sirius-press/tests/test-export.php

<?php
/**
 * URL-to-path mapping and link rewriting.
 *
 * A wrong answer here is silent. Two pages that map to one path means one
 * quietly overwrites the other, and nobody finds out until a reader opens an
 * article and gets a different one. A link rewritten wrongly means the static
 * copy points back at the origin server it exists to make optional.
 *
 * So the cases below are mostly the awkward ones: a URL that looks like a file
 * but is a page, a path the gateway would refuse, a document deep in a tree
 * linking back up to an asset at the root.
 *
 * @package SiriusPress
 */

require_once __DIR__ . '/bootstrap.php';

T::group( 'gateway path rules' );

T::is( SP_Gateway::clean_path( '/blog/hello/index.html' ), 'blog/hello/index.html', 'a leading slash is dropped' );
T::is( SP_Gateway::clean_path( 'blog//hello///index.html' ), 'blog/hello/index.html', 'repeated slashes collapse' );
T::is( SP_Gateway::clean_path( 'a/../../etc/passwd' ), '', 'traversal is refused' );
T::is( SP_Gateway::clean_path( 'blog/' ), '', 'a trailing slash is refused — the gateway stores objects, not directories' );
T::is( SP_Gateway::clean_path( '' ), '', 'an empty path is refused' );
T::is( SP_Gateway::clean_path( 'caf&eacute;/index.html' ), '', 'characters the gateway rejects are refused here' );
T::is( SP_Gateway::clean_path( str_repeat( 'a', 201 ) ), '', 'an over-long path is refused' );
T::is( SP_Gateway::clean_path( 'wp-content/themes/x/style.css' ), 'wp-content/themes/x/style.css', 'an asset path survives' );

T::is( SP_Gateway::mime_for( 'index.html' ), 'text/html; charset=utf-8', 'html content type' );
T::is( SP_Gateway::mime_for( 'app.CSS' ), 'text/css; charset=utf-8', 'extensions are matched case-insensitively' );
T::is( SP_Gateway::mime_for( 'thing.unknown' ), 'application/octet-stream', 'an unknown extension falls back' );

T::group( 'URL to bucket path' );

$cases = array(
	'/'                              => 'index.html',
	''                               => 'index.html',
	'/hello-world/'                  => 'hello-world/index.html',
	'/hello-world'                   => 'hello-world/index.html',
	'/blog/2026/09/a-post/'          => 'blog/2026/09/a-post/index.html',
	'/robots.txt'                    => 'robots.txt',
	'/wp-content/themes/x/style.css' => 'wp-content/themes/x/style.css',
	'/sitemap.xml'                   => 'sitemap.xml',
	// Ends in a slash, so it is a page whose name happens to contain a dot —
	// not a file called "about.us".
	'/about.us/'                     => 'about.us/index.html',
);
foreach ( $cases as $url => $expected ) {
	T::is( SPE_Mapper::path_for_url_path( $url ), $expected, "'{$url}' maps to '{$expected}'" );
}

T::ok( SPE_Mapper::is_asset( 'wp-content/x.css' ), 'a stylesheet is an asset' );
T::ok( SPE_Mapper::is_asset( 'a/b/photo.JPG' ), 'an uppercase extension still reads as an asset' );
T::ok( ! SPE_Mapper::is_asset( 'blog/post/index.html' ), 'a page is not an asset' );

// Distinct pages must never land on one path.
$paths = array();
foreach ( array( '/a/', '/b/', '/a/b/', '/a/b/c/', '/a.html', '/ab/' ) as $url ) {
	$paths[] = SPE_Mapper::path_for_url_path( $url );
}
T::is( count( array_unique( $paths ) ), count( $paths ), 'six distinct URLs give six distinct paths' );

// `/a/` and `/a/index.html` are the same page in WordPress and are SUPPOSED to
// share a path — this records that as intended, not as the collision above.
T::is(
	SPE_Mapper::path_for_url_path( '/a/' ),
	SPE_Mapper::path_for_url_path( '/a/index.html' ),
	'a directory URL and its explicit index are one page'
);

T::group( 'document-relative links' );

$relative = array(
	// from                        to                          expected
	array( 'index.html',           'about/index.html',         'about/index.html' ),
	array( 'about/index.html',     'index.html',               '../index.html' ),
	array( 'blog/post/index.html', 'wp-content/app.css',       '../../wp-content/app.css' ),
	array( 'blog/post/index.html', 'blog/other/index.html',    '../other/index.html' ),
	array( 'a/b/c/index.html',     'a/b/c/style.css',          'style.css' ),
	array( 'index.html',           'index.html',               'index.html' ),
);
foreach ( $relative as $case ) {
	list( $from, $to, $expected ) = $case;
	T::is( SPE_Renderer::relative( $from, $to ), $expected, "{$from} -> {$to}" );
}

T::group( 'rewriting a page' );

$html = '<!doctype html><html><head>'
	. '<link rel="stylesheet" href="https://example.test/wp-content/themes/x/style.css?ver=1.2">'
	. '<link rel="canonical" href="https://example.test/blog/hello/">'
	. '</head><body>'
	. '<a href="https://example.test/">Home</a>'
	. '<a href="https://example.test/about/">About</a>'
	. '<a href="https://elsewhere.example/offsite">Offsite</a>'
	. '<img src="//example.test/wp-content/uploads/pic.png" alt="">'
	. '<a href="https://example.test/blog/hello/#notes">Notes</a>'
	. '</body></html>';

$result = SPE_Renderer::rewrite( $html, 'blog/hello/index.html' );
$out    = $result['html'];

T::ok( false === strpos( $out, 'https://example.test' ), 'no absolute self-links remain' );
T::ok( false !== strpos( $out, 'href="../../index.html"' ), 'the home link became document-relative' );
T::ok( false !== strpos( $out, 'href="../../about/index.html"' ), 'an internal page link became relative' );
T::ok( false !== strpos( $out, 'href="https://elsewhere.example/offsite"' ), 'an external link is left alone' );
T::ok(
	false !== strpos( $out, 'href="../../wp-content/themes/x/style.css"' ),
	'the ?ver= query is dropped from an asset, which the bucket stores once'
);
T::ok( false !== strpos( $out, 'src="../../wp-content/uploads/pic.png"' ), 'a protocol-relative URL is rewritten too' );
T::ok( false !== strpos( $out, 'href="index.html#notes"' ), 'a fragment survives rewriting' );

T::ok( isset( $result['assets']['wp-content/themes/x/style.css'] ), 'the stylesheet was collected for export' );
T::ok( isset( $result['assets']['wp-content/uploads/pic.png'] ), 'the image was collected for export' );
T::is(
	$result['assets']['wp-content/themes/x/style.css'],
	'https://example.test/wp-content/themes/x/style.css',
	'a collected asset carries the URL to fetch it from'
);
T::ok( ! isset( $result['assets']['about/index.html'] ), 'a page is not collected as an asset' );

// The exporter's own marker must never survive into a published page: every
// internal link would carry it, and the published copy would then be asking
// for the export-rendered variant of every page.
$flagged  = '<a href="https://example.test/about/?sirius_export=abc123">About</a>';
$rewrote  = SPE_Renderer::rewrite( $flagged, 'index.html' )['html'];
T::ok( false === strpos( $rewrote, 'sirius_export' ), 'the export marker is stripped' );

exit( T::summary() );