tablepress / classes / class-import-legacy.php

class-import-legacy.php in TablePress – Tables in WordPress made easy 3.4, at classes/class-import-legacy.php

349 lines 9.5 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	declare(strict_types=1);
12
13	// Prohibit direct script loading.
14	defined( 'ABSPATH' ) \|\| die( 'No direct script access allowed!' );
15
16	/**
17	* TablePress Table Import Class
18	*
19	* @package TablePress
20	* @subpackage Export/Import
21	* @author Tobias Bäthge
22	* @since 1.0.0
23	*/
24	class TablePress_Import_Legacy extends TablePress_Import_Base {
25
26	/**
27	* File/Data Formats that are available for import.
28	*
29	* The constructor also adds 'html', if the PHP prerequisites are met.
30	*
31	* @since 1.0.0
32	* @var string[]
33	*/
34	public array $import_formats = array( 'csv', 'json', 'xls', 'xlsx' );
35
36	/**
37	* Whether HTML import support is available in the PHP installation on the server.
38	*
39	* @since 1.0.0
40	*/
41	public bool $html_import_support_available = false;
42
43	/**
44	* Data to be imported.
45	*
46	* @since 1.0.0
47	*/
48	protected string $import_data;
49
50	/**
51	* Imported table.
52	*
53	* @since 1.0.0
54	* @var array<string, mixed>\|false
55	*/
56	protected $imported_table = false;
57
58	/**
59	* Initialize the Import class.
60	*
61	* @since 1.0.0
62	*/
63	public function __construct() {
64	if ( class_exists( 'DOMDocument', false ) && function_exists( 'simplexml_import_dom' ) && function_exists( 'libxml_use_internal_errors' ) ) {
65	$this->html_import_support_available = true;
66	}
67
68	if ( $this->html_import_support_available ) {
69	$this->import_formats[] = 'html';
70	}
71	}
72
73	/**
74	* Import a table.
75	*
76	* @since 1.0.0
77	*
78	* @param string $format Import format.
79	* @param string $data Data to import.
80	* @return array<string, mixed>\|false Table array on success, false on error.
81	*/
82	public function import_table( string $format, string $data ) /* : array\|false */ {
83	/**
84	* Filters the data that is to be imported.
85	*
86	* @since 1.8.1
87	*
88	* @param string $data Data to import.
89	* @param string $format Import format.
90	*/
91	$this->import_data = apply_filters( 'tablepress_import_table_data', $data, $format );
92
93	if ( ! in_array( $format, array( 'xlsx', 'xls' ), true ) ) {
94	$this->fix_table_encoding();
95	}
96
97	switch ( $format ) {
98	case 'csv':
99	$this->import_csv();
100	break;
101	case 'html':
102	$this->import_html();
103	break;
104	case 'json':
105	$this->import_json();
106	break;
107	case 'xlsx':
108	$this->import_xlsx();
109	break;
110	case 'xls':
111	$this->import_xls();
112	break;
113	default:
114	return false;
115	}
116
117	if ( false === $this->imported_table ) {
118	return false;
119	}
120
121	// Make sure that cells are stored as strings.
122	array_walk_recursive(
123	$this->imported_table['data'],
124	static function ( /* string\|int\|float\|bool\|null */ &$cell_content, int $col_idx ): void {
125	$cell_content = (string) $cell_content;
126	},
127	);
128
129	return $this->imported_table;
130	}
131
132	/**
133	* Import CSV data.
134	*
135	* @since 1.0.0
136	*/
137	protected function import_csv(): void {
138	$csv_parser = TablePress::load_class( 'CSV_Parser', 'csv-parser.class.php', 'libraries' );
139	$csv_parser->load_data( $this->import_data );
140	$delimiter = $csv_parser->find_delimiter();
141	$data = $csv_parser->parse( $delimiter );
142	$this->pad_array_to_max_cols( $data );
143	$this->normalize_line_endings( $data );
144	$this->imported_table = array( 'data' => $data );
145	}
146
147	/**
148	* Import HTML data.
149	*
150	* @since 1.0.0
151	*/
152	protected function import_html(): void {
153	if ( ! $this->html_import_support_available ) {
154	return;
155	}
156
157	TablePress::load_file( 'html-parser.class.php', 'libraries' );
158	$table = HTML_Parser::parse( $this->import_data );
159
160	if ( is_wp_error( $table ) ) {
161	$this->imported_table = false;
162	return;
163	}
164
165	$this->pad_array_to_max_cols( $table['data'] );
166	$this->normalize_line_endings( $table['data'] );
167	$this->imported_table = $table;
168	}
169
170	/**
171	* Import JSON data.
172	*
173	* @since 1.0.0
174	*/
175	protected function import_json(): void {
176	$json_table = json_decode( $this->import_data, true );
177
178	// Check if JSON could be decoded.
179	if ( is_null( $json_table ) ) {
180	$json_error = json_last_error_msg();
181	$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . "</strong><br><br>JSON error: {$json_error}<br>";
182	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
183	}
184
185	// Specifically cast to an array again.
186	$json_table = (array) $json_table;
187
188	if ( isset( $json_table['data'] ) ) {
189	// JSON data contained a full export.
190	$table = $json_table;
191	} else {
192	// JSON data contained only the data of a table, but no options.
193	$table = array( 'data' => array() );
194	foreach ( $json_table as $row ) {
195	$table['data'][] = array_values( (array) $row );
196	}
197	}
198
199	$this->pad_array_to_max_cols( $table['data'] );
200	$this->imported_table = $table;
201	}
202
203	/**
204	* Import Microsoft Excel 97-2003 data.
205	*
206	* @since 1.1.0
207	*/
208	protected function import_xls(): void {
209	$excel_reader = TablePress::load_class( 'Spreadsheet_Excel_Reader', 'excel-reader.class.php', 'libraries', $this->import_data );
210
211	// Loop through Excel file and retrieve value and colspan/rowspan properties for each cell.
212	$sheet = 0; // 0 means first sheet of the Workbook
213	$table = array();
214	$num_rows = $excel_reader->rowcount( $sheet );
215	$num_columns = $excel_reader->colcount( $sheet );
216	for ( $row = 1; $row <= $num_rows; $row++ ) {
217	$table_row = array();
218	for ( $column = 1; $column <= $num_columns; $column++ ) {
219	$cell = array();
220	$cell['rowspan'] = $excel_reader->rowspan( $row, $column, $sheet );
221	$cell['colspan'] = $excel_reader->colspan( $row, $column, $sheet );
222	$cell['val'] = $excel_reader->val( $row, $column, $sheet );
223	$table_row[] = $cell;
224	}
225	$table[] = $table_row;
226	}
227
228	// Transform colspan/rowspan properties to TablePress equivalent (cell content).
229	foreach ( $table as $row_idx => $row ) {
230	foreach ( $row as $col_idx => $cell ) {
231	if ( 1 === $cell['rowspan'] && 1 === $cell['colspan'] ) {
232	continue;
233	}
234
235	if ( 1 < $cell['colspan'] ) {
236	for ( $i = 1; $i < $cell['colspan']; $i++ ) {
237	$table[ $row_idx ][ $col_idx + $i ]['val'] = '#colspan#';
238	}
239	}
240	if ( 1 < $cell['rowspan'] ) {
241	for ( $i = 1; $i < $cell['rowspan']; $i++ ) {
242	$table[ $row_idx + $i ][ $col_idx ]['val'] = '#rowspan#';
243	}
244	}
245
246	if ( 1 < $cell['rowspan'] && 1 < $cell['colspan'] ) {
247	for ( $i = 1; $i < $cell['rowspan']; $i++ ) {
248	for ( $j = 1; $j < $cell['colspan']; $j++ ) {
249	$table[ $row_idx + $i ][ $col_idx + $j ]['val'] = '#span#';
250	}
251	}
252	}
253	}
254	}
255
256	// Flatten value property to two-dimensional array.
257	foreach ( $table as &$row ) {
258	foreach ( $row as &$cell ) {
259	$cell = $cell['val'];
260	}
261	unset( $cell ); // Unset use-by-reference parameter of foreach loop.
262	}
263	unset( $row ); // Unset use-by-reference parameter of foreach loop.
264
265	$this->imported_table = array( 'data' => $table );
266	}
267
268	/**
269	* Import Microsoft Excel 2007-2019 data.
270	*
271	* @since 1.1.0
272	*/
273	protected function import_xlsx(): void {
274	TablePress::load_file( 'simplexlsx.class.php', 'libraries' );
275	$xlsx_file = \Shuchkin\SimpleXLSX::parse( $this->import_data, true );
276
277	if ( ! $xlsx_file ) {
278	$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . '</strong><br><br>' . \Shuchkin\SimpleXLSX::parseError() . '<br>';
279	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
280	}
281
282	$this->imported_table = array( 'data' => $xlsx_file->rows() );
283	}
284
285	/**
286	* Fixes the encoding to UTF-8 for the entire string that is to be imported.
287	*
288	* @since 1.0.0
289	*
290	* @link http://stevephillips.me/blog/dealing-php-and-character-encoding
291	*/
292	protected function fix_table_encoding(): void {
293	// Check and remove possible UTF-8 Byte-Order Mark (BOM).
294	$bom = pack( 'CCC', 0xef, 0xbb, 0xbf );
295	if ( str_starts_with( $this->import_data, $bom ) ) {
296	$this->import_data = substr( $this->import_data, 3 );
297	// If data has a BOM, it's UTF-8, so further checks unnecessary.
298	return;
299	}
300
301	// Require the iconv() function for the following checks.
302	if ( ! function_exists( 'iconv' ) ) {
303	return;
304	}
305
306	// Check for possible UTF-16 BOMs ("little endian" and "big endian") and try to convert the data to UTF-8.
307	if ( str_starts_with( $this->import_data, "\xFF\xFE" ) \|\| str_starts_with( $this->import_data, "\xFE\xFF" ) ) {
308	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
309	$data = @iconv( 'UTF-16', 'UTF-8', $this->import_data );
310	if ( false !== $data ) {
311	$this->import_data = $data;
312	return;
313	}
314	}
315
316	// Detect the character encoding and convert to UTF-8, if it's different.
317	if ( function_exists( 'mb_detect_encoding' ) ) {
318	$current_encoding = mb_detect_encoding( $this->import_data, 'ASCII, UTF-8, ISO-8859-1' );
319	if ( 'UTF-8' !== $current_encoding ) {
320	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
321	$data = @iconv( $current_encoding, 'UTF-8', $this->import_data ); // @phpstan-ignore argument.type
322	if ( false !== $data ) {
323	$this->import_data = $data;
324	return;
325	}
326	}
327	}
328	}
329
330	/**
331	* Replaces Windows line breaks \r\n with Unix line breaks \n.
332	*
333	* This function uses call by reference to save PHP memory on large arrays.
334	*
335	* @since 2.0.0
336	*
337	* @param array<int, array<int, string>> $an_array Array in which Windows line breaks should be replaced by Unix line breaks.
338	*/
339	protected function normalize_line_endings( array &$an_array ): void {
340	array_walk_recursive(
341	$an_array,
342	static function ( string &$cell_content, int $col_idx ): void {
343	$cell_content = str_replace( "\r\n", "\n", $cell_content );
344	},
345	);
346	}
347
348	} // class TablePress_Import_Legacy
349