tablepress / classes / class-import-legacy.php

class-import-legacy.php in TablePress – Tables in WordPress made easy 3.0, at classes/class-import-legacy.php

351 lines 9.9 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 array $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	*/
37	public bool $html_import_support_available = false;
38
39	/**
40	* Data to be imported.
41	*
42	* @since 1.0.0
43	*/
44	protected string $import_data;
45
46	/**
47	* Imported table.
48	*
49	* @since 1.0.0
50	* @var array<string, mixed>\|false
51	*/
52	protected $imported_table = false;
53
54	/**
55	* Initialize the Import class.
56	*
57	* @since 1.0.0
58	*/
59	public function __construct() {
60	if ( class_exists( 'DOMDocument', false ) && function_exists( 'simplexml_import_dom' ) && function_exists( 'libxml_use_internal_errors' ) ) {
61	$this->html_import_support_available = true;
62	}
63
64	// Initiate here, because function call not possible outside a class method.
65	$this->import_formats = array();
66	$this->import_formats['csv'] = __( 'CSV - Character-Separated Values', 'tablepress' );
67	if ( $this->html_import_support_available ) {
68	$this->import_formats['html'] = __( 'HTML - Hypertext Markup Language', 'tablepress' );
69	}
70	$this->import_formats['json'] = __( 'JSON - JavaScript Object Notation', 'tablepress' );
71	$this->import_formats['xls'] = __( 'XLS - Microsoft Excel 97-2003 (experimental)', 'tablepress' );
72	$this->import_formats['xlsx'] = __( 'XLSX - Microsoft Excel 2007-2019 (experimental)', 'tablepress' );
73	}
74
75	/**
76	* Import a table.
77	*
78	* @since 1.0.0
79	*
80	* @param string $format Import format.
81	* @param string $data Data to import.
82	* @return array<string, mixed>\|false Table array on success, false on error.
83	*/
84	public function import_table( string $format, string $data ) /* : array\|false */ {
85	/**
86	* Filters the data that is to be imported.
87	*
88	* @since 1.8.1
89	*
90	* @param string $data Data to import.
91	* @param string $format Import format.
92	*/
93	$this->import_data = apply_filters( 'tablepress_import_table_data', $data, $format );
94
95	if ( ! in_array( $format, array( 'xlsx', 'xls' ), true ) ) {
96	$this->fix_table_encoding();
97	}
98
99	switch ( $format ) {
100	case 'csv':
101	$this->import_csv();
102	break;
103	case 'html':
104	$this->import_html();
105	break;
106	case 'json':
107	$this->import_json();
108	break;
109	case 'xlsx':
110	$this->import_xlsx();
111	break;
112	case 'xls':
113	$this->import_xls();
114	break;
115	default:
116	return false;
117	}
118
119	if ( false === $this->imported_table ) {
120	return false;
121	}
122
123	// Make sure that cells are stored as strings.
124	array_walk_recursive(
125	$this->imported_table['data'],
126	static function ( /* string\|int\|float\|bool\|null */ &$cell_content, int $col_idx ): void {
127	$cell_content = (string) $cell_content;
128	},
129	);
130
131	return $this->imported_table;
132	}
133
134	/**
135	* Import CSV data.
136	*
137	* @since 1.0.0
138	*/
139	protected function import_csv(): void {
140	$csv_parser = TablePress::load_class( 'CSV_Parser', 'csv-parser.class.php', 'libraries' );
141	$csv_parser->load_data( $this->import_data );
142	$delimiter = $csv_parser->find_delimiter();
143	$data = $csv_parser->parse( $delimiter );
144	$this->pad_array_to_max_cols( $data );
145	$this->normalize_line_endings( $data );
146	$this->imported_table = array( 'data' => $data );
147	}
148
149	/**
150	* Import HTML data.
151	*
152	* @since 1.0.0
153	*/
154	protected function import_html(): void {
155	if ( ! $this->html_import_support_available ) {
156	return;
157	}
158
159	TablePress::load_file( 'html-parser.class.php', 'libraries' );
160	$table = HTML_Parser::parse( $this->import_data );
161
162	if ( is_wp_error( $table ) ) {
163	$this->imported_table = false;
164	return;
165	}
166
167	$this->pad_array_to_max_cols( $table['data'] );
168	$this->normalize_line_endings( $table['data'] );
169	$this->imported_table = $table;
170	}
171
172	/**
173	* Import JSON data.
174	*
175	* @since 1.0.0
176	*/
177	protected function import_json(): void {
178	$json_table = json_decode( $this->import_data, true );
179
180	// Check if JSON could be decoded.
181	if ( is_null( $json_table ) ) {
182	$json_error = json_last_error_msg();
183	$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . "</strong><br><br>JSON error: {$json_error}<br>";
184	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
185	}
186
187	// Specifically cast to an array again.
188	$json_table = (array) $json_table;
189
190	if ( isset( $json_table['data'] ) ) {
191	// JSON data contained a full export.
192	$table = $json_table;
193	} else {
194	// JSON data contained only the data of a table, but no options.
195	$table = array( 'data' => array() );
196	foreach ( $json_table as $row ) {
197	$table['data'][] = array_values( (array) $row );
198	}
199	}
200
201	$this->pad_array_to_max_cols( $table['data'] );
202	$this->imported_table = $table;
203	}
204
205	/**
206	* Import Microsoft Excel 97-2003 data.
207	*
208	* @since 1.1.0
209	*/
210	protected function import_xls(): void {
211	$excel_reader = TablePress::load_class( 'Spreadsheet_Excel_Reader', 'excel-reader.class.php', 'libraries', $this->import_data );
212
213	// Loop through Excel file and retrieve value and colspan/rowspan properties for each cell.
214	$sheet = 0; // 0 means first sheet of the Workbook
215	$table = array();
216	$num_rows = $excel_reader->rowcount( $sheet );
217	$num_columns = $excel_reader->colcount( $sheet );
218	for ( $row = 1; $row <= $num_rows; $row++ ) {
219	$table_row = array();
220	for ( $column = 1; $column <= $num_columns; $column++ ) {
221	$cell = array();
222	$cell['rowspan'] = $excel_reader->rowspan( $row, $column, $sheet );
223	$cell['colspan'] = $excel_reader->colspan( $row, $column, $sheet );
224	$cell['val'] = $excel_reader->val( $row, $column, $sheet );
225	$table_row[] = $cell;
226	}
227	$table[] = $table_row;
228	}
229
230	// Transform colspan/rowspan properties to TablePress equivalent (cell content).
231	foreach ( $table as $row_idx => $row ) {
232	foreach ( $row as $col_idx => $cell ) {
233	if ( 1 === $cell['rowspan'] && 1 === $cell['colspan'] ) {
234	continue;
235	}
236
237	if ( 1 < $cell['colspan'] ) {
238	for ( $i = 1; $i < $cell['colspan']; $i++ ) {
239	$table[ $row_idx ][ $col_idx + $i ]['val'] = '#colspan#';
240	}
241	}
242	if ( 1 < $cell['rowspan'] ) {
243	for ( $i = 1; $i < $cell['rowspan']; $i++ ) {
244	$table[ $row_idx + $i ][ $col_idx ]['val'] = '#rowspan#';
245	}
246	}
247
248	if ( 1 < $cell['rowspan'] && 1 < $cell['colspan'] ) {
249	for ( $i = 1; $i < $cell['rowspan']; $i++ ) {
250	for ( $j = 1; $j < $cell['colspan']; $j++ ) {
251	$table[ $row_idx + $i ][ $col_idx + $j ]['val'] = '#span#';
252	}
253	}
254	}
255	}
256	}
257
258	// Flatten value property to two-dimensional array.
259	foreach ( $table as &$row ) {
260	foreach ( $row as &$cell ) {
261	$cell = $cell['val'];
262	}
263	unset( $cell ); // Unset use-by-reference parameter of foreach loop.
264	}
265	unset( $row ); // Unset use-by-reference parameter of foreach loop.
266
267	$this->imported_table = array( 'data' => $table );
268	}
269
270	/**
271	* Import Microsoft Excel 2007-2019 data.
272	*
273	* @since 1.1.0
274	*/
275	protected function import_xlsx(): void {
276	TablePress::load_file( 'simplexlsx.class.php', 'libraries' );
277	$xlsx_file = \Shuchkin\SimpleXLSX::parse( $this->import_data, true );
278
279	if ( ! $xlsx_file ) {
280	$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . '</strong><br><br>' . \Shuchkin\SimpleXLSX::parseError() . '<br>';
281	wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) );
282	}
283
284	$this->imported_table = array( 'data' => $xlsx_file->rows() );
285	}
286
287	/**
288	* Fixes the encoding to UTF-8 for the entire string that is to be imported.
289	*
290	* @since 1.0.0
291	*
292	* @link http://stevephillips.me/blog/dealing-php-and-character-encoding
293	*/
294	protected function fix_table_encoding(): void {
295	// Check and remove possible UTF-8 Byte-Order Mark (BOM).
296	$bom = pack( 'CCC', 0xef, 0xbb, 0xbf );
297	if ( str_starts_with( $this->import_data, $bom ) ) {
298	$this->import_data = substr( $this->import_data, 3 );
299	// If data has a BOM, it's UTF-8, so further checks unnecessary.
300	return;
301	}
302
303	// Require the iconv() function for the following checks.
304	if ( ! function_exists( 'iconv' ) ) {
305	return;
306	}
307
308	// Check for possible UTF-16 BOMs ("little endian" and "big endian") and try to convert the data to UTF-8.
309	if ( str_starts_with( $this->import_data, "\xFF\xFE" ) \|\| str_starts_with( $this->import_data, "\xFE\xFF" ) ) {
310	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
311	$data = @iconv( 'UTF-16', 'UTF-8', $this->import_data );
312	if ( false !== $data ) {
313	$this->import_data = $data;
314	return;
315	}
316	}
317
318	// Detect the character encoding and convert to UTF-8, if it's different.
319	if ( function_exists( 'mb_detect_encoding' ) ) {
320	$current_encoding = mb_detect_encoding( $this->import_data, 'ASCII, UTF-8, ISO-8859-1' );
321	if ( 'UTF-8' !== $current_encoding ) {
322	// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
323	$data = @iconv( $current_encoding, 'UTF-8', $this->import_data ); // @phpstan-ignore argument.type
324	if ( false !== $data ) {
325	$this->import_data = $data;
326	return;
327	}
328	}
329	}
330	}
331
332	/**
333	* Replaces Windows line breaks \r\n with Unix line breaks \n.
334	*
335	* This function uses call by reference to save PHP memory on large arrays.
336	*
337	* @since 2.0.0
338	*
339	* @param array<int, array<int, string>> $an_array Array in which Windows line breaks should be replaced by Unix line breaks.
340	*/
341	protected function normalize_line_endings( array &$an_array ): void {
342	array_walk_recursive(
343	$an_array,
344	static function ( string &$cell_content, int $col_idx ): void {
345	$cell_content = str_replace( "\r\n", "\n", $cell_content );
346	},
347	);
348	}
349
350	} // class TablePress_Import_Legacy
351