| 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 |
* @package TablePress |
| 17 |
* @subpackage Export/Import |
| 18 |
* @author Tobias Bäthge |
| 19 |
* @since 1.0.0 |
| 20 |
*/ |
| 21 |
class TablePress_Import { |
| 22 |
|
| 23 |
/** |
| 24 |
* File/Data Formats that are available for import. |
| 25 |
* |
| 26 |
* @since 1.0.0 |
| 27 |
* @var array |
| 28 |
*/ |
| 29 |
public $import_formats = array(); |
| 30 |
|
| 31 |
/** |
| 32 |
* Whether ZIP archive support is available in the PHP installation on the server. |
| 33 |
* |
| 34 |
* @since 1.0.0 |
| 35 |
* @var bool |
| 36 |
*/ |
| 37 |
public $zip_support_available = false; |
| 38 |
|
| 39 |
/** |
| 40 |
* Whether HTML import support is available in the PHP installation on the server. |
| 41 |
* |
| 42 |
* @since 1.0.0 |
| 43 |
* @var bool |
| 44 |
*/ |
| 45 |
public $html_import_support_available = false; |
| 46 |
|
| 47 |
/** |
| 48 |
* Data to be imported. |
| 49 |
* |
| 50 |
* @since 1.0.0 |
| 51 |
* @var string |
| 52 |
*/ |
| 53 |
protected $import_data; |
| 54 |
|
| 55 |
/** |
| 56 |
* Imported table. |
| 57 |
* |
| 58 |
* @since 1.0.0 |
| 59 |
* @var array|false |
| 60 |
*/ |
| 61 |
protected $imported_table = false; |
| 62 |
|
| 63 |
/** |
| 64 |
* Initialize the Import class. |
| 65 |
* |
| 66 |
* @since 1.0.0 |
| 67 |
*/ |
| 68 |
public function __construct() { |
| 69 |
/** This filter is documented in the WordPress function unzip_file() in wp-admin/includes/file.php */ |
| 70 |
if ( class_exists( 'ZipArchive', false ) && apply_filters( 'unzip_file_use_ziparchive', true ) ) { |
| 71 |
$this->zip_support_available = true; |
| 72 |
} |
| 73 |
|
| 74 |
if ( class_exists( 'DOMDocument', false ) && function_exists( 'simplexml_import_dom' ) && function_exists( 'libxml_use_internal_errors' ) ) { |
| 75 |
$this->html_import_support_available = true; |
| 76 |
} |
| 77 |
|
| 78 |
// Initiate here, because function call not possible outside a class method. |
| 79 |
$this->import_formats = array(); |
| 80 |
$this->import_formats['csv'] = __( 'CSV - Character-Separated Values', 'tablepress' ); |
| 81 |
if ( $this->html_import_support_available ) { |
| 82 |
$this->import_formats['html'] = __( 'HTML - Hypertext Markup Language', 'tablepress' ); |
| 83 |
} |
| 84 |
$this->import_formats['json'] = __( 'JSON - JavaScript Object Notation', 'tablepress' ); |
| 85 |
$this->import_formats['xls'] = __( 'XLS - Microsoft Excel 97-2003 (experimental)', 'tablepress' ); |
| 86 |
$this->import_formats['xlsx'] = __( 'XLSX - Microsoft Excel 2007-2019 (experimental)', 'tablepress' ); |
| 87 |
} |
| 88 |
|
| 89 |
/** |
| 90 |
* Import a table. |
| 91 |
* |
| 92 |
* @since 1.0.0 |
| 93 |
* |
| 94 |
* @param string $format Import format. |
| 95 |
* @param string $data Data to import. |
| 96 |
* @return array|false Table array on success, false on error. |
| 97 |
*/ |
| 98 |
public function import_table( $format, $data ) { |
| 99 |
$this->import_data = apply_filters( 'tablepress_import_table_data', $data, $format ); |
| 100 |
|
| 101 |
if ( ! in_array( $format, array( 'xlsx', 'xls' ), true ) ) { |
| 102 |
$this->fix_table_encoding(); |
| 103 |
} |
| 104 |
|
| 105 |
switch ( $format ) { |
| 106 |
case 'csv': |
| 107 |
$this->import_csv(); |
| 108 |
break; |
| 109 |
case 'html': |
| 110 |
$this->import_html(); |
| 111 |
break; |
| 112 |
case 'json': |
| 113 |
$this->import_json(); |
| 114 |
break; |
| 115 |
case 'xlsx': |
| 116 |
$this->import_xlsx(); |
| 117 |
break; |
| 118 |
case 'xls': |
| 119 |
$this->import_xls(); |
| 120 |
break; |
| 121 |
default: |
| 122 |
return false; |
| 123 |
} |
| 124 |
|
| 125 |
return $this->imported_table; |
| 126 |
} |
| 127 |
|
| 128 |
/** |
| 129 |
* Import CSV data. |
| 130 |
* |
| 131 |
* @since 1.0.0 |
| 132 |
*/ |
| 133 |
protected function import_csv() { |
| 134 |
$csv_parser = TablePress::load_class( 'CSV_Parser', 'csv-parser.class.php', 'libraries' ); |
| 135 |
$csv_parser->load_data( $this->import_data ); |
| 136 |
$delimiter = $csv_parser->find_delimiter(); |
| 137 |
$data = $csv_parser->parse( $delimiter ); |
| 138 |
$this->imported_table = array( 'data' => $this->pad_array_to_max_cols( $data ) ); |
| 139 |
} |
| 140 |
|
| 141 |
/** |
| 142 |
* Import HTML data. |
| 143 |
* |
| 144 |
* @since 1.0.0 |
| 145 |
*/ |
| 146 |
protected function import_html() { |
| 147 |
if ( ! $this->html_import_support_available ) { |
| 148 |
return false; |
| 149 |
} |
| 150 |
|
| 151 |
if ( false === stripos( $this->import_data, '<table' ) || false === stripos( $this->import_data, '</table>' ) ) { |
| 152 |
$this->imported_table = false; |
| 153 |
return; |
| 154 |
} |
| 155 |
|
| 156 |
// Prepend XML declaration, for better encoding support. |
| 157 |
$full_html = '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>' . $this->import_data; |
| 158 |
if ( function_exists( 'libxml_disable_entity_loader' ) ) { |
| 159 |
/* |
| 160 |
* Don't expand external entities, see https://websec.io/2012/08/27/Preventing-XXE-in-PHP.html. |
| 161 |
* 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. |
| 162 |
*/ |
| 163 |
@libxml_disable_entity_loader( true ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged |
| 164 |
} |
| 165 |
// No warnings/errors raised, but stored internally. |
| 166 |
libxml_use_internal_errors( true ); |
| 167 |
$dom = new DOMDocument( '1.0', 'UTF-8' ); |
| 168 |
// No strict checking for invalid HTML. |
| 169 |
$dom->strictErrorChecking = false; // phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase |
| 170 |
$dom->loadHTML( $full_html ); |
| 171 |
if ( false === $dom ) { |
| 172 |
$this->imported_table = false; |
| 173 |
return; |
| 174 |
} |
| 175 |
$dom_tables = $dom->getElementsByTagName( 'table' ); |
| 176 |
if ( 0 === count( $dom_tables ) ) { |
| 177 |
$this->imported_table = false; |
| 178 |
return; |
| 179 |
} |
| 180 |
libxml_clear_errors(); // Clear errors so that we only catch those inside the table in the next line. |
| 181 |
$table = simplexml_import_dom( $dom_tables->item( 0 ) ); |
| 182 |
if ( false === $table ) { |
| 183 |
$this->imported_table = false; |
| 184 |
return; |
| 185 |
} |
| 186 |
|
| 187 |
$errors = libxml_get_errors(); |
| 188 |
libxml_clear_errors(); |
| 189 |
if ( ! empty( $errors ) ) { |
| 190 |
$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . '</strong><br /><br />'; |
| 191 |
foreach ( $errors as $error ) { |
| 192 |
switch ( $error->level ) { |
| 193 |
case LIBXML_ERR_WARNING: |
| 194 |
$output .= "Warning {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br />"; |
| 195 |
break; |
| 196 |
case LIBXML_ERR_ERROR: |
| 197 |
$output .= "Error {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br />"; |
| 198 |
break; |
| 199 |
case LIBXML_ERR_FATAL: |
| 200 |
$output .= "Fatal Error {$error->code}: {$error->message} in line {$error->line}, column {$error->column}<br />"; |
| 201 |
break; |
| 202 |
} |
| 203 |
} |
| 204 |
wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) ); |
| 205 |
} |
| 206 |
|
| 207 |
$html_table = array( |
| 208 |
'data' => array(), |
| 209 |
'options' => array(), |
| 210 |
); |
| 211 |
if ( isset( $table->thead ) ) { |
| 212 |
$html_table['data'] = array_merge( $html_table['data'], $this->_import_html_rows( $table->thead[0]->tr ) ); |
| 213 |
$html_table['options']['table_head'] = true; |
| 214 |
} |
| 215 |
if ( isset( $table->tbody ) ) { |
| 216 |
$html_table['data'] = array_merge( $html_table['data'], $this->_import_html_rows( $table->tbody[0]->tr ) ); |
| 217 |
} |
| 218 |
if ( isset( $table->tr ) ) { |
| 219 |
$html_table['data'] = array_merge( $html_table['data'], $this->_import_html_rows( $table->tr ) ); |
| 220 |
} |
| 221 |
if ( isset( $table->tfoot ) ) { |
| 222 |
$html_table['data'] = array_merge( $html_table['data'], $this->_import_html_rows( $table->tfoot[0]->tr ) ); |
| 223 |
$html_table['options']['table_foot'] = true; |
| 224 |
} |
| 225 |
|
| 226 |
$html_table['data'] = $this->pad_array_to_max_cols( $html_table['data'] ); |
| 227 |
$this->imported_table = $html_table; |
| 228 |
} |
| 229 |
|
| 230 |
/** |
| 231 |
* Helper for HTML import. |
| 232 |
* |
| 233 |
* @since 1.0.0 |
| 234 |
* |
| 235 |
* @param SimpleXMLElement $element XMLElement. |
| 236 |
* @return array SimpleXMLElement exported to an array. |
| 237 |
*/ |
| 238 |
protected function _import_html_rows( $element ) { |
| 239 |
$rows = array(); |
| 240 |
foreach ( $element as $row ) { |
| 241 |
$new_row = array(); |
| 242 |
foreach ( $row as $cell ) { |
| 243 |
// Get text between <td>...</td>, or <th>...</th>, possibly with attributes. |
| 244 |
if ( 1 === preg_match( '#<t(?:d|h).*?>(.*)</t(?:d|h)>#is', $cell->asXML(), $matches ) ) { |
| 245 |
/* |
| 246 |
* Decode HTML entities again, as there might be some left especially in attributes of HTML tags in the cells, |
| 247 |
* see https://secure.php.net/manual/en/simplexmlelement.asxml.php#107137. |
| 248 |
*/ |
| 249 |
$matches[1] = html_entity_decode( $matches[1], ENT_NOQUOTES, 'UTF-8' ); |
| 250 |
$new_row[] = $matches[1]; |
| 251 |
|
| 252 |
// Look for colspan and add correct number of cells. |
| 253 |
if ( 1 === preg_match( '#<t(?:d|h).*colspan="(\d+)".*?>#is', $cell->asXml(), $matches ) ) { |
| 254 |
for ( $i = 1; $i < (int) $matches[1]; $i++ ) { |
| 255 |
$new_row[] = '#colspan#'; |
| 256 |
} |
| 257 |
} |
| 258 |
} else { |
| 259 |
$new_row[] = ''; |
| 260 |
} |
| 261 |
} |
| 262 |
$rows[] = $new_row; |
| 263 |
} |
| 264 |
return $rows; |
| 265 |
} |
| 266 |
|
| 267 |
/** |
| 268 |
* Import JSON data. |
| 269 |
* |
| 270 |
* @since 1.0.0 |
| 271 |
*/ |
| 272 |
protected function import_json() { |
| 273 |
$json_table = json_decode( $this->import_data, true ); |
| 274 |
|
| 275 |
// Check if JSON could be decoded. |
| 276 |
if ( is_null( $json_table ) ) { |
| 277 |
// If possible, try to find out what error prevented the JSON from being decoded. |
| 278 |
$json_error = 'The error could not be determined.'; |
| 279 |
$json_error_msg = json_last_error_msg(); |
| 280 |
if ( false !== $json_error_msg ) { |
| 281 |
$json_error = $json_error_msg; |
| 282 |
} |
| 283 |
$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . "</strong><br /><br />JSON error: {$json_error}<br />"; |
| 284 |
wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) ); |
| 285 |
} |
| 286 |
|
| 287 |
// Specifically cast to an array again. |
| 288 |
$json_table = (array) $json_table; |
| 289 |
|
| 290 |
if ( isset( $json_table['data'] ) ) { |
| 291 |
// JSON data contained a full export. |
| 292 |
$table = $json_table; |
| 293 |
} else { |
| 294 |
// JSON data contained only the data of a table, but no options. |
| 295 |
$table = array( 'data' => array() ); |
| 296 |
foreach ( $json_table as $row ) { |
| 297 |
$table['data'][] = array_values( (array) $row ); |
| 298 |
} |
| 299 |
} |
| 300 |
|
| 301 |
$table['data'] = $this->pad_array_to_max_cols( $table['data'] ); |
| 302 |
$this->imported_table = $table; |
| 303 |
} |
| 304 |
|
| 305 |
/** |
| 306 |
* Import Microsoft Excel 97-2003 data. |
| 307 |
* |
| 308 |
* @since 1.1.0 |
| 309 |
*/ |
| 310 |
protected function import_xls() { |
| 311 |
$excel_reader = TablePress::load_class( 'Spreadsheet_Excel_Reader', 'excel-reader.class.php', 'libraries', $this->import_data ); |
| 312 |
|
| 313 |
// Loop through Excel file and retrieve value and colspan/rowspan properties for each cell. |
| 314 |
$sheet = 0; // 0 means first sheet of the Workbook |
| 315 |
$table = array(); |
| 316 |
for ( $row = 1; $row <= $excel_reader->rowcount( $sheet ); $row++ ) { |
| 317 |
$table_row = array(); |
| 318 |
for ( $col = 1; $col <= $excel_reader->colcount( $sheet ); $col++ ) { |
| 319 |
$cell = array(); |
| 320 |
$cell['rowspan'] = $excel_reader->rowspan( $row, $col, $sheet ); |
| 321 |
$cell['colspan'] = $excel_reader->colspan( $row, $col, $sheet ); |
| 322 |
$cell['val'] = $excel_reader->val( $row, $col, $sheet ); |
| 323 |
$table_row[] = $cell; |
| 324 |
} |
| 325 |
$table[] = $table_row; |
| 326 |
} |
| 327 |
|
| 328 |
// Transform colspan/rowspan properties to TablePress equivalent (cell content). |
| 329 |
foreach ( $table as $row_idx => $row ) { |
| 330 |
foreach ( $row as $col_idx => $cell ) { |
| 331 |
if ( 1 === $cell['rowspan'] && 1 === $cell['colspan'] ) { |
| 332 |
continue; |
| 333 |
} |
| 334 |
|
| 335 |
if ( 1 < $cell['colspan'] ) { |
| 336 |
for ( $i = 1; $i < $cell['colspan']; $i++ ) { |
| 337 |
$table[ $row_idx ][ $col_idx + $i ]['val'] = '#colspan#'; |
| 338 |
} |
| 339 |
} |
| 340 |
if ( 1 < $cell['rowspan'] ) { |
| 341 |
for ( $i = 1; $i < $cell['rowspan']; $i++ ) { |
| 342 |
$table[ $row_idx + $i ][ $col_idx ]['val'] = '#rowspan#'; |
| 343 |
} |
| 344 |
} |
| 345 |
|
| 346 |
if ( 1 < $cell['rowspan'] && 1 < $cell['colspan'] ) { |
| 347 |
for ( $i = 1; $i < $cell['rowspan']; $i++ ) { |
| 348 |
for ( $j = 1; $j < $cell['colspan']; $j++ ) { |
| 349 |
$table[ $row_idx + $i ][ $col_idx + $j ]['val'] = '#span#'; |
| 350 |
} |
| 351 |
} |
| 352 |
} |
| 353 |
} |
| 354 |
} |
| 355 |
|
| 356 |
// Flatten value property to two-dimensional array. |
| 357 |
$result_table = array(); |
| 358 |
foreach ( $table as $row_idx => $row ) { |
| 359 |
$table_row = array(); |
| 360 |
foreach ( $row as $col_idx => $cell ) { |
| 361 |
$table_row[] = (string) $cell['val']; |
| 362 |
} |
| 363 |
$result_table[] = $table_row; |
| 364 |
} |
| 365 |
|
| 366 |
$this->imported_table = array( 'data' => $this->pad_array_to_max_cols( $result_table ) ); |
| 367 |
} |
| 368 |
|
| 369 |
/** |
| 370 |
* Import Microsoft Excel 2007-2019 data. |
| 371 |
* |
| 372 |
* @since 1.1.0 |
| 373 |
*/ |
| 374 |
protected function import_xlsx() { |
| 375 |
TablePress::load_file( 'simplexlsx.class.php', 'libraries' ); |
| 376 |
$xlsx_file = SimpleXLSX::parse( $this->import_data, true ); |
| 377 |
|
| 378 |
if ( ! $xlsx_file ) { |
| 379 |
$output = '<strong>' . __( 'The imported file contains errors:', 'tablepress' ) . '</strong><br /><br />' . SimpleXLSX::parseError() . '<br />'; |
| 380 |
wp_die( $output, 'Import Error', array( 'response' => 200, 'back_link' => true ) ); |
| 381 |
} |
| 382 |
|
| 383 |
$this->imported_table = array( 'data' => $xlsx_file->rows() ); |
| 384 |
} |
| 385 |
|
| 386 |
/** |
| 387 |
* Make sure array is rectangular with $max_cols columns in every row. |
| 388 |
* |
| 389 |
* @since 1.0.0 |
| 390 |
* |
| 391 |
* @param array $array Two-dimensional array to be padded. |
| 392 |
* @return array Padded array. |
| 393 |
*/ |
| 394 |
public function pad_array_to_max_cols( array $array ) { |
| 395 |
$rows = count( $array ); |
| 396 |
$rows = ( $rows > 0 ) ? $rows : 1; |
| 397 |
$max_columns = $this->count_max_columns( $array ); |
| 398 |
$max_columns = ( $max_columns > 0 ) ? $max_columns : 1; |
| 399 |
// array_map wants arrays as additional parameters, so we create one with the max_columns to pad to and one with the value to use (empty string). |
| 400 |
$max_columns_array = array_fill( 1, $rows, $max_columns ); |
| 401 |
$pad_values_array = array_fill( 1, $rows, '' ); |
| 402 |
return array_map( 'array_pad', $array, $max_columns_array, $pad_values_array ); |
| 403 |
} |
| 404 |
|
| 405 |
/** |
| 406 |
* Get the highest number of columns in the rows. |
| 407 |
* |
| 408 |
* @since 1.0.0 |
| 409 |
* |
| 410 |
* @param array $array Two-dimensional array. |
| 411 |
* @return int Highest number of columns in the rows of the array. |
| 412 |
*/ |
| 413 |
protected function count_max_columns( array $array ) { |
| 414 |
$max_columns = 0; |
| 415 |
foreach ( $array as $row_idx => $row ) { |
| 416 |
$num_columns = count( $row ); |
| 417 |
$max_columns = max( $num_columns, $max_columns ); |
| 418 |
} |
| 419 |
return $max_columns; |
| 420 |
} |
| 421 |
|
| 422 |
/** |
| 423 |
* Fixes the encoding to UTF-8 for the entire string that is to be imported. |
| 424 |
* |
| 425 |
* @since 1.0.0 |
| 426 |
* |
| 427 |
* @link http://stevephillips.me/blog/dealing-php-and-character-encoding |
| 428 |
*/ |
| 429 |
protected function fix_table_encoding() { |
| 430 |
// Check and remove possible UTF-8 Byte-Order Mark (BOM). |
| 431 |
$bom = pack( 'CCC', 0xef, 0xbb, 0xbf ); |
| 432 |
if ( 0 === strncmp( $this->import_data, $bom, 3 ) ) { |
| 433 |
$this->import_data = substr( $this->import_data, 3 ); |
| 434 |
// If data has a BOM, it's UTF-8, so further checks unnecessary. |
| 435 |
return; |
| 436 |
} |
| 437 |
|
| 438 |
// Require the iconv() function for the following checks. |
| 439 |
if ( ! function_exists( 'iconv' ) ) { |
| 440 |
return; |
| 441 |
} |
| 442 |
|
| 443 |
// Check for possible UTF-16 BOMs ("little endian" and "big endian") and try to convert the data to UTF-8. |
| 444 |
if ( "\xFF\xFE" === substr( $this->import_data, 0, 2 ) || "\xFE\xFF" === substr( $this->import_data, 0, 2 ) ) { |
| 445 |
// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged |
| 446 |
$data = @iconv( 'UTF-16', 'UTF-8', $this->import_data ); |
| 447 |
if ( false !== $data ) { |
| 448 |
$this->import_data = $data; |
| 449 |
return; |
| 450 |
} |
| 451 |
} |
| 452 |
|
| 453 |
// Detect the character encoding and convert to UTF-8, if it's different. |
| 454 |
if ( function_exists( 'mb_detect_encoding' ) ) { |
| 455 |
$current_encoding = mb_detect_encoding( $this->import_data, 'ASCII, UTF-8, ISO-8859-1' ); |
| 456 |
if ( 'UTF-8' !== $current_encoding ) { |
| 457 |
// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged |
| 458 |
$data = @iconv( $current_encoding, 'UTF-8', $this->import_data ); |
| 459 |
if ( false !== $data ) { |
| 460 |
$this->import_data = $data; |
| 461 |
return; |
| 462 |
} |
| 463 |
} |
| 464 |
} |
| 465 |
} |
| 466 |
|
| 467 |
} // class TablePress_Import |
| 468 |
|
