Jeff
Jeff

Reputation: 1162

PHPExcel taking an extremely long time to read Excel file

I'm using PHPExcel 1.7.8, PHP 5.4.14, Windows 7, and an Excel 2007 spreadsheet. The spreadsheet consists of 750 rows, columns A through BW, and is about 600KB in size. This is my code for opening the spreadsheet--pretty standard:

//Include PHPExcel_IOFactory
include 'PHPExcel/IOFactory.php';
include 'PHPExcel.php';

$inputFileName = 'C:\xls\lspimport\GetLSP1.xlsx';

//  Read your Excel workbook
try {
    $inputFileType = PHPExcel_IOFactory::identify($inputFileName);
    $objReader = PHPExcel_IOFactory::createReader($inputFileType);
    $objReader->setReadDataOnly(true);
    $objPHPExcel = $objReader->load($inputFileName);
} catch(Exception $e) {
    die('Error loading file "'.pathinfo($inputFileName,PATHINFO_BASENAME).'": '.$e->getMessage());
}

//set active worksheet
$objWorksheet = $objPHPExcel->setActiveSheetIndexbyName('Sheet1');

$j = 0;

for($i = 2; $i < 3; $i++)
{
...
}

In the end, I eventually want to loop through each row in the spreadsheet, but for the time being while I perfect the script, I'm only looping through one row. The problem is, it takes 30 minutes for this script to execute. I echo'd messages after each section of code so I could see what was being processed and when, and my script basically waits for 30 minutes at this part:

$objPHPExcel = $objReader->load($inputFileName);

Have a configured something incorrectly? I can't figure out why it takes 30 minutes to load the spreadsheet. I appreciate any and all help.

Upvotes: 3

Views: 13848

Answers (4)

John
John

Reputation: 168

If you know your file is a pretty plain excel file, you can do manual reading. A .xslx file is just a zip archive with the spreadsheet values and structure stored into xml files. This script took me from the 60 seconds used on PHPExcel down to 0.18 seconds.

$zip = new ZipArchive();
$zip->open('path_to/file.xlsx');
$sheet_xml = simplexml_load_string($zip->getFromName('xl/worksheets/sheet1.xml'));
$sheet_array = json_decode(json_encode($xml), true);
$values = simplexml_load_string($zip->getFromName('xl/sharedStrings.xml'));
$values_array = json_decode(json_encode($values), true);

$end_result = array();
if ($sheet_array['sheetData']) {
    foreach ($sheet_array['sheetData']['row'] as $r => $row) {
        $end_result[$r] = array();
        foreach ($row['c'] as $c => $cell) {
            if (isset($cell['@attributes']['t'])) {
                if ($cell['@attributes']['t'] == 's') {
                    $end_result[$r][] = $values_array['si'][$cell['v']]['t'];
                } else if ($cell['@attributes']['t'] == 'e') {
                    $end_result[$r][] = '';
                }
            } else {
                $end_result[$r][] = $cell['v'];
            }
        }
    }
}

Result:

Array
(
    [0] => Array
        (
            [0] => A1
            [1] => B1
            [2] => C1
        )

    [1] => Array
        (
            [0] => A2
            [1] => B2
            [2] => C2
        )
)

This is error prone and not optimized, but it works and illustrates the basic idea. If you know your file, then you can make reading very fast. If you allow users to input the files, then you should maybe avoid it - or at least do the neccessary checks.

Upvotes: 0

targnation
targnation

Reputation: 927

I'm using the latest version of PHPExcel (1.8.1) in a Symfony project, and I also ran into time delays when using the $objReader->load($file) method. The time delays were not due to an autoloader, but to the load method itself. This method actually reads every cell in every worksheet. And since my data worksheet was 30 columns wide by 5000 rows, it took about 90 seconds to read all this information on my ancient work computer.

I assumed that the real loading/reading of cell values would occur on the fly as I requested them, but it looks like short of a pretty major re-write of the PHPExcel code, there's no real way around this initial load time delay.

Upvotes: 0

Wandering Zombie
Wandering Zombie

Reputation: 1121

Resolved (for me) - see note at bottom of this post

I'm trying to use pretty much identical code on a dedicated quad core server with 16GB of RAM, also running similar versions - PHPExcel 1.7.9 and PHP 5.4.16

Just creating an empty reader takes 50 seconds!

// $inputFileType is 'Excel5';
$objReader = PHPExcel_IOFactory::createReader($inputFileType);

Loading the spreadsheet (1 sheet, 2000 rows, 25 columns) I want to process (readonly) then takes 1802 seconds.

$objReader->setReadDataOnly(true);
$objPHPExcel = $objReader->load($inputFileName);

Of the various types of reader I consistently get timings for instantiation as shown below

foreach(array(
  'Excel2007',     // 350 seconds
  'Excel5',        //  50 seconds
  'Excel2003XML',  //  50 seconds
  'OOCalc',        //  50 seconds
  'SYLK',          //  50 seconds
  'Gnumeric',      //  50 seconds
  'HTML',          // 250 seconds
  'CSV'            //  50 seconds
  ) as $inputFileType) {
  $objReader = PHPExcel_IOFactory::createReader($inputFileType);
}

Peak memory usage was about 8MB... far less than the 250MB the script has available to it.

My suspicion WAS that PHPExcel_IOFactory::createReader($inputFileType) was calling something within a loop that's extremely slow under PHP 5.4.x ?

However the excessive time was due to how PHPExcel names its class names and corresponding file structure. It has an autoloader that converts class names such as *PHPExcel_abc_def* into PHPExcel/abc/def.php for the require statement. Although we had PHPExcel's class directory defined in our include path, our own (already defined) autoloader couldn't handle the manipulation from class name to file name required (it was looking for *PHPExcel_abc_def.php*). When a class file cannot be included, our autoloader will loop 5 times with a 10 second delay to see if the file is being updated and so might become available. So for every PHPExcel class that needed to be loaded we were introducing a delay of 50 seconds before hitting PHPExcel's own autoloader which required the file in fine.

Now that I've got that resolved PHPExcel is proving to be truly awesome.

Upvotes: 2

km6zla
km6zla

Reputation: 4877

PHPExcel has a problem with identifying where the end of your excel file is. Or rather, Excel has a hard time knowing where the end of itself is. If you touch a cell at A:1000000 it thinks it needs to read that far.

I have done 2 things in the past to fix this:

1) Cut and past the data you need into new excel file. 2) Specify the exact dimensions you want to read.

Edit How to do option 2

public function readExcelDataToArray($excelFilePath, $maxRowNumber=-1, $maxColumnNumber=-1)
{
    $objPHPExcel = PHPExcel_IOFactory::load($excelFilePath);
    $objWorksheet = $objPHPExcel->getActiveSheet();

    //Get last row and column that have data
    if ($maxRowNumber == -1){
    $lastRow = $objWorksheet->getHighestDataRow();
    } else {
        $lastRow = $maxRowNumber;
    }

    if ($maxColumnNumber == -1){
        $lastCol = $objWorksheet->getHighestDataColumn();
        //Change Column letter to column number
        $lastCol = PHPExcel_Cell::columnIndexFromString($lastCol);      
    } else {
        $lastCol = $maxColumnNumber;
    }   

    //Get Data Array
    $dataArray = array();

    for ($currentRow = 1; $currentRow <= $lastRow; $currentRow++){
        for ($currentCol = 0; $currentCol <= $lastCol; $currentCol++){
            $dataArray[$currentRow][$currentCol] = $objWorksheet->getCellByColumnAndRow($currentCol,, $currentRow)->getValue();
        }
    }
    return $dataArray;
}

Unfortunately these solutions aren't very dynamic.

Note that a modern excel file is really just a zip with an xlsx extension. I have written extensions to PHPExcel that unzip them, and modify certain xml files to get the kinds of behaviors I want.

A third suggestion for you would be to monitor the contents of each row and stop when you get an empty one.

Upvotes: 5

Related Questions