FileBasedStrategy::getStringAtIndex() - Code Metrics - Inspection of "Introduce Managers for readers" - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — develop_3.0 (#460)

by Adrien

created 2017-08-27 01:46 UTC

FileBasedStrategy::getStringAtIndex() B

↳ Parent: FileBasedStrategy

Complexity

Conditions	5
Paths	9

Size

Total Lines	31
Code Lines	16

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	14
CRAP Score	5.0488

Importance

Changes

Metric	Value
dl	0
loc	31
ccs	14
cts	16
cp	0.875
rs	8.439
c	0
b	0
f	0
cc	5
eloc	16
nc	9
nop	1
crap	5.0488

<?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);
        }
    }
}


1		<?php
2
3		namespace Box\Spout\Reader\XLSX\Manager\SharedStringsCaching;
4
5		use Box\Spout\Reader\Exception\SharedStringNotFoundException;
6		use Box\Spout\Reader\XLSX\Creator\HelperFactory;
7
8		/**
9		* Class FileBasedStrategy
10		*
11		* This class implements the file-based caching strategy for shared strings.
12		* Shared strings are stored in small files (with a max number of strings per file).
13		* This strategy is slower than an in-memory strategy but is used to avoid out of memory crashes.
14		*
15		* @package Box\Spout\Reader\XLSX\Manager\SharedStringsCaching
16		*/
17		class FileBasedStrategy implements CachingStrategyInterface
18		{
19		/** Value to use to escape the line feed character ("\n") */
20		const ESCAPED_LINE_FEED_CHARACTER = '_x000A_';
21
22		/** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper Helper to work with global functions */
23		protected $globalFunctionsHelper;
24
25		/** @var \Box\Spout\Common\Helper\FileSystemHelper Helper to perform file system operations */
26		protected $fileSystemHelper;
27
28		/** @var string Temporary folder where the temporary files will be created */
29		protected $tempFolder;
30
31		/**
32		* @var int Maximum number of strings that can be stored in one temp file
33		* @see CachingStrategyFactory::MAX_NUM_STRINGS_PER_TEMP_FILE
34		*/
35		protected $maxNumStringsPerTempFile;
36
37		/** @var resource Pointer to the last temp file a shared string was written to */
38		protected $tempFilePointer;
39
40		/**
41		* @var string Path of the temporary file whose contents is currently stored in memory
42		* @see CachingStrategyFactory::MAX_NUM_STRINGS_PER_TEMP_FILE
43		*/
44		protected $inMemoryTempFilePath;
45
46		/**
47		* @var array Contents of the temporary file that was last read
48		* @see CachingStrategyFactory::MAX_NUM_STRINGS_PER_TEMP_FILE
49		*/
50		protected $inMemoryTempFileContents;
51
52		/**
53		* @param string $tempFolder Temporary folder where the temporary files to store shared strings will be stored
54		* @param int $maxNumStringsPerTempFile Maximum number of strings that can be stored in one temp file
55		* @param HelperFactory $helperFactory Factory to create helpers
56		*/
57	8	public function __construct($tempFolder, $maxNumStringsPerTempFile, $helperFactory)
58		{
59	8	$this->fileSystemHelper = $helperFactory->createFileSystemHelper($tempFolder);
60	8	$this->tempFolder = $this->fileSystemHelper->createFolder($tempFolder, uniqid('sharedstrings'));
61
62	8	$this->maxNumStringsPerTempFile = $maxNumStringsPerTempFile;
63
64	8	$this->globalFunctionsHelper = $helperFactory->createGlobalFunctionsHelper();
65	8	$this->tempFilePointer = null;
66	8	}
67
68		/**
69		* Adds the given string to the cache.
70		*
71		* @param string $sharedString The string to be added to the cache
72		* @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
73		* @return void
74		*/
75	2	public function addStringForIndex($sharedString, $sharedStringIndex)
76		{
77	2	$tempFilePath = $this->getSharedStringTempFilePath($sharedStringIndex);
78
79	2	if (!$this->globalFunctionsHelper->file_exists($tempFilePath)) {
80	2	if ($this->tempFilePointer) {
81	1	$this->globalFunctionsHelper->fclose($this->tempFilePointer);
82		}
83	2	$this->tempFilePointer = $this->globalFunctionsHelper->fopen($tempFilePath, 'w');
84		}
85
86		// The shared string retrieval logic expects each cell data to be on one line only
87		// Encoding the line feed character allows to preserve this assumption
88	2	$lineFeedEncodedSharedString = $this->escapeLineFeed($sharedString);
89
90	2	$this->globalFunctionsHelper->fwrite($this->tempFilePointer, $lineFeedEncodedSharedString . PHP_EOL);
91	2	}
92
93		/**
94		* Returns the path for the temp file that should contain the string for the given index
95		*
96		* @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
97		* @return string The temp file path for the given index
98		*/
99	2	protected function getSharedStringTempFilePath($sharedStringIndex)
100		{
101	2	$numTempFile = intval($sharedStringIndex / $this->maxNumStringsPerTempFile);
102	2	return $this->tempFolder . '/sharedstrings' . $numTempFile;
103		}
104
105		/**
106		* Closes the cache after the last shared string was added.
107		* This prevents any additional string from being added to the cache.
108		*
109		* @return void
110		*/
111	3	public function closeCache()
112		{
113		// close pointer to the last temp file that was written
114	3	if ($this->tempFilePointer) {
115	2	$this->globalFunctionsHelper->fclose($this->tempFilePointer);
116		}
117	3	}
118
119
120		/**
121		* Returns the string located at the given index from the cache.
122		*
123		* @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
124		* @return string The shared string at the given index
125		* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If no shared string found for the given index
126		*/
127	2	public function getStringAtIndex($sharedStringIndex)
128		{
129	2	$tempFilePath = $this->getSharedStringTempFilePath($sharedStringIndex);
130	2	$indexInFile = $sharedStringIndex % $this->maxNumStringsPerTempFile;
131
132	2	if (!$this->globalFunctionsHelper->file_exists($tempFilePath)) {
133		throw new SharedStringNotFoundException("Shared string temp file not found: $tempFilePath ; for index: $sharedStringIndex");
134		}
135
136	2	if ($this->inMemoryTempFilePath !== $tempFilePath) {
137		// free memory
138	2	unset($this->inMemoryTempFileContents);
139
140	2	$this->inMemoryTempFileContents = explode(PHP_EOL, $this->globalFunctionsHelper->file_get_contents($tempFilePath));
141	2	$this->inMemoryTempFilePath = $tempFilePath;
142		}
143
144	2	$sharedString = null;
145
146		// Using isset here because it is way faster than array_key_exists...
147	2	if (isset($this->inMemoryTempFileContents[$indexInFile])) {
148	2	$escapedSharedString = $this->inMemoryTempFileContents[$indexInFile];
149	2	$sharedString = $this->unescapeLineFeed($escapedSharedString);
150		}
151
152	2	if ($sharedString === null) {
153		throw new SharedStringNotFoundException("Shared string not found for index: $sharedStringIndex");
154		}
155
156	2	return rtrim($sharedString, PHP_EOL);
157		}
158
159		/**
160		* Escapes the line feed characters (\n)
161		*
162		* @param string $unescapedString
163		* @return string
164		*/
165	2	private function escapeLineFeed($unescapedString)
166		{
167	2	return str_replace("\n", self::ESCAPED_LINE_FEED_CHARACTER, $unescapedString);
168		}
169
170		/**
171		* Unescapes the line feed characters (\n)
172		*
173		* @param string $escapedString
174		* @return string
175		*/
176	2	private function unescapeLineFeed($escapedString)
177		{
178	2	return str_replace(self::ESCAPED_LINE_FEED_CHARACTER, "\n", $escapedString);
179		}
180
181		/**
182		* Destroys the cache, freeing memory and removing any created artifacts
183		*
184		* @return void
185		*/
186	8	public function clearCache()
187		{
188	8	if ($this->tempFolder) {
189	8	$this->fileSystemHelper->deleteFolderRecursively($this->tempFolder);
190		}
191	8	}
192		}
193

box / spout

Pull Request — develop_3.0 (#460)

FileBasedStrategy::getStringAtIndex() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like