tablepress / classes / class-import-legacy.php

class-import-legacy.php in TablePress – Tables in WordPress made easy 3.3.3, at classes/class-import-legacy.php

347 lines 9.4 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	* The constructor also adds 'html', if the PHP prerequisites are met.
28	*
29	* @since 1.0.0
30	* @var string[]
31	*/
32	public array $import_formats = array( 'csv', 'json', 'xls', 'xlsx' );
33
34	/**
35	* Whether HTML import support is available in the PHP installation on the server.
36	*
37	* @since 1.0.0
38	*/
39	public bool $html_import_support_available = false;
40
41	/**
42	* Data to be imported.
43	*
44	* @since 1.0.0
45	*/
46	protected string $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	if ( $this->html_import_support_available ) {
67	$this->import_formats[] = 'html';
68	}
69	}
70
71	/**
72	* Import a table.
73	*
74	* @since 1.0.0
75	*
76	* @param string $format Import format.
77	* @param string $data Data to import.
78	* @return array<string, mixed>\|false Table array on success, false on error.
79	*/
80	public function import_table( string $format, string $data ) /* : array\|false */ {
81	/**
82	* Filters the data that is to be imported.
83	*
84	* @since 1.8.1
85	*
86	* @param string $data Data to import.
87	* @param string $format Import format.
88	*/
89	$this->import_data = apply_filters( 'tablepress_import_table_data', $data, $format );
90
91	if ( ! in_array( $format, array( 'xlsx', 'xls' ), true ) ) {
92	$this->fix_table_encoding();
93	}
94
95	switch ( $format ) {
96	case 'csv':
97	$this->import_csv();
98	break;
99	case 'html':
100	$this->import_html();
101	break;
102	case 'json':
103	$this->import_json();
104	break;
105	case 'xlsx':
106	$this->import_xlsx();
107	break;
108	case 'xls':
109	$this->import_xls();
110	break;
111	default:
112	return false;
113	}
114
115	if ( false === $this->imported_table ) {
116	return false;
117	}
118
119	// Make sure that cells are stored as strings.
120	array_walk_recursive(
121	$this->imported_table['data'],
122	static function ( /* string\|int\|float\|bool\|null */ &$cell_content, int $col_idx ): void {
123	$cell_content = (string) $cell_content;
124	},
125	);
126
127	return $this->imported_table;
128	}
129
130	/**
131	* Import CSV data.
132	*
133	* @since 1.0.0
134	*/
135	protected function import_csv(): void {
136	$csv_parser = TablePress::load_class( 'CSV_Parser', 'csv-parser.class.php', 'libraries' );
137	$csv_parser->load_data( $this->import_data );
138	$delimiter = $csv_parser->find_delimiter();
139	$data = $csv_parser->parse( $delimiter );
140	$this->pad_array_to_max_cols( $data );
141	$this->normalize_line_endings( $data );
142	$this->imported_table = array( 'data' => $data );
143	}
144
145	/**
146	* Import HTML data.
147	*
148	* @since 1.0.0
149	*/
150	protected function import_html(): void {
151	if ( ! $this->html_import_support_available ) {
152	return;
153	}
154
155	TablePress::load_file( 'html-parser.class.php', 'libraries' );
156	$table = HTML_Parser::parse( $this->import_data );
157
158	if ( is_wp_error( $table ) ) {
159	$this->imported_table = false;
160	return;
161	}
162
163	$this->pad_array_to_max_cols( $table['data'] );
164	$this->normalize_line_endings( $table['data'] );
165	$this->imported_table = $table;
166	}
167
168	/**
169	* Import JSON data.
170	*
171	* @since 1.0.0
172	*/
173	protected function import_json(): void {
174	$json_table = json_decode( $this->import_data, true );
175
176	// Check if JSON could be decoded.
177	if ( is_null( $json_table ) ) {
178	$json_error = json_last_error_msg();
179	$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . "</strong><br><br>JSON error: {$json_error}<br>";
180	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
181	}
182
183	// Specifically cast to an array again.
184	$json_table = (array) $json_table;
185
186	if ( isset( $json_table['data'] ) ) {
187	// JSON data contained a full export.
188	$table = $json_table;
189	} else {
190	// JSON data contained only the data of a table, but no options.
191	$table = array( 'data' => array() );
192	foreach ( $json_table as $row ) {
193	$table['data'][] = array_values( (array) $row );
194	}
195	}
196
197	$this->pad_array_to_max_cols( $table['data'] );
198	$this->imported_table = $table;
199	}
200
201	/**
202	* Import Microsoft Excel 97-2003 data.
203	*
204	* @since 1.1.0
205	*/
206	protected function import_xls(): void {
207	$excel_reader = TablePress::load_class( 'Spreadsheet_Excel_Reader', 'excel-reader.class.php', 'libraries', $this->import_data );
208
209	// Loop through Excel file and retrieve value and colspan/rowspan properties for each cell.
210	$sheet = 0; // 0 means first sheet of the Workbook
211	$table = array();
212	$num_rows = $excel_reader->rowcount( $sheet );
213	$num_columns = $excel_reader->colcount( $sheet );
214	for ( $row = 1; $row <= $num_rows; $row++ ) {
215	$table_row = array();
216	for ( $column = 1; $column <= $num_columns; $column++ ) {
217	$cell = array();
218	$cell['rowspan'] = $excel_reader->rowspan( $row, $column, $sheet );
219	$cell['colspan'] = $excel_reader->colspan( $row, $column, $sheet );
220	$cell['val'] = $excel_reader->val( $row, $column, $sheet );
221	$table_row[] = $cell;
222	}
223	$table[] = $table_row;
224	}
225
226	// Transform colspan/rowspan properties to TablePress equivalent (cell content).
227	foreach ( $table as $row_idx => $row ) {
228	foreach ( $row as $col_idx => $cell ) {
229	if ( 1 === $cell['rowspan'] && 1 === $cell['colspan'] ) {
230	continue;
231	}
232
233	if ( 1 < $cell['colspan'] ) {
234	for ( $i = 1; $i < $cell['colspan']; $i++ ) {
235	$table[ $row_idx ][ $col_idx + $i ]['val'] = '#colspan#';
236	}
237	}
238	if ( 1 < $cell['rowspan'] ) {
239	for ( $i = 1; $i < $cell['rowspan']; $i++ ) {
240	$table[ $row_idx + $i ][ $col_idx ]['val'] = '#rowspan#';
241	}
242	}
243
244	if ( 1 < $cell['rowspan'] && 1 < $cell['colspan'] ) {
245	for ( $i = 1; $i < $cell['rowspan']; $i++ ) {
246	for ( $j = 1; $j < $cell['colspan']; $j++ ) {
247	$table[ $row_idx + $i ][ $col_idx + $j ]['val'] = '#span#';
248	}
249	}
250	}
251	}
252	}
253
254	// Flatten value property to two-dimensional array.
255	foreach ( $table as &$row ) {
256	foreach ( $row as &$cell ) {
257	$cell = $cell['val'];
258	}
259	unset( $cell ); // Unset use-by-reference parameter of foreach loop.
260	}
261	unset( $row ); // Unset use-by-reference parameter of foreach loop.
262
263	$this->imported_table = array( 'data' => $table );
264	}
265
266	/**
267	* Import Microsoft Excel 2007-2019 data.
268	*
269	* @since 1.1.0
270	*/
271	protected function import_xlsx(): void {
272	TablePress::load_file( 'simplexlsx.class.php', 'libraries' );
273	$xlsx_file = \Shuchkin\SimpleXLSX::parse( $this->import_data, true );
274
275	if ( ! $xlsx_file ) {
276	$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . '</strong><br><br>' . \Shuchkin\SimpleXLSX::parseError() . '<br>';
277	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
278	}
279
280	$this->imported_table = array( 'data' => $xlsx_file->rows() );
281	}
282
283	/**
284	* Fixes the encoding to UTF-8 for the entire string that is to be imported.
285	*
286	* @since 1.0.0
287	*
288	* @link http://stevephillips.me/blog/dealing-php-and-character-encoding
289	*/
290	protected function fix_table_encoding(): void {
291	// Check and remove possible UTF-8 Byte-Order Mark (BOM).
292	$bom = pack( 'CCC', 0xef, 0xbb, 0xbf );
293	if ( str_starts_with( $this->import_data, $bom ) ) {
294	$this->import_data = substr( $this->import_data, 3 );
295	// If data has a BOM, it's UTF-8, so further checks unnecessary.
296	return;
297	}
298
299	// Require the iconv() function for the following checks.
300	if ( ! function_exists( 'iconv' ) ) {
301	return;
302	}
303
304	// Check for possible UTF-16 BOMs ("little endian" and "big endian") and try to convert the data to UTF-8.
305	if ( str_starts_with( $this->import_data, "\xFF\xFE" ) \|\| str_starts_with( $this->import_data, "\xFE\xFF" ) ) {
306	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
307	$data = @iconv( 'UTF-16', 'UTF-8', $this->import_data );
308	if ( false !== $data ) {
309	$this->import_data = $data;
310	return;
311	}
312	}
313
314	// Detect the character encoding and convert to UTF-8, if it's different.
315	if ( function_exists( 'mb_detect_encoding' ) ) {
316	$current_encoding = mb_detect_encoding( $this->import_data, 'ASCII, UTF-8, ISO-8859-1' );
317	if ( 'UTF-8' !== $current_encoding ) {
318	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
319	$data = @iconv( $current_encoding, 'UTF-8', $this->import_data ); // @phpstan-ignore argument.type
320	if ( false !== $data ) {
321	$this->import_data = $data;
322	return;
323	}
324	}
325	}
326	}
327
328	/**
329	* Replaces Windows line breaks \r\n with Unix line breaks \n.
330	*
331	* This function uses call by reference to save PHP memory on large arrays.
332	*
333	* @since 2.0.0
334	*
335	* @param array<int, array<int, string>> $an_array Array in which Windows line breaks should be replaced by Unix line breaks.
336	*/
337	protected function normalize_line_endings( array &$an_array ): void {
338	array_walk_recursive(
339	$an_array,
340	static function ( string &$cell_content, int $col_idx ): void {
341	$cell_content = str_replace( "\r\n", "\n", $cell_content );
342	},
343	);
344	}
345
346	} // class TablePress_Import_Legacy
347