Merge pull request #153 from box/improve_zip_for_mime_detection

Improve ZIP interface for better mime detection
This commit is contained in:
Adrien Loison 2015-12-05 19:13:47 -08:00
commit 22daea5f9a
4 changed files with 194 additions and 68 deletions

View File

@ -11,97 +11,170 @@ namespace Box\Spout\Writer\Common\Helper;
class ZipHelper
{
const ZIP_EXTENSION = '.zip';
const CONTENT_TYPES_XML_FILE_NAME = '[Content_Types].xml';
/** Controls what to do when trying to add an existing file */
const EXISTING_FILES_SKIP = 'skip';
const EXISTING_FILES_OVERWRITE = 'overwrite';
/** @var string Path of the folder where the zip file will be created */
protected $tmpFolderPath;
/** @var \ZipArchive The ZipArchive instance */
protected $zip;
/**
* Zips the root folder and streams the contents of the zip into the given stream
*
* @param string $folderPath Path to the folder to be zipped
* @param resource $streamPointer Pointer to the stream to copy the zip
* @return void
* @param string $tmpFolderPath Path of the temp folder where the zip file will be created
*/
public function zipFolderAndCopyToStream($folderPath, $streamPointer)
public function __construct($tmpFolderPath)
{
$zipFilePath = $this->getZipFilePath($folderPath);
$this->zipFolder($folderPath, $zipFilePath);
$this->copyZipToStream($zipFilePath, $streamPointer);
$this->tmpFolderPath = $tmpFolderPath;
}
/**
* Returns the already created ZipArchive instance or
* creates one if none exists.
*
* @return \ZipArchive
*/
protected function createOrGetZip()
{
if (!isset($this->zip)) {
$this->zip = new \ZipArchive();
$zipFilePath = $this->getZipFilePath();
$this->zip->open($zipFilePath, \ZipArchive::CREATE|\ZipArchive::OVERWRITE);
}
return $this->zip;
}
/**
* @param string $folderPathToZip Path to the folder to be zipped
* @return string Path where the zip file of the given folder will be created
*/
public function getZipFilePath($folderPathToZip)
public function getZipFilePath()
{
return $folderPathToZip . self::ZIP_EXTENSION;
return $this->tmpFolderPath . self::ZIP_EXTENSION;
}
/**
* Zips the given folder
* Adds the given file, located under the given root folder to the archive.
* The file will be compressed.
*
* @param string $folderPath Path of the folder to be zipped
* @param string $destinationPath Path where the zip file will be created
* Example of use:
* addFileToArchive('/tmp/xlsx/foo', 'bar/baz.xml');
* => will add the file located at '/tmp/xlsx/foo/bar/baz.xml' in the archive, but only as 'bar/baz.xml'
*
* @param string $rootFolderPath Path of the root folder that will be ignored in the archive tree.
* @param string $localFilePath Path of the file to be added, under the root folder
* @param string|void $existingFileMode Controls what to do when trying to add an existing file
* @return void
*/
public function zipFolder($folderPath, $destinationPath)
public function addFileToArchive($rootFolderPath, $localFilePath, $existingFileMode = self::EXISTING_FILES_OVERWRITE)
{
$zip = new \ZipArchive();
if ($zip->open($destinationPath, \ZipArchive::CREATE)) {
$this->addFolderToZip($zip, $folderPath);
$zip->close();
}
$this->addFileToArchiveWithCompressionMethod(
$rootFolderPath,
$localFilePath,
$existingFileMode,
\ZipArchive::CM_DEFAULT
);
}
/**
* @param \ZipArchive $zip
* @param string $folderPath Path of the folder to add to the zip
* Adds the given file, located under the given root folder to the archive.
* The file will NOT be compressed.
*
* Example of use:
* addUncompressedFileToArchive('/tmp/xlsx/foo', 'bar/baz.xml');
* => will add the file located at '/tmp/xlsx/foo/bar/baz.xml' in the archive, but only as 'bar/baz.xml'
*
* @param string $rootFolderPath Path of the root folder that will be ignored in the archive tree.
* @param string $localFilePath Path of the file to be added, under the root folder
* @param string|void $existingFileMode Controls what to do when trying to add an existing file
* @return void
*/
protected function addFolderToZip($zip, $folderPath)
public function addUncompressedFileToArchive($rootFolderPath, $localFilePath, $existingFileMode = self::EXISTING_FILES_OVERWRITE)
{
$folderRealPath = $this->getNormalizedRealPath($folderPath) . '/';
$itemIterator = new \RecursiveIteratorIterator(new \RecursiveDirectoryIterator($folderPath, \RecursiveDirectoryIterator::SKIP_DOTS), \RecursiveIteratorIterator::SELF_FIRST);
$this->addFileToArchiveWithCompressionMethod(
$rootFolderPath,
$localFilePath,
$existingFileMode,
\ZipArchive::CM_STORE
);
}
// In order to have the file's mime type detected properly, items need to be
// sorted in a particular order...
$itemsInfo = iterator_to_array($itemIterator);
usort($itemsInfo, [$this, 'sortItemsForCorrectMimeTypeDetection']);
/**
* Adds the given file, located under the given root folder to the archive.
* The file will NOT be compressed.
*
* Example of use:
* addUncompressedFileToArchive('/tmp/xlsx/foo', 'bar/baz.xml');
* => will add the file located at '/tmp/xlsx/foo/bar/baz.xml' in the archive, but only as 'bar/baz.xml'
*
* @param string $rootFolderPath Path of the root folder that will be ignored in the archive tree.
* @param string $localFilePath Path of the file to be added, under the root folder
* @param string $existingFileMode Controls what to do when trying to add an existing file
* @param int $compressionMethod The compression method
* @return void
*/
protected function addFileToArchiveWithCompressionMethod($rootFolderPath, $localFilePath, $existingFileMode, $compressionMethod)
{
$zip = $this->createOrGetZip();
foreach ($itemsInfo as $itemInfo) {
$itemRealPath = $this->getNormalizedRealPath($itemInfo->getPathname());
$itemLocalPath = str_replace($folderRealPath, '', $itemRealPath);
if (!$this->shouldSkipFile($zip, $localFilePath, $existingFileMode)) {
$normalizedFullFilePath = $this->getNormalizedRealPath($rootFolderPath . '/' . $localFilePath);
$zip->addFile($normalizedFullFilePath, $localFilePath);
if ($itemInfo->isFile()) {
$zip->addFile($itemRealPath, $itemLocalPath);
} else if ($itemInfo->isDir()) {
$zip->addEmptyDir($itemLocalPath);
if (self::canChooseCompressionMethod()) {
$zip->setCompressionName($localFilePath, $compressionMethod);
}
}
}
/**
* On order to have the file's mime type detected properly, files need to be added
* to the zip file in a particular order.
* [Content_Types].xml and files located in "xl" folder should be zipped first.
*
* @param \SplFileInfo $itemInfo1 First item to compare
* @param \SplFileInfo $itemInfo2 Second item to compare
* @return int
* @return bool Whether it is possible to choose the desired compression method to be used
*/
protected function sortItemsForCorrectMimeTypeDetection($itemInfo1, $itemInfo2)
public static function canChooseCompressionMethod()
{
// Have the "[Content_Types].xml" file be first
if ($itemInfo1->getFilename() === self::CONTENT_TYPES_XML_FILE_NAME) {
return -1;
} else if ($itemInfo2->getFilename() === self::CONTENT_TYPES_XML_FILE_NAME) {
return 1;
} else {
// Then make sure the files in the "xl" folder will go next
// by sorting items in reverse alphabetical order
return strcmp($itemInfo2->getRealPath(), $itemInfo1->getRealPath());
// setCompressionName() is a PHP7+ method...
return (method_exists(new \ZipArchive(), 'setCompressionName'));
}
/**
* @param string $folderPath Path to the folder to be zipped
* @param string|void $existingFileMode Controls what to do when trying to add an existing file
* @return void
*/
public function addFolderToArchive($folderPath, $existingFileMode = self::EXISTING_FILES_OVERWRITE)
{
$zip = $this->createOrGetZip();
$folderRealPath = $this->getNormalizedRealPath($folderPath) . '/';
$itemIterator = new \RecursiveIteratorIterator(new \RecursiveDirectoryIterator($folderPath, \RecursiveDirectoryIterator::SKIP_DOTS), \RecursiveIteratorIterator::SELF_FIRST);
foreach ($itemIterator as $itemInfo) {
$itemRealPath = $this->getNormalizedRealPath($itemInfo->getPathname());
$itemLocalPath = str_replace($folderRealPath, '', $itemRealPath);
if ($itemInfo->isFile() && !$this->shouldSkipFile($zip, $itemLocalPath, $existingFileMode)) {
$zip->addFile($itemRealPath, $itemLocalPath);
}
}
}
/**
* @param \ZipArchive $zip
* @param string $itemLocalPath
* @param string $existingFileMode
* @return bool Whether the file should be added to the archive or skipped
*/
protected function shouldSkipFile($zip, $itemLocalPath, $existingFileMode)
{
// Skip files if:
// - EXISTING_FILES_SKIP mode chosen
// - File already exists in the archive
return ($existingFileMode === self::EXISTING_FILES_SKIP && $zip->locateName($itemLocalPath) !== false);
}
/**
* Returns canonicalized absolute pathname, containing only forward slashes.
*
@ -114,16 +187,30 @@ class ZipHelper
return str_replace(DIRECTORY_SEPARATOR, '/', $realPath);
}
/**
* Closes the archive and copies it into the given stream
*
* @param resource $streamPointer Pointer to the stream to copy the zip
* @return void
*/
public function closeArchiveAndCopyToStream($streamPointer)
{
$zip = $this->createOrGetZip();
$zip->close();
unset($this->zip);
$this->copyZipToStream($streamPointer);
}
/**
* Streams the contents of the zip file into the given stream
*
* @param string $zipFilePath Path to the zip file
* @param resource $pointer Pointer to the stream to copy the zip
* @return void
*/
protected function copyZipToStream($zipFilePath, $pointer)
protected function copyZipToStream($pointer)
{
$zipFilePointer = fopen($zipFilePath, 'r');
$zipFilePointer = fopen($this->getZipFilePath(), 'r');
stream_copy_to_stream($zipFilePointer, $pointer);
fclose($zipFilePointer);
}

View File

@ -103,14 +103,12 @@ class FileSystemHelper extends \Box\Spout\Common\Helper\FileSystemHelper
protected function createManifestFile()
{
$manifestXmlFileContents = <<<EOD
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<manifest:manifest xmlns:manifest="urn:oasis:names:tc:opendocument:xmlns:manifest:1.0">
<?xml version="1.0" encoding="UTF-8"?>
<manifest:manifest xmlns:manifest="urn:oasis:names:tc:opendocument:xmlns:manifest:1.0" manifest:version="1.2">
<manifest:file-entry manifest:full-path="/" manifest:media-type="application/vnd.oasis.opendocument.spreadsheet"/>
<manifest:file-entry manifest:full-path="styles.xml" manifest:media-type="text/xml"/>
<manifest:file-entry manifest:full-path="content.xml" manifest:media-type="text/xml"/>
<manifest:file-entry manifest:full-path="meta.xml" manifest:media-type="text/xml"/>
<manifest:file-entry manifest:full-path="mimetype" manifest:media-type="text/plain"/>
<manifest:file-entry manifest:full-path="META-INF/manifest.xml" manifest:media-type="text/xml"/>
</manifest:manifest>
EOD;
@ -265,10 +263,17 @@ EOD;
*/
public function zipRootFolderAndCopyToStream($streamPointer)
{
$zipHelper = new ZipHelper();
$zipHelper->zipFolderAndCopyToStream($this->rootFolder, $streamPointer);
$zipHelper = new ZipHelper($this->rootFolder);
// In order to have the file's mime type detected properly, files need to be added
// to the zip file in a particular order.
// @see http://www.jejik.com/articles/2010/03/how_to_correctly_create_odf_documents_using_zip/
$zipHelper->addUncompressedFileToArchive($this->rootFolder, self::MIMETYPE_FILE_NAME);
$zipHelper->addFolderToArchive($this->rootFolder, ZipHelper::EXISTING_FILES_SKIP);
$zipHelper->closeArchiveAndCopyToStream($streamPointer);
// once the zip is copied, remove it
$this->deleteFile($zipHelper->getZipFilePath($this->rootFolder));
$this->deleteFile($zipHelper->getZipFilePath());
}
}

View File

@ -353,10 +353,19 @@ EOD;
*/
public function zipRootFolderAndCopyToStream($streamPointer)
{
$zipHelper = new ZipHelper();
$zipHelper->zipFolderAndCopyToStream($this->rootFolder, $streamPointer);
$zipHelper = new ZipHelper($this->rootFolder);
// In order to have the file's mime type detected properly, files need to be added
// to the zip file in a particular order.
// "[Content_Types].xml" then at least 2 files located in "xl" folder should be zipped first.
$zipHelper->addFileToArchive($this->rootFolder, self::CONTENT_TYPES_XML_FILE_NAME);
$zipHelper->addFileToArchive($this->rootFolder, self::XL_FOLDER_NAME . '/' . self::WORKBOOK_XML_FILE_NAME);
$zipHelper->addFileToArchive($this->rootFolder, self::XL_FOLDER_NAME . '/' . self::STYLES_XML_FILE_NAME);
$zipHelper->addFolderToArchive($this->rootFolder, ZipHelper::EXISTING_FILES_SKIP);
$zipHelper->closeArchiveAndCopyToStream($streamPointer);
// once the zip is copied, remove it
$this->deleteFile($zipHelper->getZipFilePath($this->rootFolder));
$this->deleteFile($zipHelper->getZipFilePath());
}
}

View File

@ -5,6 +5,7 @@ namespace Box\Spout\Writer\ODS;
use Box\Spout\Common\Type;
use Box\Spout\Reader\Wrapper\XMLReader;
use Box\Spout\TestUsingResource;
use Box\Spout\Writer\Common\Helper\ZipHelper;
use Box\Spout\Writer\WriterFactory;
/**
@ -368,6 +369,30 @@ class WriterTest extends \PHPUnit_Framework_TestCase
$this->assertValueWasWrittenToSheet($fileName, 1, 'a dream');
}
/**
* @return void
*/
public function testGeneratedFileShouldHaveTheCorrectMimeType()
{
// Only PHP7+ gives the correct mime type since it requires adding
// uncompressed files to the final archive (which support was added in PHP7)
if (!ZipHelper::canChooseCompressionMethod()) {
$this->markTestSkipped(
'The PHP version used does not support setting the compression method of archived files,
resulting in the mime type to be detected incorrectly.'
);
}
$fileName = 'test_mime_type.ods';
$resourcePath = $this->getGeneratedResourcePath($fileName);
$dataRow = ['foo'];
$this->writeToODSFile([$dataRow], $fileName);
$finfo = new \finfo(FILEINFO_MIME_TYPE);
$this->assertEquals('application/vnd.oasis.opendocument.spreadsheet', $finfo->file($resourcePath));
}
/**
* @param array $allRows
* @param string $fileName