mcndave/spout
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge d52e87d4bde2d5259425a1ba9103bdd37c006d6d into 3edb056286e9f047bcddff4350c00b00e70718b6

Browse Source
This commit is contained in:
Bob4ever 2015-07-08 00:03:38 +00:00
commit ce7a750746
2 changed files with 63 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

72
src/Spout/Reader/Helper/XLSX/SharedStringsHelper.php
View File

@ -34,6 +34,9 @@ class SharedStringsHelper
/** Value to use to escape the line feed character ("\n") */
const ESCAPED_LINE_FEED_CHARACTER = '_x000A_';
/** Disabling this will increase your memory usage but can improve your execution time */
protected $useSharedStringsFileCache;
/** @var string Path of the XLSX file being read */
protected $filePath;
@ -56,15 +59,17 @@ class SharedStringsHelper
* @var string Contents of the temporary file that was last read
* @see MAX_NUM_STRINGS_PER_TEMP_FILE
*/
protected $inMemoryTempFileContents;
protected $inMemoryContents;
/**
* @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 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->useSharedStringsFileCache = $useSharedStringsFileCache;
$rootTempFolder = ($tempFolder) ?: sys_get_temp_dir();
$this->fileSystemHelper = new FileSystemHelper($rootTempFolder);
@ -141,11 +146,19 @@ class SharedStringsHelper
$unescapedTextValue = $escaper->unescape($textValue);
// 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
$lineFeedEncodedTextValue = $this->escapeLineFeed($unescapedTextValue);
$this->writeSharedStringToTempFile($lineFeedEncodedTextValue, $sharedStringIndex);
if ($this->useSharedStringsFileCache) {
// 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
$lineFeedEncodedTextValue = $this->escapeLineFeed($unescapedTextValue);
$this->writeSharedStringToTempFile($lineFeedEncodedTextValue, $sharedStringIndex);
} else {
if (!isset($this->inMemoryContents)) {
$this->inMemoryContents = [];
}
$this->inMemoryContents[$sharedStringIndex] = $unescapedTextValue;
}
$sharedStringIndex++;
@ -262,25 +275,32 @@ class SharedStringsHelper
*/
public function getStringAtIndex($sharedStringIndex)
{
$tempFilePath = $this->getSharedStringTempFilePath($sharedStringIndex);
$indexInFile = $sharedStringIndex % self::MAX_NUM_STRINGS_PER_TEMP_FILE;
if (!file_exists($tempFilePath)) {
throw new SharedStringNotFoundException("Shared string temp file not found: $tempFilePath ; for index: $sharedStringIndex");
}
if ($this->inMemoryTempFilePath !== $tempFilePath) {
// free memory
unset($this->inMemoryTempFileContents);
$this->inMemoryTempFileContents = explode(PHP_EOL, file_get_contents($tempFilePath));
$this->inMemoryTempFilePath = $tempFilePath;
}
$sharedString = null;
if (array_key_exists($indexInFile, $this->inMemoryTempFileContents)) {
$escapedSharedString = $this->inMemoryTempFileContents[$indexInFile];
$sharedString = $this->unescapeLineFeed($escapedSharedString);
if ($this->useSharedStringsFileCache) {
$tempFilePath = $this->getSharedStringTempFilePath($sharedStringIndex);
$indexInFile = $sharedStringIndex % self::MAX_NUM_STRINGS_PER_TEMP_FILE;
if (!file_exists($tempFilePath)) {
throw new SharedStringNotFoundException("Shared string temp file not found: $tempFilePath ; for index: $sharedStringIndex");
}
if ($this->inMemoryTempFilePath !== $tempFilePath) {
// free memory
unset($this->inMemoryContents);
$this->inMemoryContents = explode(PHP_EOL, file_get_contents($tempFilePath));
$this->inMemoryTempFilePath = $tempFilePath;
}
if (array_key_exists($indexInFile, $this->inMemoryContents)) {
$escapedSharedString = $this->inMemoryContents[$indexInFile];
$sharedString = $this->unescapeLineFeed($escapedSharedString);
}
} else {
if (is_array($this->inMemoryContents) && array_key_exists($sharedStringIndex, $this->inMemoryContents)) {
$sharedString = $this->inMemoryContents[$sharedStringIndex];
}
}
if ($sharedString === null) {

18
src/Spout/Reader/XLSX.php
View File

@ -32,6 +32,9 @@ class XLSX extends AbstractReader
/** @var string Temporary folder where the temporary files will be created */
protected $tempFolder;
/** @var bool Disabling this will increase your memory usage but can improve your execution time */
protected $useSharedStringsFileCache = true;
/** @var \ZipArchive */
protected $zip;
@ -63,6 +66,19 @@ class XLSX extends AbstractReader
$this->tempFolder = $tempFolder;
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.
@ -80,7 +96,7 @@ class XLSX extends AbstractReader
$this->zip = new \ZipArchive();
if ($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()) {
// Extracts all the strings from the worksheets for easy access in the future