← All changes | classes/class-import-phpspreadsheet.php +77 -38 2.2.1 → 3.4 View file →

		@@ -7,8 +7,12 @@
7	7	* @author Tobias Bäthge
8	8	* @since 2.0.0
9	9	*/
10	10
	11	+declare(strict_types=1);
	12	+
	13	+use TablePress\Import\File;
	14	+
11	15	// Prohibit direct script loading.
12	16	defined( 'ABSPATH' ) || die( 'No direct script access allowed!' );
13	17
14	18	/**
		@@ -26,10 +30,10 @@
26	30	*
27	31	* @since 2.0.0
28	32	*/
29	33	public function __construct() {
30		- // Load PHPSpreadsheet via the Composer autoloading mechanism.
31		- TablePress::load_file( 'autoload.php', 'libraries' );
	34	+ // Load PHPSpreadsheet via its autoloading mechanism.
	35	+ TablePress::load_file( 'autoload.php', 'libraries/vendor' );
32	36	}
33	37
34	38	/**
35	39	* Imports a table from a file.
		@@ -35,15 +39,15 @@
35	39	* Imports a table from a file.
36	40	*
37	41	* @since 2.0.0
38	42	*
39		- * @param array<string, mixed> $file File to import.
	43	+ * @param File $file File to import.
40	44	* @return array<string, mixed>|WP_Error Table array on success, WP_Error on error.
41	45	*/
42		- public function import_table( array $file ) /* : array|WP_Error */ {
43		- $data = file_get_contents( $file['location'] );
	46	+ public function import_table( File $file ) /* : array|WP_Error */ {
	47	+ $data = file_get_contents( $file->location );
44	48	if ( false === $data ) {
45		- return new WP_Error( 'table_import_phpspreadsheet_data_read', '', $file['location'] );
	49	+ return new WP_Error( 'table_import_phpspreadsheet_data_read', '', $file->location );
46	50	}
47	51
48	52	// Remove a possible UTF-8 Byte-Order Mark (BOM).
49	53	$bom = pack( 'CCC', 0xef, 0xbb, 0xbf );
		@@ -51,18 +55,18 @@
51	55	$data = substr( $data, 3 );
52	56	}
53	57
54	58	if ( '' === $data ) {
55		- return new WP_Error( 'table_import_phpspreadsheet_data_empty', '', $file['location'] );
	59	+ return new WP_Error( 'table_import_phpspreadsheet_data_empty', '', $file->location );
56	60	}
57	61
58	62	$table = $this->_maybe_import_json( $data );
59		- if ( false !== $table ) {
	63	+ if ( is_array( $table ) ) {
60	64	return $table;
61	65	}
62	66
63	67	$table = $this->_maybe_import_html( $data );
64		- if ( false !== $table ) {
	68	+ if ( is_array( $table ) ) {
65	69	return $table;
66	70	}
67	71
68	72	return $this->_import_phpspreadsheet( $file );
		@@ -76,12 +80,14 @@
76	80	* @param string $data Data to import.
77	81	* @return array<string, mixed>|false Table array on success, false if the file is not a JSON file.
78	82	*/
79	83	protected function _maybe_import_json( string $data ) /* : array|false */ {
80		- // If the first non-whitespace character is not a { or [, the file is not a supported JSON file.
81		- $data = ltrim( $data );
	84	+ $data = trim( $data );
	85	+
	86	+ // If the file does not begin / end with [ / ] or { / }, it's not a supported JSON file.
82	87	$first_character = $data[0];
83		- if ( '{' !== $first_character && '[' !== $first_character ) {
	88	+ $last_character = $data[-1];
	89	+ if ( ! ( '[' === $first_character && ']' === $last_character ) && ! ( '{' === $first_character && '}' === $last_character ) ) {
84	90	return false;
85	91	}
86	92
87	93	$json_table = json_decode( $data, true );
		@@ -125,9 +131,9 @@
125	131	*
126	132	* @since 2.0.0
127	133	*
128	134	* @param string $data Data to import.
129		- * @return array<string, mixed>|false Table array on success, false if the file is not an HTML file.
	135	+ * @return array<string, mixed>|WP_Error Table array on success, WP_Error if the file is not an HTML file.
130	136	*/
131	137	protected function _maybe_import_html( string $data ) /* : array|false */ {
132	138	TablePress::load_file( 'html-parser.class.php', 'libraries' );
133	139	$table = HTML_Parser::parse( $data );
		@@ -133,9 +139,9 @@
133	139	$table = HTML_Parser::parse( $data );
134	140
135	141	// Check if the HTML code could be parsed. If not, this is probably not an HTML file.
136	142	if ( is_wp_error( $table ) ) {
137		- return false;
	143	+ return $table;
138	144	}
139	145
140	146	$this->pad_array_to_max_cols( $table['data'] );
141	147	return $table;
		@@ -145,19 +151,28 @@
145	151	* Tries to import a table via PHPSpreadsheet.
146	152	*
147	153	* @since 2.0.0
148	154	*
149		- * @param array<string, mixed> $file File to import.
	155	+ * @param File $file File to import.
150	156	* @return array<string, mixed>|WP_Error Table array on success, WP_Error on error.
151	157	*/
152		- protected function _import_phpspreadsheet( array $file ) /* : array|WP_Error */ {
	158	+ protected function _import_phpspreadsheet( File $file ) /* : array|WP_Error */ {
153	159	// Rename the temporary file, as PHPSpreadsheet tries to infer the format from the file's extension.
154		- if ( '' !== $file['extension'] ) {
155		- $temp_file = pathinfo( $file['location'] );
156		- if ( ! isset( $temp_file['extension'] ) || $file['extension'] !== $temp_file['extension'] ) {
157		- $new_location = "{$temp_file['dirname']}/{$temp_file['filename']}.{$file['extension']}"; // @phpstan-ignore-line
158		- if ( rename( $file['location'], $new_location ) ) {
159		- $file['location'] = $new_location;
	160	+ if ( '' !== $file->extension ) {
	161	+ $file_data = pathinfo( $file->location );
	162	+ if ( ! isset( $file_data['extension'] ) || $file->extension !== $file_data['extension'] ) {
	163	+ $temp_file = wp_tempnam();
	164	+ $new_location = "{$temp_file}.{$file->extension}";
	165	+ if ( $file->keep_file ) {
	166	+ // Copy the file, as the original should be kept.
	167	+ if ( copy( $file->location, $new_location ) ) {
	168	+ $file->location = $new_location;
	169	+ $file->keep_file = false; // Delete the newly created file after the import.
	170	+ }
	171	+ } else { // phpcs:ignore Universal.ControlStructures.DisallowLonelyIf.Found
	172	+ if ( rename( $file->location, $new_location ) ) {
	173	+ $file->location = $new_location;
	174	+ }
160	175	}
161	176	}
162	177	}
163	178
		@@ -163,9 +178,9 @@
163	178
164	179	try {
165	180	// Treat all cell values as strings, except for formulas (due to recognition of quoted/escaped formulas like `'=A2`).
166	181	\TablePress\PhpOffice\PhpSpreadsheet\Cell\Cell::setValueBinder( new \TablePress\PhpOffice\PhpSpreadsheet\Cell\StringValueBinder() );
167		- \TablePress\PhpOffice\PhpSpreadsheet\Cell\Cell::getValueBinder()->setFormulaConversion( false ); // @phpstan-ignore-line
	182	+ \TablePress\PhpOffice\PhpSpreadsheet\Cell\Cell::getValueBinder()->setFormulaConversion( false ); // @phpstan-ignore method.notFound
168	183
169	184	/*
170	185	* Try to detect a reader from the file extension and MIME type.
171	186	* Fall back to CSV if no reader could be determined.
		@@ -170,15 +185,24 @@
170	185	* Try to detect a reader from the file extension and MIME type.
171	186	* Fall back to CSV if no reader could be determined.
172	187	*/
173	188	try {
174		- $reader = \TablePress\PhpOffice\PhpSpreadsheet\IOFactory::createReaderForFile( $file['location'] );
	189	+ $reader = \TablePress\PhpOffice\PhpSpreadsheet\IOFactory::createReaderForFile( $file->location );
175	190	} catch ( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Exception $exception ) {
176	191	$reader = \TablePress\PhpOffice\PhpSpreadsheet\IOFactory::createReader( 'Csv' );
177		- // Append .csv to the file name, so that \TablePress\PhpOffice\PhpSpreadsheet\Reader\Csv::canRead() returns true.
178		- $new_location = $file['location'] . '.csv';
179		- if ( rename( $file['location'], $new_location ) ) {
180		- $file['location'] = $new_location;
	192	+ // Change the file extension to .csv, so that \TablePress\PhpOffice\PhpSpreadsheet\Reader\Csv::canRead() returns true.
	193	+ $temp_file = wp_tempnam();
	194	+ $new_location = "{$temp_file}.csv";
	195	+ if ( $file->keep_file ) {
	196	+ // Copy the file, as the original should be kept.
	197	+ if ( copy( $file->location, $new_location ) ) {
	198	+ $file->location = $new_location;
	199	+ $file->keep_file = false; // Delete the newly created file after the import.
	200	+ }
	201	+ } else { // phpcs:ignore Universal.ControlStructures.DisallowLonelyIf.Found
	202	+ if ( rename( $file->location, $new_location ) ) {
	203	+ $file->location = $new_location;
	204	+ }
181	205	}
182	206	}
183	207
184	208	$class_name = get_class( $reader );
		@@ -185,10 +209,10 @@
185	209	$class_type = explode( '\\', $class_name );
186	210	$detected_format = strtolower( array_pop( $class_type ) );
187	211
188	212	if ( 'csv' === $detected_format ) {
189		- $reader->setInputEncoding( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Csv::GUESS_ENCODING ); // @phpstan-ignore-line
190		- // @phpstan-ignore-next-line
	213	+ $reader->setInputEncoding( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Csv::GUESS_ENCODING ); // @phpstan-ignore method.notFound
	214	+ // @phpstan-ignore method.notFound, smaller.alwaysFalse (PHPStan thinks that the Composer minimum version will always be fulfilled.)
191	215	$reader->setEscapeCharacter( ( PHP_VERSION_ID < 70400 ) ? "\x0" : '' ); // Disable the proprietary escape mechanism of PHP's fgetcsv() in PHP >= 7.4.
192	216	}
193	217
194	218	$reader->setIncludeCharts( false );
		@@ -200,12 +224,12 @@
200	224	}
201	225
202	226	// For formats where it's supported, import only the first sheet.
203	227	if ( in_array( $detected_format, array( 'csv', 'html', 'slk' ), true ) ) {
204		- $reader->setSheetIndex( 0 ); // @phpstan-ignore-line
	228	+ $reader->setSheetIndex( 0 ); // @phpstan-ignore method.notFound
205	229	}
206	230
207		- $spreadsheet = $reader->load( $file['location'] );
	231	+ $spreadsheet = $reader->load( $file->location );
208	232	$worksheet = $spreadsheet->getActiveSheet();
209	233	$cell_collection = $worksheet->getCellCollection();
210	234	$comments = $worksheet->getComments();
211	235
		@@ -218,12 +242,12 @@
218	242	$max_col = $worksheet->getHighestColumn();
219	243	$max_row = $worksheet->getHighestRow();
220	244
221	245	// Adapted from \TablePress\PhpOffice\PhpSpreadsheet\Worksheet\Worksheet::rangeToArray().
222		- ++$max_col; // Due to for-loop with characters for columns.
	246	+ \TablePress\PhpOffice\PhpSpreadsheet\Shared\StringHelper::stringIncrement( $max_col ); // Due to for-loop with characters for columns.
223	247	for ( $row = $min_row; $row <= $max_row; $row++ ) {
224	248	$row_data = array();
225		- for ( $col = $min_col; $col !== $max_col; $col++ ) {
	249	+ for ( $col = $min_col; $col !== $max_col; \TablePress\PhpOffice\PhpSpreadsheet\Shared\StringHelper::stringIncrement( $col ) ) {
226	250	$cell_reference = $col . $row;
227	251	if ( ! $cell_collection->has( $cell_reference ) ) {
228	252	$row_data[] = '';
229	253	continue;
		@@ -245,16 +269,31 @@
245	269	}
246	270
247	271	// Apply data type formatting.
248	272	$style = $spreadsheet->getCellXfByIndex( $cell->getXfIndex() );
	273	+
	274	+ $format = $style->getNumberFormat()->getFormatCode() ?? \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_GENERAL;
	275	+
	276	+ /*
	277	+ * When cells in Excel files are formatted as "Text", quotation marks are removed, due to https://github.com/PHPOffice/PhpSpreadsheet/pull/3344.
	278	+ * Setting the format to "General" seems to prevent that.
	279	+ */
	280	+ if ( \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_TEXT === $format && ! is_numeric( $cell_data ) ) {
	281	+ $format = \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_GENERAL;
	282	+ }
	283	+
	284	+ // Fix floating point precision issues with numbers in the "General" Excel .xlsx format.
	285	+ if ( 'xlsx' === $detected_format && \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_GENERAL === $format && is_numeric( $cell_data ) ) {
	286	+ $cell_data = (string) (float) $cell_data; // Type-cast strings to float and back.
	287	+ }
249	288	$cell_data = \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::toFormattedString(
250	289	$cell_data,
251		- $style->getNumberFormat() ? $style->getNumberFormat()->getFormatCode() : \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_GENERAL, // @phpstan-ignore-line
252		- array( $this, 'format_color' )
	290	+ $format,
	291	+ array( $this, 'format_color' ),
253	292	);
254	293
255	294	if ( strlen( $cell_data ) > 1 && '=' === $cell_data[0] ) {
256		- if ( $style->getQuotePrefix() ) {
	295	+ if ( 'xlsx' === $detected_format && $style->getQuotePrefix() ) {
257	296	// Prepend a ' to quoted/escaped formulas (so that they are shown as text). This is currently not supported (at least) for the XLS format.
258	297	$cell_data = "'{$cell_data}";
259	298	} else {
260	299	// Bail early, to not add inline HTML styling around formulas, as they won't work anymore then.
		@@ -341,9 +380,9 @@
341	380	$spreadsheet->disconnectWorksheets();
342	381	unset( $comments, $cell_collection, $worksheet, $spreadsheet );
343	382
344	383	return $table;
345		- } catch ( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Exception $exception ) {
	384	+ } catch ( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Exception | \TablePress\PhpOffice\PhpSpreadsheet\Exception $exception ) {
346	385	return new WP_Error( 'table_import_phpspreadsheet_failed', '', 'Exception: ' . $exception->getMessage() );
347	386	}
348	387	}
349	388