wp-extra / src / Modules / Common / TOC / Parser.php

Parser.php in WP EXtra – One Click Optimize trunk, at src/Modules/Common/TOC/Parser.php

221 lines 5.6 KB

Raw Download Zip

1	<?php
2	/**
3	* TOC Parser
4	*
5	* Lightweight, high-performance heading extractor and anchor generator using WordPress core functions.
6	*
7	* @package WPEXtra\Modules\Common\TOC
8	*/
9
10	namespace WPEXtra\Modules\Common\TOC;
11
12	use WPEXtra\Helper;
13
14	if ( ! defined( 'ABSPATH' ) ) {
15	exit;
16	}
17
18	class Parser {
19
20	/**
21	* Unique anchor IDs tracked per request to avoid collisions.
22	*
23	* @var array<string, int>
24	*/
25	private $anchor_counts = array();
26
27	/**
28	* Parse content and extract TOC headings.
29	*
30	* @param string $content Raw HTML content.
31	* @param array $custom_options Optional runtime option overrides.
32	* @return array{content: string, headings: array, count: int}
33	*/
34	public function parse( $content, $custom_options = array() ) {
35	$this->anchor_counts = array();
36
37	if ( empty( $content ) \|\| ! is_string( $content ) ) {
38	return array(
39	'content' => $content,
40	'headings' => array(),
41	'count' => 0,
42	);
43	}
44
45	$heading_levels = isset( $custom_options['heading_levels'] )
46	? (array) $custom_options['heading_levels']
47	: (array) Helper::get_option( 'heading_levels', array( '1', '2', '3', '4', '5', '6' ) );
48
49	if ( empty( $heading_levels ) ) {
50	return array(
51	'content' => $content,
52	'headings' => array(),
53	'count' => 0,
54	);
55	}
56
57	$levels_pattern = implode( '', array_map( 'intval', $heading_levels ) );
58	$pattern = '/<h([' . $levels_pattern . '])([^>])>(.?)<\/h\1>/is';
59
60	if ( ! preg_match_all( $pattern, $content, $matches, PREG_SET_ORDER ) ) {
61	return array(
62	'content' => $content,
63	'headings' => array(),
64	'count' => 0,
65	);
66	}
67
68	$min_start = isset( $custom_options['start'] )
69	? (int) $custom_options['start']
70	: (int) Helper::get_option( 'start', 4 );
71
72	$exclude_pattern = isset( $custom_options['exclude'] )
73	? (string) $custom_options['exclude']
74	: (string) Helper::get_option( 'exclude', '' );
75
76	$exclude_classes = isset( $custom_options['exclude_by_class'] )
77	? (string) $custom_options['exclude_by_class']
78	: (string) Helper::get_option( 'exclude_by_class', '' );
79
80	$exclude_class_list = array_filter( array_map( 'trim', explode( ',', $exclude_classes ) ) );
81	$exclude_patterns = array_filter( array_map( 'trim', explode( '\|', $exclude_pattern ) ) );
82
83	$headings_flat = array();
84	$replacements = array();
85
86	foreach ( $matches as $index => $match ) {
87	$full_tag = $match[0];
88	$level = (int) $match[1];
89	$attributes = $match[2];
90	$inner_html = $match[3];
91
92	$clean_text = trim( wp_strip_all_tags( $inner_html ) );
93	if ( '' === $clean_text ) {
94	continue;
95	}
96
97	// Check exclude by class
98	if ( ! empty( $exclude_class_list ) && preg_match( '/class=["\']([^"\']+)["\']/i', $attributes, $class_match ) ) {
99	$tag_classes = preg_split( '/\s+/', $class_match[1] );
100	if ( array_intersect( $tag_classes, $exclude_class_list ) ) {
101	continue;
102	}
103	}
104
105	// Check exclude by wildcard pattern
106	if ( ! empty( $exclude_patterns ) && $this->is_excluded_by_pattern( $clean_text, $exclude_patterns ) ) {
107	continue;
108	}
109
110	// Extract or generate anchor ID
111	$anchor_id = '';
112	if ( preg_match( '/id=["\']([^"\']+)["\']/i', $attributes, $id_match ) ) {
113	$anchor_id = sanitize_title( $id_match[1] );
114	}
115
116	if ( empty( $anchor_id ) ) {
117	$anchor_id = $this->generate_unique_anchor( $clean_text );
118	$new_attrs = ' id="' . esc_attr( $anchor_id ) . '"' . $attributes;
119	$new_tag = '<h' . $level . $new_attrs . '>' . $inner_html . '</h' . $level . '>';
120	$replacements[ $full_tag ] = $new_tag;
121	}
122
123	$headings_flat[] = array(
124	'id' => $anchor_id,
125	'title' => $clean_text,
126	'inner_html' => $inner_html,
127	'level' => $level,
128	);
129	}
130
131	$total_count = count( $headings_flat );
132	if ( $total_count < $min_start ) {
133	return array(
134	'content' => $content,
135	'headings' => array(),
136	'count' => $total_count,
137	);
138	}
139
140	// Apply ID replacements to content
141	if ( ! empty( $replacements ) ) {
142	$content = str_replace( array_keys( $replacements ), array_values( $replacements ), $content );
143	}
144
145	$tree = $this->build_hierarchy_tree( $headings_flat );
146
147	return array(
148	'content' => $content,
149	'headings' => $tree,
150	'count' => $total_count,
151	);
152	}
153
154	/**
155	* Generate a unique anchor ID from heading text.
156	*
157	* @param string $text
158	* @return string
159	*/
160	private function generate_unique_anchor( $text ) {
161	$slug = sanitize_title( $text );
162	if ( empty( $slug ) ) {
163	$slug = 'heading';
164	}
165
166	if ( ! isset( $this->anchor_counts[ $slug ] ) ) {
167	$this->anchor_counts[ $slug ] = 1;
168	return $slug;
169	}
170
171	$this->anchor_counts[ $slug ]++;
172	return $slug . '-' . $this->anchor_counts[ $slug ];
173	}
174
175	/**
176	* Check if heading matches any exclusion wildcard pattern.
177	*
178	* @param string $text
179	* @param array $patterns
180	* @return bool
181	*/
182	private function is_excluded_by_pattern( $text, array $patterns ) {
183	foreach ( $patterns as $pattern ) {
184	$regex = '/^' . str_replace( '\', '.', preg_quote( $pattern, '/' ) ) . '$/iu';
185	if ( preg_match( $regex, $text ) ) {
186	return true;
187	}
188	}
189	return false;
190	}
191
192	/**
193	* Convert flat heading list into a nested hierarchical tree.
194	*
195	* @param array $headings
196	* @return array
197	*/
198	private function build_hierarchy_tree( array $headings ) {
199	$tree = array();
200	$stack = array();
201
202	foreach ( $headings as &$item ) {
203	$item['children'] = array();
204
205	while ( ! empty( $stack ) && end( $stack )['level'] >= $item['level'] ) {
206	array_pop( $stack );
207	}
208
209	if ( empty( $stack ) ) {
210	$tree[] = &$item;
211	} else {
212	$stack[ count( $stack ) - 1 ]['children'][] = &$item;
213	}
214	$stack[] = &$item;
215	}
216	unset( $item );
217
218	return $tree;
219	}
220	}
221