← All changes | libraries/html-parser.class.php +27 -23 2.0.4 → 3.4 View file →

		@@ -7,8 +7,10 @@
7	7	* @author Tobias Bäthge
8	8	* @since 2.0.0
9	9	*/
10	10
	11	+declare(strict_types=1);
	12	+
11	13	// Prohibit direct script loading.
12	14	defined( 'ABSPATH' ) || die( 'No direct script access allowed!' );
13	15
14	16	/**
		@@ -26,11 +28,11 @@
26	28	*
27	29	* @since 2.0.0
28	30	*
29	31	* @param string $html Data to be parsed.
30		- * @return array|WP_Error Array with table data and options (current table head and foot row) on success, WP_Error on error.
	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.
31	33	*/
32		- public static function parse( $html ) {
	34	+ public static function parse( string $html ) /* : array|WP_Error */ {
33	35	if ( false === stripos( $html, '<table' ) || false === stripos( $html, '</table>' ) ) {
34	36	return new WP_Error( 'table_import_html_no_table_found' );
35	37	}
36	38
		@@ -40,9 +42,9 @@
40	42	/*
41	43	* Don't expand external entities, see https://websec.io/2012/08/27/Preventing-XXE-in-PHP.html.
42	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.
43	45	*/
44		- @libxml_disable_entity_loader( true ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
	46	+ @libxml_disable_entity_loader( true ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged,Generic.PHP.DeprecatedFunctions.Deprecated
45	47	}
46	48	// No warnings/errors raised, but stored internally.
47	49	libxml_use_internal_errors( true );
48	50	$dom = new DOMDocument( '1.0', 'UTF-8' );
		@@ -47,10 +49,10 @@
47	49	libxml_use_internal_errors( true );
48	50	$dom = new DOMDocument( '1.0', 'UTF-8' );
49	51	// No strict checking for invalid HTML.
50	52	$dom->strictErrorChecking = false; // phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
51		- $dom->loadHTML( $full_html );
52		- if ( false === $dom ) {
	53	+ $result = $dom->loadHTML( $full_html );
	54	+ if ( ! $result ) {
53	55	return new WP_Error( 'table_import_html_dom_load_html_failed' );
54	56	}
55	57	$dom_tables = $dom->getElementsByTagName( 'table' );
56	58	if ( 0 === count( $dom_tables ) ) {
		@@ -56,10 +58,10 @@
56	58	if ( 0 === count( $dom_tables ) ) {
57	59	return new WP_Error( 'table_import_html_dom_get_tables' );
58	60	}
59	61	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 ) );
61		- if ( false === $table ) {
	62	+ $table = simplexml_import_dom( $dom_tables->item( 0 ) ); // @phpstan-ignore argument.type
	63	+ if ( is_null( $table ) ) {
62	64	return new WP_Error( 'table_import_html_simplexml_import_dom_failed' );
63	65	}
64	66
65	67	$errors = libxml_get_errors();
		@@ -64,19 +66,19 @@
64	66
65	67	$errors = libxml_get_errors();
66	68	libxml_clear_errors();
67	69	if ( ! empty( $errors ) ) {
68		- $output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . '</strong><br /><br />';
	70	+ $output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . '</strong><br><br>';
69	71	foreach ( $errors as $error ) {
70	72	switch ( $error->level ) {
71	73	case LIBXML_ERR_WARNING:
72		- $output .= "Warning {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br />";
	74	+ $output .= "Warning {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br>";
73	75	break;
74	76	case LIBXML_ERR_ERROR:
75		- $output .= "Error {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br />";
	77	+ $output .= "Error {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br>";
76	78	break;
77	79	case LIBXML_ERR_FATAL:
78		- $output .= "Fatal Error {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br />";
	80	+ $output .= "Fatal Error {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br>";
79	81	break;
80	82	}
81	83	}
82	84	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
		@@ -86,20 +88,22 @@
86	88	'data' => array(),
87	89	'options' => array(),
88	90	);
89	91	if ( isset( $table->thead ) ) {
90		- $html_table['data'] = array_merge( $html_table['data'], self::_import_html_rows( $table->thead[0]->tr ) );
91		- $html_table['options']['table_head'] = true;
	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 );
92	95	}
93	96	if ( isset( $table->tbody ) ) {
94		- $html_table['data'] = array_merge( $html_table['data'], self::_import_html_rows( $table->tbody[0]->tr ) );
	97	+ $html_table['data'] = array_merge( $html_table['data'], self::_import_html_rows( $table->tbody[0]->tr ) ); // @phpstan-ignore property.nonObject
95	98	}
96	99	if ( isset( $table->tr ) ) {
97	100	$html_table['data'] = array_merge( $html_table['data'], self::_import_html_rows( $table->tr ) );
98	101	}
99	102	if ( isset( $table->tfoot ) ) {
100		- $html_table['data'] = array_merge( $html_table['data'], self::_import_html_rows( $table->tfoot[0]->tr ) );
101		- $html_table['options']['table_foot'] = true;
	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 );
102	106	}
103	107
104	108	return $html_table;
105	109	}
		@@ -109,11 +113,11 @@
109	113	*
110	114	* @since 2.0.0
111	115	*
112	116	* @param SimpleXMLElement $element XMLElement.
113		- * @return array SimpleXMLElement exported to an array.
	117	+ * @return array<int, array<int, string>> SimpleXMLElement exported to an array.
114	118	*/
115		- protected static function _import_html_rows( $element ) {
	119	+ protected static function _import_html_rows( SimpleXMLElement $element ): array {
116	120	$rows = array(); // Container for the table data.
117	121	$rowspans = array(); // Container for information about rowspans in rows that follow the currently processed row.
118	122
119	123	$row_idx = 0;
		@@ -118,9 +122,9 @@
118	122
119	123	$row_idx = 0;
120	124	foreach ( $element as $row ) {
121	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).
122		- while ( isset( $rowspans[ $row_idx ] ) && count( $rowspans[ $row_idx ] ) === count( $rows[ $row_idx - 1 ] ) ) {
	126	+ while ( isset( $rowspans[ $row_idx ] ) && count( $rowspans[ $row_idx ] ) === count( $rows[ $row_idx - 1 ] ) ) { // phpcs:ignore Squiz.PHP.DisallowSizeFunctionsInLoops.Found
123	127	$rows[] = $rowspans[ $row_idx ];
124	128	++$row_idx;
125	129	}
126	130
		@@ -132,15 +136,15 @@
132	136	$new_row[] = $rowspans[ $row_idx ][ $column_idx ];
133	137	++$column_idx;
134	138	}
135	139
136		- $cell_xml = $cell->asXml();
	140	+ $cell_xml = $cell->asXML();
137	141
138	142	// Get content between <td>...</td>, or <th>...</th>, possibly with HTML.
139		- if ( 1 === preg_match( '#<t[d|h].*?>(.*)</t[d|h]>#is', $cell_xml, $matches ) ) {
	143	+ if ( false !== $cell_xml && 1 === preg_match( '#<t[d|h].*?>(.*)</t[d|h]>#is', $cell_xml, $matches ) ) {
140	144	/*
141	145	* Decode HTML entities again, as there might be some left especially in attributes of HTML tags in the cells,
142		- * see https://secure.php.net/manual/en/simplexmlelement.asxml.php#107137.
	146	+ * see https://www.php.net/manual/en/simplexmlelement.asxml.php#107137.
143	147	*/
144	148	$new_row[] = html_entity_decode( $matches[1], ENT_NOQUOTES, 'UTF-8' );
145	149
146	150	// Search for colspan and rowspan attributes in the cell's HTML tag.
		@@ -186,9 +190,9 @@
186	190	++$row_idx;
187	191	}
188	192
189	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).
190		- while ( isset( $rowspans[ $row_idx ] ) && count( $rowspans[ $row_idx ] ) === count( $rows[ $row_idx - 1 ] ) ) {
	194	+ while ( isset( $rowspans[ $row_idx ] ) && count( $rowspans[ $row_idx ] ) === count( $rows[ $row_idx - 1 ] ) ) { // phpcs:ignore Squiz.PHP.DisallowSizeFunctionsInLoops.Found
191	195	$rows[] = $rowspans[ $row_idx ];
192	196	++$row_idx;
193	197	}
194	198