Easily manipulate PhpSpreadsheet, OpenSpout and League CSV
Importing/exporting csv data is a very common task in web development. While it's a very efficient format, it's also somewhat difficult for end users that are used to Excel. This is why you often end up accepting also xlsx format as a import/export target.
Ideally, importing single sheets of csv or excel should be just a matter of changing an adapter. Thankfully, this package does just this :-)
OpenSpout: fast csv and excel import/export https://github.com/openspout/openspout
League CSV: very fast csv import/export. Can read streams. https://github.com/thephpleague/csv
PhpSpreadsheet: slow excel (xls and xlsx) and csv import/export, but more features https://github.com/PHPOffice/PhpSpreadsheet
Native php: very fast csv import/export, but limited features. Can read/output streams.
SimpleXLSX: very fast excel import/export https://github.com/shuchkin/simplexlsx https://github.com/shuchkin/simplexlsxgen
This package will prioritize installed library, by order of performance. You can also pick your preferred default adapter for each format like this:
SpreadCompat::$preferredCsvAdapter = SpreadCompat::NATIVE; // our native csv adapter is the fastest
SpreadCompat::$preferredXlsxAdapter = SpreadCompat::NATIVE; // our native xlsx adapter is the fastest
While you can use individual adapters, it's very likely you don't want to bother too much how your files are read and written. This package provides a simple facade with static methods in order to read and write files.
Please note that read methods return a Generator
. If you want an array, you need to use iterator_to_array
.
$data = iterator_to_array(SpreadCompat::read('myfile.csv'));
// or
foreach(SpreadCompat::read('myfile.csv') as $row) {
// Do something
}
This package includes a simple way to leverage output to browser type of functionnality.
Some adapters allow you to stream directly the response.
SpreadCompat::output('myfile.csv');
exit();
This package accepts options using ...opts, this means you can freely use named arguments or pass an array.
$data = iterator_to_array(SpreadCompat::read('myfile.csv', assoc: true));
// or
$data = iterator_to_array(SpreadCompat::read('myfile.csv', ...$opts));
You can also use the Options
class that regroups all available options for all adapters. Unsupported options are ignored.
$options = new Options();
$options->separator = ";";
$data = iterator_to_array(SpreadCompat::read('myfile.csv', $options));
Instead of relying on the static variables, you can choose which adapter to use:
$csvData = SpreadCompat::readString($csv, adapter: SpreadCompat::NATIVE);
// or
$options = new Options();
$options->adapter = SpreadCompat::NATIVE;
$csvData = SpreadCompat::readString($csv, $options);
This package supports only 1 worksheet, as it is meant to be able to replace csv by xlsx or vice versa
Since we can compare our solutions, there is a built in bench.php script that give the following results on my machine These results are run with:
openspout/openspout 4.24.4
phpoffice/phpspreadsheet 2.1.0
league/csv 9.16.0
shuchkin/simplexlsx 1.1.11
shuchkin/simplexlsxgen 1.4.11
Reading a file with 5000 rows:
Results for csv
LeKoala\SpreadCompat\Csv\Native : 0.0084
LeKoala\SpreadCompat\Csv\League : 0.0317
LeKoala\SpreadCompat\Csv\OpenSpout : 0.0993
LeKoala\SpreadCompat\Csv\PhpSpreadsheet : 0.6713
Results for xlsx
LeKoala\SpreadCompat\Xlsx\Native : 0.0639
LeKoala\SpreadCompat\Xlsx\Simple : 0.1476
LeKoala\SpreadCompat\Xlsx\PhpSpreadsheet : 0.7436
LeKoala\SpreadCompat\Xlsx\OpenSpout : 0.8979
Write a file with 2500 rows:
Results for csv
LeKoala\SpreadCompat\Csv\Native : 0.0065
LeKoala\SpreadCompat\Csv\League : 0.0129
LeKoala\SpreadCompat\Csv\OpenSpout : 0.0443
LeKoala\SpreadCompat\Csv\PhpSpreadsheet : 0.3668
Results for xlsx
LeKoala\SpreadCompat\Xlsx\Native : 0.0401
LeKoala\SpreadCompat\Xlsx\Simple : 0.0861
LeKoala\SpreadCompat\Xlsx\OpenSpout : 0.2555
LeKoala\SpreadCompat\Xlsx\PhpSpreadsheet : 0.674
For simple imports/exports, it's very clear that using the Native adapter is the fastest. These are not enabled by default since they might lack some compatibility feature you may require.
Stop wasting cpu cycles right now and please use the most efficient adapter :-)