Skip to content
Permalink
Branch: master
Find file Copy path
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
7950 lines (6529 sloc) 279 KB
<?php
namespace PhpOffice\PhpSpreadsheet\Reader;
use PhpOffice\PhpSpreadsheet\Cell\Coordinate;
use PhpOffice\PhpSpreadsheet\Cell\DataType;
use PhpOffice\PhpSpreadsheet\Cell\DataValidation;
use PhpOffice\PhpSpreadsheet\Exception as PhpSpreadsheetException;
use PhpOffice\PhpSpreadsheet\NamedRange;
use PhpOffice\PhpSpreadsheet\RichText\RichText;
use PhpOffice\PhpSpreadsheet\Shared\CodePage;
use PhpOffice\PhpSpreadsheet\Shared\Date;
use PhpOffice\PhpSpreadsheet\Shared\Escher;
use PhpOffice\PhpSpreadsheet\Shared\Escher\DggContainer\BstoreContainer\BSE;
use PhpOffice\PhpSpreadsheet\Shared\File;
use PhpOffice\PhpSpreadsheet\Shared\OLE;
use PhpOffice\PhpSpreadsheet\Shared\OLERead;
use PhpOffice\PhpSpreadsheet\Shared\StringHelper;
use PhpOffice\PhpSpreadsheet\Spreadsheet;
use PhpOffice\PhpSpreadsheet\Style\Alignment;
use PhpOffice\PhpSpreadsheet\Style\Borders;
use PhpOffice\PhpSpreadsheet\Style\Font;
use PhpOffice\PhpSpreadsheet\Style\NumberFormat;
use PhpOffice\PhpSpreadsheet\Style\Protection;
use PhpOffice\PhpSpreadsheet\Style\Style;
use PhpOffice\PhpSpreadsheet\Worksheet\MemoryDrawing;
use PhpOffice\PhpSpreadsheet\Worksheet\PageSetup;
use PhpOffice\PhpSpreadsheet\Worksheet\SheetView;
use PhpOffice\PhpSpreadsheet\Worksheet\Worksheet;
// Original file header of ParseXL (used as the base for this class):
// --------------------------------------------------------------------------------
// Adapted from Excel_Spreadsheet_Reader developed by users bizon153,
// trex005, and mmp11 (SourceForge.net)
// https://sourceforge.net/projects/phpexcelreader/
// Primary changes made by canyoncasa (dvc) for ParseXL 1.00 ...
// Modelled moreso after Perl Excel Parse/Write modules
// Added Parse_Excel_Spreadsheet object
// Reads a whole worksheet or tab as row,column array or as
// associated hash of indexed rows and named column fields
// Added variables for worksheet (tab) indexes and names
// Added an object call for loading individual woorksheets
// Changed default indexing defaults to 0 based arrays
// Fixed date/time and percent formats
// Includes patches found at SourceForge...
// unicode patch by nobody
// unpack("d") machine depedency patch by matchy
// boundsheet utf16 patch by bjaenichen
// Renamed functions for shorter names
// General code cleanup and rigor, including <80 column width
// Included a testcase Excel file and PHP example calls
// Code works for PHP 5.x
// Primary changes made by canyoncasa (dvc) for ParseXL 1.10 ...
// http://sourceforge.net/tracker/index.php?func=detail&aid=1466964&group_id=99160&atid=623334
// Decoding of formula conditions, results, and tokens.
// Support for user-defined named cells added as an array "namedcells"
// Patch code for user-defined named cells supports single cells only.
// NOTE: this patch only works for BIFF8 as BIFF5-7 use a different
// external sheet reference structure
class Xls extends BaseReader
{
// ParseXL definitions
const XLS_BIFF8 = 0x0600;
const XLS_BIFF7 = 0x0500;
const XLS_WORKBOOKGLOBALS = 0x0005;
const XLS_WORKSHEET = 0x0010;
// record identifiers
const XLS_TYPE_FORMULA = 0x0006;
const XLS_TYPE_EOF = 0x000a;
const XLS_TYPE_PROTECT = 0x0012;
const XLS_TYPE_OBJECTPROTECT = 0x0063;
const XLS_TYPE_SCENPROTECT = 0x00dd;
const XLS_TYPE_PASSWORD = 0x0013;
const XLS_TYPE_HEADER = 0x0014;
const XLS_TYPE_FOOTER = 0x0015;
const XLS_TYPE_EXTERNSHEET = 0x0017;
const XLS_TYPE_DEFINEDNAME = 0x0018;
const XLS_TYPE_VERTICALPAGEBREAKS = 0x001a;
const XLS_TYPE_HORIZONTALPAGEBREAKS = 0x001b;
const XLS_TYPE_NOTE = 0x001c;
const XLS_TYPE_SELECTION = 0x001d;
const XLS_TYPE_DATEMODE = 0x0022;
const XLS_TYPE_EXTERNNAME = 0x0023;
const XLS_TYPE_LEFTMARGIN = 0x0026;
const XLS_TYPE_RIGHTMARGIN = 0x0027;
const XLS_TYPE_TOPMARGIN = 0x0028;
const XLS_TYPE_BOTTOMMARGIN = 0x0029;
const XLS_TYPE_PRINTGRIDLINES = 0x002b;
const XLS_TYPE_FILEPASS = 0x002f;
const XLS_TYPE_FONT = 0x0031;
const XLS_TYPE_CONTINUE = 0x003c;
const XLS_TYPE_PANE = 0x0041;
const XLS_TYPE_CODEPAGE = 0x0042;
const XLS_TYPE_DEFCOLWIDTH = 0x0055;
const XLS_TYPE_OBJ = 0x005d;
const XLS_TYPE_COLINFO = 0x007d;
const XLS_TYPE_IMDATA = 0x007f;
const XLS_TYPE_SHEETPR = 0x0081;
const XLS_TYPE_HCENTER = 0x0083;
const XLS_TYPE_VCENTER = 0x0084;
const XLS_TYPE_SHEET = 0x0085;
const XLS_TYPE_PALETTE = 0x0092;
const XLS_TYPE_SCL = 0x00a0;
const XLS_TYPE_PAGESETUP = 0x00a1;
const XLS_TYPE_MULRK = 0x00bd;
const XLS_TYPE_MULBLANK = 0x00be;
const XLS_TYPE_DBCELL = 0x00d7;
const XLS_TYPE_XF = 0x00e0;
const XLS_TYPE_MERGEDCELLS = 0x00e5;
const XLS_TYPE_MSODRAWINGGROUP = 0x00eb;
const XLS_TYPE_MSODRAWING = 0x00ec;
const XLS_TYPE_SST = 0x00fc;
const XLS_TYPE_LABELSST = 0x00fd;
const XLS_TYPE_EXTSST = 0x00ff;
const XLS_TYPE_EXTERNALBOOK = 0x01ae;
const XLS_TYPE_DATAVALIDATIONS = 0x01b2;
const XLS_TYPE_TXO = 0x01b6;
const XLS_TYPE_HYPERLINK = 0x01b8;
const XLS_TYPE_DATAVALIDATION = 0x01be;
const XLS_TYPE_DIMENSION = 0x0200;
const XLS_TYPE_BLANK = 0x0201;
const XLS_TYPE_NUMBER = 0x0203;
const XLS_TYPE_LABEL = 0x0204;
const XLS_TYPE_BOOLERR = 0x0205;
const XLS_TYPE_STRING = 0x0207;
const XLS_TYPE_ROW = 0x0208;
const XLS_TYPE_INDEX = 0x020b;
const XLS_TYPE_ARRAY = 0x0221;
const XLS_TYPE_DEFAULTROWHEIGHT = 0x0225;
const XLS_TYPE_WINDOW2 = 0x023e;
const XLS_TYPE_RK = 0x027e;
const XLS_TYPE_STYLE = 0x0293;
const XLS_TYPE_FORMAT = 0x041e;
const XLS_TYPE_SHAREDFMLA = 0x04bc;
const XLS_TYPE_BOF = 0x0809;
const XLS_TYPE_SHEETPROTECTION = 0x0867;
const XLS_TYPE_RANGEPROTECTION = 0x0868;
const XLS_TYPE_SHEETLAYOUT = 0x0862;
const XLS_TYPE_XFEXT = 0x087d;
const XLS_TYPE_PAGELAYOUTVIEW = 0x088b;
const XLS_TYPE_UNKNOWN = 0xffff;
// Encryption type
const MS_BIFF_CRYPTO_NONE = 0;
const MS_BIFF_CRYPTO_XOR = 1;
const MS_BIFF_CRYPTO_RC4 = 2;
// Size of stream blocks when using RC4 encryption
const REKEY_BLOCK = 0x400;
/**
* Summary Information stream data.
*
* @var string
*/
private $summaryInformation;
/**
* Extended Summary Information stream data.
*
* @var string
*/
private $documentSummaryInformation;
/**
* Workbook stream data. (Includes workbook globals substream as well as sheet substreams).
*
* @var string
*/
private $data;
/**
* Size in bytes of $this->data.
*
* @var int
*/
private $dataSize;
/**
* Current position in stream.
*
* @var int
*/
private $pos;
/**
* Workbook to be returned by the reader.
*
* @var Spreadsheet
*/
private $spreadsheet;
/**
* Worksheet that is currently being built by the reader.
*
* @var Worksheet
*/
private $phpSheet;
/**
* BIFF version.
*
* @var int
*/
private $version;
/**
* Codepage set in the Excel file being read. Only important for BIFF5 (Excel 5.0 - Excel 95)
* For BIFF8 (Excel 97 - Excel 2003) this will always have the value 'UTF-16LE'.
*
* @var string
*/
private $codepage;
/**
* Shared formats.
*
* @var array
*/
private $formats;
/**
* Shared fonts.
*
* @var array
*/
private $objFonts;
/**
* Color palette.
*
* @var array
*/
private $palette;
/**
* Worksheets.
*
* @var array
*/
private $sheets;
/**
* External books.
*
* @var array
*/
private $externalBooks;
/**
* REF structures. Only applies to BIFF8.
*
* @var array
*/
private $ref;
/**
* External names.
*
* @var array
*/
private $externalNames;
/**
* Defined names.
*
* @var array
*/
private $definedname;
/**
* Shared strings. Only applies to BIFF8.
*
* @var array
*/
private $sst;
/**
* Panes are frozen? (in sheet currently being read). See WINDOW2 record.
*
* @var bool
*/
private $frozen;
/**
* Fit printout to number of pages? (in sheet currently being read). See SHEETPR record.
*
* @var bool
*/
private $isFitToPages;
/**
* Objects. One OBJ record contributes with one entry.
*
* @var array
*/
private $objs;
/**
* Text Objects. One TXO record corresponds with one entry.
*
* @var array
*/
private $textObjects;
/**
* Cell Annotations (BIFF8).
*
* @var array
*/
private $cellNotes;
/**
* The combined MSODRAWINGGROUP data.
*
* @var string
*/
private $drawingGroupData;
/**
* The combined MSODRAWING data (per sheet).
*
* @var string
*/
private $drawingData;
/**
* Keep track of XF index.
*
* @var int
*/
private $xfIndex;
/**
* Mapping of XF index (that is a cell XF) to final index in cellXf collection.
*
* @var array
*/
private $mapCellXfIndex;
/**
* Mapping of XF index (that is a style XF) to final index in cellStyleXf collection.
*
* @var array
*/
private $mapCellStyleXfIndex;
/**
* The shared formulas in a sheet. One SHAREDFMLA record contributes with one value.
*
* @var array
*/
private $sharedFormulas;
/**
* The shared formula parts in a sheet. One FORMULA record contributes with one value if it
* refers to a shared formula.
*
* @var array
*/
private $sharedFormulaParts;
/**
* The type of encryption in use.
*
* @var int
*/
private $encryption = 0;
/**
* The position in the stream after which contents are encrypted.
*
* @var int
*/
private $encryptionStartPos = false;
/**
* The current RC4 decryption object.
*
* @var Xls\RC4
*/
private $rc4Key;
/**
* The position in the stream that the RC4 decryption object was left at.
*
* @var int
*/
private $rc4Pos = 0;
/**
* The current MD5 context state.
*
* @var string
*/
private $md5Ctxt;
/**
* @var int
*/
private $textObjRef;
/**
* @var string
*/
private $baseCell;
/**
* Create a new Xls Reader instance.
*/
public function __construct()
{
parent::__construct();
}
/**
* Can the current IReader read the file?
*
* @param string $pFilename
*
* @return bool
*/
public function canRead($pFilename)
{
File::assertFile($pFilename);
try {
// Use ParseXL for the hard work.
$ole = new OLERead();
// get excel data
$ole->read($pFilename);
return true;
} catch (PhpSpreadsheetException $e) {
return false;
}
}
/**
* Reads names of the worksheets from a file, without parsing the whole file to a PhpSpreadsheet object.
*
* @param string $pFilename
*
* @throws Exception
*
* @return array
*/
public function listWorksheetNames($pFilename)
{
File::assertFile($pFilename);
$worksheetNames = [];
// Read the OLE file
$this->loadOLE($pFilename);
// total byte size of Excel data (workbook global substream + sheet substreams)
$this->dataSize = strlen($this->data);
$this->pos = 0;
$this->sheets = [];
// Parse Workbook Global Substream
while ($this->pos < $this->dataSize) {
$code = self::getUInt2d($this->data, $this->pos);
switch ($code) {
case self::XLS_TYPE_BOF:
$this->readBof();
break;
case self::XLS_TYPE_SHEET:
$this->readSheet();
break;
case self::XLS_TYPE_EOF:
$this->readDefault();
break 2;
default:
$this->readDefault();
break;
}
}
foreach ($this->sheets as $sheet) {
if ($sheet['sheetType'] != 0x00) {
// 0x00: Worksheet, 0x02: Chart, 0x06: Visual Basic module
continue;
}
$worksheetNames[] = $sheet['name'];
}
return $worksheetNames;
}
/**
* Return worksheet info (Name, Last Column Letter, Last Column Index, Total Rows, Total Columns).
*
* @param string $pFilename
*
* @throws Exception
*
* @return array
*/
public function listWorksheetInfo($pFilename)
{
File::assertFile($pFilename);
$worksheetInfo = [];
// Read the OLE file
$this->loadOLE($pFilename);
// total byte size of Excel data (workbook global substream + sheet substreams)
$this->dataSize = strlen($this->data);
// initialize
$this->pos = 0;
$this->sheets = [];
// Parse Workbook Global Substream
while ($this->pos < $this->dataSize) {
$code = self::getUInt2d($this->data, $this->pos);
switch ($code) {
case self::XLS_TYPE_BOF:
$this->readBof();
break;
case self::XLS_TYPE_SHEET:
$this->readSheet();
break;
case self::XLS_TYPE_EOF:
$this->readDefault();
break 2;
default:
$this->readDefault();
break;
}
}
// Parse the individual sheets
foreach ($this->sheets as $sheet) {
if ($sheet['sheetType'] != 0x00) {
// 0x00: Worksheet
// 0x02: Chart
// 0x06: Visual Basic module
continue;
}
$tmpInfo = [];
$tmpInfo['worksheetName'] = $sheet['name'];
$tmpInfo['lastColumnLetter'] = 'A';
$tmpInfo['lastColumnIndex'] = 0;
$tmpInfo['totalRows'] = 0;
$tmpInfo['totalColumns'] = 0;
$this->pos = $sheet['offset'];
while ($this->pos <= $this->dataSize - 4) {
$code = self::getUInt2d($this->data, $this->pos);
switch ($code) {
case self::XLS_TYPE_RK:
case self::XLS_TYPE_LABELSST:
case self::XLS_TYPE_NUMBER:
case self::XLS_TYPE_FORMULA:
case self::XLS_TYPE_BOOLERR:
case self::XLS_TYPE_LABEL:
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
$rowIndex = self::getUInt2d($recordData, 0) + 1;
$columnIndex = self::getUInt2d($recordData, 2);
$tmpInfo['totalRows'] = max($tmpInfo['totalRows'], $rowIndex);
$tmpInfo['lastColumnIndex'] = max($tmpInfo['lastColumnIndex'], $columnIndex);
break;
case self::XLS_TYPE_BOF:
$this->readBof();
break;
case self::XLS_TYPE_EOF:
$this->readDefault();
break 2;
default:
$this->readDefault();
break;
}
}
$tmpInfo['lastColumnLetter'] = Coordinate::stringFromColumnIndex($tmpInfo['lastColumnIndex'] + 1);
$tmpInfo['totalColumns'] = $tmpInfo['lastColumnIndex'] + 1;
$worksheetInfo[] = $tmpInfo;
}
return $worksheetInfo;
}
/**
* Loads PhpSpreadsheet from file.
*
* @param string $pFilename
*
* @throws Exception
*
* @return Spreadsheet
*/
public function load($pFilename)
{
// Read the OLE file
$this->loadOLE($pFilename);
// Initialisations
$this->spreadsheet = new Spreadsheet();
$this->spreadsheet->removeSheetByIndex(0); // remove 1st sheet
if (!$this->readDataOnly) {
$this->spreadsheet->removeCellStyleXfByIndex(0); // remove the default style
$this->spreadsheet->removeCellXfByIndex(0); // remove the default style
}
// Read the summary information stream (containing meta data)
$this->readSummaryInformation();
// Read the Additional document summary information stream (containing application-specific meta data)
$this->readDocumentSummaryInformation();
// total byte size of Excel data (workbook global substream + sheet substreams)
$this->dataSize = strlen($this->data);
// initialize
$this->pos = 0;
$this->codepage = 'CP1252';
$this->formats = [];
$this->objFonts = [];
$this->palette = [];
$this->sheets = [];
$this->externalBooks = [];
$this->ref = [];
$this->definedname = [];
$this->sst = [];
$this->drawingGroupData = '';
$this->xfIndex = '';
$this->mapCellXfIndex = [];
$this->mapCellStyleXfIndex = [];
// Parse Workbook Global Substream
while ($this->pos < $this->dataSize) {
$code = self::getUInt2d($this->data, $this->pos);
switch ($code) {
case self::XLS_TYPE_BOF:
$this->readBof();
break;
case self::XLS_TYPE_FILEPASS:
$this->readFilepass();
break;
case self::XLS_TYPE_CODEPAGE:
$this->readCodepage();
break;
case self::XLS_TYPE_DATEMODE:
$this->readDateMode();
break;
case self::XLS_TYPE_FONT:
$this->readFont();
break;
case self::XLS_TYPE_FORMAT:
$this->readFormat();
break;
case self::XLS_TYPE_XF:
$this->readXf();
break;
case self::XLS_TYPE_XFEXT:
$this->readXfExt();
break;
case self::XLS_TYPE_STYLE:
$this->readStyle();
break;
case self::XLS_TYPE_PALETTE:
$this->readPalette();
break;
case self::XLS_TYPE_SHEET:
$this->readSheet();
break;
case self::XLS_TYPE_EXTERNALBOOK:
$this->readExternalBook();
break;
case self::XLS_TYPE_EXTERNNAME:
$this->readExternName();
break;
case self::XLS_TYPE_EXTERNSHEET:
$this->readExternSheet();
break;
case self::XLS_TYPE_DEFINEDNAME:
$this->readDefinedName();
break;
case self::XLS_TYPE_MSODRAWINGGROUP:
$this->readMsoDrawingGroup();
break;
case self::XLS_TYPE_SST:
$this->readSst();
break;
case self::XLS_TYPE_EOF:
$this->readDefault();
break 2;
default:
$this->readDefault();
break;
}
}
// Resolve indexed colors for font, fill, and border colors
// Cannot be resolved already in XF record, because PALETTE record comes afterwards
if (!$this->readDataOnly) {
foreach ($this->objFonts as $objFont) {
if (isset($objFont->colorIndex)) {
$color = Xls\Color::map($objFont->colorIndex, $this->palette, $this->version);
$objFont->getColor()->setRGB($color['rgb']);
}
}
foreach ($this->spreadsheet->getCellXfCollection() as $objStyle) {
// fill start and end color
$fill = $objStyle->getFill();
if (isset($fill->startcolorIndex)) {
$startColor = Xls\Color::map($fill->startcolorIndex, $this->palette, $this->version);
$fill->getStartColor()->setRGB($startColor['rgb']);
}
if (isset($fill->endcolorIndex)) {
$endColor = Xls\Color::map($fill->endcolorIndex, $this->palette, $this->version);
$fill->getEndColor()->setRGB($endColor['rgb']);
}
// border colors
$top = $objStyle->getBorders()->getTop();
$right = $objStyle->getBorders()->getRight();
$bottom = $objStyle->getBorders()->getBottom();
$left = $objStyle->getBorders()->getLeft();
$diagonal = $objStyle->getBorders()->getDiagonal();
if (isset($top->colorIndex)) {
$borderTopColor = Xls\Color::map($top->colorIndex, $this->palette, $this->version);
$top->getColor()->setRGB($borderTopColor['rgb']);
}
if (isset($right->colorIndex)) {
$borderRightColor = Xls\Color::map($right->colorIndex, $this->palette, $this->version);
$right->getColor()->setRGB($borderRightColor['rgb']);
}
if (isset($bottom->colorIndex)) {
$borderBottomColor = Xls\Color::map($bottom->colorIndex, $this->palette, $this->version);
$bottom->getColor()->setRGB($borderBottomColor['rgb']);
}
if (isset($left->colorIndex)) {
$borderLeftColor = Xls\Color::map($left->colorIndex, $this->palette, $this->version);
$left->getColor()->setRGB($borderLeftColor['rgb']);
}
if (isset($diagonal->colorIndex)) {
$borderDiagonalColor = Xls\Color::map($diagonal->colorIndex, $this->palette, $this->version);
$diagonal->getColor()->setRGB($borderDiagonalColor['rgb']);
}
}
}
// treat MSODRAWINGGROUP records, workbook-level Escher
if (!$this->readDataOnly && $this->drawingGroupData) {
$escherWorkbook = new Escher();
$reader = new Xls\Escher($escherWorkbook);
$escherWorkbook = $reader->load($this->drawingGroupData);
}
// Parse the individual sheets
foreach ($this->sheets as $sheet) {
if ($sheet['sheetType'] != 0x00) {
// 0x00: Worksheet, 0x02: Chart, 0x06: Visual Basic module
continue;
}
// check if sheet should be skipped
if (isset($this->loadSheetsOnly) && !in_array($sheet['name'], $this->loadSheetsOnly)) {
continue;
}
// add sheet to PhpSpreadsheet object
$this->phpSheet = $this->spreadsheet->createSheet();
// Use false for $updateFormulaCellReferences to prevent adjustment of worksheet references in formula
// cells... during the load, all formulae should be correct, and we're simply bringing the worksheet
// name in line with the formula, not the reverse
$this->phpSheet->setTitle($sheet['name'], false, false);
$this->phpSheet->setSheetState($sheet['sheetState']);
$this->pos = $sheet['offset'];
// Initialize isFitToPages. May change after reading SHEETPR record.
$this->isFitToPages = false;
// Initialize drawingData
$this->drawingData = '';
// Initialize objs
$this->objs = [];
// Initialize shared formula parts
$this->sharedFormulaParts = [];
// Initialize shared formulas
$this->sharedFormulas = [];
// Initialize text objs
$this->textObjects = [];
// Initialize cell annotations
$this->cellNotes = [];
$this->textObjRef = -1;
while ($this->pos <= $this->dataSize - 4) {
$code = self::getUInt2d($this->data, $this->pos);
switch ($code) {
case self::XLS_TYPE_BOF:
$this->readBof();
break;
case self::XLS_TYPE_PRINTGRIDLINES:
$this->readPrintGridlines();
break;
case self::XLS_TYPE_DEFAULTROWHEIGHT:
$this->readDefaultRowHeight();
break;
case self::XLS_TYPE_SHEETPR:
$this->readSheetPr();
break;
case self::XLS_TYPE_HORIZONTALPAGEBREAKS:
$this->readHorizontalPageBreaks();
break;
case self::XLS_TYPE_VERTICALPAGEBREAKS:
$this->readVerticalPageBreaks();
break;
case self::XLS_TYPE_HEADER:
$this->readHeader();
break;
case self::XLS_TYPE_FOOTER:
$this->readFooter();
break;
case self::XLS_TYPE_HCENTER:
$this->readHcenter();
break;
case self::XLS_TYPE_VCENTER:
$this->readVcenter();
break;
case self::XLS_TYPE_LEFTMARGIN:
$this->readLeftMargin();
break;
case self::XLS_TYPE_RIGHTMARGIN:
$this->readRightMargin();
break;
case self::XLS_TYPE_TOPMARGIN:
$this->readTopMargin();
break;
case self::XLS_TYPE_BOTTOMMARGIN:
$this->readBottomMargin();
break;
case self::XLS_TYPE_PAGESETUP:
$this->readPageSetup();
break;
case self::XLS_TYPE_PROTECT:
$this->readProtect();
break;
case self::XLS_TYPE_SCENPROTECT:
$this->readScenProtect();
break;
case self::XLS_TYPE_OBJECTPROTECT:
$this->readObjectProtect();
break;
case self::XLS_TYPE_PASSWORD:
$this->readPassword();
break;
case self::XLS_TYPE_DEFCOLWIDTH:
$this->readDefColWidth();
break;
case self::XLS_TYPE_COLINFO:
$this->readColInfo();
break;
case self::XLS_TYPE_DIMENSION:
$this->readDefault();
break;
case self::XLS_TYPE_ROW:
$this->readRow();
break;
case self::XLS_TYPE_DBCELL:
$this->readDefault();
break;
case self::XLS_TYPE_RK:
$this->readRk();
break;
case self::XLS_TYPE_LABELSST:
$this->readLabelSst();
break;
case self::XLS_TYPE_MULRK:
$this->readMulRk();
break;
case self::XLS_TYPE_NUMBER:
$this->readNumber();
break;
case self::XLS_TYPE_FORMULA:
$this->readFormula();
break;
case self::XLS_TYPE_SHAREDFMLA:
$this->readSharedFmla();
break;
case self::XLS_TYPE_BOOLERR:
$this->readBoolErr();
break;
case self::XLS_TYPE_MULBLANK:
$this->readMulBlank();
break;
case self::XLS_TYPE_LABEL:
$this->readLabel();
break;
case self::XLS_TYPE_BLANK:
$this->readBlank();
break;
case self::XLS_TYPE_MSODRAWING:
$this->readMsoDrawing();
break;
case self::XLS_TYPE_OBJ:
$this->readObj();
break;
case self::XLS_TYPE_WINDOW2:
$this->readWindow2();
break;
case self::XLS_TYPE_PAGELAYOUTVIEW:
$this->readPageLayoutView();
break;
case self::XLS_TYPE_SCL:
$this->readScl();
break;
case self::XLS_TYPE_PANE:
$this->readPane();
break;
case self::XLS_TYPE_SELECTION:
$this->readSelection();
break;
case self::XLS_TYPE_MERGEDCELLS:
$this->readMergedCells();
break;
case self::XLS_TYPE_HYPERLINK:
$this->readHyperLink();
break;
case self::XLS_TYPE_DATAVALIDATIONS:
$this->readDataValidations();
break;
case self::XLS_TYPE_DATAVALIDATION:
$this->readDataValidation();
break;
case self::XLS_TYPE_SHEETLAYOUT:
$this->readSheetLayout();
break;
case self::XLS_TYPE_SHEETPROTECTION:
$this->readSheetProtection();
break;
case self::XLS_TYPE_RANGEPROTECTION:
$this->readRangeProtection();
break;
case self::XLS_TYPE_NOTE:
$this->readNote();
break;
case self::XLS_TYPE_TXO:
$this->readTextObject();
break;
case self::XLS_TYPE_CONTINUE:
$this->readContinue();
break;
case self::XLS_TYPE_EOF:
$this->readDefault();
break 2;
default:
$this->readDefault();
break;
}
}
// treat MSODRAWING records, sheet-level Escher
if (!$this->readDataOnly && $this->drawingData) {
$escherWorksheet = new Escher();
$reader = new Xls\Escher($escherWorksheet);
$escherWorksheet = $reader->load($this->drawingData);
// get all spContainers in one long array, so they can be mapped to OBJ records
$allSpContainers = $escherWorksheet->getDgContainer()->getSpgrContainer()->getAllSpContainers();
}
// treat OBJ records
foreach ($this->objs as $n => $obj) {
// the first shape container never has a corresponding OBJ record, hence $n + 1
if (isset($allSpContainers[$n + 1]) && is_object($allSpContainers[$n + 1])) {
$spContainer = $allSpContainers[$n + 1];
// we skip all spContainers that are a part of a group shape since we cannot yet handle those
if ($spContainer->getNestingLevel() > 1) {
continue;
}
// calculate the width and height of the shape
list($startColumn, $startRow) = Coordinate::coordinateFromString($spContainer->getStartCoordinates());
list($endColumn, $endRow) = Coordinate::coordinateFromString($spContainer->getEndCoordinates());
$startOffsetX = $spContainer->getStartOffsetX();
$startOffsetY = $spContainer->getStartOffsetY();
$endOffsetX = $spContainer->getEndOffsetX();
$endOffsetY = $spContainer->getEndOffsetY();
$width = \PhpOffice\PhpSpreadsheet\Shared\Xls::getDistanceX($this->phpSheet, $startColumn, $startOffsetX, $endColumn, $endOffsetX);
$height = \PhpOffice\PhpSpreadsheet\Shared\Xls::getDistanceY($this->phpSheet, $startRow, $startOffsetY, $endRow, $endOffsetY);
// calculate offsetX and offsetY of the shape
$offsetX = $startOffsetX * \PhpOffice\PhpSpreadsheet\Shared\Xls::sizeCol($this->phpSheet, $startColumn) / 1024;
$offsetY = $startOffsetY * \PhpOffice\PhpSpreadsheet\Shared\Xls::sizeRow($this->phpSheet, $startRow) / 256;
switch ($obj['otObjType']) {
case 0x19:
// Note
if (isset($this->cellNotes[$obj['idObjID']])) {
$cellNote = $this->cellNotes[$obj['idObjID']];
if (isset($this->textObjects[$obj['idObjID']])) {
$textObject = $this->textObjects[$obj['idObjID']];
$this->cellNotes[$obj['idObjID']]['objTextData'] = $textObject;
}
}
break;
case 0x08:
// picture
// get index to BSE entry (1-based)
$BSEindex = $spContainer->getOPT(0x0104);
// If there is no BSE Index, we will fail here and other fields are not read.
// Fix by checking here.
// TODO: Why is there no BSE Index? Is this a new Office Version? Password protected field?
// More likely : a uncompatible picture
if (!$BSEindex) {
continue 2;
}
$BSECollection = $escherWorkbook->getDggContainer()->getBstoreContainer()->getBSECollection();
$BSE = $BSECollection[$BSEindex - 1];
$blipType = $BSE->getBlipType();
// need check because some blip types are not supported by Escher reader such as EMF
if ($blip = $BSE->getBlip()) {
$ih = imagecreatefromstring($blip->getData());
$drawing = new MemoryDrawing();
$drawing->setImageResource($ih);
// width, height, offsetX, offsetY
$drawing->setResizeProportional(false);
$drawing->setWidth($width);
$drawing->setHeight($height);
$drawing->setOffsetX($offsetX);
$drawing->setOffsetY($offsetY);
switch ($blipType) {
case BSE::BLIPTYPE_JPEG:
$drawing->setRenderingFunction(MemoryDrawing::RENDERING_JPEG);
$drawing->setMimeType(MemoryDrawing::MIMETYPE_JPEG);
break;
case BSE::BLIPTYPE_PNG:
$drawing->setRenderingFunction(MemoryDrawing::RENDERING_PNG);
$drawing->setMimeType(MemoryDrawing::MIMETYPE_PNG);
break;
}
$drawing->setWorksheet($this->phpSheet);
$drawing->setCoordinates($spContainer->getStartCoordinates());
}
break;
default:
// other object type
break;
}
}
}
// treat SHAREDFMLA records
if ($this->version == self::XLS_BIFF8) {
foreach ($this->sharedFormulaParts as $cell => $baseCell) {
list($column, $row) = Coordinate::coordinateFromString($cell);
if (($this->getReadFilter() !== null) && $this->getReadFilter()->readCell($column, $row, $this->phpSheet->getTitle())) {
$formula = $this->getFormulaFromStructure($this->sharedFormulas[$baseCell], $cell);
$this->phpSheet->getCell($cell)->setValueExplicit('=' . $formula, DataType::TYPE_FORMULA);
}
}
}
if (!empty($this->cellNotes)) {
foreach ($this->cellNotes as $note => $noteDetails) {
if (!isset($noteDetails['objTextData'])) {
if (isset($this->textObjects[$note])) {
$textObject = $this->textObjects[$note];
$noteDetails['objTextData'] = $textObject;
} else {
$noteDetails['objTextData']['text'] = '';
}
}
$cellAddress = str_replace('$', '', $noteDetails['cellRef']);
$this->phpSheet->getComment($cellAddress)->setAuthor($noteDetails['author'])->setText($this->parseRichText($noteDetails['objTextData']['text']));
}
}
}
// add the named ranges (defined names)
foreach ($this->definedname as $definedName) {
if ($definedName['isBuiltInName']) {
switch ($definedName['name']) {
case pack('C', 0x06):
// print area
// in general, formula looks like this: Foo!$C$7:$J$66,Bar!$A$1:$IV$2
$ranges = explode(',', $definedName['formula']); // FIXME: what if sheetname contains comma?
$extractedRanges = [];
foreach ($ranges as $range) {
// $range should look like one of these
// Foo!$C$7:$J$66
// Bar!$A$1:$IV$2
$explodes = Worksheet::extractSheetTitle($range, true);
$sheetName = trim($explodes[0], "'");
if (count($explodes) == 2) {
if (strpos($explodes[1], ':') === false) {
$explodes[1] = $explodes[1] . ':' . $explodes[1];
}
$extractedRanges[] = str_replace('$', '', $explodes[1]); // C7:J66
}
}
if ($docSheet = $this->spreadsheet->getSheetByName($sheetName)) {
$docSheet->getPageSetup()->setPrintArea(implode(',', $extractedRanges)); // C7:J66,A1:IV2
}
break;
case pack('C', 0x07):
// print titles (repeating rows)
// Assuming BIFF8, there are 3 cases
// 1. repeating rows
// formula looks like this: Sheet!$A$1:$IV$2
// rows 1-2 repeat
// 2. repeating columns
// formula looks like this: Sheet!$A$1:$B$65536
// columns A-B repeat
// 3. both repeating rows and repeating columns
// formula looks like this: Sheet!$A$1:$B$65536,Sheet!$A$1:$IV$2
$ranges = explode(',', $definedName['formula']); // FIXME: what if sheetname contains comma?
foreach ($ranges as $range) {
// $range should look like this one of these
// Sheet!$A$1:$B$65536
// Sheet!$A$1:$IV$2
if (strpos($range, '!') !== false) {
$explodes = Worksheet::extractSheetTitle($range, true);
if ($docSheet = $this->spreadsheet->getSheetByName($explodes[0])) {
$extractedRange = $explodes[1];
$extractedRange = str_replace('$', '', $extractedRange);
$coordinateStrings = explode(':', $extractedRange);
if (count($coordinateStrings) == 2) {
list($firstColumn, $firstRow) = Coordinate::coordinateFromString($coordinateStrings[0]);
list($lastColumn, $lastRow) = Coordinate::coordinateFromString($coordinateStrings[1]);
if ($firstColumn == 'A' and $lastColumn == 'IV') {
// then we have repeating rows
$docSheet->getPageSetup()->setRowsToRepeatAtTop([$firstRow, $lastRow]);
} elseif ($firstRow == 1 and $lastRow == 65536) {
// then we have repeating columns
$docSheet->getPageSetup()->setColumnsToRepeatAtLeft([$firstColumn, $lastColumn]);
}
}
}
}
}
break;
}
} else {
// Extract range
if (strpos($definedName['formula'], '!') !== false) {
$explodes = Worksheet::extractSheetTitle($definedName['formula'], true);
if (($docSheet = $this->spreadsheet->getSheetByName($explodes[0])) ||
($docSheet = $this->spreadsheet->getSheetByName(trim($explodes[0], "'")))) {
$extractedRange = $explodes[1];
$extractedRange = str_replace('$', '', $extractedRange);
$localOnly = ($definedName['scope'] == 0) ? false : true;
$scope = ($definedName['scope'] == 0) ? null : $this->spreadsheet->getSheetByName($this->sheets[$definedName['scope'] - 1]['name']);
$this->spreadsheet->addNamedRange(new NamedRange((string) $definedName['name'], $docSheet, $extractedRange, $localOnly, $scope));
}
}
// Named Value
// TODO Provide support for named values
}
}
$this->data = null;
return $this->spreadsheet;
}
/**
* Read record data from stream, decrypting as required.
*
* @param string $data Data stream to read from
* @param int $pos Position to start reading from
* @param int $len Record data length
*
* @return string Record data
*/
private function readRecordData($data, $pos, $len)
{
$data = substr($data, $pos, $len);
// File not encrypted, or record before encryption start point
if ($this->encryption == self::MS_BIFF_CRYPTO_NONE || $pos < $this->encryptionStartPos) {
return $data;
}
$recordData = '';
if ($this->encryption == self::MS_BIFF_CRYPTO_RC4) {
$oldBlock = floor($this->rc4Pos / self::REKEY_BLOCK);
$block = floor($pos / self::REKEY_BLOCK);
$endBlock = floor(($pos + $len) / self::REKEY_BLOCK);
// Spin an RC4 decryptor to the right spot. If we have a decryptor sitting
// at a point earlier in the current block, re-use it as we can save some time.
if ($block != $oldBlock || $pos < $this->rc4Pos || !$this->rc4Key) {
$this->rc4Key = $this->makeKey($block, $this->md5Ctxt);
$step = $pos % self::REKEY_BLOCK;
} else {
$step = $pos - $this->rc4Pos;
}
$this->rc4Key->RC4(str_repeat("\0", $step));
// Decrypt record data (re-keying at the end of every block)
while ($block != $endBlock) {
$step = self::REKEY_BLOCK - ($pos % self::REKEY_BLOCK);
$recordData .= $this->rc4Key->RC4(substr($data, 0, $step));
$data = substr($data, $step);
$pos += $step;
$len -= $step;
++$block;
$this->rc4Key = $this->makeKey($block, $this->md5Ctxt);
}
$recordData .= $this->rc4Key->RC4(substr($data, 0, $len));
// Keep track of the position of this decryptor.
// We'll try and re-use it later if we can to speed things up
$this->rc4Pos = $pos + $len;
} elseif ($this->encryption == self::MS_BIFF_CRYPTO_XOR) {
throw new Exception('XOr encryption not supported');
}
return $recordData;
}
/**
* Use OLE reader to extract the relevant data streams from the OLE file.
*
* @param string $pFilename
*/
private function loadOLE($pFilename)
{
// OLE reader
$ole = new OLERead();
// get excel data,
$ole->read($pFilename);
// Get workbook data: workbook stream + sheet streams
$this->data = $ole->getStream($ole->wrkbook);
// Get summary information data
$this->summaryInformation = $ole->getStream($ole->summaryInformation);
// Get additional document summary information data
$this->documentSummaryInformation = $ole->getStream($ole->documentSummaryInformation);
}
/**
* Read summary information.
*/
private function readSummaryInformation()
{
if (!isset($this->summaryInformation)) {
return;
}
// offset: 0; size: 2; must be 0xFE 0xFF (UTF-16 LE byte order mark)
// offset: 2; size: 2;
// offset: 4; size: 2; OS version
// offset: 6; size: 2; OS indicator
// offset: 8; size: 16
// offset: 24; size: 4; section count
$secCount = self::getInt4d($this->summaryInformation, 24);
// offset: 28; size: 16; first section's class id: e0 85 9f f2 f9 4f 68 10 ab 91 08 00 2b 27 b3 d9
// offset: 44; size: 4
$secOffset = self::getInt4d($this->summaryInformation, 44);
// section header
// offset: $secOffset; size: 4; section length
$secLength = self::getInt4d($this->summaryInformation, $secOffset);
// offset: $secOffset+4; size: 4; property count
$countProperties = self::getInt4d($this->summaryInformation, $secOffset + 4);
// initialize code page (used to resolve string values)
$codePage = 'CP1252';
// offset: ($secOffset+8); size: var
// loop through property decarations and properties
for ($i = 0; $i < $countProperties; ++$i) {
// offset: ($secOffset+8) + (8 * $i); size: 4; property ID
$id = self::getInt4d($this->summaryInformation, ($secOffset + 8) + (8 * $i));
// Use value of property id as appropriate
// offset: ($secOffset+12) + (8 * $i); size: 4; offset from beginning of section (48)
$offset = self::getInt4d($this->summaryInformation, ($secOffset + 12) + (8 * $i));
$type = self::getInt4d($this->summaryInformation, $secOffset + $offset);
// initialize property value
$value = null;
// extract property value based on property type
switch ($type) {
case 0x02: // 2 byte signed integer
$value = self::getUInt2d($this->summaryInformation, $secOffset + 4 + $offset);
break;
case 0x03: // 4 byte signed integer
$value = self::getInt4d($this->summaryInformation, $secOffset + 4 + $offset);
break;
case 0x13: // 4 byte unsigned integer
// not needed yet, fix later if necessary
break;
case 0x1E: // null-terminated string prepended by dword string length
$byteLength = self::getInt4d($this->summaryInformation, $secOffset + 4 + $offset);
$value = substr($this->summaryInformation, $secOffset + 8 + $offset, $byteLength);
$value = StringHelper::convertEncoding($value, 'UTF-8', $codePage);
$value = rtrim($value);
break;
case 0x40: // Filetime (64-bit value representing the number of 100-nanosecond intervals since January 1, 1601)
// PHP-time
$value = OLE::OLE2LocalDate(substr($this->summaryInformation, $secOffset + 4 + $offset, 8));
break;
case 0x47: // Clipboard format
// not needed yet, fix later if necessary
break;
}
switch ($id) {
case 0x01: // Code Page
$codePage = CodePage::numberToName($value);
break;
case 0x02: // Title
$this->spreadsheet->getProperties()->setTitle($value);
break;
case 0x03: // Subject
$this->spreadsheet->getProperties()->setSubject($value);
break;
case 0x04: // Author (Creator)
$this->spreadsheet->getProperties()->setCreator($value);
break;
case 0x05: // Keywords
$this->spreadsheet->getProperties()->setKeywords($value);
break;
case 0x06: // Comments (Description)
$this->spreadsheet->getProperties()->setDescription($value);
break;
case 0x07: // Template
// Not supported by PhpSpreadsheet
break;
case 0x08: // Last Saved By (LastModifiedBy)
$this->spreadsheet->getProperties()->setLastModifiedBy($value);
break;
case 0x09: // Revision
// Not supported by PhpSpreadsheet
break;
case 0x0A: // Total Editing Time
// Not supported by PhpSpreadsheet
break;
case 0x0B: // Last Printed
// Not supported by PhpSpreadsheet
break;
case 0x0C: // Created Date/Time
$this->spreadsheet->getProperties()->setCreated($value);
break;
case 0x0D: // Modified Date/Time
$this->spreadsheet->getProperties()->setModified($value);
break;
case 0x0E: // Number of Pages
// Not supported by PhpSpreadsheet
break;
case 0x0F: // Number of Words
// Not supported by PhpSpreadsheet
break;
case 0x10: // Number of Characters
// Not supported by PhpSpreadsheet
break;
case 0x11: // Thumbnail
// Not supported by PhpSpreadsheet
break;
case 0x12: // Name of creating application
// Not supported by PhpSpreadsheet
break;
case 0x13: // Security
// Not supported by PhpSpreadsheet
break;
}
}
}
/**
* Read additional document summary information.
*/
private function readDocumentSummaryInformation()
{
if (!isset($this->documentSummaryInformation)) {
return;
}
// offset: 0; size: 2; must be 0xFE 0xFF (UTF-16 LE byte order mark)
// offset: 2; size: 2;
// offset: 4; size: 2; OS version
// offset: 6; size: 2; OS indicator
// offset: 8; size: 16
// offset: 24; size: 4; section count
$secCount = self::getInt4d($this->documentSummaryInformation, 24);
// offset: 28; size: 16; first section's class id: 02 d5 cd d5 9c 2e 1b 10 93 97 08 00 2b 2c f9 ae
// offset: 44; size: 4; first section offset
$secOffset = self::getInt4d($this->documentSummaryInformation, 44);
// section header
// offset: $secOffset; size: 4; section length
$secLength = self::getInt4d($this->documentSummaryInformation, $secOffset);
// offset: $secOffset+4; size: 4; property count
$countProperties = self::getInt4d($this->documentSummaryInformation, $secOffset + 4);
// initialize code page (used to resolve string values)
$codePage = 'CP1252';
// offset: ($secOffset+8); size: var
// loop through property decarations and properties
for ($i = 0; $i < $countProperties; ++$i) {
// offset: ($secOffset+8) + (8 * $i); size: 4; property ID
$id = self::getInt4d($this->documentSummaryInformation, ($secOffset + 8) + (8 * $i));
// Use value of property id as appropriate
// offset: 60 + 8 * $i; size: 4; offset from beginning of section (48)
$offset = self::getInt4d($this->documentSummaryInformation, ($secOffset + 12) + (8 * $i));
$type = self::getInt4d($this->documentSummaryInformation, $secOffset + $offset);
// initialize property value
$value = null;
// extract property value based on property type
switch ($type) {
case 0x02: // 2 byte signed integer
$value = self::getUInt2d($this->documentSummaryInformation, $secOffset + 4 + $offset);
break;
case 0x03: // 4 byte signed integer
$value = self::getInt4d($this->documentSummaryInformation, $secOffset + 4 + $offset);
break;
case 0x0B: // Boolean
$value = self::getUInt2d($this->documentSummaryInformation, $secOffset + 4 + $offset);
$value = ($value == 0 ? false : true);
break;
case 0x13: // 4 byte unsigned integer
// not needed yet, fix later if necessary
break;
case 0x1E: // null-terminated string prepended by dword string length
$byteLength = self::getInt4d($this->documentSummaryInformation, $secOffset + 4 + $offset);
$value = substr($this->documentSummaryInformation, $secOffset + 8 + $offset, $byteLength);
$value = StringHelper::convertEncoding($value, 'UTF-8', $codePage);
$value = rtrim($value);
break;
case 0x40: // Filetime (64-bit value representing the number of 100-nanosecond intervals since January 1, 1601)
// PHP-Time
$value = OLE::OLE2LocalDate(substr($this->documentSummaryInformation, $secOffset + 4 + $offset, 8));
break;
case 0x47: // Clipboard format
// not needed yet, fix later if necessary
break;
}
switch ($id) {
case 0x01: // Code Page
$codePage = CodePage::numberToName($value);
break;
case 0x02: // Category
$this->spreadsheet->getProperties()->setCategory($value);
break;
case 0x03: // Presentation Target
// Not supported by PhpSpreadsheet
break;
case 0x04: // Bytes
// Not supported by PhpSpreadsheet
break;
case 0x05: // Lines
// Not supported by PhpSpreadsheet
break;
case 0x06: // Paragraphs
// Not supported by PhpSpreadsheet
break;
case 0x07: // Slides
// Not supported by PhpSpreadsheet
break;
case 0x08: // Notes
// Not supported by PhpSpreadsheet
break;
case 0x09: // Hidden Slides
// Not supported by PhpSpreadsheet
break;
case 0x0A: // MM Clips
// Not supported by PhpSpreadsheet
break;
case 0x0B: // Scale Crop
// Not supported by PhpSpreadsheet
break;
case 0x0C: // Heading Pairs
// Not supported by PhpSpreadsheet
break;
case 0x0D: // Titles of Parts
// Not supported by PhpSpreadsheet
break;
case 0x0E: // Manager
$this->spreadsheet->getProperties()->setManager($value);
break;
case 0x0F: // Company
$this->spreadsheet->getProperties()->setCompany($value);
break;
case 0x10: // Links up-to-date
// Not supported by PhpSpreadsheet
break;
}
}
}
/**
* Reads a general type of BIFF record. Does nothing except for moving stream pointer forward to next record.
*/
private function readDefault()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
// move stream pointer to next record
$this->pos += 4 + $length;
}
/**
* The NOTE record specifies a comment associated with a particular cell. In Excel 95 (BIFF7) and earlier versions,
* this record stores a note (cell note). This feature was significantly enhanced in Excel 97.
*/
private function readNote()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
if ($this->readDataOnly) {
return;
}
$cellAddress = $this->readBIFF8CellAddress(substr($recordData, 0, 4));
if ($this->version == self::XLS_BIFF8) {
$noteObjID = self::getUInt2d($recordData, 6);
$noteAuthor = self::readUnicodeStringLong(substr($recordData, 8));
$noteAuthor = $noteAuthor['value'];
$this->cellNotes[$noteObjID] = [
'cellRef' => $cellAddress,
'objectID' => $noteObjID,
'author' => $noteAuthor,
];
} else {
$extension = false;
if ($cellAddress == '$B$65536') {
// If the address row is -1 and the column is 0, (which translates as $B$65536) then this is a continuation
// note from the previous cell annotation. We're not yet handling this, so annotations longer than the
// max 2048 bytes will probably throw a wobbly.
$row = self::getUInt2d($recordData, 0);
$extension = true;
$cellAddress = array_pop(array_keys($this->phpSheet->getComments()));
}
$cellAddress = str_replace('$', '', $cellAddress);
$noteLength = self::getUInt2d($recordData, 4);
$noteText = trim(substr($recordData, 6));
if ($extension) {
// Concatenate this extension with the currently set comment for the cell
$comment = $this->phpSheet->getComment($cellAddress);
$commentText = $comment->getText()->getPlainText();
$comment->setText($this->parseRichText($commentText . $noteText));
} else {
// Set comment for the cell
$this->phpSheet->getComment($cellAddress)->setText($this->parseRichText($noteText));
// ->setAuthor($author)
}
}
}
/**
* The TEXT Object record contains the text associated with a cell annotation.
*/
private function readTextObject()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
if ($this->readDataOnly) {
return;
}
// recordData consists of an array of subrecords looking like this:
// grbit: 2 bytes; Option Flags
// rot: 2 bytes; rotation
// cchText: 2 bytes; length of the text (in the first continue record)
// cbRuns: 2 bytes; length of the formatting (in the second continue record)
// followed by the continuation records containing the actual text and formatting
$grbitOpts = self::getUInt2d($recordData, 0);
$rot = self::getUInt2d($recordData, 2);
$cchText = self::getUInt2d($recordData, 10);
$cbRuns = self::getUInt2d($recordData, 12);
$text = $this->getSplicedRecordData();
$textByte = $text['spliceOffsets'][1] - $text['spliceOffsets'][0] - 1;
$textStr = substr($text['recordData'], $text['spliceOffsets'][0] + 1, $textByte);
// get 1 byte
$is16Bit = ord($text['recordData'][0]);
// it is possible to use a compressed format,
// which omits the high bytes of all characters, if they are all zero
if (($is16Bit & 0x01) === 0) {
$textStr = StringHelper::ConvertEncoding($textStr, 'UTF-8', 'ISO-8859-1');
} else {
$textStr = $this->decodeCodepage($textStr);
}
$this->textObjects[$this->textObjRef] = [
'text' => $textStr,
'format' => substr($text['recordData'], $text['spliceOffsets'][1], $cbRuns),
'alignment' => $grbitOpts,
'rotation' => $rot,
];
}
/**
* Read BOF.
*/
private function readBof()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = substr($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
// offset: 2; size: 2; type of the following data
$substreamType = self::getUInt2d($recordData, 2);
switch ($substreamType) {
case self::XLS_WORKBOOKGLOBALS:
$version = self::getUInt2d($recordData, 0);
if (($version != self::XLS_BIFF8) && ($version != self::XLS_BIFF7)) {
throw new Exception('Cannot read this Excel file. Version is too old.');
}
$this->version = $version;
break;
case self::XLS_WORKSHEET:
// do not use this version information for anything
// it is unreliable (OpenOffice doc, 5.8), use only version information from the global stream
break;
default:
// substream, e.g. chart
// just skip the entire substream
do {
$code = self::getUInt2d($this->data, $this->pos);
$this->readDefault();
} while ($code != self::XLS_TYPE_EOF && $this->pos < $this->dataSize);
break;
}
}
/**
* FILEPASS.
*
* This record is part of the File Protection Block. It
* contains information about the read/write password of the
* file. All record contents following this record will be
* encrypted.
*
* -- "OpenOffice.org's Documentation of the Microsoft
* Excel File Format"
*
* The decryption functions and objects used from here on in
* are based on the source of Spreadsheet-ParseExcel:
* https://metacpan.org/release/Spreadsheet-ParseExcel
*/
private function readFilepass()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
if ($length != 54) {
throw new Exception('Unexpected file pass record length');
}
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
if (!$this->verifyPassword('VelvetSweatshop', substr($recordData, 6, 16), substr($recordData, 22, 16), substr($recordData, 38, 16), $this->md5Ctxt)) {
throw new Exception('Decryption password incorrect');
}
$this->encryption = self::MS_BIFF_CRYPTO_RC4;
// Decryption required from the record after next onwards
$this->encryptionStartPos = $this->pos + self::getUInt2d($this->data, $this->pos + 2);
}
/**
* Make an RC4 decryptor for the given block.
*
* @param int $block Block for which to create decrypto
* @param string $valContext MD5 context state
*
* @return Xls\RC4
*/
private function makeKey($block, $valContext)
{
$pwarray = str_repeat("\0", 64);
for ($i = 0; $i < 5; ++$i) {
$pwarray[$i] = $valContext[$i];
}
$pwarray[5] = chr($block & 0xff);
$pwarray[6] = chr(($block >> 8) & 0xff);
$pwarray[7] = chr(($block >> 16) & 0xff);
$pwarray[8] = chr(($block >> 24) & 0xff);
$pwarray[9] = "\x80";
$pwarray[56] = "\x48";
$md5 = new Xls\MD5();
$md5->add($pwarray);
$s = $md5->getContext();
return new Xls\RC4($s);
}
/**
* Verify RC4 file password.
*
* @param string $password Password to check
* @param string $docid Document id
* @param string $salt_data Salt data
* @param string $hashedsalt_data Hashed salt data
* @param string $valContext Set to the MD5 context of the value
*
* @return bool Success
*/
private function verifyPassword($password, $docid, $salt_data, $hashedsalt_data, &$valContext)
{
$pwarray = str_repeat("\0", 64);
$iMax = strlen($password);
for ($i = 0; $i < $iMax; ++$i) {
$o = ord(substr($password, $i, 1));
$pwarray[2 * $i] = chr($o & 0xff);
$pwarray[2 * $i + 1] = chr(($o >> 8) & 0xff);
}
$pwarray[2 * $i] = chr(0x80);
$pwarray[56] = chr(($i << 4) & 0xff);
$md5 = new Xls\MD5();
$md5->add($pwarray);
$mdContext1 = $md5->getContext();
$offset = 0;
$keyoffset = 0;
$tocopy = 5;
$md5->reset();
while ($offset != 16) {
if ((64 - $offset) < 5) {
$tocopy = 64 - $offset;
}
for ($i = 0; $i <= $tocopy; ++$i) {
$pwarray[$offset + $i] = $mdContext1[$keyoffset + $i];
}
$offset += $tocopy;
if ($offset == 64) {
$md5->add($pwarray);
$keyoffset = $tocopy;
$tocopy = 5 - $tocopy;
$offset = 0;
continue;
}
$keyoffset = 0;
$tocopy = 5;
for ($i = 0; $i < 16; ++$i) {
$pwarray[$offset + $i] = $docid[$i];
}
$offset += 16;
}
$pwarray[16] = "\x80";
for ($i = 0; $i < 47; ++$i) {
$pwarray[17 + $i] = "\0";
}
$pwarray[56] = "\x80";
$pwarray[57] = "\x0a";
$md5->add($pwarray);
$valContext = $md5->getContext();
$key = $this->makeKey(0, $valContext);
$salt = $key->RC4($salt_data);
$hashedsalt = $key->RC4($hashedsalt_data);
$salt .= "\x80" . str_repeat("\0", 47);
$salt[56] = "\x80";
$md5->reset();
$md5->add($salt);
$mdContext2 = $md5->getContext();
return $mdContext2 == $hashedsalt;
}
/**
* CODEPAGE.
*
* This record stores the text encoding used to write byte
* strings, stored as MS Windows code page identifier.
*
* -- "OpenOffice.org's Documentation of the Microsoft
* Excel File Format"
*/
private function readCodepage()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
// offset: 0; size: 2; code page identifier
$codepage = self::getUInt2d($recordData, 0);
$this->codepage = CodePage::numberToName($codepage);
}
/**
* DATEMODE.
*
* This record specifies the base date for displaying date
* values. All dates are stored as count of days past this
* base date. In BIFF2-BIFF4 this record is part of the
* Calculation Settings Block. In BIFF5-BIFF8 it is
* stored in the Workbook Globals Substream.
*
* -- "OpenOffice.org's Documentation of the Microsoft
* Excel File Format"
*/
private function readDateMode()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
// offset: 0; size: 2; 0 = base 1900, 1 = base 1904
Date::setExcelCalendar(Date::CALENDAR_WINDOWS_1900);
if (ord($recordData[0]) == 1) {
Date::setExcelCalendar(Date::CALENDAR_MAC_1904);
}
}
/**
* Read a FONT record.
*/
private function readFont()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
if (!$this->readDataOnly) {
$objFont = new Font();
// offset: 0; size: 2; height of the font (in twips = 1/20 of a point)
$size = self::getUInt2d($recordData, 0);
$objFont->setSize($size / 20);
// offset: 2; size: 2; option flags
// bit: 0; mask 0x0001; bold (redundant in BIFF5-BIFF8)
// bit: 1; mask 0x0002; italic
$isItalic = (0x0002 & self::getUInt2d($recordData, 2)) >> 1;
if ($isItalic) {
$objFont->setItalic(true);
}
// bit: 2; mask 0x0004; underlined (redundant in BIFF5-BIFF8)
// bit: 3; mask 0x0008; strikethrough
$isStrike = (0x0008 & self::getUInt2d($recordData, 2)) >> 3;
if ($isStrike) {
$objFont->setStrikethrough(true);
}
// offset: 4; size: 2; colour index
$colorIndex = self::getUInt2d($recordData, 4);
$objFont->colorIndex = $colorIndex;
// offset: 6; size: 2; font weight
$weight = self::getUInt2d($recordData, 6);
switch ($weight) {
case 0x02BC:
$objFont->setBold(true);
break;
}
// offset: 8; size: 2; escapement type
$escapement = self::getUInt2d($recordData, 8);
switch ($escapement) {
case 0x0001:
$objFont->setSuperscript(true);
break;
case 0x0002:
$objFont->setSubscript(true);
break;
}
// offset: 10; size: 1; underline type
$underlineType = ord($recordData[10]);
switch ($underlineType) {
case 0x00:
break; // no underline
case 0x01:
$objFont->setUnderline(Font::UNDERLINE_SINGLE);
break;
case 0x02:
$objFont->setUnderline(Font::UNDERLINE_DOUBLE);
break;
case 0x21:
$objFont->setUnderline(Font::UNDERLINE_SINGLEACCOUNTING);
break;
case 0x22:
$objFont->setUnderline(Font::UNDERLINE_DOUBLEACCOUNTING);
break;
}
// offset: 11; size: 1; font family
// offset: 12; size: 1; character set
// offset: 13; size: 1; not used
// offset: 14; size: var; font name
if ($this->version == self::XLS_BIFF8) {
$string = self::readUnicodeStringShort(substr($recordData, 14));
} else {
$string = $this->readByteStringShort(substr($recordData, 14));
}
$objFont->setName($string['value']);
$this->objFonts[] = $objFont;
}
}
/**
* FORMAT.
*
* This record contains information about a number format.
* All FORMAT records occur together in a sequential list.
*
* In BIFF2-BIFF4 other records referencing a FORMAT record
* contain a zero-based index into this list. From BIFF5 on
* the FORMAT record contains the index itself that will be
* used by other records.
*
* -- "OpenOffice.org's Documentation of the Microsoft
* Excel File Format"
*/
private function readFormat()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
if (!$this->readDataOnly) {
$indexCode = self::getUInt2d($recordData, 0);
if ($this->version == self::XLS_BIFF8) {
$string = self::readUnicodeStringLong(substr($recordData, 2));
} else {
// BIFF7
$string = $this->readByteStringShort(substr($recordData, 2));
}
$formatString = $string['value'];
$this->formats[$indexCode] = $formatString;
}
}
/**
* XF - Extended Format.
*
* This record contains formatting information for cells, rows, columns or styles.
* According to https://support.microsoft.com/en-us/help/147732 there are always at least 15 cell style XF
* and 1 cell XF.
* Inspection of Excel files generated by MS Office Excel shows that XF records 0-14 are cell style XF
* and XF record 15 is a cell XF
* We only read the first cell style XF and skip the remaining cell style XF records
* We read all cell XF records.
*
* -- "OpenOffice.org's Documentation of the Microsoft
* Excel File Format"
*/
private function readXf()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
$objStyle = new Style();
if (!$this->readDataOnly) {
// offset: 0; size: 2; Index to FONT record
if (self::getUInt2d($recordData, 0) < 4) {
$fontIndex = self::getUInt2d($recordData, 0);
} else {
// this has to do with that index 4 is omitted in all BIFF versions for some strange reason
// check the OpenOffice documentation of the FONT record
$fontIndex = self::getUInt2d($recordData, 0) - 1;
}
$objStyle->setFont($this->objFonts[$fontIndex]);
// offset: 2; size: 2; Index to FORMAT record
$numberFormatIndex = self::getUInt2d($recordData, 2);
if (isset($this->formats[$numberFormatIndex])) {
// then we have user-defined format code
$numberFormat = ['formatCode' => $this->formats[$numberFormatIndex]];
} elseif (($code = NumberFormat::builtInFormatCode($numberFormatIndex)) !== '') {
// then we have built-in format code
$numberFormat = ['formatCode' => $code];
} else {
// we set the general format code
$numberFormat = ['formatCode' => 'General'];
}
$objStyle->getNumberFormat()->setFormatCode($numberFormat['formatCode']);
// offset: 4; size: 2; XF type, cell protection, and parent style XF
// bit 2-0; mask 0x0007; XF_TYPE_PROT
$xfTypeProt = self::getUInt2d($recordData, 4);
// bit 0; mask 0x01; 1 = cell is locked
$isLocked = (0x01 & $xfTypeProt) >> 0;
$objStyle->getProtection()->setLocked($isLocked ? Protection::PROTECTION_INHERIT : Protection::PROTECTION_UNPROTECTED);
// bit 1; mask 0x02; 1 = Formula is hidden
$isHidden = (0x02 & $xfTypeProt) >> 1;
$objStyle->getProtection()->setHidden($isHidden ? Protection::PROTECTION_PROTECTED : Protection::PROTECTION_UNPROTECTED);
// bit 2; mask 0x04; 0 = Cell XF, 1 = Cell Style XF
$isCellStyleXf = (0x04 & $xfTypeProt) >> 2;
// offset: 6; size: 1; Alignment and text break
// bit 2-0, mask 0x07; horizontal alignment
$horAlign = (0x07 & ord($recordData[6])) >> 0;
switch ($horAlign) {
case 0:
$objStyle->getAlignment()->setHorizontal(Alignment::HORIZONTAL_GENERAL);
break;
case 1:
$objStyle->getAlignment()->setHorizontal(Alignment::HORIZONTAL_LEFT);
break;
case 2:
$objStyle->getAlignment()->setHorizontal(Alignment::HORIZONTAL_CENTER);
break;
case 3:
$objStyle->getAlignment()->setHorizontal(Alignment::HORIZONTAL_RIGHT);
break;
case 4:
$objStyle->getAlignment()->setHorizontal(Alignment::HORIZONTAL_FILL);
break;
case 5:
$objStyle->getAlignment()->setHorizontal(Alignment::HORIZONTAL_JUSTIFY);
break;
case 6:
$objStyle->getAlignment()->setHorizontal(Alignment::HORIZONTAL_CENTER_CONTINUOUS);
break;
}
// bit 3, mask 0x08; wrap text
$wrapText = (0x08 & ord($recordData[6])) >> 3;
switch ($wrapText) {
case 0:
$objStyle->getAlignment()->setWrapText(false);
break;
case 1:
$objStyle->getAlignment()->setWrapText(true);
break;
}
// bit 6-4, mask 0x70; vertical alignment
$vertAlign = (0x70 & ord($recordData[6])) >> 4;
switch ($vertAlign) {
case 0:
$objStyle->getAlignment()->setVertical(Alignment::VERTICAL_TOP);
break;
case 1:
$objStyle->getAlignment()->setVertical(Alignment::VERTICAL_CENTER);
break;
case 2:
$objStyle->getAlignment()->setVertical(Alignment::VERTICAL_BOTTOM);
break;
case 3:
$objStyle->getAlignment()->setVertical(Alignment::VERTICAL_JUSTIFY);
break;
}
if ($this->version == self::XLS_BIFF8) {
// offset: 7; size: 1; XF_ROTATION: Text rotation angle
$angle = ord($recordData[7]);
$rotation = 0;
if ($angle <= 90) {
$rotation = $angle;
} elseif ($angle <= 180) {
$rotation = 90 - $angle;
} elseif ($angle == 255) {
$rotation = -165;
}
$objStyle->getAlignment()->setTextRotation($rotation);
// offset: 8; size: 1; Indentation, shrink to cell size, and text direction
// bit: 3-0; mask: 0x0F; indent level
$indent = (0x0F & ord($recordData[8])) >> 0;
$objStyle->getAlignment()->setIndent($indent);
// bit: 4; mask: 0x10; 1 = shrink content to fit into cell
$shrinkToFit = (0x10 & ord($recordData[8])) >> 4;
switch ($shrinkToFit) {
case 0:
$objStyle->getAlignment()->setShrinkToFit(false);
break;
case 1:
$objStyle->getAlignment()->setShrinkToFit(true);
break;
}
// offset: 9; size: 1; Flags used for attribute groups
// offset: 10; size: 4; Cell border lines and background area
// bit: 3-0; mask: 0x0000000F; left style
if ($bordersLeftStyle = Xls\Style\Border::lookup((0x0000000F & self::getInt4d($recordData, 10)) >> 0)) {
$objStyle->getBorders()->getLeft()->setBorderStyle($bordersLeftStyle);
}
// bit: 7-4; mask: 0x000000F0; right style
if ($bordersRightStyle = Xls\Style\Border::lookup((0x000000F0 & self::getInt4d($recordData, 10)) >> 4)) {
$objStyle->getBorders()->getRight()->setBorderStyle($bordersRightStyle);
}
// bit: 11-8; mask: 0x00000F00; top style
if ($bordersTopStyle = Xls\Style\Border::lookup((0x00000F00 & self::getInt4d($recordData, 10)) >> 8)) {
$objStyle->getBorders()->getTop()->setBorderStyle($bordersTopStyle);
}
// bit: 15-12; mask: 0x0000F000; bottom style
if ($bordersBottomStyle = Xls\Style\Border::lookup((0x0000F000 & self::getInt4d($recordData, 10)) >> 12)) {
$objStyle->getBorders()->getBottom()->setBorderStyle($bordersBottomStyle);
}
// bit: 22-16; mask: 0x007F0000; left color
$objStyle->getBorders()->getLeft()->colorIndex = (0x007F0000 & self::getInt4d($recordData, 10)) >> 16;
// bit: 29-23; mask: 0x3F800000; right color
$objStyle->getBorders()->getRight()->colorIndex = (0x3F800000 & self::getInt4d($recordData, 10)) >> 23;
// bit: 30; mask: 0x40000000; 1 = diagonal line from top left to right bottom
$diagonalDown = (0x40000000 & self::getInt4d($recordData, 10)) >> 30 ? true : false;
// bit: 31; mask: 0x80000000; 1 = diagonal line from bottom left to top right
$diagonalUp = (0x80000000 & self::getInt4d($recordData, 10)) >> 31 ? true : false;
if ($diagonalUp == false && $diagonalDown == false) {
$objStyle->getBorders()->setDiagonalDirection(Borders::DIAGONAL_NONE);
} elseif ($diagonalUp == true && $diagonalDown == false) {
$objStyle->getBorders()->setDiagonalDirection(Borders::DIAGONAL_UP);
} elseif ($diagonalUp == false && $diagonalDown == true) {
$objStyle->getBorders()->setDiagonalDirection(Borders::DIAGONAL_DOWN);
} elseif ($diagonalUp == true && $diagonalDown == true) {
$objStyle->getBorders()->setDiagonalDirection(Borders::DIAGONAL_BOTH);
}
// offset: 14; size: 4;
// bit: 6-0; mask: 0x0000007F; top color
$objStyle->getBorders()->getTop()->colorIndex = (0x0000007F & self::getInt4d($recordData, 14)) >> 0;
// bit: 13-7; mask: 0x00003F80; bottom color
$objStyle->getBorders()->getBottom()->colorIndex = (0x00003F80 & self::getInt4d($recordData, 14)) >> 7;
// bit: 20-14; mask: 0x001FC000; diagonal color
$objStyle->getBorders()->getDiagonal()->colorIndex = (0x001FC000 & self::getInt4d($recordData, 14)) >> 14;
// bit: 24-21; mask: 0x01E00000; diagonal style
if ($bordersDiagonalStyle = Xls\Style\Border::lookup((0x01E00000 & self::getInt4d($recordData, 14)) >> 21)) {
$objStyle->getBorders()->getDiagonal()->setBorderStyle($bordersDiagonalStyle);
}
// bit: 31-26; mask: 0xFC000000 fill pattern
if ($fillType = Xls\Style\FillPattern::lookup((0xFC000000 & self::getInt4d($recordData, 14)) >> 26)) {
$objStyle->getFill()->setFillType($fillType);
}
// offset: 18; size: 2; pattern and background colour
// bit: 6-0; mask: 0x007F; color index for pattern color
$objStyle->getFill()->startcolorIndex = (0x007F & self::getUInt2d($recordData, 18)) >> 0;
// bit: 13-7; mask: 0x3F80; color index for pattern background
$objStyle->getFill()->endcolorIndex = (0x3F80 & self::getUInt2d($recordData, 18)) >> 7;
} else {
// BIFF5
// offset: 7; size: 1; Text orientation and flags
$orientationAndFlags = ord($recordData[7]);
// bit: 1-0; mask: 0x03; XF_ORIENTATION: Text orientation
$xfOrientation = (0x03 & $orientationAndFlags) >> 0;
switch ($xfOrientation) {
case 0:
$objStyle->getAlignment()->setTextRotation(0);
break;
case 1:
$objStyle->getAlignment()->setTextRotation(-165);
break;
case 2:
$objStyle->getAlignment()->setTextRotation(90);
break;
case 3:
$objStyle->getAlignment()->setTextRotation(-90);
break;
}
// offset: 8; size: 4; cell border lines and background area
$borderAndBackground = self::getInt4d($recordData, 8);
// bit: 6-0; mask: 0x0000007F; color index for pattern color
$objStyle->getFill()->startcolorIndex = (0x0000007F & $borderAndBackground) >> 0;
// bit: 13-7; mask: 0x00003F80; color index for pattern background
$objStyle->getFill()->endcolorIndex = (0x00003F80 & $borderAndBackground) >> 7;
// bit: 21-16; mask: 0x003F0000; fill pattern
$objStyle->getFill()->setFillType(Xls\Style\FillPattern::lookup((0x003F0000 & $borderAndBackground) >> 16));
// bit: 24-22; mask: 0x01C00000; bottom line style
$objStyle->getBorders()->getBottom()->setBorderStyle(Xls\Style\Border::lookup((0x01C00000 & $borderAndBackground) >> 22));
// bit: 31-25; mask: 0xFE000000; bottom line color
$objStyle->getBorders()->getBottom()->colorIndex = (0xFE000000 & $borderAndBackground) >> 25;
// offset: 12; size: 4; cell border lines
$borderLines = self::getInt4d($recordData, 12);
// bit: 2-0; mask: 0x00000007; top line style
$objStyle->getBorders()->getTop()->setBorderStyle(Xls\Style\Border::lookup((0x00000007 & $borderLines) >> 0));
// bit: 5-3; mask: 0x00000038; left line style
$objStyle->getBorders()->getLeft()->setBorderStyle(Xls\Style\Border::lookup((0x00000038 & $borderLines) >> 3));
// bit: 8-6; mask: 0x000001C0; right line style
$objStyle->getBorders()->getRight()->setBorderStyle(Xls\Style\Border::lookup((0x000001C0 & $borderLines) >> 6));
// bit: 15-9; mask: 0x0000FE00; top line color index
$objStyle->getBorders()->getTop()->colorIndex = (0x0000FE00 & $borderLines) >> 9;
// bit: 22-16; mask: 0x007F0000; left line color index
$objStyle->getBorders()->getLeft()->colorIndex = (0x007F0000 & $borderLines) >> 16;
// bit: 29-23; mask: 0x3F800000; right line color index
$objStyle->getBorders()->getRight()->colorIndex = (0x3F800000 & $borderLines) >> 23;
}
// add cellStyleXf or cellXf and update mapping
if ($isCellStyleXf) {
// we only read one style XF record which is always the first
if ($this->xfIndex == 0) {
$this->spreadsheet->addCellStyleXf($objStyle);
$this->mapCellStyleXfIndex[$this->xfIndex] = 0;
}
} else {
// we read all cell XF records
$this->spreadsheet->addCellXf($objStyle);
$this->mapCellXfIndex[$this->xfIndex] = count($this->spreadsheet->getCellXfCollection()) - 1;
}
// update XF index for when we read next record
++$this->xfIndex;
}
}
private function readXfExt()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
if (!$this->readDataOnly) {
// offset: 0; size: 2; 0x087D = repeated header
// offset: 2; size: 2
// offset: 4; size: 8; not used
// offset: 12; size: 2; record version
// offset: 14; size: 2; index to XF record which this record modifies
$ixfe = self::getUInt2d($recordData, 14);
// offset: 16; size: 2; not used
// offset: 18; size: 2; number of extension properties that follow
$cexts = self::getUInt2d($recordData, 18);
// start reading the actual extension data
$offset = 20;
while ($offset < $length) {
// extension type
$extType = self::getUInt2d($recordData, $offset);
// extension length
$cb = self::getUInt2d($recordData, $offset + 2);
// extension data
$extData = substr($recordData, $offset + 4, $cb);
switch ($extType) {
case 4: // fill start color
$xclfType = self::getUInt2d($extData, 0); // color type
$xclrValue = substr($extData, 4, 4); // color value (value based on color type)
if ($xclfType == 2) {
$rgb = sprintf('%02X%02X%02X', ord($xclrValue[0]), ord($xclrValue[1]), ord($xclrValue[2]));
// modify the relevant style property
if (isset($this->mapCellXfIndex[$ixfe])) {
$fill = $this->spreadsheet->getCellXfByIndex($this->mapCellXfIndex[$ixfe])->getFill();
$fill->getStartColor()->setRGB($rgb);
unset($fill->startcolorIndex); // normal color index does not apply, discard
}
}
break;
case 5: // fill end color
$xclfType = self::getUInt2d($extData, 0); // color type
$xclrValue = substr($extData, 4, 4); // color value (value based on color type)
if ($xclfType == 2) {
$rgb = sprintf('%02X%02X%02X', ord($xclrValue[0]), ord($xclrValue[1]), ord($xclrValue[2]));
// modify the relevant style property
if (isset($this->mapCellXfIndex[$ixfe])) {
$fill = $this->spreadsheet->getCellXfByIndex($this->mapCellXfIndex[$ixfe])->getFill();
$fill->getEndColor()->setRGB($rgb);
unset($fill->endcolorIndex); // normal color index does not apply, discard
}
}
break;
case 7: // border color top
$xclfType = self::getUInt2d($extData, 0); // color type
$xclrValue = substr($extData, 4, 4); // color value (value based on color type)
if ($xclfType == 2) {
$rgb = sprintf('%02X%02X%02X', ord($xclrValue[0]), ord($xclrValue[1]), ord($xclrValue[2]));
// modify the relevant style property
if (isset($this->mapCellXfIndex[$ixfe])) {
$top = $this->spreadsheet->getCellXfByIndex($this->mapCellXfIndex[$ixfe])->getBorders()->getTop();
$top->getColor()->setRGB($rgb);
unset($top->colorIndex); // normal color index does not apply, discard
}
}
break;
case 8: // border color bottom
$xclfType = self::getUInt2d($extData, 0); // color type
$xclrValue = substr($extData, 4, 4); // color value (value based on color type)
if ($xclfType == 2) {
$rgb = sprintf('%02X%02X%02X', ord($xclrValue[0]), ord($xclrValue[1]), ord($xclrValue[2]));
// modify the relevant style property
if (isset($this->mapCellXfIndex[$ixfe])) {
$bottom = $this->spreadsheet->getCellXfByIndex($this->mapCellXfIndex[$ixfe])->getBorders()->getBottom();
$bottom->getColor()->setRGB($rgb);
unset($bottom->colorIndex); // normal color index does not apply, discard
}
}
break;
case 9: // border color left
$xclfType = self::getUInt2d($extData, 0); // color type
$xclrValue = substr($extData, 4, 4); // color value (value based on color type)
if ($xclfType == 2) {
$rgb = sprintf('%02X%02X%02X', ord($xclrValue[0]), ord($xclrValue[1]), ord($xclrValue[2]));
// modify the relevant style property
if (isset($this->mapCellXfIndex[$ixfe])) {
$left = $this->spreadsheet->getCellXfByIndex($this->mapCellXfIndex[$ixfe])->getBorders()->getLeft();
$left->getColor()->setRGB($rgb);
unset($left->colorIndex); // normal color index does not apply, discard
}
}
break;
case 10: // border color right
$xclfType = self::getUInt2d($extData, 0); // color type
$xclrValue = substr($extData, 4, 4); // color value (value based on color type)
if ($xclfType == 2) {
$rgb = sprintf('%02X%02X%02X', ord($xclrValue[0]), ord($xclrValue[1]), ord($xclrValue[2]));
// modify the relevant style property
if (isset($this->mapCellXfIndex[$ixfe])) {
$right = $this->spreadsheet->getCellXfByIndex($this->mapCellXfIndex[$ixfe])->getBorders()->getRight();
$right->getColor()->setRGB($rgb);
unset($right->colorIndex); // normal color index does not apply, discard
}
}
break;
case 11: // border color diagonal
$xclfType = self::getUInt2d($extData, 0); // color type
$xclrValue = substr($extData, 4, 4); // color value (value based on color type)
if ($xclfType == 2) {
$rgb = sprintf('%02X%02X%02X', ord($xclrValue[0]), ord($xclrValue[1]), ord($xclrValue[2]));
// modify the relevant style property
if (isset($this->mapCellXfIndex[$ixfe])) {
$diagonal = $this->spreadsheet->getCellXfByIndex($this->mapCellXfIndex[$ixfe])->getBorders()->getDiagonal();
$diagonal->getColor()->setRGB($rgb);
unset($diagonal->colorIndex); // normal color index does not apply, discard
}
}
break;
case 13: // font color
$xclfType = self::getUInt2d($extData, 0); // color type
$xclrValue = substr($extData, 4, 4); // color value (value based on color type)
if ($xclfType == 2) {
$rgb = sprintf('%02X%02X%02X', ord($xclrValue[0]), ord($xclrValue[1]), ord($xclrValue[2]));
// modify the relevant style property
if (isset($this->mapCellXfIndex[$ixfe])) {
$font = $this->spreadsheet->getCellXfByIndex($this->mapCellXfIndex[$ixfe])->getFont();
$font->getColor()->setRGB($rgb);
unset($font->colorIndex); // normal color index does not apply, discard
}
}
break;
}
$offset += $cb;
}
}
}
/**
* Read STYLE record.
*/
private function readStyle()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
if (!$this->readDataOnly) {
// offset: 0; size: 2; index to XF record and flag for built-in style
$ixfe = self::getUInt2d($recordData, 0);
// bit: 11-0; mask 0x0FFF; index to XF record
$xfIndex = (0x0FFF & $ixfe) >> 0;
// bit: 15; mask 0x8000; 0 = user-defined style, 1 = built-in style
$isBuiltIn = (bool) ((0x8000 & $ixfe) >> 15);
if ($isBuiltIn) {
// offset: 2; size: 1; identifier for built-in style
$builtInId = ord($recordData[2]);
switch ($builtInId) {
case 0x00:
// currently, we are not using this for anything
break;
default:
break;
}
}
// user-defined; not supported by PhpSpreadsheet
}
}
/**
* Read PALETTE record.
*/
private function readPalette()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
if (!$this->readDataOnly) {
// offset: 0; size: 2; number of following colors
$nm = self::getUInt2d($recordData, 0);
// list of RGB colors
for ($i = 0; $i < $nm; ++$i) {
$rgb = substr($recordData, 2 + 4 * $i, 4);
$this->palette[] = self::readRGB($rgb);
}
}
}
/**
* SHEET.
*
* This record is located in the Workbook Globals
* Substream and represents a sheet inside the workbook.
* One SHEET record is written for each sheet. It stores the
* sheet name and a stream offset to the BOF record of the
* respective Sheet Substream within the Workbook Stream.
*
* -- "OpenOffice.org's Documentation of the Microsoft
* Excel File Format"
*/
private function readSheet()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// offset: 0; size: 4; absolute stream position of the BOF record of the sheet
// NOTE: not encrypted
$rec_offset = self::getInt4d($this->data, $this->pos + 4);
// move stream pointer to next record
$this->pos += 4 + $length;
// offset: 4; size: 1; sheet state
switch (ord($recordData[4])) {
case 0x00:
$sheetState = Worksheet::SHEETSTATE_VISIBLE;
break;
case 0x01:
$sheetState = Worksheet::SHEETSTATE_HIDDEN;
break;
case 0x02:
$sheetState = Worksheet::SHEETSTATE_VERYHIDDEN;
break;
default:
$sheetState = Worksheet::SHEETSTATE_VISIBLE;
break;
}
// offset: 5; size: 1; sheet type
$sheetType = ord($recordData[5]);
// offset: 6; size: var; sheet name
if ($this->version == self::XLS_BIFF8) {
$string = self::readUnicodeStringShort(substr($recordData, 6));
$rec_name = $string['value'];
} elseif ($this->version == self::XLS_BIFF7) {
$string = $this->readByteStringShort(substr($recordData, 6));
$rec_name = $string['value'];
}
$this->sheets[] = [
'name' => $rec_name,
'offset' => $rec_offset,
'sheetState' => $sheetState,
'sheetType' => $sheetType,
];
}
/**
* Read EXTERNALBOOK record.
*/
private function readExternalBook()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
// offset within record data
$offset = 0;
// there are 4 types of records
if (strlen($recordData) > 4) {
// external reference
// offset: 0; size: 2; number of sheet names ($nm)
$nm = self::getUInt2d($recordData, 0);
$offset += 2;
// offset: 2; size: var; encoded URL without sheet name (Unicode string, 16-bit length)
$encodedUrlString = self::readUnicodeStringLong(substr($recordData, 2));
$offset += $encodedUrlString['size'];
// offset: var; size: var; list of $nm sheet names (Unicode strings, 16-bit length)
$externalSheetNames = [];
for ($i = 0; $i < $nm; ++$i) {
$externalSheetNameString = self::readUnicodeStringLong(substr($recordData, $offset));
$externalSheetNames[] = $externalSheetNameString['value'];
$offset += $externalSheetNameString['size'];
}
// store the record data
$this->externalBooks[] = [
'type' => 'external',
'encodedUrl' => $encodedUrlString['value'],
'externalSheetNames' => $externalSheetNames,
];
} elseif (substr($recordData, 2, 2) == pack('CC', 0x01, 0x04)) {
// internal reference
// offset: 0; size: 2; number of sheet in this document
// offset: 2; size: 2; 0x01 0x04
$this->externalBooks[] = [
'type' => 'internal',
];
} elseif (substr($recordData, 0, 4) == pack('vCC', 0x0001, 0x01, 0x3A)) {
// add-in function
// offset: 0; size: 2; 0x0001
$this->externalBooks[] = [
'type' => 'addInFunction',
];
} elseif (substr($recordData, 0, 2) == pack('v', 0x0000)) {
// DDE links, OLE links
// offset: 0; size: 2; 0x0000
// offset: 2; size: var; encoded source document name
$this->externalBooks[] = [
'type' => 'DDEorOLE',
];
}
}
/**
* Read EXTERNNAME record.
*/
private function readExternName()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
// external sheet references provided for named cells
if ($this->version == self::XLS_BIFF8) {
// offset: 0; size: 2; options
$options = self::getUInt2d($recordData, 0);
// offset: 2; size: 2;
// offset: 4; size: 2; not used
// offset: 6; size: var
$nameString = self::readUnicodeStringShort(substr($recordData, 6));
// offset: var; size: var; formula data
$offset = 6 + $nameString['size'];
$formula = $this->getFormulaFromStructure(substr($recordData, $offset));
$this->externalNames[] = [
'name' => $nameString['value'],
'formula' => $formula,
];
}
}
/**
* Read EXTERNSHEET record.
*/
private function readExternSheet()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
// external sheet references provided for named cells
if ($this->version == self::XLS_BIFF8) {
// offset: 0; size: 2; number of following ref structures
$nm = self::getUInt2d($recordData, 0);
for ($i = 0; $i < $nm; ++$i) {
$this->ref[] = [
// offset: 2 + 6 * $i; index to EXTERNALBOOK record
'externalBookIndex' => self::getUInt2d($recordData, 2 + 6 * $i),
// offset: 4 + 6 * $i; index to first sheet in EXTERNALBOOK record
'firstSheetIndex' => self::getUInt2d($recordData, 4 + 6 * $i),
// offset: 6 + 6 * $i; index to last sheet in EXTERNALBOOK record
'lastSheetIndex' => self::getUInt2d($recordData, 6 + 6 * $i),
];
}
}
}
/**
* DEFINEDNAME.
*
* This record is part of a Link Table. It contains the name
* and the token array of an internal defined name. Token
* arrays of defined names contain tokens with aberrant
* token classes.
*
* -- "OpenOffice.org's Documentation of the Microsoft
* Excel File Format"
*/
private function readDefinedName()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
$recordData = $this->readRecordData($this->data, $this->pos + 4, $length);
// move stream pointer to next record
$this->pos += 4 + $length;
if ($this->version == self::XLS_BIFF8) {
// retrieves named cells
// offset: 0; size: 2; option flags
$opts = self::getUInt2d($recordData, 0);
// bit: 5; mask: 0x0020; 0 = user-defined name, 1 = built-in-name
$isBuiltInName = (0x0020 & $opts) >> 5;
// offset: 2; size: 1; keyboard shortcut
// offset: 3; size: 1; length of the name (character count)
$nlen = ord($recordData[3]);
// offset: 4; size: 2; size of the formula data (it can happen that this is zero)
// note: there can also be additional data, this is not included in $flen
$flen = self::getUInt2d($recordData, 4);
// offset: 8; size: 2; 0=Global name, otherwise index to sheet (1-based)
$scope = self::getUInt2d($recordData, 8);
// offset: 14; size: var; Name (Unicode string without length field)
$string = self::readUnicodeString(substr($recordData, 14), $nlen);
// offset: var; size: $flen; formula data
$offset = 14 + $string['size'];
$formulaStructure = pack('v', $flen) . substr($recordData, $offset);
try {
$formula = $this->getFormulaFromStructure($formulaStructure);
} catch (PhpSpreadsheetException $e) {
$formula = '';
}
$this->definedname[] = [
'isBuiltInName' => $isBuiltInName,
'name' => $string['value'],
'formula' => $formula,
'scope' => $scope,
];
}
}
/**
* Read MSODRAWINGGROUP record.
*/
private function readMsoDrawingGroup()
{
$length = self::getUInt2d($this->data, $this->pos + 2);
// get spliced record data
$splicedRecordData = $this->getSplicedRecordData();
$recordData = $splicedRecordData['recordData'];
$this->drawingGroupData .= $recordData;
}
/**
* SST - Shared String Table.
*
* This record contains a list of all strings used anywhere
* in the workbook. Each string occurs only once. The
* workbook uses indexes into the list to reference the
* strings.
*
* -- "OpenOffice.org's Documentation of the Microsoft
* Excel File Format"
*/
private function readSst()
{
// offset within (spliced) record data
$pos = 0;
// get spliced record data
$splicedRecordData = $this->getSplicedRecordData();
$recordData = $splicedRecordData['recordData'];
$spliceOffsets = $splicedRecordData['spliceOffsets'];
// offset: 0; size: 4; total number of strings in the workbook
$pos += 4;
// offset: 4; size: 4; number of following strings ($nm)
$nm = self::getInt4d($recordData, 4);
$pos += 4;
// loop through the Unicode strings (16-bit length)
for ($i = 0; $i < $nm; ++$i) {
// number of characters in the Unicode string
$numChars = self::getUInt2d($recordData, $pos);
$pos += 2;
// option flags
$optionFlags = ord($recordData[$pos]);
++$pos;
// bit: 0; mask: 0x01; 0 = compressed; 1 = uncompressed
$isCompressed = (($optionFlags & 0x01) == 0);
// bit: 2; mask: 0x02; 0 = ordinary; 1 = Asian phonetic
$hasAsian = (($optionFlags & 0x04) != 0);
// bit: 3; mask: 0x03; 0 = ordinary; 1 = Rich-Text
$hasRichText = (($optionFlags & 0x08) != 0);
if ($hasRichText) {
// number of Rich-Text formatting runs
$formattingRuns = self::getUInt2d($recordData, $pos);
$pos += 2;
}
if ($hasAsian) {
// size of Asian phonetic setting
$extendedRunLength = self::getInt4d($recordData, $pos);
$pos += 4;
}
// expected byte length of character array if not split
$len = ($isCompressed) ? $numChars : $numChars * 2;
// look up limit position
foreach ($spliceOffsets as $spliceOffset) {
// it can happen that the string is empty, therefore we need
// <= and not just <
if ($pos <= $spliceOffset) {
$limitpos = $spliceOffset;
break;
}
}
if ($pos + $len <= $limitpos) {