Added method for XLSX reader to disable the sharedStringsFileCache

This commit is contained in:
Bob4ever 2015-06-12 09:53:38 +02:00
parent 2a5f4f973e
commit d52e87d4bd
2 changed files with 34 additions and 16 deletions

View File

@ -34,8 +34,8 @@ class SharedStringsHelper
/** Value to use to escape the line feed character ("\n") */ /** Value to use to escape the line feed character ("\n") */
const ESCAPED_LINE_FEED_CHARACTER = '_x000A_'; const ESCAPED_LINE_FEED_CHARACTER = '_x000A_';
/** This will increase your memory usage but can improve your execution time */ /** Disabling this will increase your memory usage but can improve your execution time */
static public $KEEP_ALL_STRINGS_IN_MEMORY = false; protected $useSharedStringsFileCache;
/** @var string Path of the XLSX file being read */ /** @var string Path of the XLSX file being read */
protected $filePath; protected $filePath;
@ -64,10 +64,12 @@ class SharedStringsHelper
/** /**
* @param string $filePath Path of the XLSX file being read * @param string $filePath Path of the XLSX file being read
* @param string|void $tempFolder Temporary folder where the temporary files to store shared strings will be stored * @param string|void $tempFolder Temporary folder where the temporary files to store shared strings will be stored
* @param bool|void $useSharedStringsFileCache Disabling this will increase your memory usage but can improve your execution time
*/ */
public function __construct($filePath, $tempFolder = null) public function __construct($filePath, $tempFolder = null, $useSharedStringsFileCache = true)
{ {
$this->filePath = $filePath; $this->filePath = $filePath;
$this->useSharedStringsFileCache = $useSharedStringsFileCache;
$rootTempFolder = ($tempFolder) ?: sys_get_temp_dir(); $rootTempFolder = ($tempFolder) ?: sys_get_temp_dir();
$this->fileSystemHelper = new FileSystemHelper($rootTempFolder); $this->fileSystemHelper = new FileSystemHelper($rootTempFolder);
@ -144,18 +146,18 @@ class SharedStringsHelper
$unescapedTextValue = $escaper->unescape($textValue); $unescapedTextValue = $escaper->unescape($textValue);
if (self::$KEEP_ALL_STRINGS_IN_MEMORY) { if ($this->useSharedStringsFileCache) {
if (!Isset($this->inMemoryContents)) {
$this->inMemoryContents = [];
}
$this->inMemoryContents[$sharedStringIndex] = $unescapedTextValue;
} else {
// The shared string retrieval logic expects each cell data to be on one line only // The shared string retrieval logic expects each cell data to be on one line only
// Encoding the line feed character allows to preserve this assumption // Encoding the line feed character allows to preserve this assumption
$lineFeedEncodedTextValue = $this->escapeLineFeed($unescapedTextValue); $lineFeedEncodedTextValue = $this->escapeLineFeed($unescapedTextValue);
$this->writeSharedStringToTempFile($lineFeedEncodedTextValue, $sharedStringIndex); $this->writeSharedStringToTempFile($lineFeedEncodedTextValue, $sharedStringIndex);
} else {
if (!isset($this->inMemoryContents)) {
$this->inMemoryContents = [];
}
$this->inMemoryContents[$sharedStringIndex] = $unescapedTextValue;
} }
$sharedStringIndex++; $sharedStringIndex++;
@ -276,11 +278,7 @@ class SharedStringsHelper
{ {
$sharedString = null; $sharedString = null;
if (self::$KEEP_ALL_STRINGS_IN_MEMORY) { if ($this->useSharedStringsFileCache) {
if (isset($this->inMemoryContents) && array_key_exists($sharedStringIndex, $this->inMemoryContents)) {
$sharedString = $this->inMemoryContents[$sharedStringIndex];
}
} else {
$tempFilePath = $this->getSharedStringTempFilePath($sharedStringIndex); $tempFilePath = $this->getSharedStringTempFilePath($sharedStringIndex);
$indexInFile = $sharedStringIndex % self::MAX_NUM_STRINGS_PER_TEMP_FILE; $indexInFile = $sharedStringIndex % self::MAX_NUM_STRINGS_PER_TEMP_FILE;
@ -300,6 +298,10 @@ class SharedStringsHelper
$escapedSharedString = $this->inMemoryContents[$indexInFile]; $escapedSharedString = $this->inMemoryContents[$indexInFile];
$sharedString = $this->unescapeLineFeed($escapedSharedString); $sharedString = $this->unescapeLineFeed($escapedSharedString);
} }
} else {
if (is_array($this->inMemoryContents) && array_key_exists($sharedStringIndex, $this->inMemoryContents)) {
$sharedString = $this->inMemoryContents[$sharedStringIndex];
}
} }
if ($sharedString === null) { if ($sharedString === null) {

View File

@ -25,6 +25,9 @@ class XLSX extends AbstractReader
/** @var string Temporary folder where the temporary files will be created */ /** @var string Temporary folder where the temporary files will be created */
protected $tempFolder; protected $tempFolder;
/** @var bool Disabling this will increase your memory usage but can improve your execution time */
protected $useSharedStringsFileCache = true;
/** @var \ZipArchive */ /** @var \ZipArchive */
protected $zip; protected $zip;
@ -56,6 +59,19 @@ class XLSX extends AbstractReader
return $this; return $this;
} }
/**
* Disabling the shared strings file cache will increase your memory usage but can improve your execution time.
* The shared strings file cache is active by default.
*
* @param bool $useSharedStringsFileCache
* @return XLSX
*/
public function setUseSharedStringsFileCache($useSharedStringsFileCache)
{
$this->useSharedStringsFileCache = $useSharedStringsFileCache;
return $this;
}
/** /**
* Opens the file at the given file path to make it ready to be read. * Opens the file at the given file path to make it ready to be read.
* It also parses the sharedStrings.xml file to get all the shared strings available in memory * It also parses the sharedStrings.xml file to get all the shared strings available in memory
@ -72,7 +88,7 @@ class XLSX extends AbstractReader
$this->zip = new \ZipArchive(); $this->zip = new \ZipArchive();
if ($this->zip->open($filePath) === true) { if ($this->zip->open($filePath) === true) {
$this->sharedStringsHelper = new SharedStringsHelper($filePath, $this->tempFolder); $this->sharedStringsHelper = new SharedStringsHelper($filePath, $this->tempFolder, $this->useSharedStringsFileCache);
if ($this->sharedStringsHelper->hasSharedStrings()) { if ($this->sharedStringsHelper->hasSharedStrings()) {
// Extracts all the strings from the worksheets for easy access in the future // Extracts all the strings from the worksheets for easy access in the future