tablepress / libraries / html-parser.class.php

html-parser.class.php in TablePress – Tables in WordPress made easy 3.3.2, at libraries/html-parser.class.php

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