| 1 |
<?php |
| 2 |
|
| 3 |
namespace TablePress\PhpOffice\PhpSpreadsheet\Reader; |
| 4 |
|
| 5 |
use TablePress\PhpOffice\PhpSpreadsheet\Exception as PhpSpreadsheetException; |
| 6 |
use TablePress\PhpOffice\PhpSpreadsheet\Shared\CodePage; |
| 7 |
use TablePress\PhpOffice\PhpSpreadsheet\Shared\File; |
| 8 |
use TablePress\PhpOffice\PhpSpreadsheet\Shared\OLERead; |
| 9 |
use TablePress\PhpOffice\PhpSpreadsheet\Shared\StringHelper; |
| 10 |
use TablePress\PhpOffice\PhpSpreadsheet\Style\Border; |
| 11 |
|
| 12 |
class XlsBase extends BaseReader |
| 13 |
{ |
| 14 |
protected const HIGH_ORDER_BIT = 0x80 << 24; |
| 15 |
protected const FC000000 = 0xFC << 24; |
| 16 |
protected const FE000000 = 0xFE << 24; |
| 17 |
|
| 18 |
// ParseXL definitions |
| 19 |
const XLS_BIFF8 = 0x0600; |
| 20 |
const XLS_BIFF7 = 0x0500; |
| 21 |
const XLS_WORKBOOKGLOBALS = 0x0005; |
| 22 |
const XLS_WORKSHEET = 0x0010; |
| 23 |
|
| 24 |
// record identifiers |
| 25 |
const XLS_TYPE_FORMULA = 0x0006; |
| 26 |
const XLS_TYPE_EOF = 0x000A; |
| 27 |
const XLS_TYPE_PROTECT = 0x0012; |
| 28 |
const XLS_TYPE_PASSWORD = 0x0013; |
| 29 |
const XLS_TYPE_HEADER = 0x0014; |
| 30 |
const XLS_TYPE_FOOTER = 0x0015; |
| 31 |
const XLS_TYPE_EXTERNSHEET = 0x0017; |
| 32 |
const XLS_TYPE_DEFINEDNAME = 0x0018; |
| 33 |
const XLS_TYPE_VERTICALPAGEBREAKS = 0x001A; |
| 34 |
const XLS_TYPE_HORIZONTALPAGEBREAKS = 0x001B; |
| 35 |
const XLS_TYPE_NOTE = 0x001C; |
| 36 |
const XLS_TYPE_SELECTION = 0x001D; |
| 37 |
const XLS_TYPE_DATEMODE = 0x0022; |
| 38 |
const XLS_TYPE_EXTERNNAME = 0x0023; |
| 39 |
const XLS_TYPE_LEFTMARGIN = 0x0026; |
| 40 |
const XLS_TYPE_RIGHTMARGIN = 0x0027; |
| 41 |
const XLS_TYPE_TOPMARGIN = 0x0028; |
| 42 |
const XLS_TYPE_BOTTOMMARGIN = 0x0029; |
| 43 |
const XLS_TYPE_PRINTGRIDLINES = 0x002B; |
| 44 |
const XLS_TYPE_FILEPASS = 0x002F; |
| 45 |
const XLS_TYPE_FONT = 0x0031; |
| 46 |
const XLS_TYPE_CONTINUE = 0x003C; |
| 47 |
const XLS_TYPE_PANE = 0x0041; |
| 48 |
const XLS_TYPE_CODEPAGE = 0x0042; |
| 49 |
const XLS_TYPE_DEFCOLWIDTH = 0x0055; |
| 50 |
const XLS_TYPE_OBJ = 0x005D; |
| 51 |
const XLS_TYPE_OBJECTPROTECT = 0x0063; |
| 52 |
const XLS_TYPE_COLINFO = 0x007D; |
| 53 |
const XLS_TYPE_IMDATA = 0x007F; |
| 54 |
const XLS_TYPE_SHEETPR = 0x0081; |
| 55 |
const XLS_TYPE_HCENTER = 0x0083; |
| 56 |
const XLS_TYPE_VCENTER = 0x0084; |
| 57 |
const XLS_TYPE_SHEET = 0x0085; |
| 58 |
const XLS_TYPE_PALETTE = 0x0092; |
| 59 |
const XLS_TYPE_SCL = 0x00A0; |
| 60 |
const XLS_TYPE_PAGESETUP = 0x00A1; |
| 61 |
const XLS_TYPE_MULRK = 0x00BD; |
| 62 |
const XLS_TYPE_MULBLANK = 0x00BE; |
| 63 |
const XLS_TYPE_DBCELL = 0x00D7; |
| 64 |
const XLS_TYPE_SCENPROTECT = 0x00DD; |
| 65 |
const XLS_TYPE_XF = 0x00E0; |
| 66 |
const XLS_TYPE_MERGEDCELLS = 0x00E5; |
| 67 |
const XLS_TYPE_MSODRAWINGGROUP = 0x00EB; |
| 68 |
const XLS_TYPE_MSODRAWING = 0x00EC; |
| 69 |
const XLS_TYPE_SST = 0x00FC; |
| 70 |
const XLS_TYPE_LABELSST = 0x00FD; |
| 71 |
const XLS_TYPE_EXTSST = 0x00FF; |
| 72 |
const XLS_TYPE_EXTERNALBOOK = 0x01AE; |
| 73 |
const XLS_TYPE_CFHEADER = 0x01B0; |
| 74 |
const XLS_TYPE_CFRULE = 0x01B1; |
| 75 |
const XLS_TYPE_DATAVALIDATIONS = 0x01B2; |
| 76 |
const XLS_TYPE_TXO = 0x01B6; |
| 77 |
const XLS_TYPE_HYPERLINK = 0x01B8; |
| 78 |
const XLS_TYPE_DATAVALIDATION = 0x01BE; |
| 79 |
const XLS_TYPE_DIMENSION = 0x0200; |
| 80 |
const XLS_TYPE_BLANK = 0x0201; |
| 81 |
const XLS_TYPE_NUMBER = 0x0203; |
| 82 |
const XLS_TYPE_LABEL = 0x0204; |
| 83 |
const XLS_TYPE_BOOLERR = 0x0205; |
| 84 |
const XLS_TYPE_STRING = 0x0207; |
| 85 |
const XLS_TYPE_ROW = 0x0208; |
| 86 |
const XLS_TYPE_INDEX = 0x020B; |
| 87 |
const XLS_TYPE_ARRAY = 0x0221; |
| 88 |
const XLS_TYPE_DEFAULTROWHEIGHT = 0x0225; |
| 89 |
const XLS_TYPE_WINDOW2 = 0x023E; |
| 90 |
const XLS_TYPE_RK = 0x027E; |
| 91 |
const XLS_TYPE_STYLE = 0x0293; |
| 92 |
const XLS_TYPE_FORMAT = 0x041E; |
| 93 |
const XLS_TYPE_SHAREDFMLA = 0x04BC; |
| 94 |
const XLS_TYPE_BOF = 0x0809; |
| 95 |
const XLS_TYPE_SHEETLAYOUT = 0x0862; |
| 96 |
const XLS_TYPE_SHEETPROTECTION = 0x0867; |
| 97 |
const XLS_TYPE_RANGEPROTECTION = 0x0868; |
| 98 |
const XLS_TYPE_XFEXT = 0x087D; |
| 99 |
const XLS_TYPE_PAGELAYOUTVIEW = 0x088B; |
| 100 |
const XLS_TYPE_UNKNOWN = 0xFFFF; |
| 101 |
|
| 102 |
// Encryption type |
| 103 |
const MS_BIFF_CRYPTO_NONE = 0; |
| 104 |
const MS_BIFF_CRYPTO_XOR = 1; |
| 105 |
const MS_BIFF_CRYPTO_RC4 = 2; |
| 106 |
|
| 107 |
// Size of stream blocks when using RC4 encryption |
| 108 |
const REKEY_BLOCK = 0x400; |
| 109 |
|
| 110 |
// should be consistent with Writer\Xls\Style\CellBorder |
| 111 |
const BORDER_STYLE_MAP = [ |
| 112 |
Border::BORDER_NONE, // => 0x00, |
| 113 |
Border::BORDER_THIN, // => 0x01, |
| 114 |
Border::BORDER_MEDIUM, // => 0x02, |
| 115 |
Border::BORDER_DASHED, // => 0x03, |
| 116 |
Border::BORDER_DOTTED, // => 0x04, |
| 117 |
Border::BORDER_THICK, // => 0x05, |
| 118 |
Border::BORDER_DOUBLE, // => 0x06, |
| 119 |
Border::BORDER_HAIR, // => 0x07, |
| 120 |
Border::BORDER_MEDIUMDASHED, // => 0x08, |
| 121 |
Border::BORDER_DASHDOT, // => 0x09, |
| 122 |
Border::BORDER_MEDIUMDASHDOT, // => 0x0A, |
| 123 |
Border::BORDER_DASHDOTDOT, // => 0x0B, |
| 124 |
Border::BORDER_MEDIUMDASHDOTDOT, // => 0x0C, |
| 125 |
Border::BORDER_SLANTDASHDOT, // => 0x0D, |
| 126 |
Border::BORDER_OMIT, // => 0x0E, |
| 127 |
Border::BORDER_OMIT, // => 0x0F, |
| 128 |
]; |
| 129 |
|
| 130 |
/** |
| 131 |
* Codepage set in the Excel file being read. Only important for BIFF5 (Excel 5.0 - Excel 95) |
| 132 |
* For BIFF8 (Excel 97 - Excel 2003) this will always have the value 'UTF-16LE'. |
| 133 |
*/ |
| 134 |
protected string $codepage = ''; |
| 135 |
|
| 136 |
public function setCodepage(string $codepage): void |
| 137 |
{ |
| 138 |
if (CodePage::validate($codepage) === false) { |
| 139 |
throw new PhpSpreadsheetException('Unknown codepage: ' . $codepage); |
| 140 |
} |
| 141 |
|
| 142 |
$this->codepage = $codepage; |
| 143 |
} |
| 144 |
|
| 145 |
public function getCodepage(): string |
| 146 |
{ |
| 147 |
return $this->codepage; |
| 148 |
} |
| 149 |
|
| 150 |
/** |
| 151 |
* Can the current IReader read the file? |
| 152 |
*/ |
| 153 |
public function canRead(string $filename): bool |
| 154 |
{ |
| 155 |
if (File::testFileNoThrow($filename) === false) { |
| 156 |
return false; |
| 157 |
} |
| 158 |
|
| 159 |
try { |
| 160 |
// Use ParseXL for the hard work. |
| 161 |
$ole = new OLERead(); |
| 162 |
|
| 163 |
// get excel data |
| 164 |
$ole->read($filename); |
| 165 |
if ($ole->wrkbook === null) { |
| 166 |
throw new Exception('The filename ' . $filename . ' is not recognised as a Spreadsheet file'); |
| 167 |
} |
| 168 |
|
| 169 |
return true; |
| 170 |
} catch (PhpSpreadsheetException $exception) { |
| 171 |
return false; |
| 172 |
} |
| 173 |
} |
| 174 |
|
| 175 |
/** |
| 176 |
* Extract RGB color |
| 177 |
* OpenOffice.org's Documentation of the Microsoft Excel File Format, section 2.5.4. |
| 178 |
* |
| 179 |
* @param string $rgb Encoded RGB value (4 bytes) |
| 180 |
* |
| 181 |
* @return array{rgb: string} |
| 182 |
*/ |
| 183 |
protected static function readRGB(string $rgb): array |
| 184 |
{ |
| 185 |
// offset: 0; size 1; Red component |
| 186 |
$r = ord($rgb[0]); |
| 187 |
|
| 188 |
// offset: 1; size: 1; Green component |
| 189 |
$g = ord($rgb[1]); |
| 190 |
|
| 191 |
// offset: 2; size: 1; Blue component |
| 192 |
$b = ord($rgb[2]); |
| 193 |
|
| 194 |
// HEX notation, e.g. 'FF00FC' |
| 195 |
$rgb = sprintf('%02X%02X%02X', $r, $g, $b); |
| 196 |
|
| 197 |
return ['rgb' => $rgb]; |
| 198 |
} |
| 199 |
|
| 200 |
/** |
| 201 |
* Extracts an Excel Unicode short string (8-bit string length) |
| 202 |
* OpenOffice documentation: 2.5.3 |
| 203 |
* function will automatically find out where the Unicode string ends. |
| 204 |
* |
| 205 |
* @return array{value: string, size: int} |
| 206 |
*/ |
| 207 |
protected static function readUnicodeStringShort(string $subData): array |
| 208 |
{ |
| 209 |
// offset: 0: size: 1; length of the string (character count) |
| 210 |
$characterCount = ord($subData[0]); |
| 211 |
|
| 212 |
$string = self::readUnicodeString((string) substr($subData, 1), $characterCount); |
| 213 |
|
| 214 |
// add 1 for the string length |
| 215 |
++$string['size']; |
| 216 |
|
| 217 |
return $string; |
| 218 |
} |
| 219 |
|
| 220 |
/** |
| 221 |
* Extracts an Excel Unicode long string (16-bit string length) |
| 222 |
* OpenOffice documentation: 2.5.3 |
| 223 |
* this function is under construction, needs to support rich text, and Asian phonetic settings. |
| 224 |
* |
| 225 |
* @return array{value: string, size: int} |
| 226 |
*/ |
| 227 |
protected static function readUnicodeStringLong(string $subData): array |
| 228 |
{ |
| 229 |
// offset: 0: size: 2; length of the string (character count) |
| 230 |
$characterCount = self::getUInt2d($subData, 0); |
| 231 |
|
| 232 |
$string = self::readUnicodeString((string) substr($subData, 2), $characterCount); |
| 233 |
|
| 234 |
// add 2 for the string length |
| 235 |
$string['size'] += 2; |
| 236 |
|
| 237 |
return $string; |
| 238 |
} |
| 239 |
|
| 240 |
/** |
| 241 |
* Read Unicode string with no string length field, but with known character count |
| 242 |
* this function is under construction, needs to support rich text, and Asian phonetic settings |
| 243 |
* OpenOffice.org's Documentation of the Microsoft Excel File Format, section 2.5.3. |
| 244 |
* |
| 245 |
* @return array{value: string, size: int} |
| 246 |
*/ |
| 247 |
protected static function readUnicodeString(string $subData, int $characterCount): array |
| 248 |
{ |
| 249 |
// offset: 0: size: 1; option flags |
| 250 |
// bit: 0; mask: 0x01; character compression (0 = compressed 8-bit, 1 = uncompressed 16-bit) |
| 251 |
$isCompressed = !((0x01 & ord($subData[0])) >> 0); |
| 252 |
|
| 253 |
// bit: 2; mask: 0x04; Asian phonetic settings |
| 254 |
//$hasAsian = (0x04) & ord($subData[0]) >> 2; |
| 255 |
|
| 256 |
// bit: 3; mask: 0x08; Rich-Text settings |
| 257 |
//$hasRichText = (0x08) & ord($subData[0]) >> 3; |
| 258 |
|
| 259 |
// offset: 1: size: var; character array |
| 260 |
// this offset assumes richtext and Asian phonetic settings are off which is generally wrong |
| 261 |
// needs to be fixed |
| 262 |
$value = self::encodeUTF16((string) substr($subData, 1, $isCompressed ? $characterCount : 2 * $characterCount), $isCompressed); |
| 263 |
|
| 264 |
return [ |
| 265 |
'value' => $value, |
| 266 |
'size' => $isCompressed ? 1 + $characterCount : 1 + 2 * $characterCount, // the size in bytes including the option flags |
| 267 |
]; |
| 268 |
} |
| 269 |
|
| 270 |
/** |
| 271 |
* Convert UTF-8 string to string surounded by double quotes. Used for explicit string tokens in formulas. |
| 272 |
* Example: hello"world --> "hello""world". |
| 273 |
* |
| 274 |
* @param string $value UTF-8 encoded string |
| 275 |
*/ |
| 276 |
protected static function UTF8toExcelDoubleQuoted(string $value): string |
| 277 |
{ |
| 278 |
return '"' . str_replace('"', '""', $value) . '"'; |
| 279 |
} |
| 280 |
|
| 281 |
/** |
| 282 |
* Reads first 8 bytes of a string and return IEEE 754 float. |
| 283 |
* |
| 284 |
* @param string $data Binary string that is at least 8 bytes long |
| 285 |
* @return int|float |
| 286 |
*/ |
| 287 |
protected static function extractNumber(string $data) |
| 288 |
{ |
| 289 |
$rknumhigh = self::getInt4d($data, 4); |
| 290 |
$rknumlow = self::getInt4d($data, 0); |
| 291 |
$sign = ($rknumhigh & self::HIGH_ORDER_BIT) >> 31; |
| 292 |
$exp = (($rknumhigh & 0x7FF00000) >> 20) - 1023; |
| 293 |
$mantissa = (0x100000 | ($rknumhigh & 0x000FFFFF)); |
| 294 |
$mantissalow1 = ($rknumlow & self::HIGH_ORDER_BIT) >> 31; |
| 295 |
$mantissalow2 = ($rknumlow & 0x7FFFFFFF); |
| 296 |
$value = $mantissa / 2 ** (20 - $exp); |
| 297 |
|
| 298 |
if ($mantissalow1 != 0) { |
| 299 |
$value += 1 / 2 ** (21 - $exp); |
| 300 |
} |
| 301 |
|
| 302 |
if ($mantissalow2 != 0) { |
| 303 |
$value += $mantissalow2 / 2 ** (52 - $exp); |
| 304 |
} |
| 305 |
if ($sign) { |
| 306 |
$value *= -1; |
| 307 |
} |
| 308 |
|
| 309 |
return $value; |
| 310 |
} |
| 311 |
|
| 312 |
/** |
| 313 |
* @return float|int |
| 314 |
*/ |
| 315 |
protected static function getIEEE754(int $rknum) |
| 316 |
{ |
| 317 |
if (($rknum & 0x02) != 0) { |
| 318 |
$value = $rknum >> 2; |
| 319 |
} else { |
| 320 |
// changes by mmp, info on IEEE754 encoding from |
| 321 |
// research.microsoft.com/~hollasch/cgindex/coding/ieeefloat.html |
| 322 |
// The RK format calls for using only the most significant 30 bits |
| 323 |
// of the 64 bit floating point value. The other 34 bits are assumed |
| 324 |
// to be 0 so we use the upper 30 bits of $rknum as follows... |
| 325 |
$sign = ($rknum & self::HIGH_ORDER_BIT) >> 31; |
| 326 |
$exp = ($rknum & 0x7FF00000) >> 20; |
| 327 |
$mantissa = (0x100000 | ($rknum & 0x000FFFFC)); |
| 328 |
$value = $mantissa / 2 ** (20 - ($exp - 1023)); |
| 329 |
if ($sign) { |
| 330 |
$value = -1 * $value; |
| 331 |
} |
| 332 |
//end of changes by mmp |
| 333 |
} |
| 334 |
if (($rknum & 0x01) != 0) { |
| 335 |
$value /= 100; |
| 336 |
} |
| 337 |
|
| 338 |
return $value; |
| 339 |
} |
| 340 |
|
| 341 |
/** |
| 342 |
* Get UTF-8 string from (compressed or uncompressed) UTF-16 string. |
| 343 |
*/ |
| 344 |
protected static function encodeUTF16(string $string, bool $compressed = false): string |
| 345 |
{ |
| 346 |
if ($compressed) { |
| 347 |
$string = self::uncompressByteString($string); |
| 348 |
} |
| 349 |
|
| 350 |
return StringHelper::convertEncoding($string, 'UTF-8', 'UTF-16LE'); |
| 351 |
} |
| 352 |
|
| 353 |
/** |
| 354 |
* Convert UTF-16 string in compressed notation to uncompressed form. Only used for BIFF8. |
| 355 |
*/ |
| 356 |
protected static function uncompressByteString(string $string): string |
| 357 |
{ |
| 358 |
$uncompressedString = ''; |
| 359 |
$strLen = strlen($string); |
| 360 |
for ($i = 0; $i < $strLen; ++$i) { |
| 361 |
$uncompressedString .= $string[$i] . "\0"; |
| 362 |
} |
| 363 |
|
| 364 |
return $uncompressedString; |
| 365 |
} |
| 366 |
|
| 367 |
/** |
| 368 |
* Convert string to UTF-8. Only used for BIFF5. |
| 369 |
*/ |
| 370 |
protected function decodeCodepage(string $string): string |
| 371 |
{ |
| 372 |
return StringHelper::convertEncoding($string, 'UTF-8', $this->codepage); |
| 373 |
} |
| 374 |
|
| 375 |
protected static function confirmPos(string $data, int $pos): void |
| 376 |
{ |
| 377 |
if ($pos >= strlen($data)) { |
| 378 |
throw new PhpSpreadsheetException('File appears to be corrupt'); // @codeCoverageIgnore |
| 379 |
} |
| 380 |
} |
| 381 |
|
| 382 |
/** |
| 383 |
* Read 16-bit unsigned integer. |
| 384 |
*/ |
| 385 |
public static function getUInt2d(string $data, int $pos): int |
| 386 |
{ |
| 387 |
self::confirmPos($data, $pos + 1); |
| 388 |
|
| 389 |
return ord($data[$pos]) | (ord($data[$pos + 1]) << 8); |
| 390 |
} |
| 391 |
|
| 392 |
/** |
| 393 |
* Read 16-bit signed integer. |
| 394 |
*/ |
| 395 |
public static function getInt2d(string $data, int $pos): int |
| 396 |
{ |
| 397 |
self::confirmPos($data, $pos + 1); |
| 398 |
|
| 399 |
return unpack('s', $data[$pos] . $data[$pos + 1])[1]; // @phpstan-ignore offsetAccess.nonOffsetAccessible (I don't know how this statement works), return.type (ditto) |
| 400 |
} |
| 401 |
|
| 402 |
/** |
| 403 |
* Read 32-bit signed integer. |
| 404 |
*/ |
| 405 |
public static function getInt4d(string $data, int $pos): int |
| 406 |
{ |
| 407 |
self::confirmPos($data, $pos + 3); |
| 408 |
|
| 409 |
// FIX: represent numbers correctly on 64-bit system |
| 410 |
// http://sourceforge.net/tracker/index.php?func=detail&aid=1487372&group_id=99160&atid=623334 |
| 411 |
// Changed by Andreas Rehm 2006 to ensure correct result of the <<24 block on 32 and 64bit systems |
| 412 |
$_or_24 = ord($data[$pos + 3]); |
| 413 |
if ($_or_24 >= 128) { |
| 414 |
// negative number |
| 415 |
$_ord_24 = -abs((256 - $_or_24) << 24); |
| 416 |
} else { |
| 417 |
$_ord_24 = ($_or_24 & 127) << 24; |
| 418 |
} |
| 419 |
|
| 420 |
return ord($data[$pos]) | (ord($data[$pos + 1]) << 8) | (ord($data[$pos + 2]) << 16) | $_ord_24; |
| 421 |
} |
| 422 |
} |
| 423 |
|
