tablepress / libraries / html-parser.class.php

html-parser.class.php in TablePress – Tables in WordPress made easy 3.4, at libraries/html-parser.class.php

203 lines 7.7 KB

Raw Download Zip

1	<?php
2	/**
3	* HTML Parsing class for TablePress, used for import of HTML files.
4	*
5	* @package TablePress
6	* @subpackage Import
7	* @author Tobias Bäthge
8	* @since 2.0.0
9	*/
10
11	declare(strict_types=1);
12
13	// Prohibit direct script loading.
14	defined( 'ABSPATH' ) \|\| die( 'No direct script access allowed!' );
15
16	/**
17	* HTML Parsing class
18	*
19	* @package TablePress
20	* @subpackage Import
21	* @author Tobias Bäthge
22	* @since 2.0.0
23	*/
24	abstract class HTML_Parser {
25
26	/**
27	* Parses HTML string into a two-dimensional array, maybe with options.
28	*
29	* @since 2.0.0
30	*
31	* @param string $html Data to be parsed.
32	* @return array<string, mixed>\|WP_Error Array with table data and options (current table head and foot row) on success, WP_Error on error.
33	*/
34	public static function parse( string $html ) /* : array\|WP_Error */ {
35	if ( false === stripos( $html, '<table' ) \|\| false === stripos( $html, '</table>' ) ) {
36	return new WP_Error( 'table_import_html_no_table_found' );
37	}
38
39	// Prepend XML declaration, for better encoding support.
40	$full_html = '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>' . $html;
41	if ( function_exists( 'libxml_disable_entity_loader' ) ) {
42	/*
43	* Don't expand external entities, see https://websec.io/2012/08/27/Preventing-XXE-in-PHP.html.
44	* Silence warnings as the function is deprecated in PHP 8, but can be necessary with LIBXML_NOENT being defined, see https://core.trac.wordpress.org/changeset/50714.
45	*/
46	@libxml_disable_entity_loader( true ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged,Generic.PHP.DeprecatedFunctions.Deprecated
47	}
48	// No warnings/errors raised, but stored internally.
49	libxml_use_internal_errors( true );
50	$dom = new DOMDocument( '1.0', 'UTF-8' );
51	// No strict checking for invalid HTML.
52	$dom->strictErrorChecking = false; // phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
53	$result = $dom->loadHTML( $full_html );
54	if ( ! $result ) {
55	return new WP_Error( 'table_import_html_dom_load_html_failed' );
56	}
57	$dom_tables = $dom->getElementsByTagName( 'table' );
58	if ( 0 === count( $dom_tables ) ) {
59	return new WP_Error( 'table_import_html_dom_get_tables' );
60	}
61	libxml_clear_errors(); // Clear errors so that we only catch those inside the table in the next line.
62	$table = simplexml_import_dom( $dom_tables->item( 0 ) ); // @phpstan-ignore argument.type
63	if ( is_null( $table ) ) {
64	return new WP_Error( 'table_import_html_simplexml_import_dom_failed' );
65	}
66
67	$errors = libxml_get_errors();
68	libxml_clear_errors();
69	if ( ! empty( $errors ) ) {
70	$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . '</strong><br><br>';
71	foreach ( $errors as $error ) {
72	switch ( $error->level ) {
73	case LIBXML_ERR_WARNING:
74	$output .= "Warning {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br>";
75	break;
76	case LIBXML_ERR_ERROR:
77	$output .= "Error {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br>";
78	break;
79	case LIBXML_ERR_FATAL:
80	$output .= "Fatal Error {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br>";
81	break;
82	}
83	}
84	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
85	}
86
87	$html_table = array(
88	'data' => array(),
89	'options' => array(),
90	);
91	if ( isset( $table->thead ) ) {
92	$head_rows = self::_import_html_rows( $table->thead[0]->tr ); // @phpstan-ignore property.nonObject
93	$html_table['data'] = array_merge( $html_table['data'], $head_rows );
94	$html_table['options']['table_head'] = count( $head_rows );
95	}
96	if ( isset( $table->tbody ) ) {
97	$html_table['data'] = array_merge( $html_table['data'], self::_import_html_rows( $table->tbody[0]->tr ) ); // @phpstan-ignore property.nonObject
98	}
99	if ( isset( $table->tr ) ) {
100	$html_table['data'] = array_merge( $html_table['data'], self::_import_html_rows( $table->tr ) );
101	}
102	if ( isset( $table->tfoot ) ) {
103	$foot_rows = self::_import_html_rows( $table->tfoot[0]->tr ); // @phpstan-ignore property.nonObject
104	$html_table['data'] = array_merge( $html_table['data'], $foot_rows );
105	$html_table['options']['table_foot'] = count( $foot_rows );
106	}
107
108	return $html_table;
109	}
110
111	/**
112	* Converts table HTML rows to an array.
113	*
114	* @since 2.0.0
115	*
116	* @param SimpleXMLElement $element XMLElement.
117	* @return array<int, array<int, string>> SimpleXMLElement exported to an array.
118	*/
119	protected static function _import_html_rows( SimpleXMLElement $element ): array {
120	$rows = array(); // Container for the table data.
121	$rowspans = array(); // Container for information about rowspans in rows that follow the currently processed row.
122
123	$row_idx = 0;
124	foreach ( $element as $row ) {
125	// If all cells in a row should be merged with the cells in the row above, add the trigger word to each of them (should be very rare).
126	while ( isset( $rowspans[ $row_idx ] ) && count( $rowspans[ $row_idx ] ) === count( $rows[ $row_idx - 1 ] ) ) { // phpcs:ignore Squiz.PHP.DisallowSizeFunctionsInLoops.Found
127	$rows[] = $rowspans[ $row_idx ];
128	++$row_idx;
129	}
130
131	$new_row = array();
132	$column_idx = 0;
133	foreach ( $row as $cell ) {
134	// If a cell in a row should be merged with the cell above it, add the trigger word to it.
135	while ( isset( $rowspans[ $row_idx ][ $column_idx ] ) ) {
136	$new_row[] = $rowspans[ $row_idx ][ $column_idx ];
137	++$column_idx;
138	}
139
140	$cell_xml = $cell->asXML();
141
142	// Get content between <td>...</td>, or <th>...</th>, possibly with HTML.
143	if ( false !== $cell_xml && 1 === preg_match( '#<t[d\|h].?>(.)</t[d\|h]>#is', $cell_xml, $matches ) ) {
144	/*
145	* Decode HTML entities again, as there might be some left especially in attributes of HTML tags in the cells,
146	* see https://www.php.net/manual/en/simplexmlelement.asxml.php#107137.
147	*/
148	$new_row[] = html_entity_decode( $matches[1], ENT_NOQUOTES, 'UTF-8' );
149
150	// Search for colspan and rowspan attributes in the cell's HTML tag.
151	$colspan = 1;
152	$rowspan = 1;
153	if ( 1 === preg_match( '#<t[d\|h].colspan=["\']?(\d+)["\']?.?>#is', $cell_xml, $matches ) ) {
154	$colspan = (int) $matches[1];
155	}
156	if ( 1 === preg_match( '#<t[d\|h].rowspan=["\']?(\d+)["\']?.?>#is', $cell_xml, $matches ) ) {
157	$rowspan = (int) $matches[1];
158	}
159
160	// Add cells with the colspan trigger word, if merged cells across columns were found.
161	for ( $i = 1; $i < $colspan; $i++ ) {
162	$new_row[] = '#colspan#';
163	}
164
165	// If merged cells across rows were found, add trigger words to a temporary variable.
166	for ( $i = 1; $i < $rowspan; $i++ ) {
167	if ( ! isset( $rowspans[ $row_idx + $i ] ) ) {
168	$rowspans[ $row_idx + $i ] = array();
169	}
170	$rowspans[ $row_idx + $i ][ $column_idx ] = '#rowspan#';
171	for ( $j = 1; $j < $colspan; $j++ ) {
172	$rowspans[ $row_idx + $i ][ $column_idx + $j ] = '#span#';
173	}
174	}
175	} else {
176	// Add an empty cell if no content could be extracted from the cell's HTML tag.
177	$new_row[] = '';
178	}
179
180	++$column_idx;
181	}
182
183	// After the last cell in a row: If a cell in a row should be merged with the cell above it, add the trigger word to it.
184	while ( isset( $rowspans[ $row_idx ][ $column_idx ] ) ) {
185	$new_row[] = $rowspans[ $row_idx ][ $column_idx ];
186	++$column_idx;
187	}
188
189	$rows[] = $new_row;
190	++$row_idx;
191	}
192
193	// After the last data row: If all cells in a row should be merged with the cells in the row above, add the trigger word to each of them (should be very rare).
194	while ( isset( $rowspans[ $row_idx ] ) && count( $rowspans[ $row_idx ] ) === count( $rows[ $row_idx - 1 ] ) ) { // phpcs:ignore Squiz.PHP.DisallowSizeFunctionsInLoops.Found
195	$rows[] = $rowspans[ $row_idx ];
196	++$row_idx;
197	}
198
199	return $rows;
200	}
201
202	} // class HTML_Parser
203