← All changes | classes/class-import-phpspreadsheet.php +72 -35 2.2 → 3.0 View file →

		@@ -7,8 +7,10 @@
7	7	* @author Tobias Bäthge
8	8	* @since 2.0.0
9	9	*/
10	10
	11	+use TablePress\Import\File;
	12	+
11	13	// Prohibit direct script loading.
12	14	defined( 'ABSPATH' ) || die( 'No direct script access allowed!' );
13	15
14	16	/**
		@@ -26,9 +28,9 @@
26	28	*
27	29	* @since 2.0.0
28	30	*/
29	31	public function __construct() {
30		- // Load PHPSpreadsheet via the Composer autoloading mechanism.
	32	+ // Load PHPSpreadsheet via its autoloading mechanism.
31	33	TablePress::load_file( 'autoload.php', 'libraries' );
32	34	}
33	35
34	36	/**
		@@ -35,15 +37,15 @@
35	37	* Imports a table from a file.
36	38	*
37	39	* @since 2.0.0
38	40	*
39		- * @param array<string, mixed> $file File to import.
	41	+ * @param File $file File to import.
40	42	* @return array<string, mixed>|WP_Error Table array on success, WP_Error on error.
41	43	*/
42		- public function import_table( array $file ) /* : array|WP_Error */ {
43		- $data = file_get_contents( $file['location'] );
	44	+ public function import_table( File $file ) /* : array|WP_Error */ {
	45	+ $data = file_get_contents( $file->location );
44	46	if ( false === $data ) {
45		- return new WP_Error( 'table_import_phpspreadsheet_data_read', '', $file['location'] );
	47	+ return new WP_Error( 'table_import_phpspreadsheet_data_read', '', $file->location );
46	48	}
47	49
48	50	// Remove a possible UTF-8 Byte-Order Mark (BOM).
49	51	$bom = pack( 'CCC', 0xef, 0xbb, 0xbf );
		@@ -51,18 +53,18 @@
51	53	$data = substr( $data, 3 );
52	54	}
53	55
54	56	if ( '' === $data ) {
55		- return new WP_Error( 'table_import_phpspreadsheet_data_empty', '', $file['location'] );
	57	+ return new WP_Error( 'table_import_phpspreadsheet_data_empty', '', $file->location );
56	58	}
57	59
58	60	$table = $this->_maybe_import_json( $data );
59		- if ( false !== $table ) {
	61	+ if ( is_array( $table ) ) {
60	62	return $table;
61	63	}
62	64
63	65	$table = $this->_maybe_import_html( $data );
64		- if ( false !== $table ) {
	66	+ if ( is_array( $table ) ) {
65	67	return $table;
66	68	}
67	69
68	70	return $this->_import_phpspreadsheet( $file );
		@@ -76,12 +78,14 @@
76	78	* @param string $data Data to import.
77	79	* @return array<string, mixed>|false Table array on success, false if the file is not a JSON file.
78	80	*/
79	81	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 );
	82	+ $data = trim( $data );
	83	+
	84	+ // If the file does not begin / end with [ / ] or { / }, it's not a supported JSON file.
82	85	$first_character = $data[0];
83		- if ( '{' !== $first_character && '[' !== $first_character ) {
	86	+ $last_character = $data[-1];
	87	+ if ( ! ( '[' === $first_character && ']' === $last_character ) && ! ( '{' === $first_character && '}' === $last_character ) ) {
84	88	return false;
85	89	}
86	90
87	91	$json_table = json_decode( $data, true );
		@@ -125,9 +129,9 @@
125	129	*
126	130	* @since 2.0.0
127	131	*
128	132	* @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.
	133	+ * @return array<string, mixed>|WP_Error Table array on success, WP_Error if the file is not an HTML file.
130	134	*/
131	135	protected function _maybe_import_html( string $data ) /* : array|false */ {
132	136	TablePress::load_file( 'html-parser.class.php', 'libraries' );
133	137	$table = HTML_Parser::parse( $data );
		@@ -133,9 +137,9 @@
133	137	$table = HTML_Parser::parse( $data );
134	138
135	139	// Check if the HTML code could be parsed. If not, this is probably not an HTML file.
136	140	if ( is_wp_error( $table ) ) {
137		- return false;
	141	+ return $table;
138	142	}
139	143
140	144	$this->pad_array_to_max_cols( $table['data'] );
141	145	return $table;
		@@ -145,19 +149,28 @@
145	149	* Tries to import a table via PHPSpreadsheet.
146	150	*
147	151	* @since 2.0.0
148	152	*
149		- * @param array<string, mixed> $file File to import.
	153	+ * @param File $file File to import.
150	154	* @return array<string, mixed>|WP_Error Table array on success, WP_Error on error.
151	155	*/
152		- protected function _import_phpspreadsheet( array $file ) /* : array|WP_Error */ {
	156	+ protected function _import_phpspreadsheet( File $file ) /* : array|WP_Error */ {
153	157	// 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;
	158	+ if ( '' !== $file->extension ) {
	159	+ $file_data = pathinfo( $file->location );
	160	+ if ( ! isset( $file_data['extension'] ) || $file->extension !== $file_data['extension'] ) {
	161	+ $temp_file = wp_tempnam();
	162	+ $new_location = "{$temp_file}.{$file->extension}";
	163	+ if ( $file->keep_file ) {
	164	+ // Copy the file, as the original should be kept.
	165	+ if ( copy( $file->location, $new_location ) ) {
	166	+ $file->location = $new_location;
	167	+ $file->keep_file = false; // Delete the newly created file after the import.
	168	+ }
	169	+ } else { // phpcs:ignore Universal.ControlStructures.DisallowLonelyIf.Found
	170	+ if ( rename( $file->location, $new_location ) ) {
	171	+ $file->location = $new_location;
	172	+ }
160	173	}
161	174	}
162	175	}
163	176
		@@ -163,9 +176,9 @@
163	176
164	177	try {
165	178	// Treat all cell values as strings, except for formulas (due to recognition of quoted/escaped formulas like `'=A2`).
166	179	\TablePress\PhpOffice\PhpSpreadsheet\Cell\Cell::setValueBinder( new \TablePress\PhpOffice\PhpSpreadsheet\Cell\StringValueBinder() );
167		- \TablePress\PhpOffice\PhpSpreadsheet\Cell\Cell::getValueBinder()->setFormulaConversion( false ); // @phpstan-ignore-line
	180	+ \TablePress\PhpOffice\PhpSpreadsheet\Cell\Cell::getValueBinder()->setFormulaConversion( false ); // @phpstan-ignore method.notFound
168	181
169	182	/*
170	183	* Try to detect a reader from the file extension and MIME type.
171	184	* Fall back to CSV if no reader could be determined.
		@@ -170,15 +183,24 @@
170	183	* Try to detect a reader from the file extension and MIME type.
171	184	* Fall back to CSV if no reader could be determined.
172	185	*/
173	186	try {
174		- $reader = \TablePress\PhpOffice\PhpSpreadsheet\IOFactory::createReaderForFile( $file['location'] );
	187	+ $reader = \TablePress\PhpOffice\PhpSpreadsheet\IOFactory::createReaderForFile( $file->location );
175	188	} catch ( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Exception $exception ) {
176	189	$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;
	190	+ // Change the file extension to .csv, so that \TablePress\PhpOffice\PhpSpreadsheet\Reader\Csv::canRead() returns true.
	191	+ $temp_file = wp_tempnam();
	192	+ $new_location = "{$temp_file}.csv";
	193	+ if ( $file->keep_file ) {
	194	+ // Copy the file, as the original should be kept.
	195	+ if ( copy( $file->location, $new_location ) ) {
	196	+ $file->location = $new_location;
	197	+ $file->keep_file = false; // Delete the newly created file after the import.
	198	+ }
	199	+ } else { // phpcs:ignore Universal.ControlStructures.DisallowLonelyIf.Found
	200	+ if ( rename( $file->location, $new_location ) ) {
	201	+ $file->location = $new_location;
	202	+ }
181	203	}
182	204	}
183	205
184	206	$class_name = get_class( $reader );
		@@ -185,10 +207,10 @@
185	207	$class_type = explode( '\\', $class_name );
186	208	$detected_format = strtolower( array_pop( $class_type ) );
187	209
188	210	if ( 'csv' === $detected_format ) {
189		- $reader->setInputEncoding( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Csv::GUESS_ENCODING ); // @phpstan-ignore-line
190		- // @phpstan-ignore-next-line
	211	+ $reader->setInputEncoding( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Csv::GUESS_ENCODING ); // @phpstan-ignore method.notFound
	212	+ // @phpstan-ignore method.notFound
191	213	$reader->setEscapeCharacter( ( PHP_VERSION_ID < 70400 ) ? "\x0" : '' ); // Disable the proprietary escape mechanism of PHP's fgetcsv() in PHP >= 7.4.
192	214	}
193	215
194	216	$reader->setIncludeCharts( false );
		@@ -200,12 +222,12 @@
200	222	}
201	223
202	224	// For formats where it's supported, import only the first sheet.
203	225	if ( in_array( $detected_format, array( 'csv', 'html', 'slk' ), true ) ) {
204		- $reader->setSheetIndex( 0 ); // @phpstan-ignore-line
	226	+ $reader->setSheetIndex( 0 ); // @phpstan-ignore method.notFound
205	227	}
206	228
207		- $spreadsheet = $reader->load( $file['location'] );
	229	+ $spreadsheet = $reader->load( $file->location );
208	230	$worksheet = $spreadsheet->getActiveSheet();
209	231	$cell_collection = $worksheet->getCellCollection();
210	232	$comments = $worksheet->getComments();
211	233
		@@ -245,16 +267,31 @@
245	267	}
246	268
247	269	// Apply data type formatting.
248	270	$style = $spreadsheet->getCellXfByIndex( $cell->getXfIndex() );
	271	+
	272	+ $format = $style->getNumberFormat()->getFormatCode() ?? \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_GENERAL;
	273	+
	274	+ /*
	275	+ * When cells in Excel files are formatted as "Text", quotation marks are removed, due to https://github.com/PHPOffice/PhpSpreadsheet/pull/3344.
	276	+ * Setting the format to "General" seems to prevent that.
	277	+ */
	278	+ if ( \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_TEXT === $format && ! is_numeric( $cell_data ) ) {
	279	+ $format = \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_GENERAL;
	280	+ }
	281	+
	282	+ // Fix floating point precision issues with numbers in the "General" Excel .xlsx format.
	283	+ if ( 'xlsx' === $detected_format && \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_GENERAL === $format && is_numeric( $cell_data ) ) {
	284	+ $cell_data = (string) (float) $cell_data; // Type-cast strings to float and back.
	285	+ }
249	286	$cell_data = \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::toFormattedString(
250	287	$cell_data,
251		- $style->getNumberFormat() ? $style->getNumberFormat()->getFormatCode() : \TablePress\PhpOffice\PhpSpreadsheet\Style\NumberFormat::FORMAT_GENERAL, // @phpstan-ignore-line
252		- array( $this, 'format_color' )
	288	+ $format,
	289	+ array( $this, 'format_color' ),
253	290	);
254	291
255	292	if ( strlen( $cell_data ) > 1 && '=' === $cell_data[0] ) {
256		- if ( $style->getQuotePrefix() ) {
	293	+ if ( 'xlsx' === $detected_format && $style->getQuotePrefix() ) {
257	294	// 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	295	$cell_data = "'{$cell_data}";
259	296	} else {
260	297	// Bail early, to not add inline HTML styling around formulas, as they won't work anymore then.
		@@ -341,9 +378,9 @@
341	378	$spreadsheet->disconnectWorksheets();
342	379	unset( $comments, $cell_collection, $worksheet, $spreadsheet );
343	380
344	381	return $table;
345		- } catch ( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Exception $exception ) {
	382	+ } catch ( \TablePress\PhpOffice\PhpSpreadsheet\Reader\Exception | \TablePress\PhpOffice\PhpSpreadsheet\Exception $exception ) {
346	383	return new WP_Error( 'table_import_phpspreadsheet_failed', '', 'Exception: ' . $exception->getMessage() );
347	384	}
348	385	}
349	386