tablepress / classes / class-import-legacy.php

class-import-legacy.php in TablePress – Tables in WordPress made easy 2.2.1, at classes/class-import-legacy.php

353 lines 9.8 KB

Raw Download Zip

1	<?php
2	/**
3	* TablePress Table Import Class
4	*
5	* @package TablePress
6	* @subpackage Export/Import
7	* @author Tobias Bäthge
8	* @since 1.0.0
9	*/
10
11	// Prohibit direct script loading.
12	defined( 'ABSPATH' ) \|\| die( 'No direct script access allowed!' );
13
14	/**
15	* TablePress Table Import Class
16	*
17	* @package TablePress
18	* @subpackage Export/Import
19	* @author Tobias Bäthge
20	* @since 1.0.0
21	*/
22	class TablePress_Import_Legacy extends TablePress_Import_Base {
23
24	/**
25	* File/Data Formats that are available for import.
26	*
27	* @since 1.0.0
28	* @var array<string, string>
29	*/
30	public $import_formats = array();
31
32	/**
33	* Whether HTML import support is available in the PHP installation on the server.
34	*
35	* @since 1.0.0
36	* @var bool
37	*/
38	public $html_import_support_available = false;
39
40	/**
41	* Data to be imported.
42	*
43	* @since 1.0.0
44	* @var string
45	*/
46	protected $import_data;
47
48	/**
49	* Imported table.
50	*
51	* @since 1.0.0
52	* @var array<string, mixed>\|false
53	*/
54	protected $imported_table = false;
55
56	/**
57	* Initialize the Import class.
58	*
59	* @since 1.0.0
60	*/
61	public function __construct() {
62	if ( class_exists( 'DOMDocument', false ) && function_exists( 'simplexml_import_dom' ) && function_exists( 'libxml_use_internal_errors' ) ) {
63	$this->html_import_support_available = true;
64	}
65
66	// Initiate here, because function call not possible outside a class method.
67	$this->import_formats = array();
68	$this->import_formats['csv'] = __( 'CSV - Character-Separated Values', 'tablepress' );
69	if ( $this->html_import_support_available ) {
70	$this->import_formats['html'] = __( 'HTML - Hypertext Markup Language', 'tablepress' );
71	}
72	$this->import_formats['json'] = __( 'JSON - JavaScript Object Notation', 'tablepress' );
73	$this->import_formats['xls'] = __( 'XLS - Microsoft Excel 97-2003 (experimental)', 'tablepress' );
74	$this->import_formats['xlsx'] = __( 'XLSX - Microsoft Excel 2007-2019 (experimental)', 'tablepress' );
75	}
76
77	/**
78	* Import a table.
79	*
80	* @since 1.0.0
81	*
82	* @param string $format Import format.
83	* @param string $data Data to import.
84	* @return array<string, mixed>\|false Table array on success, false on error.
85	*/
86	public function import_table( string $format, string $data ) /* : array\|false */ {
87	/**
88	* Filters the data that is to be imported.
89	*
90	* @since 1.8.1
91	*
92	* @param string $data Data to import.
93	* @param string $format Import format.
94	*/
95	$this->import_data = apply_filters( 'tablepress_import_table_data', $data, $format );
96
97	if ( ! in_array( $format, array( 'xlsx', 'xls' ), true ) ) {
98	$this->fix_table_encoding();
99	}
100
101	switch ( $format ) {
102	case 'csv':
103	$this->import_csv();
104	break;
105	case 'html':
106	$this->import_html();
107	break;
108	case 'json':
109	$this->import_json();
110	break;
111	case 'xlsx':
112	$this->import_xlsx();
113	break;
114	case 'xls':
115	$this->import_xls();
116	break;
117	default:
118	return false;
119	}
120
121	if ( false === $this->imported_table ) {
122	return false;
123	}
124
125	// Make sure that cells are stored as strings.
126	array_walk_recursive(
127	$this->imported_table['data'],
128	static function ( /* string\|int\|float\|bool\|null */ &$cell_content, int $col_idx ): void {
129	$cell_content = (string) $cell_content;
130	}
131	);
132
133	return $this->imported_table;
134	}
135
136	/**
137	* Import CSV data.
138	*
139	* @since 1.0.0
140	*/
141	protected function import_csv(): void {
142	$csv_parser = TablePress::load_class( 'CSV_Parser', 'csv-parser.class.php', 'libraries' );
143	$csv_parser->load_data( $this->import_data );
144	$delimiter = $csv_parser->find_delimiter();
145	$data = $csv_parser->parse( $delimiter );
146	$this->pad_array_to_max_cols( $data );
147	$this->normalize_line_endings( $data );
148	$this->imported_table = array( 'data' => $data );
149	}
150
151	/**
152	* Import HTML data.
153	*
154	* @since 1.0.0
155	*/
156	protected function import_html(): void {
157	if ( ! $this->html_import_support_available ) {
158	return;
159	}
160
161	TablePress::load_file( 'html-parser.class.php', 'libraries' );
162	$table = HTML_Parser::parse( $this->import_data );
163
164	if ( is_wp_error( $table ) ) {
165	$this->imported_table = false;
166	return;
167	}
168
169	$this->pad_array_to_max_cols( $table['data'] );
170	$this->normalize_line_endings( $table['data'] );
171	$this->imported_table = $table;
172	}
173
174	/**
175	* Import JSON data.
176	*
177	* @since 1.0.0
178	*/
179	protected function import_json(): void {
180	$json_table = json_decode( $this->import_data, true );
181
182	// Check if JSON could be decoded.
183	if ( is_null( $json_table ) ) {
184	$json_error = json_last_error_msg();
185	$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . "</strong><br /><br />JSON error: {$json_error}<br />";
186	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
187	}
188
189	// Specifically cast to an array again.
190	$json_table = (array) $json_table;
191
192	if ( isset( $json_table['data'] ) ) {
193	// JSON data contained a full export.
194	$table = $json_table;
195	} else {
196	// JSON data contained only the data of a table, but no options.
197	$table = array( 'data' => array() );
198	foreach ( $json_table as $row ) {
199	$table['data'][] = array_values( (array) $row );
200	}
201	}
202
203	$this->pad_array_to_max_cols( $table['data'] );
204	$this->imported_table = $table;
205	}
206
207	/**
208	* Import Microsoft Excel 97-2003 data.
209	*
210	* @since 1.1.0
211	*/
212	protected function import_xls(): void {
213	$excel_reader = TablePress::load_class( 'Spreadsheet_Excel_Reader', 'excel-reader.class.php', 'libraries', $this->import_data );
214
215	// Loop through Excel file and retrieve value and colspan/rowspan properties for each cell.
216	$sheet = 0; // 0 means first sheet of the Workbook
217	$table = array();
218	$num_rows = $excel_reader->rowcount( $sheet );
219	$num_columns = $excel_reader->colcount( $sheet );
220	for ( $row = 1; $row <= $num_rows; $row++ ) {
221	$table_row = array();
222	for ( $column = 1; $column <= $num_columns; $column++ ) {
223	$cell = array();
224	$cell['rowspan'] = $excel_reader->rowspan( $row, $column, $sheet );
225	$cell['colspan'] = $excel_reader->colspan( $row, $column, $sheet );
226	$cell['val'] = $excel_reader->val( $row, $column, $sheet );
227	$table_row[] = $cell;
228	}
229	$table[] = $table_row;
230	}
231
232	// Transform colspan/rowspan properties to TablePress equivalent (cell content).
233	foreach ( $table as $row_idx => $row ) {
234	foreach ( $row as $col_idx => $cell ) {
235	if ( 1 === $cell['rowspan'] && 1 === $cell['colspan'] ) {
236	continue;
237	}
238
239	if ( 1 < $cell['colspan'] ) {
240	for ( $i = 1; $i < $cell['colspan']; $i++ ) {
241	$table[ $row_idx ][ $col_idx + $i ]['val'] = '#colspan#';
242	}
243	}
244	if ( 1 < $cell['rowspan'] ) {
245	for ( $i = 1; $i < $cell['rowspan']; $i++ ) {
246	$table[ $row_idx + $i ][ $col_idx ]['val'] = '#rowspan#';
247	}
248	}
249
250	if ( 1 < $cell['rowspan'] && 1 < $cell['colspan'] ) {
251	for ( $i = 1; $i < $cell['rowspan']; $i++ ) {
252	for ( $j = 1; $j < $cell['colspan']; $j++ ) {
253	$table[ $row_idx + $i ][ $col_idx + $j ]['val'] = '#span#';
254	}
255	}
256	}
257	}
258	}
259
260	// Flatten value property to two-dimensional array.
261	foreach ( $table as &$row ) {
262	foreach ( $row as &$cell ) {
263	$cell = $cell['val'];
264	}
265	unset( $cell );
266	}
267	unset( $row );
268
269	$this->imported_table = array( 'data' => $table );
270	}
271
272	/**
273	* Import Microsoft Excel 2007-2019 data.
274	*
275	* @since 1.1.0
276	*/
277	protected function import_xlsx(): void {
278	TablePress::load_file( 'simplexlsx.class.php', 'libraries' );
279	$xlsx_file = \Shuchkin\SimpleXLSX::parse( $this->import_data, true );
280
281	if ( ! $xlsx_file ) {
282	$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . '</strong><br /><br />' . \Shuchkin\SimpleXLSX::parseError() . '<br />';
283	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
284	}
285
286	$this->imported_table = array( 'data' => $xlsx_file->rows() );
287	}
288
289	/**
290	* Fixes the encoding to UTF-8 for the entire string that is to be imported.
291	*
292	* @since 1.0.0
293	*
294	* @link http://stevephillips.me/blog/dealing-php-and-character-encoding
295	*/
296	protected function fix_table_encoding(): void {
297	// Check and remove possible UTF-8 Byte-Order Mark (BOM).
298	$bom = pack( 'CCC', 0xef, 0xbb, 0xbf );
299	if ( str_starts_with( $this->import_data, $bom ) ) {
300	$this->import_data = substr( $this->import_data, 3 );
301	// If data has a BOM, it's UTF-8, so further checks unnecessary.
302	return;
303	}
304
305	// Require the iconv() function for the following checks.
306	if ( ! function_exists( 'iconv' ) ) {
307	return;
308	}
309
310	// Check for possible UTF-16 BOMs ("little endian" and "big endian") and try to convert the data to UTF-8.
311	if ( str_starts_with( $this->import_data, "\xFF\xFE" ) \|\| str_starts_with( $this->import_data, "\xFE\xFF" ) ) {
312	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
313	$data = @iconv( 'UTF-16', 'UTF-8', $this->import_data );
314	if ( false !== $data ) {
315	$this->import_data = $data;
316	return;
317	}
318	}
319
320	// Detect the character encoding and convert to UTF-8, if it's different.
321	if ( function_exists( 'mb_detect_encoding' ) ) {
322	$current_encoding = mb_detect_encoding( $this->import_data, 'ASCII, UTF-8, ISO-8859-1' );
323	if ( 'UTF-8' !== $current_encoding ) {
324	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
325	$data = @iconv( $current_encoding, 'UTF-8', $this->import_data ); // @phpstan-ignore-line
326	if ( false !== $data ) {
327	$this->import_data = $data;
328	return;
329	}
330	}
331	}
332	}
333
334	/**
335	* Replaces Windows line breaks \r\n with Unix line breaks \n.
336	*
337	* This function uses call by reference to save PHP memory on large arrays.
338	*
339	* @since 2.0.0
340	*
341	* @param string[] $an_array Array in which Windows line breaks should be replaced by Unix line breaks.
342	*/
343	protected function normalize_line_endings( array &$an_array ): void {
344	array_walk_recursive(
345	$an_array,
346	static function ( string &$cell_content, int $col_idx ): void {
347	$cell_content = str_replace( "\r\n", "\n", $cell_content );
348	}
349	);
350	}
351
352	} // class TablePress_Import_Legacy
353