php编程实现获取excel文档内容的代码实例

<?php
/* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4: */
/**
* a class for reading microsoft excel spreadsheets.
*
* originally developed by vadim tkachenko under the name phpexcelreader.
* (http://sourceforge.net/projects/phpexcelreader)
* based on the java version by andy khan (http://www.3ppt.com). now
* maintained by david sanders. reads only biff 7 and biff 8 formats.
*
* php versions 4 and 5
*
* license: this source file is subject to version 3.0 of the php license
* that is available through the world-wide-web at the following uri:
* http://www.php.net/license/3_0.txt. if you did not receive a copy of
* the php license and are unable to obtain it through the web, please
* send a note to license@php.net so we can mail you a copy immediately.
*
* @category spreadsheet
* @package spreadsheet_excel_reader
* @author vadim tkachenko <vt@apachephp.com>
* @license http://www.php.net/license/3_0.txt php license 3.0
* @version cvs: $id: reader.php 19 2007-03-13 12:42:41z shangxiao $
* @link http://pear.php.net/package/spreadsheet_excel_reader
* @see ole, spreadsheet_excel_writer
*/
//require_once 'pear.php';
//require_once 'oleread.php';
//oleread.php的内容start=======================================================================================================
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)
{
$value = ord($data[$pos]) | (ord($data[$pos+1]) << 8) | (ord($data[$pos+2]) << 16) | (ord($data[$pos+3]) << 24);
if ($value>=4294967294)
{
$value=-2;
}
return $value;
}
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->sbdstartblockn";
$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 = $pos data=$datan";
$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 = $numblocksn";
//byte[] streamdata = new byte[numblocks * big_block_size];
//print_r($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;
}
}
}
//oleread.php的内容end=======================================================================================================
//require_once 'ole.php';
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', 86400);
//define('spreadsheet_excel_reader_msinaday', 24 * 60 * 60);
//define('spreadsheet_excel_reader_def_num_format', "%.2f");
define('spreadsheet_excel_reader_def_num_format', "%s");
/*
* place includes, constant defines and $_global settings here.
* make sure they have appropriate docblocks to avoid phpdocumentor
* construing they are documented by the page-level docblock.
*/
/**
* a class for reading microsoft excel spreadsheets.
*
* originally developed by vadim tkachenko under the name phpexcelreader.
* (http://sourceforge.net/projects/phpexcelreader)
* based on the java version by andy khan (http://www.3ppt.com). now
* maintained by david sanders. reads only biff 7 and biff 8 formats.
*
* @category spreadsheet
* @package spreadsheet_excel_reader
* @author vadim tkachenko <vt@shoujidy.com>
* @copyright 2010-2011 the php group
* @license http://www.php.net/license/3_0.txt php license 3.0
* @version release: @package_version@
* @link http://pear.php.net/package/packagename
* @see ole, spreadsheet_excel_writer
*/
class readexcelsystem{
/**
* array of worksheets found
*
* @var array
* @access public
*/
var $boundsheets = array();
/**
* array of format records found
*
* @var array
* @access public
*/
var $formatrecords = array();
/**
* todo
*
* @var array
* @access public
*/
var $sst = array();
/**
* array of worksheets
*
* the data is stored in 'cells' and the meta-data is stored in an array
* called 'cellsinfo'
*
* example:
*
* $sheets --> 'cells' --> row --> column --> interpreted value
* --> 'cellsinfo' --> row --> column --> 'type' - can be 'date', 'number', or 'unknown'
* --> 'raw' - the raw data that excel stores for that data cell
*
* @var array
* @access public
*/
var $sheets = array();
/**
* the data returned by ole
*
* @var string
* @access public
*/
var $data;
/**
* ole object for reading the file
*
* @var ole object
* @access private
*/
var $_ole;
/**
* default encoding
*
* @var string
* @access private
*/
var $_defaultencoding;
/**
* default number format
*
* @var integer
* @access private
*/
var $_defaultformat = spreadsheet_excel_reader_def_num_format;
/**
* todo
* list of formats to use for each column
*
* @var array
* @access private
*/
var $_columnsformat = array();
/**
* todo
*
* @var integer
* @access private
*/
var $_rowoffset = 1;
/**
* todo
*
* @var integer
* @access private
*/
var $_coloffset = 1;
/**
* list of default date formats used by excel
*
* @var array
* @access public
*/
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");
/**
* default number formats used by excel
*
* @var array
* @access public
*/
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";
// }}}
// {{{ spreadsheet_excel_reader()
/**
* constructor
*
* some basic initialisation
*/
function __construct()
{
$this->_ole =& new oleread();
$this->setutfencoder('iconv');
}
// }}}
// {{{ setoutputencoding()
/**
* set the encoding method
*
* @param string encoding to use
* @access public
*/
function setoutputencoding($encoding)
{
$this->_defaultencoding = $encoding;
}
// }}}
// {{{ setutfencoder()
/**
* $encoder = 'iconv' or 'mb'
* set iconv if you would like use 'iconv' for encode utf-16le to your encoding
* set mb if you would like use 'mb_convert_encoding' for encode utf-16le to your encoding
*
* @access public
* @param string encoding type to use. either 'iconv' or 'mb'
*/
function setutfencoder($encoder = 'iconv')
{
$this->_encoderfunction = '';
if ($encoder == 'iconv') {
$this->_encoderfunction = function_exists('iconv') ? 'iconv' : '';
} elseif ($encoder == 'mb') {
$this->_encoderfunction = function_exists('mb_convert_encoding') ?
'mb_convert_encoding' :
'';
}
}
// }}}
// {{{ setrowcoloffset()
/**
* todo
*
* @access public
* @param offset
*/
function setrowcoloffset($ioffset)
{
$this->_rowoffset = $ioffset;
$this->_coloffset = $ioffset;
}
// }}}
// {{{ setdefaultformat()
/**
* set the default number format
*
* @access public
* @param default format
*/
function setdefaultformat($sformat)
{
$this->_defaultformat = $sformat;
}
// }}}
// {{{ setcolumnformat()
/**
* force a column to use a certain format
*
* @access public
* @param integer column number
* @param string format
*/
function setcolumnformat($column, $sformat)
{
$this->_columnsformat[$column] = $sformat;
}
// }}}
// {{{ read()
/**
* read the spreadsheet file using ole, then parse
*
* @access public
* @param filename
* @todo return a valid value
*/
function read($sfilename)
{
/*
require_once 'ole.php';
$ole = new ole();
$ole->read($sfilename);
foreach ($ole->_list as $i => $pps) {
if (($pps->name == 'workbook' || $pps->name == 'book') &&
$pps->size >= small_block_threshold) {
$this->data = $ole->getdata($i, 0, $ole->getdatalength($i));
} elseif ($pps->name == 'root entry') {
$this->data = $ole->getdata($i, 0, $ole->getdatalength($i));
}
//var_dump(strlen($ole->getdata($i, 0, $ole->getdatalength($i))), $pps->name, md5($this->data), $ole->getdatalength($i));
}
//exit;
$this->_parse();
return sizeof($this->sheets) > 0;
*/
$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");
}
*/
//echo "data =".$this->data;
//$this->readrecords();
$this->_parse();
}
// }}}
// {{{ _parse()
/**
* parse a workbook
*
* @access private
* @return bool
*/
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_sstn";
$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 "izavratn";
}
$spos += $len;
}
}
$retstr = ($asciiencoding) ? $retstr : $this->_encodeutf16($retstr);
// echo "str $i = $retstrn";
if ($richstring){
$spos += 4 * $formattingruns;
}
// for extended strings, skip over the extended string data
if ($extendedstring) {
$spos += $extendedrunlength;
}
//if ($retstr == 'derby'){
// echo "bbn";
//}
$this->sst[]=$retstr;
}
/*$continuerecords = array();
while ($this->getnextcode() == type_continue) {
$continuerecords[] = &$this->nextrecord();
}
//echo " 1 type_sstn";
$this->sharestrings = new sstrecord($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_namen";
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.formatn";
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.nineteenfourn";
$this->nineteenfour = (ord($this->data[$pos+4]) == 1);
break;
case spreadsheet_excel_reader_type_boundsheet:
//echo "type.boundsheetn";
$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;
}
/**
* parse a worksheet
*
* @access private
* @param todo
* @todo fix return codes
*/
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)." $coden";
unset($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 $fcn";
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 $stringn";
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)." $stringn";
}
//mulrkrecord($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 $stringn";
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 $stringn";
}
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'];
}
/**
* check whether the current record read is a date
*
* @param todo
* @return boolean true if date, false otherwise
*/
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;
$this->rectype = 'unknown';
}
return false;
}
}
//}}}
//{{{ createdate()
/**
* convert the raw excel date into a human readable format
*
* dates in excel are stored as number of seconds from an epoch. on
* windows, the epoch is 30/12/1899 and on mac it's 01/01/1904
*
* @access private
* @param integer the raw excel value to convert
* @return array first element is the converted date, the second element is number a unix timestamp
*/
function createdate($numvalue)
{
if ($numvalue > 1) {
$utcdays = $numvalue - ($this->nineteenfour ? spreadsheet_excel_reader_utcoffsetdays1904 : spreadsheet_excel_reader_utcoffsetdays);
$utcvalue = round(($utcdays+1) * spreadsheet_excel_reader_msinaday);
$string = date ($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 $stringn";
$this->sheets[$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)
{
$value = ord($data[$pos]) | (ord($data[$pos+1]) << 8) | (ord($data[$pos+2]) << 16) | (ord($data[$pos+3]) << 24);
if ($value>=4294967294)
{
$value=-2;
}
return $value;
}
}
/*
* local variables:
* tab-width: 4
* c-basic-offset: 4
* c-hanging-comment-ender-p: nil
* end:
*/
?>

<?php
$data = new readexcelsystem();
$data->setoutputencoding('utf-8');
$data->read($location);
print_r($data->sheets[0]['cells'] );
?>
$data->sheets[0]['cells']