Below is the code which demonstrates the importing the excel to database using php
File 1: index.php
<html>
<head>
<title>::READ EXCEL UPLOADED::</title>
</head>
<body>
<form name="maytasway" enctype="multipart/form-data" action="UploadExcel.php" method="post">
<input type="hidden" name="MAX_FILE_SIZE" value="2000000" />
<table width="600">
<tr>
<td>CSV File Name</td>
<td><input type="file" name="file" /></td>
<td> </td>
</tr>
<tr>
<td> </td>
<td><input type="submit" value="Submit" /></td>
<td> </td>
</tr>
</table>
</form>
</body>
</html>
File 2: UploadExcel.php
<?php
require_once 'Excel/reader.php';
require_once 'Excel/Date.php';
$data = new Spreadsheet_Excel_Reader();
error_reporting(E_ALL ^ E_NOTICE);
$link =mysql_connect('localhost','root','root');
if (!$link) {
die('Could not connect: ' . mysql_error());
}
echo 'Connected successfully<br/>';
mysql_select_db('myDB');
if ( $_FILES['file']['tmp_name'] )
{
$fp = fopen($_FILES['file']['tmp_name'], 'r');
$firstline=false;
$res="";
$cnt=0;
$data->read($_FILES['file']['tmp_name']);
for ($j = 1; $j <= $data->sheets[0]['numRows']-1; $j++)
{
//THIS COMMENTED CODE SHOWS THE WAY TO PARSE THE DATE FIELD
//if(is_numeric($doj))
//{
//echo $fname.'->'.$doj.'-FIRST<br/>';
//$doj = PHPExcel_Shared_Date::ExcelToPHP($doj);
//$doj =date("d/m/Y",$doj);
//}
//$doj =date("Y/m/d",$doj);
//the $val now contain a value like 1243641600 and pass this $val to php date function,
//$d =date("d/m/Y",$val);
//echo $fname.'->'.$doj.'-SECOND<br/>';
//$doj = PHPExcel_Shared_Date::ExcelToPHPObject($doj);
//$doj =date("d/m/Y",$doj);
//echo $fname.'->'.$doj.'<br/>';
//$date1 = date_parse($doj);
//$date2 = date_parse($doj);
// echo gmmktime(0, 0, 0, $date1['month'], $date1['day'], $date1['year']) .'<br/>';
//$doj=strtotime($doj);
//$doj = PHPExcel_Shared_Date::ExcelToPHP($doj);
//$doj =date("d/m/Y",$doj);
//echo $fname.'->'.$doj.'<br/>';
// Numbers of days between January 1, 1900 and 1970 (including 19 leap years)
//echo excel2timestamp($doj);
$ID=$data->sheets[0]['cells'][$j+1][1];
$NAME=$data->sheets[0]['cells'][$j+1][2];
$EMPLOYEEID=$data->sheets[0]['cells'][$j+1][3];
$query="insert into employee (id,name,empid) values ('".$ID."','".$NAME."','".$EMPLOYEEID."')";
//echo $query.'<br/>';
mysql_query($query) or die("Error in insert");
$cnt++;
}//for
}
echo "$cnt fields affected";
mysql_close($link);
?>
File3: Date.php
<?php
class PHPExcel_Shared_Date
{
/** constants */
const CALENDAR_WINDOWS_1900 = 1900; // Base date of 1st Jan 1900 = 1.0
const CALENDAR_MAC_1904 = 1904; // Base date of 2nd Jan 1904 = 1.0
private static $ExcelBaseDate = self::CALENDAR_WINDOWS_1900;
public static $dateTimeObjectType = 'DateTime';
/**
* Set the Excel calendar (Windows 1900 or Mac 1904)
*
* @param integer $baseDate Excel base date
* @return boolean Success or failure
*/
public static function setExcelCalendar($baseDate) {
if (($baseDate == self::CALENDAR_WINDOWS_1900) ||
($baseDate == self::CALENDAR_MAC_1904)) {
self::$ExcelBaseDate = $baseDate;
return True;
}
return False;
} // function setExcelCalendar()
/**
* Return the Excel calendar (Windows 1900 or Mac 1904)
*
* @return integer $baseDate Excel base date
*/
public static function getExcelCalendar() {
return self::$ExcelBaseDate;
} // function getExcelCalendar()
/**
* Convert a date from Excel to PHP
*
* @param long $dateValue Excel date/time value
* @return long PHP serialized date/time
*/
public static function ExcelToPHP($dateValue = 0) {
if (self::$ExcelBaseDate == self::CALENDAR_WINDOWS_1900) {
$myExcelBaseDate = 25569;
// Adjust for the spurious 29-Feb-1900 (Day 60)
if ($dateValue < 60) {
--$myExcelBaseDate;
}
} else {
$myExcelBaseDate = 24107;
}
// Perform conversion
if ($dateValue >= 1) {
$utcDays = $dateValue - $myExcelBaseDate;
$returnValue = round($utcDays * 24 * 60 * 60);
if (($returnValue <= PHP_INT_MAX) && ($returnValue >= -PHP_INT_MAX)) {
$returnValue = (integer) $returnValue;
}
} else {
$hours = round($dateValue * 24);
$mins = round($dateValue * 24 * 60) - round($hours * 60);
$secs = round($dateValue * 24 * 60 * 60) - round($hours * 60 * 60) - round($mins * 60);
$returnValue = (integer) gmmktime($hours, $mins, $secs);
}
// Return
return $returnValue;
} // function ExcelToPHP()
/**
* Convert a date from Excel to a PHP Date/Time object
*
* @param long $dateValue Excel date/time value
* @return long PHP date/time object
*/
public static function ExcelToPHPObject($dateValue = 0) {
$dateTime = self::ExcelToPHP($dateValue);
$days = floor($dateTime / 86400);
$time = round((($dateTime / 86400) - $days) * 86400);
$hours = round($time / 3600);
$minutes = round($time / 60) - ($hours * 60);
$seconds = round($time) - ($hours * 3600) - ($minutes * 60);
$dateObj = date_create('1-Jan-1970+'.$days.' days');
$dateObj->setTime($hours,$minutes,$seconds);
return $dateObj;
} // function ExcelToPHPObject()
/**
* Convert a date from PHP to Excel
*
* @param mixed $dateValue PHP serialized date/time or date object
* @return mixed Excel date/time value
* or boolean False on failure
*/
public static function PHPToExcel($dateValue = 0) {
$saveTimeZone = date_default_timezone_get();
date_default_timezone_set('UTC');
$retValue = False;
if ((is_object($dateValue)) && ($dateValue instanceof self::$dateTimeObjectType)) {
$retValue = self::FormattedPHPToExcel( $dateValue->format('Y'), $dateValue->format('m'), $dateValue->format('d'),
$dateValue->format('H'), $dateValue->format('i'), $dateValue->format('s')
);
} elseif (is_numeric($dateValue)) {
$retValue = self::FormattedPHPToExcel( date('Y',$dateValue), date('m',$dateValue), date('d',$dateValue),
date('H',$dateValue), date('i',$dateValue), date('s',$dateValue)
);
}
date_default_timezone_set($saveTimeZone);
return $retValue;
} // function PHPToExcel()
/**
* FormattedPHPToExcel
*
* @param long $year
* @param long $month
* @param long $day
* @param long $hours
* @param long $minutes
* @param long $seconds
* @return long Excel date/time value
*/
public static function FormattedPHPToExcel($year, $month, $day, $hours=0, $minutes=0, $seconds=0) {
if (self::$ExcelBaseDate == self::CALENDAR_WINDOWS_1900) {
//
// Fudge factor for the erroneous fact that the year 1900 is treated as a Leap Year in MS Excel
// This affects every date following 28th February 1900
//
$excel1900isLeapYear = True;
if (($year == 1900) && ($month <= 2)) { $excel1900isLeapYear = False; }
$myExcelBaseDate = 2415020;
} else {
$myExcelBaseDate = 2416481;
$excel1900isLeapYear = False;
}
// Julian base date Adjustment
if ($month > 2) {
$month = $month - 3;
} else {
$month = $month + 9;
--$year;
}
// Calculate the Julian Date, then subtract the Excel base date (JD 2415020 = 31-Dec-1899 Giving Excel Date of 0)
$century = substr($year,0,2);
$decade = substr($year,2,2);
$excelDate = floor((146097 * $century) / 4) + floor((1461 * $decade) / 4) + floor((153 * $month + 2) / 5) + $day + 1721119 - $myExcelBaseDate + $excel1900isLeapYear;
$excelTime = (($hours * 3600) + ($minutes * 60) + $seconds) / 86400;
return (float) $excelDate + $excelTime;
} // function FormattedPHPToExcel()
/**
* Is a given cell a date/time?
*
* @param PHPExcel_Cell $pCell
* @return boolean
*/
public static function isDateTime(PHPExcel_Cell $pCell) {
return self::isDateTimeFormat($pCell->getParent()->getStyle($pCell->getCoordinate())->getNumberFormat());
} // function isDateTime()
/**
* Is a given number format a date/time?
*
* @param PHPExcel_Style_NumberFormat $pFormat
* @return boolean
*/
public static function isDateTimeFormat(PHPExcel_Style_NumberFormat $pFormat) {
return self::isDateTimeFormatCode($pFormat->getFormatCode());
} // function isDateTimeFormat()
private static $possibleDateFormatCharacters = 'ymdHs';
/**
* Is a given number format code a date/time?
*
* @param string $pFormatCode
* @return boolean
*/
public static function isDateTimeFormatCode($pFormatCode = '') {
// Switch on formatcode
switch ($pFormatCode) {
case PHPExcel_Style_NumberFormat::FORMAT_DATE_YYYYMMDD:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_YYYYMMDD2:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_DDMMYYYY:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_DMYSLASH:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_DMYMINUS:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_DMMINUS:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_MYMINUS:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_DATETIME:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_TIME1:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_TIME2:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_TIME3:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_TIME4:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_TIME5:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_TIME6:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_TIME7:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_TIME8:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_YYYYMMDDSLASH:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_XLSX14:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_XLSX15:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_XLSX16:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_XLSX17:
case PHPExcel_Style_NumberFormat::FORMAT_DATE_XLSX22:
return true;
}
// Typically number, currency or accounting (or occasionally fraction) formats
if ((substr($pFormatCode,0,1) == '_') || (substr($pFormatCode,0,2) == '0 ')) {
return false;
}
// Try checking for any of the date formatting characters that don't appear within square braces
if (preg_match('/(^|\])[^\[]*['.self::$possibleDateFormatCharacters.']/i',$pFormatCode)) {
// We might also have a format mask containing quoted strings...
// we don't want to test for any of our characters within the quoted blocks
if (strpos($pFormatCode,'"') !== false) {
$i = false;
foreach(explode('"',$pFormatCode) as $subVal) {
// Only test in alternate array entries (the non-quoted blocks)
if (($i = !$i) && (preg_match('/(^|\])[^\[]*['.self::$possibleDateFormatCharacters.']/i',$subVal))) {
return true;
}
}
return false;
}
return true;
}
// No date...
return false;
} // function isDateTimeFormatCode()
/**
* Convert a date/time string to Excel time
*
* @param string $dateValue Examples: '2009-12-31', '2009-12-31 15:59', '2009-12-31 15:59:10'
* @return float|false Excel date/time serial value
*/
public static function stringToExcel($dateValue = '') {
if (strlen($dateValue) < 2)
return false;
if (!preg_match('/^(\d{1,4}[ \.\/\-][A-Z]{3,9}([ \.\/\-]\d{1,4})?|[A-Z]{3,9}[ \.\/\-]\d{1,4}([ \.\/\-]\d{1,4})?|\d{1,4}[ \.\/\-]\d{1,4}([ \.\/\-]\d{1,4})?)( \d{1,2}:\d{1,2}(:\d{1,2})?)?$/iu', $dateValue))
return false;
$dateValueNew = PHPExcel_Calculation_DateTime::DATEVALUE($dateValue);
if ($dateValueNew === PHPExcel_Calculation_Functions::VALUE()) {
return false;
} else {
if (strpos($dateValue, ':') !== false) {
$timeValue = PHPExcel_Calculation_DateTime::TIMEVALUE($dateValue);
if ($timeValue === PHPExcel_Calculation_Functions::VALUE()) {
return false;
}
$dateValueNew += $timeValue;
}
return $dateValueNew;
}
}
}
?>
File 4: reader.php
<?php
//require_once 'PEAR.php';
require_once 'oleread.inc';
//define('Spreadsheet_Excel_Reader_HAVE_ICONV', function_exists('iconv'));
//define('Spreadsheet_Excel_Reader_HAVE_MB', function_exists('mb_convert_encoding'));
define('Spreadsheet_Excel_Reader_BIFF8', 0x600);
define('Spreadsheet_Excel_Reader_BIFF7', 0x500);
define('Spreadsheet_Excel_Reader_WorkbookGlobals', 0x5);
define('Spreadsheet_Excel_Reader_Worksheet', 0x10);
define('Spreadsheet_Excel_Reader_Type_BOF', 0x809);
define('Spreadsheet_Excel_Reader_Type_EOF', 0x0a);
define('Spreadsheet_Excel_Reader_Type_BOUNDSHEET', 0x85);
define('Spreadsheet_Excel_Reader_Type_DIMENSION', 0x200);
define('Spreadsheet_Excel_Reader_Type_ROW', 0x208);
define('Spreadsheet_Excel_Reader_Type_DBCELL', 0xd7);
define('Spreadsheet_Excel_Reader_Type_FILEPASS', 0x2f);
define('Spreadsheet_Excel_Reader_Type_NOTE', 0x1c);
define('Spreadsheet_Excel_Reader_Type_TXO', 0x1b6);
define('Spreadsheet_Excel_Reader_Type_RK', 0x7e);
define('Spreadsheet_Excel_Reader_Type_RK2', 0x27e);
define('Spreadsheet_Excel_Reader_Type_MULRK', 0xbd);
define('Spreadsheet_Excel_Reader_Type_MULBLANK', 0xbe);
define('Spreadsheet_Excel_Reader_Type_INDEX', 0x20b);
define('Spreadsheet_Excel_Reader_Type_SST', 0xfc);
define('Spreadsheet_Excel_Reader_Type_EXTSST', 0xff);
define('Spreadsheet_Excel_Reader_Type_CONTINUE', 0x3c);
define('Spreadsheet_Excel_Reader_Type_LABEL', 0x204);
define('Spreadsheet_Excel_Reader_Type_LABELSST', 0xfd);
define('Spreadsheet_Excel_Reader_Type_NUMBER', 0x203);
define('Spreadsheet_Excel_Reader_Type_NAME', 0x18);
define('Spreadsheet_Excel_Reader_Type_ARRAY', 0x221);
define('Spreadsheet_Excel_Reader_Type_STRING', 0x207);
define('Spreadsheet_Excel_Reader_Type_FORMULA', 0x406);
define('Spreadsheet_Excel_Reader_Type_FORMULA2', 0x6);
define('Spreadsheet_Excel_Reader_Type_FORMAT', 0x41e);
define('Spreadsheet_Excel_Reader_Type_XF', 0xe0);
define('Spreadsheet_Excel_Reader_Type_BOOLERR', 0x205);
define('Spreadsheet_Excel_Reader_Type_UNKNOWN', 0xffff);
define('Spreadsheet_Excel_Reader_Type_NINETEENFOUR', 0x22);
define('Spreadsheet_Excel_Reader_Type_MERGEDCELLS', 0xE5);
define('Spreadsheet_Excel_Reader_utcOffsetDays' , 25569);
define('Spreadsheet_Excel_Reader_utcOffsetDays1904', 24107);
define('Spreadsheet_Excel_Reader_msInADay', 24 * 60 * 60);
//define('Spreadsheet_Excel_Reader_DEF_NUM_FORMAT', "%.2f");
define('Spreadsheet_Excel_Reader_DEF_NUM_FORMAT', "%s");
// function file_get_contents for PHP < 4.3.0
// Thanks Marian Steinbach for this function
if (!function_exists('file_get_contents')) {
function file_get_contents($filename, $use_include_path = 0) {
$data = '';
$file = @fopen($filename, "rb", $use_include_path);
if ($file) {
while (!feof($file)) $data .= fread($file, 1024);
fclose($file);
} else {
// There was a problem opening the file
$data = FALSE;
}
return $data;
}
}
//class Spreadsheet_Excel_Reader extends PEAR {
class Spreadsheet_Excel_Reader {
var $boundsheets = array();
var $formatRecords = array();
var $sst = array();
var $sheets = array();
var $data;
var $pos;
var $_ole;
var $_defaultEncoding;
var $_defaultFormat = Spreadsheet_Excel_Reader_DEF_NUM_FORMAT;
var $_columnsFormat = array();
var $_rowoffset = 1;
var $_coloffset = 1;
var $dateFormats = array (
0xe => "d/m/Y",
0xf => "d-M-Y",
0x10 => "d-M",
0x11 => "M-Y",
0x12 => "h:i a",
0x13 => "h:i:s a",
0x14 => "H:i",
0x15 => "H:i:s",
0x16 => "d/m/Y H:i",
0x2d => "i:s",
0x2e => "H:i:s",
0x2f => "i:s.S");
var $numberFormats = array(
0x1 => "%1.0f", // "0"
0x2 => "%1.2f", // "0.00",
0x3 => "%1.0f", //"#,##0",
0x4 => "%1.2f", //"#,##0.00",
0x5 => "%1.0f", /*"$#,##0;($#,##0)",*/
0x6 => '$%1.0f', /*"$#,##0;($#,##0)",*/
0x7 => '$%1.2f', //"$#,##0.00;($#,##0.00)",
0x8 => '$%1.2f', //"$#,##0.00;($#,##0.00)",
0x9 => '%1.0f%%', // "0%"
0xa => '%1.2f%%', // "0.00%"
0xb => '%1.2f', // 0.00E00",
0x25 => '%1.0f', // "#,##0;(#,##0)",
0x26 => '%1.0f', //"#,##0;(#,##0)",
0x27 => '%1.2f', //"#,##0.00;(#,##0.00)",
0x28 => '%1.2f', //"#,##0.00;(#,##0.00)",
0x29 => '%1.0f', //"#,##0;(#,##0)",
0x2a => '$%1.0f', //"$#,##0;($#,##0)",
0x2b => '%1.2f', //"#,##0.00;(#,##0.00)",
0x2c => '$%1.2f', //"$#,##0.00;($#,##0.00)",
0x30 => '%1.0f'); //"##0.0E0";
function Spreadsheet_Excel_Reader(){
$this->_ole =& new OLERead();
$this->setUTFEncoder('iconv');
}
function setOutputEncoding($Encoding){
$this->_defaultEncoding = $Encoding;
}
/**
* $encoder = 'iconv' or 'mb'
* set iconv if you would like use 'iconv' for encode UTF-16LE to your encoding
* set mb if you would like use 'mb_convert_encoding' for encode UTF-16LE to your encoding
*/
function setUTFEncoder($encoder = 'iconv'){
$this->_encoderFunction = '';
if ($encoder == 'iconv'){
$this->_encoderFunction = function_exists('iconv') ? 'iconv' : '';
}elseif ($encoder == 'mb') {
$this->_encoderFunction = function_exists('mb_convert_encoding') ? 'mb_convert_encoding' : '';
}
}
function setRowColOffset($iOffset){
$this->_rowoffset = $iOffset;
$this->_coloffset = $iOffset;
}
function setDefaultFormat($sFormat){
$this->_defaultFormat = $sFormat;
}
function setColumnFormat($column, $sFormat){
$this->_columnsFormat[$column] = $sFormat;
}
function read($sFileName) {
$errlevel = error_reporting();
error_reporting($errlevel ^ E_NOTICE);
$res = $this->_ole->read($sFileName);
// oops, something goes wrong (Darko Miljanovic)
if($res === false) {
// check error code
if($this->_ole->error == 1) {
// bad file
die('The filename ' . $sFileName . ' is not readable');
}
// check other error codes here (eg bad fileformat, etc...)
}
$this->data = $this->_ole->getWorkBook();
/*
$res = $this->_ole->read($sFileName);
if ($this->isError($res)) {
// var_dump($res);
return $this->raiseError($res);
}
$total = $this->_ole->ppsTotal();
for ($i = 0; $i < $total; $i++) {
if ($this->_ole->isFile($i)) {
$type = unpack("v", $this->_ole->getData($i, 0, 2));
if ($type[''] == 0x0809) { // check if it's a BIFF stream
$this->_index = $i;
$this->data = $this->_ole->getData($i, 0, $this->_ole->getDataLength($i));
break;
}
}
}
if ($this->_index === null) {
return $this->raiseError("$file doesn't seem to be an Excel file");
}
*/
//var_dump($this->data);
//echo "data =".$this->data;
$this->pos = 0;
//$this->readRecords();
$this->_parse();
error_reporting($errlevel);
}
function _parse(){
$pos = 0;
$code = ord($this->data[$pos]) | ord($this->data[$pos+1])<<8;
$length = ord($this->data[$pos+2]) | ord($this->data[$pos+3])<<8;
$version = ord($this->data[$pos + 4]) | ord($this->data[$pos + 5])<<8;
$substreamType = ord($this->data[$pos + 6]) | ord($this->data[$pos + 7])<<8;
//echo "Start parse code=".base_convert($code,10,16)." version=".base_convert($version,10,16)." substreamType=".base_convert($substreamType,10,16).""."\n";
if (($version != Spreadsheet_Excel_Reader_BIFF8) && ($version != Spreadsheet_Excel_Reader_BIFF7)) {
return false;
}
if ($substreamType != Spreadsheet_Excel_Reader_WorkbookGlobals){
return false;
}
//print_r($rec);
$pos += $length + 4;
$code = ord($this->data[$pos]) | ord($this->data[$pos+1])<<8;
$length = ord($this->data[$pos+2]) | ord($this->data[$pos+3])<<8;
while ($code != Spreadsheet_Excel_Reader_Type_EOF){
switch ($code) {
case Spreadsheet_Excel_Reader_Type_SST:
//echo "Type_SST\n";
$spos = $pos + 4;
$limitpos = $spos + $length;
$uniqueStrings = $this->_GetInt4d($this->data, $spos+4);
$spos += 8;
for ($i = 0; $i < $uniqueStrings; $i++) {
// Read in the number of characters
if ($spos == $limitpos) {
$opcode = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$conlength = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
if ($opcode != 0x3c) {
return -1;
}
$spos += 4;
$limitpos = $spos + $conlength;
}
$numChars = ord($this->data[$spos]) | (ord($this->data[$spos+1]) << 8);
//echo "i = $i pos = $pos numChars = $numChars ";
$spos += 2;
$optionFlags = ord($this->data[$spos]);
$spos++;
$asciiEncoding = (($optionFlags & 0x01) == 0) ;
$extendedString = ( ($optionFlags & 0x04) != 0);
// See if string contains formatting information
$richString = ( ($optionFlags & 0x08) != 0);
if ($richString) {
// Read in the crun
$formattingRuns = ord($this->data[$spos]) | (ord($this->data[$spos+1]) << 8);
$spos += 2;
}
if ($extendedString) {
// Read in cchExtRst
$extendedRunLength = $this->_GetInt4d($this->data, $spos);
$spos += 4;
}
$len = ($asciiEncoding)? $numChars : $numChars*2;
if ($spos + $len < $limitpos) {
$retstr = substr($this->data, $spos, $len);
$spos += $len;
}else{
// found countinue
$retstr = substr($this->data, $spos, $limitpos - $spos);
$bytesRead = $limitpos - $spos;
$charsLeft = $numChars - (($asciiEncoding) ? $bytesRead : ($bytesRead / 2));
$spos = $limitpos;
while ($charsLeft > 0){
$opcode = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$conlength = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
if ($opcode != 0x3c) {
return -1;
}
$spos += 4;
$limitpos = $spos + $conlength;
$option = ord($this->data[$spos]);
$spos += 1;
if ($asciiEncoding && ($option == 0)) {
$len = min($charsLeft, $limitpos - $spos); // min($charsLeft, $conlength);
$retstr .= substr($this->data, $spos, $len);
$charsLeft -= $len;
$asciiEncoding = true;
}elseif (!$asciiEncoding && ($option != 0)){
$len = min($charsLeft * 2, $limitpos - $spos); // min($charsLeft, $conlength);
$retstr .= substr($this->data, $spos, $len);
$charsLeft -= $len/2;
$asciiEncoding = false;
}elseif (!$asciiEncoding && ($option == 0)) {
// Bummer - the string starts off as Unicode, but after the
// continuation it is in straightforward ASCII encoding
$len = min($charsLeft, $limitpos - $spos); // min($charsLeft, $conlength);
for ($j = 0; $j < $len; $j++) {
$retstr .= $this->data[$spos + $j].chr(0);
}
$charsLeft -= $len;
$asciiEncoding = false;
}else{
$newstr = '';
for ($j = 0; $j < strlen($retstr); $j++) {
$newstr = $retstr[$j].chr(0);
}
$retstr = $newstr;
$len = min($charsLeft * 2, $limitpos - $spos); // min($charsLeft, $conlength);
$retstr .= substr($this->data, $spos, $len);
$charsLeft -= $len/2;
$asciiEncoding = false;
//echo "Izavrat\n";
}
$spos += $len;
}
}
$retstr = ($asciiEncoding) ? $retstr : $this->_encodeUTF16($retstr);
// echo "Str $i = $retstr\n";
if ($richString){
$spos += 4 * $formattingRuns;
}
// For extended strings, skip over the extended string data
if ($extendedString) {
$spos += $extendedRunLength;
}
//if ($retstr == 'Derby'){
// echo "bb\n";
//}
$this->sst[]=$retstr;
}
/*$continueRecords = array();
while ($this->getNextCode() == Type_CONTINUE) {
$continueRecords[] = &$this->nextRecord();
}
//echo " 1 Type_SST\n";
$this->shareStrings = new SSTRecord($r, $continueRecords);
//print_r($this->shareStrings->strings);
*/
// echo 'SST read: '.($time_end-$time_start)."\n";
break;
case Spreadsheet_Excel_Reader_Type_FILEPASS:
return false;
break;
case Spreadsheet_Excel_Reader_Type_NAME:
//echo "Type_NAME\n";
break;
case Spreadsheet_Excel_Reader_Type_FORMAT:
$indexCode = ord($this->data[$pos+4]) | ord($this->data[$pos+5]) << 8;
if ($version == Spreadsheet_Excel_Reader_BIFF8) {
$numchars = ord($this->data[$pos+6]) | ord($this->data[$pos+7]) << 8;
if (ord($this->data[$pos+8]) == 0){
$formatString = substr($this->data, $pos+9, $numchars);
} else {
$formatString = substr($this->data, $pos+9, $numchars*2);
}
} else {
$numchars = ord($this->data[$pos+6]);
$formatString = substr($this->data, $pos+7, $numchars*2);
}
$this->formatRecords[$indexCode] = $formatString;
// echo "Type.FORMAT\n";
break;
case Spreadsheet_Excel_Reader_Type_XF:
//global $dateFormats, $numberFormats;
$indexCode = ord($this->data[$pos+6]) | ord($this->data[$pos+7]) << 8;
//echo "\nType.XF ".count($this->formatRecords['xfrecords'])." $indexCode ";
if (array_key_exists($indexCode, $this->dateFormats)) {
//echo "isdate ".$dateFormats[$indexCode];
$this->formatRecords['xfrecords'][] = array(
'type' => 'date',
'format' => $this->dateFormats[$indexCode]
);
}elseif (array_key_exists($indexCode, $this->numberFormats)) {
//echo "isnumber ".$this->numberFormats[$indexCode];
$this->formatRecords['xfrecords'][] = array(
'type' => 'number',
'format' => $this->numberFormats[$indexCode]
);
}else{
$isdate = FALSE;
if ($indexCode > 0){
if (isset($this->formatRecords[$indexCode]))
$formatstr = $this->formatRecords[$indexCode];
//echo '.other.';
//echo "\ndate-time=$formatstr=\n";
if ($formatstr)
if (preg_match("/[^hmsday\/\-:\s]/i", $formatstr) == 0) { // found day and time format
$isdate = TRUE;
$formatstr = str_replace('mm', 'i', $formatstr);
$formatstr = str_replace('h', 'H', $formatstr);
//echo "\ndate-time $formatstr \n";
}
}
if ($isdate){
$this->formatRecords['xfrecords'][] = array(
'type' => 'date',
'format' => $formatstr,
);
}else{
$this->formatRecords['xfrecords'][] = array(
'type' => 'other',
'format' => '',
'code' => $indexCode
);
}
}
//echo "\n";
break;
case Spreadsheet_Excel_Reader_Type_NINETEENFOUR:
//echo "Type.NINETEENFOUR\n";
$this->nineteenFour = (ord($this->data[$pos+4]) == 1);
break;
case Spreadsheet_Excel_Reader_Type_BOUNDSHEET:
//echo "Type.BOUNDSHEET\n";
$rec_offset = $this->_GetInt4d($this->data, $pos+4);
$rec_typeFlag = ord($this->data[$pos+8]);
$rec_visibilityFlag = ord($this->data[$pos+9]);
$rec_length = ord($this->data[$pos+10]);
if ($version == Spreadsheet_Excel_Reader_BIFF8){
$chartype = ord($this->data[$pos+11]);
if ($chartype == 0){
$rec_name = substr($this->data, $pos+12, $rec_length);
} else {
$rec_name = $this->_encodeUTF16(substr($this->data, $pos+12, $rec_length*2));
}
}elseif ($version == Spreadsheet_Excel_Reader_BIFF7){
$rec_name = substr($this->data, $pos+11, $rec_length);
}
$this->boundsheets[] = array('name'=>$rec_name,
'offset'=>$rec_offset);
break;
}
//echo "Code = ".base_convert($r['code'],10,16)."\n";
$pos += $length + 4;
$code = ord($this->data[$pos]) | ord($this->data[$pos+1])<<8;
$length = ord($this->data[$pos+2]) | ord($this->data[$pos+3])<<8;
//$r = &$this->nextRecord();
//echo "1 Code = ".base_convert($r['code'],10,16)."\n";
}
foreach ($this->boundsheets as $key=>$val){
$this->sn = $key;
$this->_parsesheet($val['offset']);
}
return true;
}
function _parsesheet($spos){
$cont = true;
// read BOF
$code = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$length = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
$version = ord($this->data[$spos + 4]) | ord($this->data[$spos + 5])<<8;
$substreamType = ord($this->data[$spos + 6]) | ord($this->data[$spos + 7])<<8;
if (($version != Spreadsheet_Excel_Reader_BIFF8) && ($version != Spreadsheet_Excel_Reader_BIFF7)) {
return -1;
}
if ($substreamType != Spreadsheet_Excel_Reader_Worksheet){
return -2;
}
//echo "Start parse code=".base_convert($code,10,16)." version=".base_convert($version,10,16)." substreamType=".base_convert($substreamType,10,16).""."\n";
$spos += $length + 4;
//var_dump($this->formatRecords);
//echo "code $code $length";
while($cont) {
//echo "mem= ".memory_get_usage()."\n";
// $r = &$this->file->nextRecord();
$lowcode = ord($this->data[$spos]);
if ($lowcode == Spreadsheet_Excel_Reader_Type_EOF) break;
$code = $lowcode | ord($this->data[$spos+1])<<8;
$length = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
$spos += 4;
$this->sheets[$this->sn]['maxrow'] = $this->_rowoffset - 1;
$this->sheets[$this->sn]['maxcol'] = $this->_coloffset - 1;
//echo "Code=".base_convert($code,10,16)." $code\n";
unset($this->rectype);
$this->multiplier = 1; // need for format with %
switch ($code) {
case Spreadsheet_Excel_Reader_Type_DIMENSION:
//echo 'Type_DIMENSION ';
if (!isset($this->numRows)) {
if (($length == 10) || ($version == Spreadsheet_Excel_Reader_BIFF7)){
$this->sheets[$this->sn]['numRows'] = ord($this->data[$spos+2]) | ord($this->data[$spos+3]) << 8;
$this->sheets[$this->sn]['numCols'] = ord($this->data[$spos+6]) | ord($this->data[$spos+7]) << 8;
} else {
$this->sheets[$this->sn]['numRows'] = ord($this->data[$spos+4]) | ord($this->data[$spos+5]) << 8;
$this->sheets[$this->sn]['numCols'] = ord($this->data[$spos+10]) | ord($this->data[$spos+11]) << 8;
}
}
//echo 'numRows '.$this->numRows.' '.$this->numCols."\n";
break;
case Spreadsheet_Excel_Reader_Type_MERGEDCELLS:
$cellRanges = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
for ($i = 0; $i < $cellRanges; $i++) {
$fr = ord($this->data[$spos + 8*$i + 2]) | ord($this->data[$spos + 8*$i + 3])<<8;
$lr = ord($this->data[$spos + 8*$i + 4]) | ord($this->data[$spos + 8*$i + 5])<<8;
$fc = ord($this->data[$spos + 8*$i + 6]) | ord($this->data[$spos + 8*$i + 7])<<8;
$lc = ord($this->data[$spos + 8*$i + 8]) | ord($this->data[$spos + 8*$i + 9])<<8;
//$this->sheets[$this->sn]['mergedCells'][] = array($fr + 1, $fc + 1, $lr + 1, $lc + 1);
if ($lr - $fr > 0) {
$this->sheets[$this->sn]['cellsInfo'][$fr+1][$fc+1]['rowspan'] = $lr - $fr + 1;
}
if ($lc - $fc > 0) {
$this->sheets[$this->sn]['cellsInfo'][$fr+1][$fc+1]['colspan'] = $lc - $fc + 1;
}
}
//echo "Merged Cells $cellRanges $lr $fr $lc $fc\n";
break;
case Spreadsheet_Excel_Reader_Type_RK:
case Spreadsheet_Excel_Reader_Type_RK2:
//echo 'Spreadsheet_Excel_Reader_Type_RK'."\n";
$row = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$column = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
$rknum = $this->_GetInt4d($this->data, $spos + 6);
$numValue = $this->_GetIEEE754($rknum);
//echo $numValue." ";
if ($this->isDate($spos)) {
list($string, $raw) = $this->createDate($numValue);
}else{
$raw = $numValue;
if (isset($this->_columnsFormat[$column + 1])){
$this->curformat = $this->_columnsFormat[$column + 1];
}
$string = sprintf($this->curformat, $numValue * $this->multiplier);
//$this->addcell(RKRecord($r));
}
$this->addcell($row, $column, $string, $raw);
//echo "Type_RK $row $column $string $raw {$this->curformat}\n";
break;
case Spreadsheet_Excel_Reader_Type_LABELSST:
$row = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$column = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
$xfindex = ord($this->data[$spos+4]) | ord($this->data[$spos+5])<<8;
$index = $this->_GetInt4d($this->data, $spos + 6);
//var_dump($this->sst);
$this->addcell($row, $column, $this->sst[$index]);
//echo "LabelSST $row $column $string\n";
break;
case Spreadsheet_Excel_Reader_Type_MULRK:
$row = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$colFirst = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
$colLast = ord($this->data[$spos + $length - 2]) | ord($this->data[$spos + $length - 1])<<8;
$columns = $colLast - $colFirst + 1;
$tmppos = $spos+4;
for ($i = 0; $i < $columns; $i++) {
$numValue = $this->_GetIEEE754($this->_GetInt4d($this->data, $tmppos + 2));
if ($this->isDate($tmppos-4)) {
list($string, $raw) = $this->createDate($numValue);
}else{
$raw = $numValue;
if (isset($this->_columnsFormat[$colFirst + $i + 1])){
$this->curformat = $this->_columnsFormat[$colFirst + $i + 1];
}
$string = sprintf($this->curformat, $numValue * $this->multiplier);
}
//$rec['rknumbers'][$i]['xfindex'] = ord($rec['data'][$pos]) | ord($rec['data'][$pos+1]) << 8;
$tmppos += 6;
$this->addcell($row, $colFirst + $i, $string, $raw);
//echo "MULRK $row ".($colFirst + $i)." $string\n";
}
//MulRKRecord($r);
// Get the individual cell records from the multiple record
//$num = ;
break;
case Spreadsheet_Excel_Reader_Type_NUMBER:
$row = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$column = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
$tmp = unpack("ddouble", substr($this->data, $spos + 6, 8)); // It machine machine dependent
if ($this->isDate($spos)) {
list($string, $raw) = $this->createDate($tmp['double']);
// $this->addcell(DateRecord($r, 1));
}else{
//$raw = $tmp[''];
if (isset($this->_columnsFormat[$column + 1])){
$this->curformat = $this->_columnsFormat[$column + 1];
}
$raw = $this->createNumber($spos);
$string = sprintf($this->curformat, $raw * $this->multiplier);
// $this->addcell(NumberRecord($r));
}
$this->addcell($row, $column, $string, $raw);
//echo "Number $row $column $string\n";
break;
case Spreadsheet_Excel_Reader_Type_FORMULA:
case Spreadsheet_Excel_Reader_Type_FORMULA2:
$row = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$column = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
if ((ord($this->data[$spos+6])==0) && (ord($this->data[$spos+12])==255) && (ord($this->data[$spos+13])==255)) {
//String formula. Result follows in a STRING record
//echo "FORMULA $row $column Formula with a string<br>\n";
} elseif ((ord($this->data[$spos+6])==1) && (ord($this->data[$spos+12])==255) && (ord($this->data[$spos+13])==255)) {
//Boolean formula. Result is in +2; 0=false,1=true
} elseif ((ord($this->data[$spos+6])==2) && (ord($this->data[$spos+12])==255) && (ord($this->data[$spos+13])==255)) {
//Error formula. Error code is in +2;
} elseif ((ord($this->data[$spos+6])==3) && (ord($this->data[$spos+12])==255) && (ord($this->data[$spos+13])==255)) {
//Formula result is a null string.
} else {
// result is a number, so first 14 bytes are just like a _NUMBER record
$tmp = unpack("ddouble", substr($this->data, $spos + 6, 8)); // It machine machine dependent
if ($this->isDate($spos)) {
list($string, $raw) = $this->createDate($tmp['double']);
// $this->addcell(DateRecord($r, 1));
}else{
//$raw = $tmp[''];
if (isset($this->_columnsFormat[$column + 1])){
$this->curformat = $this->_columnsFormat[$column + 1];
}
$raw = $this->createNumber($spos);
$string = sprintf($this->curformat, $raw * $this->multiplier);
// $this->addcell(NumberRecord($r));
}
$this->addcell($row, $column, $string, $raw);
//echo "Number $row $column $string\n";
}
break;
case Spreadsheet_Excel_Reader_Type_BOOLERR:
$row = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$column = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
$string = ord($this->data[$spos+6]);
$this->addcell($row, $column, $string);
//echo 'Type_BOOLERR '."\n";
break;
case Spreadsheet_Excel_Reader_Type_ROW:
case Spreadsheet_Excel_Reader_Type_DBCELL:
case Spreadsheet_Excel_Reader_Type_MULBLANK:
break;
case Spreadsheet_Excel_Reader_Type_LABEL:
$row = ord($this->data[$spos]) | ord($this->data[$spos+1])<<8;
$column = ord($this->data[$spos+2]) | ord($this->data[$spos+3])<<8;
$this->addcell($row, $column, substr($this->data, $spos + 8, ord($this->data[$spos + 6]) | ord($this->data[$spos + 7])<<8));
// $this->addcell(LabelRecord($r));
break;
case Spreadsheet_Excel_Reader_Type_EOF:
$cont = false;
break;
default:
//echo ' unknown :'.base_convert($r['code'],10,16)."\n";
break;
}
$spos += $length;
}
if (!isset($this->sheets[$this->sn]['numRows']))
$this->sheets[$this->sn]['numRows'] = $this->sheets[$this->sn]['maxrow'];
if (!isset($this->sheets[$this->sn]['numCols']))
$this->sheets[$this->sn]['numCols'] = $this->sheets[$this->sn]['maxcol'];
}
function isDate($spos){
//$xfindex = GetInt2d(, 4);
$xfindex = ord($this->data[$spos+4]) | ord($this->data[$spos+5]) << 8;
//echo 'check is date '.$xfindex.' '.$this->formatRecords['xfrecords'][$xfindex]['type']."\n";
//var_dump($this->formatRecords['xfrecords'][$xfindex]);
if ($this->formatRecords['xfrecords'][$xfindex]['type'] == 'date') {
$this->curformat = $this->formatRecords['xfrecords'][$xfindex]['format'];
$this->rectype = 'date';
return true;
} else {
if ($this->formatRecords['xfrecords'][$xfindex]['type'] == 'number') {
$this->curformat = $this->formatRecords['xfrecords'][$xfindex]['format'];
$this->rectype = 'number';
if (($xfindex == 0x9) || ($xfindex == 0xa)){
$this->multiplier = 100;
}
}else{
$this->curformat = $this->_defaultFormat;
$this->rectype = 'unknown';
}
return false;
}
}
function createDate($numValue){
if ($numValue > 1){
$utcDays = $numValue - ($this->nineteenFour ? Spreadsheet_Excel_Reader_utcOffsetDays1904 : Spreadsheet_Excel_Reader_utcOffsetDays);
$utcValue = round($utcDays * Spreadsheet_Excel_Reader_msInADay);
// $utcValue = round(($utcDays+1) * SPREADSHEET_EXCEL_READER_MSINADAY - 93600);
$string = date ($this->curformat, $utcValue);
$raw = $utcValue;
}else{
$raw = $numValue;
$hours = floor($numValue * 24);
$mins = floor($numValue * 24 * 60) - $hours * 60;
$secs = floor($numValue * Spreadsheet_Excel_Reader_msInADay) - $hours * 60 * 60 - $mins * 60;
$string = date ($this->curformat, mktime($hours, $mins, $secs));
}
return array($string, $raw);
}
function createNumber($spos){
$rknumhigh = $this->_GetInt4d($this->data, $spos + 10);
$rknumlow = $this->_GetInt4d($this->data, $spos + 6);
//for ($i=0; $i<8; $i++) { echo ord($this->data[$i+$spos+6]) . " "; } echo "<br>";
$sign = ($rknumhigh & 0x80000000) >> 31;
$exp = ($rknumhigh & 0x7ff00000) >> 20;
$mantissa = (0x100000 | ($rknumhigh & 0x000fffff));
$mantissalow1 = ($rknumlow & 0x80000000) >> 31;
$mantissalow2 = ($rknumlow & 0x7fffffff);
$value = $mantissa / pow( 2 , (20- ($exp - 1023)));
if ($mantissalow1 != 0) $value += 1 / pow (2 , (21 - ($exp - 1023)));
$value += $mantissalow2 / pow (2 , (52 - ($exp - 1023)));
//echo "Sign = $sign, Exp = $exp, mantissahighx = $mantissa, mantissalow1 = $mantissalow1, mantissalow2 = $mantissalow2<br>\n";
if ($sign) {$value = -1 * $value;}
return $value;
}
function addcell($row, $col, $string, $raw = ''){
//echo "ADD cel $row-$col $string\n";
$this->sheets[$this->sn]['maxrow'] = max($this->sheets[$this->sn]['maxrow'], $row + $this->_rowoffset);
$this->sheets[$this->sn]['maxcol'] = max($this->sheets[$this->sn]['maxcol'], $col + $this->_coloffset);
$this->sheets[$this->sn]['cells'][$row + $this->_rowoffset][$col + $this->_coloffset] = $string;
if ($raw)
$this->sheets[$this->sn]['cellsInfo'][$row + $this->_rowoffset][$col + $this->_coloffset]['raw'] = $raw;
if (isset($this->rectype))
$this->sheets[$this->sn]['cellsInfo'][$row + $this->_rowoffset][$col + $this->_coloffset]['type'] = $this->rectype;
}
function _GetIEEE754($rknum){
if (($rknum & 0x02) != 0) {
$value = $rknum >> 2;
} else {
//mmp
// first comment out the previously existing 7 lines of code here
// $tmp = unpack("d", pack("VV", 0, ($rknum & 0xfffffffc)));
// //$value = $tmp[''];
// if (array_key_exists(1, $tmp)) {
// $value = $tmp[1];
// } else {
// $value = $tmp[''];
// }
// I got my info on IEEE754 encoding from
// http://research.microsoft.com/~hollasch/cgindex/coding/ieeefloat.html
// The RK format calls for using only the most significant 30 bits of the
// 64 bit floating point value. The other 34 bits are assumed to be 0
// So, we use the upper 30 bits of $rknum as follows...
$sign = ($rknum & 0x80000000) >> 31;
$exp = ($rknum & 0x7ff00000) >> 20;
$mantissa = (0x100000 | ($rknum & 0x000ffffc));
$value = $mantissa / pow( 2 , (20- ($exp - 1023)));
if ($sign) {$value = -1 * $value;}
//end of changes by mmp
}
if (($rknum & 0x01) != 0) {
$value /= 100;
}
return $value;
}
function _encodeUTF16($string){
$result = $string;
if ($this->_defaultEncoding){
switch ($this->_encoderFunction){
case 'iconv' : $result = iconv('UTF-16LE', $this->_defaultEncoding, $string);
break;
case 'mb_convert_encoding' : $result = mb_convert_encoding($string, $this->_defaultEncoding, 'UTF-16LE' );
break;
}
}
return $result;
}
function _GetInt4d($data, $pos) {
return ord($data[$pos]) | (ord($data[$pos+1]) << 8) | (ord($data[$pos+2]) << 16) | (ord($data[$pos+3]) << 24);
}
}
?>
File 5: oleread.inc
<?php
define('NUM_BIG_BLOCK_DEPOT_BLOCKS_POS', 0x2c);
define('SMALL_BLOCK_DEPOT_BLOCK_POS', 0x3c);
define('ROOT_START_BLOCK_POS', 0x30);
define('BIG_BLOCK_SIZE', 0x200);
define('SMALL_BLOCK_SIZE', 0x40);
define('EXTENSION_BLOCK_POS', 0x44);
define('NUM_EXTENSION_BLOCK_POS', 0x48);
define('PROPERTY_STORAGE_BLOCK_SIZE', 0x80);
define('BIG_BLOCK_DEPOT_BLOCKS_POS', 0x4c);
define('SMALL_BLOCK_THRESHOLD', 0x1000);
// property storage offsets
define('SIZE_OF_NAME_POS', 0x40);
define('TYPE_POS', 0x42);
define('START_BLOCK_POS', 0x74);
define('SIZE_POS', 0x78);
define('IDENTIFIER_OLE', pack("CCCCCCCC",0xd0,0xcf,0x11,0xe0,0xa1,0xb1,0x1a,0xe1));
//echo 'ROOT_START_BLOCK_POS = '.ROOT_START_BLOCK_POS."\n";
//echo bin2hex($data[ROOT_START_BLOCK_POS])."\n";
//echo "a=";
//echo $data[ROOT_START_BLOCK_POS];
//function log
function GetInt4d($data, $pos) {
return ord($data[$pos]) | (ord($data[$pos+1]) << 8) | (ord($data[$pos+2]) << 16) | (ord($data[$pos+3]) << 24);
}
class OLERead {
var $data = '';
function OLERead(){
}
function read($sFileName){
// check if file exist and is readable (Darko Miljanovic)
if(!is_readable($sFileName)) {
$this->error = 1;
return false;
}
$this->data = @file_get_contents($sFileName);
if (!$this->data) {
$this->error = 1;
return false;
}
//echo IDENTIFIER_OLE;
//echo 'start';
if (substr($this->data, 0, 8) != IDENTIFIER_OLE) {
$this->error = 1;
return false;
}
$this->numBigBlockDepotBlocks = GetInt4d($this->data, NUM_BIG_BLOCK_DEPOT_BLOCKS_POS);
$this->sbdStartBlock = GetInt4d($this->data, SMALL_BLOCK_DEPOT_BLOCK_POS);
$this->rootStartBlock = GetInt4d($this->data, ROOT_START_BLOCK_POS);
$this->extensionBlock = GetInt4d($this->data, EXTENSION_BLOCK_POS);
$this->numExtensionBlocks = GetInt4d($this->data, NUM_EXTENSION_BLOCK_POS);
/*
echo $this->numBigBlockDepotBlocks." ";
echo $this->sbdStartBlock." ";
echo $this->rootStartBlock." ";
echo $this->extensionBlock." ";
echo $this->numExtensionBlocks." ";
*/
//echo "sbdStartBlock = $this->sbdStartBlock\n";
$bigBlockDepotBlocks = array();
$pos = BIG_BLOCK_DEPOT_BLOCKS_POS;
// echo "pos = $pos";
$bbdBlocks = $this->numBigBlockDepotBlocks;
if ($this->numExtensionBlocks != 0) {
$bbdBlocks = (BIG_BLOCK_SIZE - BIG_BLOCK_DEPOT_BLOCKS_POS)/4;
}
for ($i = 0; $i < $bbdBlocks; $i++) {
$bigBlockDepotBlocks[$i] = GetInt4d($this->data, $pos);
$pos += 4;
}
for ($j = 0; $j < $this->numExtensionBlocks; $j++) {
$pos = ($this->extensionBlock + 1) * BIG_BLOCK_SIZE;
$blocksToRead = min($this->numBigBlockDepotBlocks - $bbdBlocks, BIG_BLOCK_SIZE / 4 - 1);
for ($i = $bbdBlocks; $i < $bbdBlocks + $blocksToRead; $i++) {
$bigBlockDepotBlocks[$i] = GetInt4d($this->data, $pos);
$pos += 4;
}
$bbdBlocks += $blocksToRead;
if ($bbdBlocks < $this->numBigBlockDepotBlocks) {
$this->extensionBlock = GetInt4d($this->data, $pos);
}
}
// var_dump($bigBlockDepotBlocks);
// readBigBlockDepot
$pos = 0;
$index = 0;
$this->bigBlockChain = array();
for ($i = 0; $i < $this->numBigBlockDepotBlocks; $i++) {
$pos = ($bigBlockDepotBlocks[$i] + 1) * BIG_BLOCK_SIZE;
//echo "pos = $pos";
for ($j = 0 ; $j < BIG_BLOCK_SIZE / 4; $j++) {
$this->bigBlockChain[$index] = GetInt4d($this->data, $pos);
$pos += 4 ;
$index++;
}
}
//var_dump($this->bigBlockChain);
//echo '=====2';
// readSmallBlockDepot();
$pos = 0;
$index = 0;
$sbdBlock = $this->sbdStartBlock;
$this->smallBlockChain = array();
while ($sbdBlock != -2) {
$pos = ($sbdBlock + 1) * BIG_BLOCK_SIZE;
for ($j = 0; $j < BIG_BLOCK_SIZE / 4; $j++) {
$this->smallBlockChain[$index] = GetInt4d($this->data, $pos);
$pos += 4;
$index++;
}
$sbdBlock = $this->bigBlockChain[$sbdBlock];
}
// readData(rootStartBlock)
$block = $this->rootStartBlock;
$pos = 0;
$this->entry = $this->__readData($block);
/*
while ($block != -2) {
$pos = ($block + 1) * BIG_BLOCK_SIZE;
$this->entry = $this->entry.substr($this->data, $pos, BIG_BLOCK_SIZE);
$block = $this->bigBlockChain[$block];
}
*/
//echo '==='.$this->entry."===";
$this->__readPropertySets();
}
function __readData($bl) {
$block = $bl;
$pos = 0;
$data = '';
while ($block != -2) {
$pos = ($block + 1) * BIG_BLOCK_SIZE;
$data = $data.substr($this->data, $pos, BIG_BLOCK_SIZE);
//echo "pos = $pos data=$data\n";
$block = $this->bigBlockChain[$block];
}
return $data;
}
function __readPropertySets(){
$offset = 0;
//var_dump($this->entry);
while ($offset < strlen($this->entry)) {
$d = substr($this->entry, $offset, PROPERTY_STORAGE_BLOCK_SIZE);
$nameSize = ord($d[SIZE_OF_NAME_POS]) | (ord($d[SIZE_OF_NAME_POS+1]) << 8);
$type = ord($d[TYPE_POS]);
//$maxBlock = strlen($d) / BIG_BLOCK_SIZE - 1;
$startBlock = GetInt4d($d, START_BLOCK_POS);
$size = GetInt4d($d, SIZE_POS);
$name = '';
for ($i = 0; $i < $nameSize ; $i++) {
$name .= $d[$i];
}
$name = str_replace("\x00", "", $name);
$this->props[] = array (
'name' => $name,
'type' => $type,
'startBlock' => $startBlock,
'size' => $size);
if (($name == "Workbook") || ($name == "Book")) {
$this->wrkbook = count($this->props) - 1;
}
if ($name == "Root Entry") {
$this->rootentry = count($this->props) - 1;
}
//echo "name ==$name=\n";
$offset += PROPERTY_STORAGE_BLOCK_SIZE;
}
}
function getWorkBook(){
if ($this->props[$this->wrkbook]['size'] < SMALL_BLOCK_THRESHOLD){
// getSmallBlockStream(PropertyStorage ps)
$rootdata = $this->__readData($this->props[$this->rootentry]['startBlock']);
$streamData = '';
$block = $this->props[$this->wrkbook]['startBlock'];
//$count = 0;
$pos = 0;
while ($block != -2) {
$pos = $block * SMALL_BLOCK_SIZE;
$streamData .= substr($rootdata, $pos, SMALL_BLOCK_SIZE);
$block = $this->smallBlockChain[$block];
}
return $streamData;
}else{
$numBlocks = $this->props[$this->wrkbook]['size'] / BIG_BLOCK_SIZE;
if ($this->props[$this->wrkbook]['size'] % BIG_BLOCK_SIZE != 0) {
$numBlocks++;
}
if ($numBlocks == 0) return '';
//echo "numBlocks = $numBlocks\n";
//byte[] streamData = new byte[numBlocks * BIG_BLOCK_SIZE];
//print_r($this->wrkbook);
$streamData = '';
$block = $this->props[$this->wrkbook]['startBlock'];
//$count = 0;
$pos = 0;
//echo "block = $block";
while ($block != -2) {
$pos = ($block + 1) * BIG_BLOCK_SIZE;
$streamData .= substr($this->data, $pos, BIG_BLOCK_SIZE);
$block = $this->bigBlockChain[$block];
}
//echo 'stream'.$streamData;
return $streamData;
}
}
}
?>
