| 1 |
<?php |
| 2 |
|
| 3 |
namespace PhpOffice\PhpSpreadsheet\Reader; |
| 4 |
|
| 5 |
use PhpOffice\PhpSpreadsheet\Calculation\Calculation; |
| 6 |
use PhpOffice\PhpSpreadsheet\Cell\Cell; |
| 7 |
use PhpOffice\PhpSpreadsheet\Cell\Coordinate; |
| 8 |
use PhpOffice\PhpSpreadsheet\Reader\Csv\Delimiter; |
| 9 |
use PhpOffice\PhpSpreadsheet\Reader\Exception as ReaderException; |
| 10 |
use PhpOffice\PhpSpreadsheet\Shared\StringHelper; |
| 11 |
use PhpOffice\PhpSpreadsheet\Spreadsheet; |
| 12 |
use PhpOffice\PhpSpreadsheet\Style\NumberFormat; |
| 13 |
use Throwable; |
| 14 |
|
| 15 |
class Csv extends BaseReader |
| 16 |
{ |
| 17 |
const DEFAULT_FALLBACK_ENCODING = 'CP1252'; |
| 18 |
const GUESS_ENCODING = 'guess'; |
| 19 |
const UTF8_BOM = "\xEF\xBB\xBF"; |
| 20 |
const UTF8_BOM_LEN = 3; |
| 21 |
const UTF16BE_BOM = "\xfe\xff"; |
| 22 |
const UTF16BE_BOM_LEN = 2; |
| 23 |
const UTF16BE_LF = "\x00\x0a"; |
| 24 |
const UTF16LE_BOM = "\xff\xfe"; |
| 25 |
const UTF16LE_BOM_LEN = 2; |
| 26 |
const UTF16LE_LF = "\x0a\x00"; |
| 27 |
const UTF32BE_BOM = "\x00\x00\xfe\xff"; |
| 28 |
const UTF32BE_BOM_LEN = 4; |
| 29 |
const UTF32BE_LF = "\x00\x00\x00\x0a"; |
| 30 |
const UTF32LE_BOM = "\xff\xfe\x00\x00"; |
| 31 |
const UTF32LE_BOM_LEN = 4; |
| 32 |
const UTF32LE_LF = "\x0a\x00\x00\x00"; |
| 33 |
|
| 34 |
/** |
| 35 |
* Input encoding. |
| 36 |
* |
| 37 |
* @var string |
| 38 |
*/ |
| 39 |
private $inputEncoding = 'UTF-8'; |
| 40 |
|
| 41 |
/** |
| 42 |
* Fallback encoding if guess strikes out. |
| 43 |
* |
| 44 |
* @var string |
| 45 |
*/ |
| 46 |
private $fallbackEncoding = self::DEFAULT_FALLBACK_ENCODING; |
| 47 |
|
| 48 |
/** |
| 49 |
* Delimiter. |
| 50 |
* |
| 51 |
* @var ?string |
| 52 |
*/ |
| 53 |
private $delimiter; |
| 54 |
|
| 55 |
/** |
| 56 |
* Enclosure. |
| 57 |
* |
| 58 |
* @var string |
| 59 |
*/ |
| 60 |
private $enclosure = '"'; |
| 61 |
|
| 62 |
/** |
| 63 |
* Sheet index to read. |
| 64 |
* |
| 65 |
* @var int |
| 66 |
*/ |
| 67 |
private $sheetIndex = 0; |
| 68 |
|
| 69 |
/** |
| 70 |
* Load rows contiguously. |
| 71 |
* |
| 72 |
* @var bool |
| 73 |
*/ |
| 74 |
private $contiguous = false; |
| 75 |
|
| 76 |
/** |
| 77 |
* The character that can escape the enclosure. |
| 78 |
* |
| 79 |
* @var ?string |
| 80 |
*/ |
| 81 |
private $escapeCharacter; |
| 82 |
|
| 83 |
/** |
| 84 |
* The character that will be supplied to fgetcsv |
| 85 |
* when escapeCharacter is null. |
| 86 |
* It is anticipated that it will conditionally be set |
| 87 |
* to null-string for Php9 and above. |
| 88 |
*/ |
| 89 |
private static string $defaultEscapeCharacter = PHP_VERSION_ID < 90000 ? '\\' : ''; |
| 90 |
|
| 91 |
/** |
| 92 |
* Callback for setting defaults in construction. |
| 93 |
* |
| 94 |
* @var ?callable |
| 95 |
*/ |
| 96 |
private static $constructorCallback; |
| 97 |
|
| 98 |
/** |
| 99 |
* Attempt autodetect line endings (deprecated after PHP8.1)? |
| 100 |
* |
| 101 |
* @var bool |
| 102 |
*/ |
| 103 |
private $testAutodetect = true; |
| 104 |
|
| 105 |
/** |
| 106 |
* @var bool |
| 107 |
*/ |
| 108 |
protected $castFormattedNumberToNumeric = false; |
| 109 |
|
| 110 |
/** |
| 111 |
* @var bool |
| 112 |
*/ |
| 113 |
protected $preserveNumericFormatting = false; |
| 114 |
|
| 115 |
/** @var bool */ |
| 116 |
private $preserveNullString = false; |
| 117 |
|
| 118 |
/** |
| 119 |
* Create a new CSV Reader instance. |
| 120 |
*/ |
| 121 |
public function __construct() |
| 122 |
{ |
| 123 |
parent::__construct(); |
| 124 |
$callback = self::$constructorCallback; |
| 125 |
if ($callback !== null) { |
| 126 |
$callback($this); |
| 127 |
} |
| 128 |
} |
| 129 |
|
| 130 |
/** |
| 131 |
* Set a callback to change the defaults. |
| 132 |
* |
| 133 |
* The callback must accept the Csv Reader object as the first parameter, |
| 134 |
* and it should return void. |
| 135 |
*/ |
| 136 |
public static function setConstructorCallback(?callable $callback): void |
| 137 |
{ |
| 138 |
self::$constructorCallback = $callback; |
| 139 |
} |
| 140 |
|
| 141 |
public static function getConstructorCallback(): ?callable |
| 142 |
{ |
| 143 |
return self::$constructorCallback; |
| 144 |
} |
| 145 |
|
| 146 |
public function setInputEncoding(string $encoding): self |
| 147 |
{ |
| 148 |
$this->inputEncoding = $encoding; |
| 149 |
|
| 150 |
return $this; |
| 151 |
} |
| 152 |
|
| 153 |
public function getInputEncoding(): string |
| 154 |
{ |
| 155 |
return $this->inputEncoding; |
| 156 |
} |
| 157 |
|
| 158 |
public function setFallbackEncoding(string $fallbackEncoding): self |
| 159 |
{ |
| 160 |
$this->fallbackEncoding = $fallbackEncoding; |
| 161 |
|
| 162 |
return $this; |
| 163 |
} |
| 164 |
|
| 165 |
public function getFallbackEncoding(): string |
| 166 |
{ |
| 167 |
return $this->fallbackEncoding; |
| 168 |
} |
| 169 |
|
| 170 |
/** |
| 171 |
* Move filepointer past any BOM marker. |
| 172 |
*/ |
| 173 |
protected function skipBOM(): void |
| 174 |
{ |
| 175 |
rewind($this->fileHandle); |
| 176 |
|
| 177 |
if (fgets($this->fileHandle, self::UTF8_BOM_LEN + 1) !== self::UTF8_BOM) { |
| 178 |
rewind($this->fileHandle); |
| 179 |
} |
| 180 |
} |
| 181 |
|
| 182 |
/** |
| 183 |
* Identify any separator that is explicitly set in the file. |
| 184 |
*/ |
| 185 |
protected function checkSeparator(): void |
| 186 |
{ |
| 187 |
$line = fgets($this->fileHandle); |
| 188 |
if ($line === false) { |
| 189 |
return; |
| 190 |
} |
| 191 |
|
| 192 |
if ((strlen(trim($line, "\r\n")) == 5) && (stripos($line, 'sep=') === 0)) { |
| 193 |
$this->delimiter = substr($line, 4, 1); |
| 194 |
|
| 195 |
return; |
| 196 |
} |
| 197 |
|
| 198 |
$this->skipBOM(); |
| 199 |
} |
| 200 |
|
| 201 |
/** |
| 202 |
* Infer the separator if it isn't explicitly set in the file or specified by the user. |
| 203 |
*/ |
| 204 |
protected function inferSeparator(): void |
| 205 |
{ |
| 206 |
if ($this->delimiter !== null) { |
| 207 |
return; |
| 208 |
} |
| 209 |
|
| 210 |
$inferenceEngine = new Delimiter($this->fileHandle, $this->escapeCharacter ?? self::$defaultEscapeCharacter, $this->enclosure); |
| 211 |
|
| 212 |
// If number of lines is 0, nothing to infer : fall back to the default |
| 213 |
if ($inferenceEngine->linesCounted() === 0) { |
| 214 |
$this->delimiter = $inferenceEngine->getDefaultDelimiter(); |
| 215 |
$this->skipBOM(); |
| 216 |
|
| 217 |
return; |
| 218 |
} |
| 219 |
|
| 220 |
$this->delimiter = $inferenceEngine->infer(); |
| 221 |
|
| 222 |
// If no delimiter could be detected, fall back to the default |
| 223 |
if ($this->delimiter === null) { |
| 224 |
$this->delimiter = $inferenceEngine->getDefaultDelimiter(); |
| 225 |
} |
| 226 |
|
| 227 |
$this->skipBOM(); |
| 228 |
} |
| 229 |
|
| 230 |
/** |
| 231 |
* Return worksheet info (Name, Last Column Letter, Last Column Index, Total Rows, Total Columns). |
| 232 |
*/ |
| 233 |
public function listWorksheetInfo(string $filename): array |
| 234 |
{ |
| 235 |
// Open file |
| 236 |
$this->openFileOrMemory($filename); |
| 237 |
$fileHandle = $this->fileHandle; |
| 238 |
|
| 239 |
// Skip BOM, if any |
| 240 |
$this->skipBOM(); |
| 241 |
$this->checkSeparator(); |
| 242 |
$this->inferSeparator(); |
| 243 |
|
| 244 |
$worksheetInfo = []; |
| 245 |
$worksheetInfo[0]['worksheetName'] = 'Worksheet'; |
| 246 |
$worksheetInfo[0]['lastColumnLetter'] = 'A'; |
| 247 |
$worksheetInfo[0]['lastColumnIndex'] = 0; |
| 248 |
$worksheetInfo[0]['totalRows'] = 0; |
| 249 |
$worksheetInfo[0]['totalColumns'] = 0; |
| 250 |
|
| 251 |
// Loop through each line of the file in turn |
| 252 |
$rowData = self::getCsv($fileHandle, 0, $this->delimiter ?? '', $this->enclosure, $this->escapeCharacter); |
| 253 |
while (is_array($rowData)) { |
| 254 |
++$worksheetInfo[0]['totalRows']; |
| 255 |
$worksheetInfo[0]['lastColumnIndex'] = max($worksheetInfo[0]['lastColumnIndex'], count($rowData) - 1); |
| 256 |
$rowData = self::getCsv($fileHandle, 0, $this->delimiter ?? '', $this->enclosure, $this->escapeCharacter); |
| 257 |
} |
| 258 |
|
| 259 |
$worksheetInfo[0]['lastColumnLetter'] = Coordinate::stringFromColumnIndex($worksheetInfo[0]['lastColumnIndex'] + 1); |
| 260 |
$worksheetInfo[0]['totalColumns'] = $worksheetInfo[0]['lastColumnIndex'] + 1; |
| 261 |
|
| 262 |
// Close file |
| 263 |
fclose($fileHandle); |
| 264 |
|
| 265 |
return $worksheetInfo; |
| 266 |
} |
| 267 |
|
| 268 |
/** |
| 269 |
* Loads Spreadsheet from file. |
| 270 |
*/ |
| 271 |
protected function loadSpreadsheetFromFile(string $filename): Spreadsheet |
| 272 |
{ |
| 273 |
// Create new Spreadsheet |
| 274 |
$spreadsheet = new Spreadsheet(); |
| 275 |
|
| 276 |
// Load into this instance |
| 277 |
return $this->loadIntoExisting($filename, $spreadsheet); |
| 278 |
} |
| 279 |
|
| 280 |
/** |
| 281 |
* Loads Spreadsheet from string. |
| 282 |
*/ |
| 283 |
public function loadSpreadsheetFromString(string $contents): Spreadsheet |
| 284 |
{ |
| 285 |
// Create new Spreadsheet |
| 286 |
$spreadsheet = new Spreadsheet(); |
| 287 |
|
| 288 |
// Load into this instance |
| 289 |
return $this->loadStringOrFile('data://text/plain,' . urlencode($contents), $spreadsheet, true); |
| 290 |
} |
| 291 |
|
| 292 |
private function openFileOrMemory(string $filename): void |
| 293 |
{ |
| 294 |
// Open file |
| 295 |
$fhandle = $this->canRead($filename); |
| 296 |
if (!$fhandle) { |
| 297 |
throw new Exception($filename . ' is an Invalid Spreadsheet file.'); |
| 298 |
} |
| 299 |
if ($this->inputEncoding === 'UTF-8') { |
| 300 |
$encoding = self::guessEncodingBom($filename); |
| 301 |
if ($encoding !== '') { |
| 302 |
$this->inputEncoding = $encoding; |
| 303 |
} |
| 304 |
} |
| 305 |
if ($this->inputEncoding === self::GUESS_ENCODING) { |
| 306 |
$this->inputEncoding = self::guessEncoding($filename, $this->fallbackEncoding); |
| 307 |
} |
| 308 |
$this->openFile($filename); |
| 309 |
if ($this->inputEncoding !== 'UTF-8') { |
| 310 |
fclose($this->fileHandle); |
| 311 |
$entireFile = file_get_contents($filename); |
| 312 |
$fileHandle = fopen('php://memory', 'r+b'); |
| 313 |
if ($fileHandle !== false && $entireFile !== false) { |
| 314 |
$this->fileHandle = $fileHandle; |
| 315 |
$data = StringHelper::convertEncoding($entireFile, 'UTF-8', $this->inputEncoding); |
| 316 |
fwrite($this->fileHandle, $data); |
| 317 |
$this->skipBOM(); |
| 318 |
} |
| 319 |
} |
| 320 |
} |
| 321 |
|
| 322 |
public function setTestAutoDetect(bool $value): self |
| 323 |
{ |
| 324 |
$this->testAutodetect = $value; |
| 325 |
|
| 326 |
return $this; |
| 327 |
} |
| 328 |
|
| 329 |
private function setAutoDetect(?string $value): ?string |
| 330 |
{ |
| 331 |
$retVal = null; |
| 332 |
if ($value !== null && $this->testAutodetect && PHP_VERSION_ID < 90000) { |
| 333 |
$retVal2 = @ini_set('auto_detect_line_endings', $value); |
| 334 |
if (is_string($retVal2)) { |
| 335 |
$retVal = $retVal2; |
| 336 |
} |
| 337 |
} |
| 338 |
|
| 339 |
return $retVal; |
| 340 |
} |
| 341 |
|
| 342 |
public function castFormattedNumberToNumeric( |
| 343 |
bool $castFormattedNumberToNumeric, |
| 344 |
bool $preserveNumericFormatting = false |
| 345 |
): void { |
| 346 |
$this->castFormattedNumberToNumeric = $castFormattedNumberToNumeric; |
| 347 |
$this->preserveNumericFormatting = $preserveNumericFormatting; |
| 348 |
} |
| 349 |
|
| 350 |
/** |
| 351 |
* Open data uri for reading. |
| 352 |
*/ |
| 353 |
private function openDataUri(string $filename): void |
| 354 |
{ |
| 355 |
$fileHandle = fopen($filename, 'rb'); |
| 356 |
if ($fileHandle === false) { |
| 357 |
// @codeCoverageIgnoreStart |
| 358 |
throw new ReaderException('Could not open file ' . $filename . ' for reading.'); |
| 359 |
// @codeCoverageIgnoreEnd |
| 360 |
} |
| 361 |
|
| 362 |
$this->fileHandle = $fileHandle; |
| 363 |
} |
| 364 |
|
| 365 |
/** |
| 366 |
* Loads PhpSpreadsheet from file into PhpSpreadsheet instance. |
| 367 |
*/ |
| 368 |
public function loadIntoExisting(string $filename, Spreadsheet $spreadsheet): Spreadsheet |
| 369 |
{ |
| 370 |
return $this->loadStringOrFile($filename, $spreadsheet, false); |
| 371 |
} |
| 372 |
|
| 373 |
/** |
| 374 |
* Loads PhpSpreadsheet from file into PhpSpreadsheet instance. |
| 375 |
*/ |
| 376 |
private function loadStringOrFile(string $filename, Spreadsheet $spreadsheet, bool $dataUri): Spreadsheet |
| 377 |
{ |
| 378 |
// Deprecated in Php8.1 |
| 379 |
$iniset = $this->setAutoDetect('1'); |
| 380 |
|
| 381 |
try { |
| 382 |
$this->loadStringOrFile2($filename, $spreadsheet, $dataUri); |
| 383 |
$this->setAutoDetect($iniset); |
| 384 |
} catch (Throwable $e) { |
| 385 |
$this->setAutoDetect($iniset); |
| 386 |
|
| 387 |
throw $e; |
| 388 |
} |
| 389 |
|
| 390 |
return $spreadsheet; |
| 391 |
} |
| 392 |
|
| 393 |
private function loadStringOrFile2(string $filename, Spreadsheet $spreadsheet, bool $dataUri): void |
| 394 |
{ |
| 395 |
// Open file |
| 396 |
if ($dataUri) { |
| 397 |
$this->openDataUri($filename); |
| 398 |
} else { |
| 399 |
$this->openFileOrMemory($filename); |
| 400 |
} |
| 401 |
$fileHandle = $this->fileHandle; |
| 402 |
|
| 403 |
// Skip BOM, if any |
| 404 |
$this->skipBOM(); |
| 405 |
$this->checkSeparator(); |
| 406 |
$this->inferSeparator(); |
| 407 |
|
| 408 |
// Create new PhpSpreadsheet object |
| 409 |
while ($spreadsheet->getSheetCount() <= $this->sheetIndex) { |
| 410 |
$spreadsheet->createSheet(); |
| 411 |
} |
| 412 |
$sheet = $spreadsheet->setActiveSheetIndex($this->sheetIndex); |
| 413 |
|
| 414 |
// Set our starting row based on whether we're in contiguous mode or not |
| 415 |
$currentRow = 1; |
| 416 |
$outRow = 0; |
| 417 |
|
| 418 |
// Loop through each line of the file in turn |
| 419 |
$rowData = self::getCsv($fileHandle, 0, $this->delimiter ?? '', $this->enclosure, $this->escapeCharacter); |
| 420 |
$valueBinder = Cell::getValueBinder(); |
| 421 |
$preserveBooleanString = method_exists($valueBinder, 'getBooleanConversion') && $valueBinder->getBooleanConversion(); |
| 422 |
while (is_array($rowData)) { |
| 423 |
$noOutputYet = true; |
| 424 |
$columnLetter = 'A'; |
| 425 |
foreach ($rowData as $rowDatum) { |
| 426 |
$this->convertBoolean($rowDatum, $preserveBooleanString); |
| 427 |
$numberFormatMask = $this->convertFormattedNumber($rowDatum); |
| 428 |
if (($rowDatum !== '' || $this->preserveNullString) && $this->readFilter->readCell($columnLetter, $currentRow)) { |
| 429 |
if ($this->contiguous) { |
| 430 |
if ($noOutputYet) { |
| 431 |
$noOutputYet = false; |
| 432 |
++$outRow; |
| 433 |
} |
| 434 |
} else { |
| 435 |
$outRow = $currentRow; |
| 436 |
} |
| 437 |
// Set basic styling for the value (Note that this could be overloaded by styling in a value binder) |
| 438 |
$sheet->getCell($columnLetter . $outRow)->getStyle() |
| 439 |
->getNumberFormat() |
| 440 |
->setFormatCode($numberFormatMask); |
| 441 |
// Set cell value |
| 442 |
$sheet->getCell($columnLetter . $outRow)->setValue($rowDatum); |
| 443 |
} |
| 444 |
++$columnLetter; |
| 445 |
} |
| 446 |
$rowData = self::getCsv($fileHandle, 0, $this->delimiter ?? '', $this->enclosure, $this->escapeCharacter); |
| 447 |
++$currentRow; |
| 448 |
} |
| 449 |
|
| 450 |
// Close file |
| 451 |
fclose($fileHandle); |
| 452 |
} |
| 453 |
|
| 454 |
/** |
| 455 |
* Convert string true/false to boolean, and null to null-string. |
| 456 |
* |
| 457 |
* @param mixed $rowDatum |
| 458 |
*/ |
| 459 |
private function convertBoolean(&$rowDatum, bool $preserveBooleanString): void |
| 460 |
{ |
| 461 |
if (is_string($rowDatum) && !$preserveBooleanString) { |
| 462 |
if (strcasecmp(Calculation::getTRUE(), $rowDatum) === 0 || strcasecmp('true', $rowDatum) === 0) { |
| 463 |
$rowDatum = true; |
| 464 |
} elseif (strcasecmp(Calculation::getFALSE(), $rowDatum) === 0 || strcasecmp('false', $rowDatum) === 0) { |
| 465 |
$rowDatum = false; |
| 466 |
} |
| 467 |
} else { |
| 468 |
$rowDatum = $rowDatum ?? ''; |
| 469 |
} |
| 470 |
} |
| 471 |
|
| 472 |
/** |
| 473 |
* Convert numeric strings to int or float values. |
| 474 |
* |
| 475 |
* @param mixed $rowDatum |
| 476 |
*/ |
| 477 |
private function convertFormattedNumber(&$rowDatum): string |
| 478 |
{ |
| 479 |
$numberFormatMask = NumberFormat::FORMAT_GENERAL; |
| 480 |
if ($this->castFormattedNumberToNumeric === true && is_string($rowDatum)) { |
| 481 |
$numeric = str_replace( |
| 482 |
[StringHelper::getThousandsSeparator(), StringHelper::getDecimalSeparator()], |
| 483 |
['', '.'], |
| 484 |
$rowDatum |
| 485 |
); |
| 486 |
|
| 487 |
if (is_numeric($numeric)) { |
| 488 |
$decimalPos = strpos($rowDatum, StringHelper::getDecimalSeparator()); |
| 489 |
if ($this->preserveNumericFormatting === true) { |
| 490 |
$numberFormatMask = (strpos($rowDatum, StringHelper::getThousandsSeparator()) !== false) |
| 491 |
? '#,##0' : '0'; |
| 492 |
if ($decimalPos !== false) { |
| 493 |
$decimals = strlen($rowDatum) - $decimalPos - 1; |
| 494 |
$numberFormatMask .= '.' . str_repeat('0', min($decimals, 6)); |
| 495 |
} |
| 496 |
} |
| 497 |
|
| 498 |
$rowDatum = ($decimalPos !== false) ? (float) $numeric : (int) $numeric; |
| 499 |
} |
| 500 |
} |
| 501 |
|
| 502 |
return $numberFormatMask; |
| 503 |
} |
| 504 |
|
| 505 |
public function getDelimiter(): ?string |
| 506 |
{ |
| 507 |
return $this->delimiter; |
| 508 |
} |
| 509 |
|
| 510 |
public function setDelimiter(?string $delimiter): self |
| 511 |
{ |
| 512 |
$this->delimiter = $delimiter; |
| 513 |
|
| 514 |
return $this; |
| 515 |
} |
| 516 |
|
| 517 |
public function getEnclosure(): string |
| 518 |
{ |
| 519 |
return $this->enclosure; |
| 520 |
} |
| 521 |
|
| 522 |
public function setEnclosure(string $enclosure): self |
| 523 |
{ |
| 524 |
if ($enclosure == '') { |
| 525 |
$enclosure = '"'; |
| 526 |
} |
| 527 |
$this->enclosure = $enclosure; |
| 528 |
|
| 529 |
return $this; |
| 530 |
} |
| 531 |
|
| 532 |
public function getSheetIndex(): int |
| 533 |
{ |
| 534 |
return $this->sheetIndex; |
| 535 |
} |
| 536 |
|
| 537 |
public function setSheetIndex(int $indexValue): self |
| 538 |
{ |
| 539 |
$this->sheetIndex = $indexValue; |
| 540 |
|
| 541 |
return $this; |
| 542 |
} |
| 543 |
|
| 544 |
public function setContiguous(bool $contiguous): self |
| 545 |
{ |
| 546 |
$this->contiguous = $contiguous; |
| 547 |
|
| 548 |
return $this; |
| 549 |
} |
| 550 |
|
| 551 |
public function getContiguous(): bool |
| 552 |
{ |
| 553 |
return $this->contiguous; |
| 554 |
} |
| 555 |
|
| 556 |
/** |
| 557 |
* Php9 intends to drop support for this parameter in fgetcsv. |
| 558 |
* Not yet ready to mark deprecated in order to give users |
| 559 |
* a migration path. |
| 560 |
*/ |
| 561 |
public function setEscapeCharacter(string $escapeCharacter): self |
| 562 |
{ |
| 563 |
if (PHP_VERSION_ID >= 90000 && $escapeCharacter !== '') { |
| 564 |
throw new ReaderException('Escape character must be null string for Php9+'); |
| 565 |
} |
| 566 |
|
| 567 |
$this->escapeCharacter = $escapeCharacter; |
| 568 |
|
| 569 |
return $this; |
| 570 |
} |
| 571 |
|
| 572 |
public function getEscapeCharacter(): string |
| 573 |
{ |
| 574 |
return $this->escapeCharacter ?? self::$defaultEscapeCharacter; |
| 575 |
} |
| 576 |
|
| 577 |
/** |
| 578 |
* Can the current IReader read the file? |
| 579 |
*/ |
| 580 |
public function canRead(string $filename): bool |
| 581 |
{ |
| 582 |
// Check if file exists |
| 583 |
try { |
| 584 |
$this->openFile($filename); |
| 585 |
} catch (ReaderException $e) { |
| 586 |
return false; |
| 587 |
} |
| 588 |
|
| 589 |
fclose($this->fileHandle); |
| 590 |
|
| 591 |
// Trust file extension if any |
| 592 |
$extension = strtolower(/** @scrutinizer ignore-type */ pathinfo($filename, PATHINFO_EXTENSION)); |
| 593 |
if (in_array($extension, ['csv', 'tsv'])) { |
| 594 |
return true; |
| 595 |
} |
| 596 |
|
| 597 |
// Attempt to guess mimetype |
| 598 |
$type = mime_content_type($filename); |
| 599 |
$supportedTypes = [ |
| 600 |
'application/csv', |
| 601 |
'text/csv', |
| 602 |
'text/plain', |
| 603 |
'inode/x-empty', |
| 604 |
]; |
| 605 |
|
| 606 |
return in_array($type, $supportedTypes, true); |
| 607 |
} |
| 608 |
|
| 609 |
private static function guessEncodingTestNoBom(string &$encoding, string &$contents, string $compare, string $setEncoding): void |
| 610 |
{ |
| 611 |
if ($encoding === '') { |
| 612 |
$pos = strpos($contents, $compare); |
| 613 |
if ($pos !== false && $pos % strlen($compare) === 0) { |
| 614 |
$encoding = $setEncoding; |
| 615 |
} |
| 616 |
} |
| 617 |
} |
| 618 |
|
| 619 |
private static function guessEncodingNoBom(string $filename): string |
| 620 |
{ |
| 621 |
$encoding = ''; |
| 622 |
$contents = file_get_contents($filename); |
| 623 |
self::guessEncodingTestNoBom($encoding, $contents, self::UTF32BE_LF, 'UTF-32BE'); |
| 624 |
self::guessEncodingTestNoBom($encoding, $contents, self::UTF32LE_LF, 'UTF-32LE'); |
| 625 |
self::guessEncodingTestNoBom($encoding, $contents, self::UTF16BE_LF, 'UTF-16BE'); |
| 626 |
self::guessEncodingTestNoBom($encoding, $contents, self::UTF16LE_LF, 'UTF-16LE'); |
| 627 |
if ($encoding === '' && preg_match('//u', $contents) === 1) { |
| 628 |
$encoding = 'UTF-8'; |
| 629 |
} |
| 630 |
|
| 631 |
return $encoding; |
| 632 |
} |
| 633 |
|
| 634 |
private static function guessEncodingTestBom(string &$encoding, string $first4, string $compare, string $setEncoding): void |
| 635 |
{ |
| 636 |
if ($encoding === '') { |
| 637 |
if ($compare === substr($first4, 0, strlen($compare))) { |
| 638 |
$encoding = $setEncoding; |
| 639 |
} |
| 640 |
} |
| 641 |
} |
| 642 |
|
| 643 |
public static function guessEncodingBom(string $filename, ?string $convertString = null): string |
| 644 |
{ |
| 645 |
$encoding = ''; |
| 646 |
$first4 = $convertString ?? (string) file_get_contents($filename, false, null, 0, 4); |
| 647 |
self::guessEncodingTestBom($encoding, $first4, self::UTF8_BOM, 'UTF-8'); |
| 648 |
self::guessEncodingTestBom($encoding, $first4, self::UTF16BE_BOM, 'UTF-16BE'); |
| 649 |
self::guessEncodingTestBom($encoding, $first4, self::UTF32BE_BOM, 'UTF-32BE'); |
| 650 |
self::guessEncodingTestBom($encoding, $first4, self::UTF32LE_BOM, 'UTF-32LE'); |
| 651 |
self::guessEncodingTestBom($encoding, $first4, self::UTF16LE_BOM, 'UTF-16LE'); |
| 652 |
|
| 653 |
return $encoding; |
| 654 |
} |
| 655 |
|
| 656 |
public static function guessEncoding(string $filename, string $dflt = self::DEFAULT_FALLBACK_ENCODING): string |
| 657 |
{ |
| 658 |
$encoding = self::guessEncodingBom($filename); |
| 659 |
if ($encoding === '') { |
| 660 |
$encoding = self::guessEncodingNoBom($filename); |
| 661 |
} |
| 662 |
|
| 663 |
return ($encoding === '') ? $dflt : $encoding; |
| 664 |
} |
| 665 |
|
| 666 |
public function setPreserveNullString(bool $value): self |
| 667 |
{ |
| 668 |
$this->preserveNullString = $value; |
| 669 |
|
| 670 |
return $this; |
| 671 |
} |
| 672 |
|
| 673 |
public function getPreserveNullString(): bool |
| 674 |
{ |
| 675 |
return $this->preserveNullString; |
| 676 |
} |
| 677 |
|
| 678 |
/** |
| 679 |
* Php8.4 deprecates use of anything other than null string |
| 680 |
* as escape Character. |
| 681 |
* |
| 682 |
* @param resource $stream |
| 683 |
* |
| 684 |
* @return array<int,?string>|false |
| 685 |
*/ |
| 686 |
private static function getCsv( |
| 687 |
$stream, |
| 688 |
?int $length = null, |
| 689 |
string $separator = ',', |
| 690 |
string $enclosure = '"', |
| 691 |
?string $escape = null |
| 692 |
) { |
| 693 |
$escape = $escape ?? self::$defaultEscapeCharacter; |
| 694 |
if (PHP_VERSION_ID >= 80400 && $escape !== '') { |
| 695 |
return @fgetcsv($stream, $length, $separator, $enclosure, $escape); |
| 696 |
} |
| 697 |
|
| 698 |
return fgetcsv($stream, $length, $separator, $enclosure, $escape); |
| 699 |
} |
| 700 |
|
| 701 |
public static function affectedByPhp9( |
| 702 |
string $filename, |
| 703 |
string $inputEncoding = 'UTF-8', |
| 704 |
?string $delimiter = null, |
| 705 |
string $enclosure = '"', |
| 706 |
string $escapeCharacter = '\\' |
| 707 |
): bool { |
| 708 |
if (PHP_VERSION_ID < 70400 || PHP_VERSION_ID >= 90000) { |
| 709 |
throw new ReaderException('Function valid only for Php7.4 or Php8'); // @codeCoverageIgnore |
| 710 |
} |
| 711 |
$reader1 = new self(); |
| 712 |
$reader1->setInputEncoding($inputEncoding) |
| 713 |
->setTestAutoDetect(true) |
| 714 |
->setEscapeCharacter($escapeCharacter) |
| 715 |
->setDelimiter($delimiter) |
| 716 |
->setEnclosure($enclosure); |
| 717 |
$spreadsheet1 = $reader1->load($filename); |
| 718 |
$sheet1 = $spreadsheet1->getActiveSheet(); |
| 719 |
$array1 = $sheet1->toArray(null, false, false); |
| 720 |
$spreadsheet1->disconnectWorksheets(); |
| 721 |
|
| 722 |
$reader2 = new self(); |
| 723 |
$reader2->setInputEncoding($inputEncoding) |
| 724 |
->setTestAutoDetect(false) |
| 725 |
->setEscapeCharacter('') |
| 726 |
->setDelimiter($delimiter) |
| 727 |
->setEnclosure($enclosure); |
| 728 |
$spreadsheet2 = $reader2->load($filename); |
| 729 |
$sheet2 = $spreadsheet2->getActiveSheet(); |
| 730 |
$array2 = $sheet2->toArray(null, false, false); |
| 731 |
$spreadsheet2->disconnectWorksheets(); |
| 732 |
|
| 733 |
return $array1 !== $array2; |
| 734 |
} |
| 735 |
} |
| 736 |
|
