Introduce Managers for readers

Some helper classes were more managers than helpers...

src/Spout/Reader/XLSX/Manager/SharedStringsCaching/CachingStrategyFactory.php

@@ -0,0 +1,139 @@
+<?php
+
+namespace Box\Spout\Reader\XLSX\Manager\SharedStringsCaching;
+
+use Box\Spout\Reader\XLSX\Creator\HelperFactory;
+
+/**
+ * Class CachingStrategyFactory
+ *
+ * @package Box\Spout\Reader\XLSX\Manager\SharedStringsCaching
+ */
+class CachingStrategyFactory
+{
+    /**
+     * The memory amount needed to store a string was obtained empirically from this data:
+     *
+     *        ------------------------------------
+     *        | Number of chars⁺ | Memory needed |
+     *        ------------------------------------
+     *        |           3,000  |         1 MB  |
+     *        |          15,000  |         2 MB  |
+     *        |          30,000  |         5 MB  |
+     *        |          75,000  |        11 MB  |
+     *        |         150,000  |        21 MB  |
+     *        |         300,000  |        43 MB  |
+     *        |         750,000  |       105 MB  |
+     *        |       1,500,000  |       210 MB  |
+     *        |       2,250,000  |       315 MB  |
+     *        |       3,000,000  |       420 MB  |
+     *        |       4,500,000  |       630 MB  |
+     *        ------------------------------------
+     *
+     *        ⁺ All characters were 1 byte long
+     *
+     * This gives a linear graph where each 1-byte character requires about 150 bytes to be stored.
+     * Given that some characters can take up to 4 bytes, we need 600 bytes per character to be safe.
+     * Also, there is on average about 20 characters per cell (this is entirely empirical data...).
+     *
+     * This means that in order to store one shared string in memory, the memory amount needed is:
+     *   => 20 * 600 ≈ 12KB
+     */
+    const AMOUNT_MEMORY_NEEDED_PER_STRING_IN_KB = 12;
+
+    /**
+     * To avoid running out of memory when extracting a huge number of shared strings, they can be saved to temporary files
+     * instead of in memory. Then, when accessing a string, the corresponding file contents will be loaded in memory
+     * and the string will be quickly retrieved.
+     * The performance bottleneck is not when creating these temporary files, but rather when loading their content.
+     * Because the contents of the last loaded file stays in memory until another file needs to be loaded, it works
+     * best when the indexes of the shared strings are sorted in the sheet data.
+     * 10,000 was chosen because it creates small files that are fast to be loaded in memory.
+     */
+    const MAX_NUM_STRINGS_PER_TEMP_FILE = 10000;
+
+
+    /**
+     * Returns the best caching strategy, given the number of unique shared strings
+     * and the amount of memory available.
+     *
+     * @param int|null $sharedStringsUniqueCount Number of unique shared strings (NULL if unknown)
+     * @param string $tempFolder Temporary folder where the temporary files to store shared strings will be stored
+     * @param HelperFactory $helperFactory Factory to create helpers
+     * @return CachingStrategyInterface The best caching strategy
+     */
+    public function createBestCachingStrategy($sharedStringsUniqueCount, $tempFolder, $helperFactory)
+    {
+        if ($this->isInMemoryStrategyUsageSafe($sharedStringsUniqueCount)) {
+            return new InMemoryStrategy($sharedStringsUniqueCount);
+        } else {
+            return new FileBasedStrategy($tempFolder, self::MAX_NUM_STRINGS_PER_TEMP_FILE, $helperFactory);
+        }
+    }
+
+    /**
+     * Returns whether it is safe to use in-memory caching, given the number of unique shared strings
+     * and the amount of memory available.
+     *
+     * @param int|null $sharedStringsUniqueCount Number of unique shared strings (NULL if unknown)
+     * @return bool
+     */
+    protected function isInMemoryStrategyUsageSafe($sharedStringsUniqueCount)
+    {
+        // if the number of shared strings in unknown, do not use "in memory" strategy
+        if ($sharedStringsUniqueCount === null) {
+            return false;
+        }
+
+        $memoryAvailable = $this->getMemoryLimitInKB();
+
+        if ($memoryAvailable === -1) {
+            // if cannot get memory limit or if memory limit set as unlimited, don't trust and play safe
+            return ($sharedStringsUniqueCount < self::MAX_NUM_STRINGS_PER_TEMP_FILE);
+        } else {
+            $memoryNeeded = $sharedStringsUniqueCount * self::AMOUNT_MEMORY_NEEDED_PER_STRING_IN_KB;
+            return ($memoryAvailable > $memoryNeeded);
+        }
+    }
+
+    /**
+     * Returns the PHP "memory_limit" in Kilobytes
+     *
+     * @return float
+     */
+    protected function getMemoryLimitInKB()
+    {
+        $memoryLimitFormatted = $this->getMemoryLimitFromIni();
+        $memoryLimitFormatted = strtolower(trim($memoryLimitFormatted));
+
+        // No memory limit
+        if ($memoryLimitFormatted === '-1') {
+            return -1;
+        }
+
+        if (preg_match('/(\d+)([bkmgt])b?/', $memoryLimitFormatted, $matches)) {
+            $amount = intval($matches[1]);
+            $unit = $matches[2];
+
+            switch ($unit) {
+                case 'b': return ($amount / 1024);
+                case 'k': return $amount;
+                case 'm': return ($amount * 1024);
+                case 'g': return ($amount * 1024 * 1024);
+                case 't': return ($amount * 1024 * 1024 * 1024);
+            }
+        }
+
+        return -1;
+    }
+
+    /**
+     * Returns the formatted "memory_limit" value
+     *
+     * @return string
+     */
+    protected function getMemoryLimitFromIni()
+    {
+        return ini_get('memory_limit');
+    }
+}

src/Spout/Reader/XLSX/Manager/SharedStringsCaching/CachingStrategyInterface.php

@@ -0,0 +1,44 @@
+<?php
+
+namespace Box\Spout\Reader\XLSX\Manager\SharedStringsCaching;
+
+/**
+ * Interface CachingStrategyInterface
+ *
+ * @package Box\Spout\Reader\XLSX\Manager\SharedStringsCaching
+ */
+interface CachingStrategyInterface
+{
+    /**
+     * Adds the given string to the cache.
+     *
+     * @param string $sharedString The string to be added to the cache
+     * @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
+     * @return void
+     */
+    public function addStringForIndex($sharedString, $sharedStringIndex);
+
+    /**
+     * Closes the cache after the last shared string was added.
+     * This prevents any additional string from being added to the cache.
+     *
+     * @return void
+     */
+    public function closeCache();
+
+    /**
+     * Returns the string located at the given index from the cache.
+     *
+     * @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
+     * @return string The shared string at the given index
+     * @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If no shared string found for the given index
+     */
+    public function getStringAtIndex($sharedStringIndex);
+
+    /**
+     * Destroys the cache, freeing memory and removing any created artifacts
+     *
+     * @return void
+     */
+    public function clearCache();
+}

src/Spout/Reader/XLSX/Manager/SharedStringsCaching/FileBasedStrategy.php

@@ -0,0 +1,192 @@
+<?php
+
+namespace Box\Spout\Reader\XLSX\Manager\SharedStringsCaching;
+
+use Box\Spout\Reader\Exception\SharedStringNotFoundException;
+use Box\Spout\Reader\XLSX\Creator\HelperFactory;
+
+/**
+ * Class FileBasedStrategy
+ *
+ * This class implements the file-based caching strategy for shared strings.
+ * Shared strings are stored in small files (with a max number of strings per file).
+ * This strategy is slower than an in-memory strategy but is used to avoid out of memory crashes.
+ *
+ * @package Box\Spout\Reader\XLSX\Manager\SharedStringsCaching
+ */
+class FileBasedStrategy implements CachingStrategyInterface
+{
+    /** Value to use to escape the line feed character ("\n") */
+    const ESCAPED_LINE_FEED_CHARACTER = '_x000A_';
+
+    /** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper Helper to work with global functions */
+    protected $globalFunctionsHelper;
+
+    /** @var \Box\Spout\Common\Helper\FileSystemHelper Helper to perform file system operations */
+    protected $fileSystemHelper;
+
+    /** @var string Temporary folder where the temporary files will be created */
+    protected $tempFolder;
+
+    /**
+     * @var int Maximum number of strings that can be stored in one temp file
+     * @see CachingStrategyFactory::MAX_NUM_STRINGS_PER_TEMP_FILE
+     */
+    protected $maxNumStringsPerTempFile;
+
+    /** @var resource Pointer to the last temp file a shared string was written to */
+    protected $tempFilePointer;
+
+    /**
+     * @var string Path of the temporary file whose contents is currently stored in memory
+     * @see CachingStrategyFactory::MAX_NUM_STRINGS_PER_TEMP_FILE
+     */
+    protected $inMemoryTempFilePath;
+
+    /**
+     * @var array Contents of the temporary file that was last read
+     * @see CachingStrategyFactory::MAX_NUM_STRINGS_PER_TEMP_FILE
+     */
+    protected $inMemoryTempFileContents;
+
+    /**
+     * @param string $tempFolder Temporary folder where the temporary files to store shared strings will be stored
+     * @param int $maxNumStringsPerTempFile Maximum number of strings that can be stored in one temp file
+     * @param HelperFactory $helperFactory Factory to create helpers
+     */
+    public function __construct($tempFolder, $maxNumStringsPerTempFile, $helperFactory)
+    {
+        $this->fileSystemHelper = $helperFactory->createFileSystemHelper($tempFolder);
+        $this->tempFolder = $this->fileSystemHelper->createFolder($tempFolder, uniqid('sharedstrings'));
+
+        $this->maxNumStringsPerTempFile = $maxNumStringsPerTempFile;
+
+        $this->globalFunctionsHelper = $helperFactory->createGlobalFunctionsHelper();
+        $this->tempFilePointer = null;
+    }
+
+    /**
+     * Adds the given string to the cache.
+     *
+     * @param string $sharedString The string to be added to the cache
+     * @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
+     * @return void
+     */
+    public function addStringForIndex($sharedString, $sharedStringIndex)
+    {
+        $tempFilePath = $this->getSharedStringTempFilePath($sharedStringIndex);
+
+        if (!$this->globalFunctionsHelper->file_exists($tempFilePath)) {
+            if ($this->tempFilePointer) {
+                $this->globalFunctionsHelper->fclose($this->tempFilePointer);
+            }
+            $this->tempFilePointer = $this->globalFunctionsHelper->fopen($tempFilePath, 'w');
+        }
+
+        // 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
+        $lineFeedEncodedSharedString = $this->escapeLineFeed($sharedString);
+
+        $this->globalFunctionsHelper->fwrite($this->tempFilePointer, $lineFeedEncodedSharedString . PHP_EOL);
+    }
+
+    /**
+     * Returns the path for the temp file that should contain the string for the given index
+     *
+     * @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
+     * @return string The temp file path for the given index
+     */
+    protected function getSharedStringTempFilePath($sharedStringIndex)
+    {
+        $numTempFile = intval($sharedStringIndex / $this->maxNumStringsPerTempFile);
+        return $this->tempFolder . '/sharedstrings' . $numTempFile;
+    }
+
+    /**
+     * Closes the cache after the last shared string was added.
+     * This prevents any additional string from being added to the cache.
+     *
+     * @return void
+     */
+    public function closeCache()
+    {
+        // close pointer to the last temp file that was written
+        if ($this->tempFilePointer) {
+            $this->globalFunctionsHelper->fclose($this->tempFilePointer);
+        }
+    }
+
+
+    /**
+     * Returns the string located at the given index from the cache.
+     *
+     * @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
+     * @return string The shared string at the given index
+     * @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If no shared string found for the given index
+     */
+    public function getStringAtIndex($sharedStringIndex)
+    {
+        $tempFilePath = $this->getSharedStringTempFilePath($sharedStringIndex);
+        $indexInFile = $sharedStringIndex % $this->maxNumStringsPerTempFile;
+
+        if (!$this->globalFunctionsHelper->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, $this->globalFunctionsHelper->file_get_contents($tempFilePath));
+            $this->inMemoryTempFilePath = $tempFilePath;
+        }
+
+        $sharedString = null;
+
+        // Using isset here because it is way faster than array_key_exists...
+        if (isset($this->inMemoryTempFileContents[$indexInFile])) {
+            $escapedSharedString = $this->inMemoryTempFileContents[$indexInFile];
+            $sharedString = $this->unescapeLineFeed($escapedSharedString);
+        }
+
+        if ($sharedString === null) {
+            throw new SharedStringNotFoundException("Shared string not found for index: $sharedStringIndex");
+        }
+
+        return rtrim($sharedString, PHP_EOL);
+    }
+
+    /**
+     * Escapes the line feed characters (\n)
+     *
+     * @param string $unescapedString
+     * @return string
+     */
+    private function escapeLineFeed($unescapedString)
+    {
+        return str_replace("\n", self::ESCAPED_LINE_FEED_CHARACTER, $unescapedString);
+    }
+
+    /**
+     * Unescapes the line feed characters (\n)
+     *
+     * @param string $escapedString
+     * @return string
+     */
+    private function unescapeLineFeed($escapedString)
+    {
+        return str_replace(self::ESCAPED_LINE_FEED_CHARACTER, "\n", $escapedString);
+    }
+
+    /**
+     * Destroys the cache, freeing memory and removing any created artifacts
+     *
+     * @return void
+     */
+    public function clearCache()
+    {
+        if ($this->tempFolder) {
+            $this->fileSystemHelper->deleteFolderRecursively($this->tempFolder);
+        }
+    }
+}

src/Spout/Reader/XLSX/Manager/SharedStringsCaching/InMemoryStrategy.php

@@ -0,0 +1,83 @@
+<?php
+
+namespace Box\Spout\Reader\XLSX\Manager\SharedStringsCaching;
+
+use Box\Spout\Reader\Exception\SharedStringNotFoundException;
+
+/**
+ * Class InMemoryStrategy
+ *
+ * This class implements the in-memory caching strategy for shared strings.
+ * This strategy is used when the number of unique strings is low, compared to the memory available.
+ *
+ * @package Box\Spout\Reader\XLSX\Manager\SharedStringsCaching
+ */
+class InMemoryStrategy implements CachingStrategyInterface
+{
+    /** @var \SplFixedArray Array used to cache the shared strings */
+    protected $inMemoryCache;
+
+    /** @var bool Whether the cache has been closed */
+    protected $isCacheClosed;
+
+    /**
+     * @param int $sharedStringsUniqueCount Number of unique shared strings
+     */
+    public function __construct($sharedStringsUniqueCount)
+    {
+        $this->inMemoryCache = new \SplFixedArray($sharedStringsUniqueCount);
+        $this->isCacheClosed = false;
+    }
+
+    /**
+     * Adds the given string to the cache.
+     *
+     * @param string $sharedString The string to be added to the cache
+     * @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
+     * @return void
+     */
+    public function addStringForIndex($sharedString, $sharedStringIndex)
+    {
+        if (!$this->isCacheClosed) {
+            $this->inMemoryCache->offsetSet($sharedStringIndex, $sharedString);
+        }
+    }
+
+    /**
+     * Closes the cache after the last shared string was added.
+     * This prevents any additional string from being added to the cache.
+     *
+     * @return void
+     */
+    public function closeCache()
+    {
+        $this->isCacheClosed = true;
+    }
+
+    /**
+     * Returns the string located at the given index from the cache.
+     *
+     * @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
+     * @return string The shared string at the given index
+     * @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If no shared string found for the given index
+     */
+    public function getStringAtIndex($sharedStringIndex)
+    {
+        try {
+            return $this->inMemoryCache->offsetGet($sharedStringIndex);
+        } catch (\RuntimeException $e) {
+            throw new SharedStringNotFoundException("Shared string not found for index: $sharedStringIndex");
+        }
+    }
+
+    /**
+     * Destroys the cache, freeing memory and removing any created artifacts
+     *
+     * @return void
+     */
+    public function clearCache()
+    {
+        unset($this->inMemoryCache);
+        $this->isCacheClosed = false;
+    }
+}