webspellchecker / tests / test-content-cleaner.php

test-content-cleaner.php in WProofreader spell & grammar check plugin for WordPress trunk, at tests/test-content-cleaner.php

145 lines 6.1 KB

Raw Download Zip

1	<?php
2	/**
3	* Standalone test runner for WProofreader_Content_Cleaner.
4	*
5	* Usage (inside a WP container): php tests/test-content-cleaner.php /path/to/wp-load.php
6	* Exits non-zero on failure.
7	*/
8
9	$wp_load = $argv[1] ?? '/var/www/web/wp/wp-load.php';
10	require $wp_load;
11
12	if ( ! class_exists( 'WProofreader_Content_Cleaner' ) ) {
13	require dirname( __DIR__ ) . '/includes/class-wproofreader-content-cleaner.php';
14	}
15
16	$failures = 0;
17
18	function check( string $name, bool $condition, string $detail = '' ) {
19	global $failures;
20	if ( $condition ) {
21	echo "PASS {$name}\n";
22	} else {
23	$failures++;
24	echo "FAIL {$name}" . ( $detail ? " — {$detail}" : '' ) . "\n";
25	}
26	}
27
28	function clean( string $html ): string {
29	return WProofreader_Content_Cleaner::clean_content( $html );
30	}
31
32	function hook_path( string $html ): string {
33	$data = WProofreader_Content_Cleaner::clean_post_data(
34	array( 'post_content' => wp_slash( $html ), 'post_status' => 'private' ),
35	array()
36	);
37	return wp_unslash( $data['post_content'] );
38	}
39
40	// 1. Clean content without artifact markers is untouched (wsc- URLs included).
41	$plain = '<p>Hello <a href="https://webspellchecker.com/wsc-proofreader/">link</a></p>';
42	check( 'clean content untouched', clean( $plain ) === $plain );
43
44	// 2. Simple artifact span is unwrapped.
45	check(
46	'simple span unwrapped',
47	clean( '<p><span class="wsc-spelling-problem">word</span> rest</p>' ) === '<p>word rest</p>'
48	);
49
50	// 3. Multi-class artifact span with extra attributes.
51	check(
52	'multi-class span with attributes',
53	clean( '<p><span data-wsc-id="9" class="wsc-spelling-problem wsc-problem-42" style="background:red">word</span></p>' ) === '<p>word</p>'
54	);
55
56	// 4. Nested artifact spans.
57	check(
58	'nested artifact spans',
59	clean( '<p><span class="wsc-grammar-problem">a <span class="wsc-spelling-problem">b</span> c</span></p>' ) === '<p>a b c</p>'
60	);
61
62	// 5. Artifact span nested inside a regular span: regular span preserved byte-identical.
63	check(
64	'regular span preserved',
65	clean( "<p><span class='keep' data-x='1'>a <span class=\"wsc-spelling-problem\">b</span></span></p>" ) === "<p><span class='keep' data-x='1'>a b</span></p>"
66	);
67
68	// 6. rangySelectionBoundary removal.
69	check(
70	'rangy boundary removed',
71	clean( "<p>a<span class=\"rangySelectionBoundary\">\xE2\x80\x8B</span>b</p>" ) === "<p>a\xE2\x80\x8Bb</p>"
72	);
73
74	// 7. Unclosed artifact opener: opener dropped, rest intact.
75	check(
76	'unclosed artifact opener',
77	clean( '<p><span class="wsc-spelling-problem">word</p>' ) === '<p>word</p>'
78	);
79
80	// 8. Quoting / entities / boolean attributes outside artifacts stay byte-identical.
81	$tricky = '<figure class=\'"wp-block-table\' custom><a href=\'"https://x"\' rel=\'"nofollow"\'>x</a>'
82	. '<span class="wsc-spelling-problem">w</span> & <br><svg viewBox="0 0 1 1"><rect/></svg></figure>';
83	$expected = '<figure class=\'"wp-block-table\' custom><a href=\'"https://x"\' rel=\'"nofollow"\'>x</a>'
84	. 'w & <br><svg viewBox="0 0 1 1"><rect/></svg></figure>';
85	check( 'byte-identical outside splices', clean( $tricky ) === $expected );
86
87	// 9. Gutenberg block comments survive.
88	$blocks = "<!-- wp:paragraph -->\n<p><span class=\"wsc-grammar-problem\">Текст українською</span> 🚀</p>\n<!-- /wp:paragraph -->";
89	check(
90	'block comments and unicode survive',
91	clean( $blocks ) === "<!-- wp:paragraph -->\n<p>Текст українською 🚀</p>\n<!-- /wp:paragraph -->"
92	);
93
94	// 10. Slashed hook path: regression for the 3.1.1 corruption bug.
95	// Realistic Gutenberg fixture: quoted block attributes, existing " entities,
96	// and a wsc- URL that used to false-positive the old substring guard.
97	$baseline = '<!-- wp:heading {"level":2,"anchor":"h-custom"} -->' . "\n"
98	. '<h2 class="wp-block-heading" id="h-custom">Say "hi"</h2>' . "\n"
99	. '<!-- /wp:heading -->' . "\n"
100	. '<!-- wp:image {"id":42,"sizeSlug":"large","linkDestination":"custom"} -->' . "\n"
101	. '<figure class="wp-block-image size-large"><a href="https://webspellchecker.com/wsc-proofreader/">'
102	. '<img src="https://x.test/a.png" alt="alt with "quotes""/></a></figure>' . "\n"
103	. '<!-- /wp:image -->' . "\n"
104	. '<!-- wp:paragraph -->' . "\n"
105	. '<p>Plain <a href="https://webspellchecker.com/wsc-proofreader/" rel="noopener">link</a> text.</p>' . "\n"
106	. '<!-- /wp:paragraph -->';
107	if ( file_exists( '/tmp/before_editor_save.html' ) ) {
108	$real = file_get_contents( '/tmp/before_editor_save.html' );
109	check( 'slashed hook path lossless on real post baseline', hook_path( $real ) === $real );
110	}
111	check( 'slashed hook path lossless on clean content', hook_path( $baseline ) === $baseline );
112
113	// 11. Slashed hook path with artifacts: cleaned, no " inflation.
114	$dirty = str_replace( '<p>', '<p><span class="wsc-spelling-problem">x</span> ', $baseline );
115	$cleaned = hook_path( $dirty );
116	check( 'slashed hook path removes artifacts', strpos( $cleaned, 'wsc-spelling-problem' ) === false );
117	check(
118	'slashed hook path adds no quot entities',
119	substr_count( $cleaned, '"' ) === substr_count( $baseline, '"' )
120	);
121
122	// 11b. Double invocation (Gutenberg REST save + meta-box-loader re-save) is idempotent.
123	$pass1 = hook_path( $dirty );
124	$pass2 = hook_path( $pass1 );
125	check( 'double hook run idempotent', $pass2 === $pass1 );
126	check( 'double hook run adds no quot entities', substr_count( $pass2, '"' ) === substr_count( $baseline, '"' ) );
127
128	// 12. Autosave/revision statuses are no longer skipped.
129	$autosave = WProofreader_Content_Cleaner::clean_post_data(
130	array( 'post_content' => wp_slash( '<p><span class="wsc-spelling-problem">w</span></p>' ), 'post_status' => 'inherit' ),
131	array()
132	);
133	check( 'inherit status cleaned', wp_unslash( $autosave['post_content'] ) === '<p>w</p>' );
134
135	// 13. Empty content untouched.
136	$empty = WProofreader_Content_Cleaner::clean_post_data( array( 'post_content' => '' ), array() );
137	check( 'empty content untouched', '' === $empty['post_content'] );
138
139	// 14. Span with artifact class substring in another class name must NOT match.
140	$lookalike = '<p><span class="wsc-spelling-problem-note">keep wrapper</span></p>';
141	check( 'class lookalike preserved', clean( $lookalike ) === $lookalike );
142
143	echo $failures ? "\n{$failures} FAILURE(S)\n" : "\nALL PASS\n";
144	exit( $failures ? 1 : 0 );
145