

XPath expression parser and compiler for Schematron rule files in PHP.

The library tokenizes and parses XPath 1.0 expressions into an AST, compiles them into executable PHP code, and can generate one PHP rule class per sch:assert / sch:report from a .sch file. It powers the pure-PHP EN 16931 / Factur-X / ZUGFeRD Schematron validation (no Java required at runtime).

Requirements

• PHP ^8.4
• ext-dom, ext-mbstring

Installation

composer require mmnewmedia/schematron-parser

Overview

MMNewmedia\Schematron\Ast\         XPath tokenizer and AST value objects
MMNewmedia\Schematron\Parser\      XPathAstParser (token stream -> AST)
MMNewmedia\Schematron\Compiler\    ExpressionCompiler, XPathAstCompiler, XPath function library
MMNewmedia\Schematron\             SchematronCompilerService (.sch -> PHP rule classes)

Parsing and compiling XPath expressions

use MMNewmedia\Schematron\Ast\XPathTokenizer;
use MMNewmedia\Schematron\Compiler\ExpressionCompiler;
use MMNewmedia\Schematron\Compiler\ExpressionNormalizer;
use MMNewmedia\Schematron\Compiler\FunctionLoader;
use MMNewmedia\Schematron\Compiler\XPathAstCompiler;
use MMNewmedia\Schematron\Parser\XPathAstParser;

$functionLoader = new FunctionLoader(
    dirname(__DIR__) . '/vendor/mmnewmedia/schematron-parser/src/Schematron/Compiler/Functions',
    'MMNewmedia\Schematron\Compiler\Functions',
);

$compiler = new ExpressionCompiler(
    new XPathAstParser(new XPathTokenizer()),
    new XPathAstCompiler($functionLoader->load()),
    new ExpressionNormalizer(),
);

// XPath -> PHP expression evaluating against a context node ($context by default)
$php = $compiler->compile('substring-before(ram:ID, "-")', '$node');

// XPath -> PHP expression evaluating to a node-set (DOMNodeList)
$nodeSet = $compiler->compileNodeSet('.//ram:SpecifiedLineTradeTransaction', '$node');

XPathAstCompiler also accepts the AST directly ($parser->parse(...) + $compiler->compile(...)), and individual XPath functions can be compiled with a plain new XPathAstCompiler([]) if no function library is needed.

The compiler ships implementations for the XPath 1.0 core functions plus the extensions used by EN 16931 schematron (upper-case, lower-case, matches, tokenize, distinct-values, starts-with, ends-with, exists, empty, xs:date, xs:decimal, …) under Compiler/Functions/. Custom functions implement XPathFunctionInterface (or NodeSetAwareFunctionInterface) and are picked up by FunctionLoader.

The discovered class list is cached per directory/namespace, so repeated load() calls do not rescan the filesystem. Call FunctionLoader::clearCache() if the functions directory changes at runtime.

Generating rule classes from a .sch file

use MMNewmedia\Schematron\SchematronCompilerService;

$service = new SchematronCompilerService($compiler, '/path/to/output/Rules/Cii');
$service->compile('/path/to/EN16931-CII-validation-preprocessed.sch', 'MyProject\Schematron\Rules\Cii');

Each sch:assert / sch:report becomes a final class with:

public static function validate(DOMXPath $xpath, DOMNode $context): bool;
public static function message(): string;
public static function context(DOMXPath $xpath, DOMNode $context): iterable; // only when needed

CLI

php bin/compile-schematron.php <source.sch> <outputDir> <namespace>

Compiles a single Schematron file: every sch:assert / sch:report is written as a PHP rule class into <outputDir>, using <namespace> as the generated class namespace.

Codedb lookups

Some Factur-X rules resolve codelist entries at runtime via CodedbCodeList::has(), which scans codedb XML files. Point it at your files before evaluating generated rules:

use MMNewmedia\Schematron\Compiler\CodedbCodeList;

CodedbCodeList::setBaseDirectories(['/path/to/zugferd/_generated', '/path/to/zugferd/_preprocessed']);

Without configuration it looks in <base>/data/zugferd/{_generated,_preprocessed,_XSLT}.

Tests

composer install
composer test

Static analysis

composer psalm

Psalm runs with errorLevel="5" (mirroring the original project configuration).

License

This project is licensed under the BSD 3-Clause License. See the LICENSE file for the full text.