SharedStringsManager::shouldPreserveWhitespace() - 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 01:51 UTC

SharedStringsManager::shouldPreserveWhitespace() A

↳ Parent: SharedStringsManager

Complexity

Conditions	1
Paths	1

Size

Total Lines	5
Code Lines	3

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	3
CRAP Score	1

Importance

Changes

Metric	Value
dl	0
loc	5
ccs	3
cts	3
cp	1
rs	9.4285
c	0
b	0
f	0
cc	1
eloc	3
nc	1
nop	1
crap	1

<?php

namespace Box\Spout\Reader\XLSX\Manager;

use Box\Spout\Common\Exception\IOException;
use Box\Spout\Reader\Exception\XMLProcessingException;
use Box\Spout\Reader\Wrapper\XMLReader;
use Box\Spout\Reader\XLSX\Creator\EntityFactory;
use Box\Spout\Reader\XLSX\Creator\HelperFactory;
use Box\Spout\Reader\XLSX\Manager\SharedStringsCaching\CachingStrategyFactory;
use Box\Spout\Reader\XLSX\Manager\SharedStringsCaching\CachingStrategyInterface;

/**
 * Class SharedStringsManager
 * This class manages the shared strings defined in the associated XML file
 *
 * @package Box\Spout\Reader\XLSX\Manager
 */
class SharedStringsManager
{
    /** Path of sharedStrings XML file inside the XLSX file */
    const SHARED_STRINGS_XML_FILE_PATH = 'xl/sharedStrings.xml';

    /** Main namespace for the sharedStrings.xml file */
    const MAIN_NAMESPACE_FOR_SHARED_STRINGS_XML = 'http://schemas.openxmlformats.org/spreadsheetml/2006/main';

    /** Definition of XML nodes names used to parse data */
    const XML_NODE_SST = 'sst';
    const XML_NODE_SI = 'si';
    const XML_NODE_R = 'r';
    const XML_NODE_T = 't';

    /** Definition of XML attributes used to parse data */
    const XML_ATTRIBUTE_COUNT = 'count';
    const XML_ATTRIBUTE_UNIQUE_COUNT = 'uniqueCount';
    const XML_ATTRIBUTE_XML_SPACE = 'xml:space';
    const XML_ATTRIBUTE_VALUE_PRESERVE = 'preserve';

    /** @var string Path of the XLSX file being read */
    protected $filePath;

    /** @var string Temporary folder where the temporary files to store shared strings will be stored */
    protected $tempFolder;

    /** @var EntityFactory Factory to create entities */
    protected $entityFactory;

    /** @var HelperFactory $helperFactory Factory to create helpers */
    protected $helperFactory;

    /** @var CachingStrategyFactory Factory to create shared strings caching strategies */
    protected $cachingStrategyFactory;

    /** @var CachingStrategyInterface The best caching strategy for storing shared strings */
    protected $cachingStrategy;

    /**
     * @param string $filePath Path of the XLSX file being read
     * @param string $tempFolder Temporary folder where the temporary files to store shared strings will be stored
     * @param EntityFactory $entityFactory Factory to create entities
     * @param HelperFactory $helperFactory Factory to create helpers
     * @param CachingStrategyFactory $cachingStrategyFactory Factory to create shared strings caching strategies
     */
    public function __construct($filePath, $tempFolder, $entityFactory, $helperFactory, $cachingStrategyFactory)
    {
        $this->filePath = $filePath;
        $this->tempFolder = $tempFolder;
        $this->entityFactory = $entityFactory;
        $this->helperFactory = $helperFactory;
        $this->cachingStrategyFactory = $cachingStrategyFactory;
    }

    /**
     * Returns whether the XLSX file contains a shared strings XML file
     *
     * @return bool
     */
    public function hasSharedStrings()
    {
        $hasSharedStrings = false;
        $zip = $this->entityFactory->createZipArchive();

        if ($zip->open($this->filePath) === true) {
            $hasSharedStrings = ($zip->locateName(self::SHARED_STRINGS_XML_FILE_PATH) !== false);
            $zip->close();
        }

        return $hasSharedStrings;
    }

    /**
     * Builds an in-memory array containing all the shared strings of the sheet.
     * All the strings are stored in a XML file, located at 'xl/sharedStrings.xml'.
     * It is then accessed by the sheet data, via the string index in the built table.
     *
     * More documentation available here: http://msdn.microsoft.com/en-us/library/office/gg278314.aspx
     *
     * The XML file can be really big with sheets containing a lot of data. That is why
     * we need to use a XML reader that provides streaming like the XMLReader library.
     *
     * @return void
     * @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml can't be read
     */
    public function extractSharedStrings()
    {
        $xmlReader = $this->entityFactory->createXMLReader();
        $sharedStringIndex = 0;

        if ($xmlReader->openFileInZip($this->filePath, self::SHARED_STRINGS_XML_FILE_PATH) === false) {
            throw new IOException('Could not open "' . self::SHARED_STRINGS_XML_FILE_PATH . '".');
        }

        try {
            $sharedStringsUniqueCount = $this->getSharedStringsUniqueCount($xmlReader);
            $this->cachingStrategy = $this->getBestSharedStringsCachingStrategy($sharedStringsUniqueCount);

            $xmlReader->readUntilNodeFound(self::XML_NODE_SI);

            while ($xmlReader->name === self::XML_NODE_SI) {
                $this->processSharedStringsItem($xmlReader, $sharedStringIndex);
                $sharedStringIndex++;

                // jump to the next '<si>' tag
                $xmlReader->next(self::XML_NODE_SI);
            }

            $this->cachingStrategy->closeCache();

        } catch (XMLProcessingException $exception) {
            throw new IOException("The sharedStrings.xml file is invalid and cannot be read. [{$exception->getMessage()}]");
        }

        $xmlReader->close();
    }

    /**
     * Returns the shared strings unique count, as specified in <sst> tag.
     *
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader instance
     * @return int|null Number of unique shared strings in the sharedStrings.xml file
     * @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml is invalid and can't be read
     */
    protected function getSharedStringsUniqueCount($xmlReader)
    {
        $xmlReader->next(self::XML_NODE_SST);

        // Iterate over the "sst" elements to get the actual "sst ELEMENT" (skips any DOCTYPE)
        while ($xmlReader->name === self::XML_NODE_SST && $xmlReader->nodeType !== XMLReader::ELEMENT) {
            $xmlReader->read();
        }

        $uniqueCount = $xmlReader->getAttribute(self::XML_ATTRIBUTE_UNIQUE_COUNT);

        // some software do not add the "uniqueCount" attribute but only use the "count" one
        // @see https://github.com/box/spout/issues/254
        if ($uniqueCount === null) {
            $uniqueCount = $xmlReader->getAttribute(self::XML_ATTRIBUTE_COUNT);
        }

        return ($uniqueCount !== null) ? intval($uniqueCount) : null;
    }

    /**
     * Returns the best shared strings caching strategy.
     *
     * @param int|null $sharedStringsUniqueCount Number of unique shared strings (NULL if unknown)
     * @return CachingStrategyInterface
     */
    protected function getBestSharedStringsCachingStrategy($sharedStringsUniqueCount)
    {
        return $this->cachingStrategyFactory
                ->createBestCachingStrategy($sharedStringsUniqueCount, $this->tempFolder, $this->helperFactory);
    }

    /**
     * Processes the shared strings item XML node which the given XML reader is positioned on.
     *
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XML Reader positioned on a "<si>" node
     * @param int $sharedStringIndex Index of the processed shared strings item
     * @return void
     */
    protected function processSharedStringsItem($xmlReader, $sharedStringIndex)
    {
        $sharedStringValue = '';

        // NOTE: expand() will automatically decode all XML entities of the child nodes
        $siNode = $xmlReader->expand();
        $textNodes = $siNode->getElementsByTagName(self::XML_NODE_T);

        foreach ($textNodes as $textNode) {
            if ($this->shouldExtractTextNodeValue($textNode)) {
                $textNodeValue = $textNode->nodeValue;
                $shouldPreserveWhitespace = $this->shouldPreserveWhitespace($textNode);

                $sharedStringValue .= ($shouldPreserveWhitespace) ? $textNodeValue : trim($textNodeValue);
            }
        }

        $this->cachingStrategy->addStringForIndex($sharedStringValue, $sharedStringIndex);
    }

    /**
     * Not all text nodes' values must be extracted.
     * Some text nodes are part of a node describing the pronunciation for instance.
     * We'll only consider the nodes whose parents are "<si>" or "<r>".
     *
     * @param \DOMElement $textNode Text node to check
     * @return bool Whether the given text node's value must be extracted
     */
    protected function shouldExtractTextNodeValue($textNode)
    {
        $parentTagName = $textNode->parentNode->localName;
        return ($parentTagName === self::XML_NODE_SI || $parentTagName === self::XML_NODE_R);
    }

    /**
     * If the text node has the attribute 'xml:space="preserve"', then preserve whitespace.
     *
     * @param \DOMElement $textNode The text node element (<t>) whose whitespace may be preserved
     * @return bool Whether whitespace should be preserved
     */
    protected function shouldPreserveWhitespace($textNode)
    {
        $spaceValue = $textNode->getAttribute(self::XML_ATTRIBUTE_XML_SPACE);
        return ($spaceValue === self::XML_ATTRIBUTE_VALUE_PRESERVE);
    }

    /**
     * Returns the shared string at the given index, using the previously chosen caching strategy.
     *
     * @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)
    {
        return $this->cachingStrategy->getStringAtIndex($sharedStringIndex);
    }

    /**
     * Destroys the cache, freeing memory and removing any created artifacts
     *
     * @return void
     */
    public function cleanup()
    {
        if ($this->cachingStrategy) {
            $this->cachingStrategy->clearCache();
        }
    }
}


1		<?php
2
3		namespace Box\Spout\Reader\XLSX\Manager;
4
5		use Box\Spout\Common\Exception\IOException;
6		use Box\Spout\Reader\Exception\XMLProcessingException;
7		use Box\Spout\Reader\Wrapper\XMLReader;
8		use Box\Spout\Reader\XLSX\Creator\EntityFactory;
9		use Box\Spout\Reader\XLSX\Creator\HelperFactory;
10		use Box\Spout\Reader\XLSX\Manager\SharedStringsCaching\CachingStrategyFactory;
11		use Box\Spout\Reader\XLSX\Manager\SharedStringsCaching\CachingStrategyInterface;
12
13		/**
14		* Class SharedStringsManager
15		* This class manages the shared strings defined in the associated XML file
16		*
17		* @package Box\Spout\Reader\XLSX\Manager
18		*/
19		class SharedStringsManager
20		{
21		/** Path of sharedStrings XML file inside the XLSX file */
22		const SHARED_STRINGS_XML_FILE_PATH = 'xl/sharedStrings.xml';
23
24		/** Main namespace for the sharedStrings.xml file */
25		const MAIN_NAMESPACE_FOR_SHARED_STRINGS_XML = 'http://schemas.openxmlformats.org/spreadsheetml/2006/main';
26
27		/** Definition of XML nodes names used to parse data */
28		const XML_NODE_SST = 'sst';
29		const XML_NODE_SI = 'si';
30		const XML_NODE_R = 'r';
31		const XML_NODE_T = 't';
32
33		/** Definition of XML attributes used to parse data */
34		const XML_ATTRIBUTE_COUNT = 'count';
35		const XML_ATTRIBUTE_UNIQUE_COUNT = 'uniqueCount';
36		const XML_ATTRIBUTE_XML_SPACE = 'xml:space';
37		const XML_ATTRIBUTE_VALUE_PRESERVE = 'preserve';
38
39		/** @var string Path of the XLSX file being read */
40		protected $filePath;
41
42		/** @var string Temporary folder where the temporary files to store shared strings will be stored */
43		protected $tempFolder;
44
45		/** @var EntityFactory Factory to create entities */
46		protected $entityFactory;
47
48		/** @var HelperFactory $helperFactory Factory to create helpers */
49		protected $helperFactory;
50
51		/** @var CachingStrategyFactory Factory to create shared strings caching strategies */
52		protected $cachingStrategyFactory;
53
54		/** @var CachingStrategyInterface The best caching strategy for storing shared strings */
55		protected $cachingStrategy;
56
57		/**
58		* @param string $filePath Path of the XLSX file being read
59		* @param string $tempFolder Temporary folder where the temporary files to store shared strings will be stored
60		* @param EntityFactory $entityFactory Factory to create entities
61		* @param HelperFactory $helperFactory Factory to create helpers
62		* @param CachingStrategyFactory $cachingStrategyFactory Factory to create shared strings caching strategies
63		*/
64	41	public function __construct($filePath, $tempFolder, $entityFactory, $helperFactory, $cachingStrategyFactory)
65		{
66	41	$this->filePath = $filePath;
67	41	$this->tempFolder = $tempFolder;
68	41	$this->entityFactory = $entityFactory;
69	41	$this->helperFactory = $helperFactory;
70	41	$this->cachingStrategyFactory = $cachingStrategyFactory;
71	41	}
72
73		/**
74		* Returns whether the XLSX file contains a shared strings XML file
75		*
76		* @return bool
77		*/
78	35	public function hasSharedStrings()
79		{
80	35	$hasSharedStrings = false;
81	35	$zip = $this->entityFactory->createZipArchive();
82
83	35	if ($zip->open($this->filePath) === true) {
84	35	$hasSharedStrings = ($zip->locateName(self::SHARED_STRINGS_XML_FILE_PATH) !== false);
85	35	$zip->close();
86		}
87
88	35	return $hasSharedStrings;
89		}
90
91		/**
92		* Builds an in-memory array containing all the shared strings of the sheet.
93		* All the strings are stored in a XML file, located at 'xl/sharedStrings.xml'.
94		* It is then accessed by the sheet data, via the string index in the built table.
95		*
96		* More documentation available here: http://msdn.microsoft.com/en-us/library/office/gg278314.aspx
97		*
98		* The XML file can be really big with sheets containing a lot of data. That is why
99		* we need to use a XML reader that provides streaming like the XMLReader library.
100		*
101		* @return void
102		* @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml can't be read
103		*/
104	35	public function extractSharedStrings()
105		{
106	35	$xmlReader = $this->entityFactory->createXMLReader();
107	35	$sharedStringIndex = 0;
108
109	35	if ($xmlReader->openFileInZip($this->filePath, self::SHARED_STRINGS_XML_FILE_PATH) === false) {
110		throw new IOException('Could not open "' . self::SHARED_STRINGS_XML_FILE_PATH . '".');
111		}
112
113		try {
114	35	$sharedStringsUniqueCount = $this->getSharedStringsUniqueCount($xmlReader);
115	34	$this->cachingStrategy = $this->getBestSharedStringsCachingStrategy($sharedStringsUniqueCount);
116
117	34	$xmlReader->readUntilNodeFound(self::XML_NODE_SI);
118
119	34	while ($xmlReader->name === self::XML_NODE_SI) {
120	24	$this->processSharedStringsItem($xmlReader, $sharedStringIndex);
121	24	$sharedStringIndex++;
122
123		// jump to the next '<si>' tag
124	24	$xmlReader->next(self::XML_NODE_SI);
125		}
126
127	34	$this->cachingStrategy->closeCache();
128
129	1	} catch (XMLProcessingException $exception) {
130	1	throw new IOException("The sharedStrings.xml file is invalid and cannot be read. [{$exception->getMessage()}]");
131		}
132
133	34	$xmlReader->close();
134	34	}
135
136		/**
137		* Returns the shared strings unique count, as specified in <sst> tag.
138		*
139		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader instance
140		* @return int\|null Number of unique shared strings in the sharedStrings.xml file
141		* @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml is invalid and can't be read
142		*/
143	35	protected function getSharedStringsUniqueCount($xmlReader)
144		{
145	35	$xmlReader->next(self::XML_NODE_SST);
146
147		// Iterate over the "sst" elements to get the actual "sst ELEMENT" (skips any DOCTYPE)
148	34	while ($xmlReader->name === self::XML_NODE_SST && $xmlReader->nodeType !== XMLReader::ELEMENT) {
149	1	$xmlReader->read();
150		}
151
152	34	$uniqueCount = $xmlReader->getAttribute(self::XML_ATTRIBUTE_UNIQUE_COUNT);
153
154		// some software do not add the "uniqueCount" attribute but only use the "count" one
155		// @see https://github.com/box/spout/issues/254
156	34	if ($uniqueCount === null) {
157	3	$uniqueCount = $xmlReader->getAttribute(self::XML_ATTRIBUTE_COUNT);
158		}
159
160	34	return ($uniqueCount !== null) ? intval($uniqueCount) : null;
161		}
162
163		/**
164		* Returns the best shared strings caching strategy.
165		*
166		* @param int\|null $sharedStringsUniqueCount Number of unique shared strings (NULL if unknown)
167		* @return CachingStrategyInterface
168		*/
169	34	protected function getBestSharedStringsCachingStrategy($sharedStringsUniqueCount)
170		{
171	34	return $this->cachingStrategyFactory
172	34	->createBestCachingStrategy($sharedStringsUniqueCount, $this->tempFolder, $this->helperFactory);
173		}
174
175		/**
176		* Processes the shared strings item XML node which the given XML reader is positioned on.
177		*
178		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XML Reader positioned on a "<si>" node
179		* @param int $sharedStringIndex Index of the processed shared strings item
180		* @return void
181		*/
182	24	protected function processSharedStringsItem($xmlReader, $sharedStringIndex)
183		{
184	24	$sharedStringValue = '';
185
186		// NOTE: expand() will automatically decode all XML entities of the child nodes
187	24	$siNode = $xmlReader->expand();
188	24	$textNodes = $siNode->getElementsByTagName(self::XML_NODE_T);
189
190	24	foreach ($textNodes as $textNode) {
191	24	if ($this->shouldExtractTextNodeValue($textNode)) {
192	24	$textNodeValue = $textNode->nodeValue;
193	24	$shouldPreserveWhitespace = $this->shouldPreserveWhitespace($textNode);
194
195	24	$sharedStringValue .= ($shouldPreserveWhitespace) ? $textNodeValue : trim($textNodeValue);
196		}
197		}
198
199	24	$this->cachingStrategy->addStringForIndex($sharedStringValue, $sharedStringIndex);
200	24	}
201
202		/**
203		* Not all text nodes' values must be extracted.
204		* Some text nodes are part of a node describing the pronunciation for instance.
205		* We'll only consider the nodes whose parents are "<si>" or "<r>".
206		*
207		* @param \DOMElement $textNode Text node to check
208		* @return bool Whether the given text node's value must be extracted
209		*/
210	24	protected function shouldExtractTextNodeValue($textNode)
211		{
212	24	$parentTagName = $textNode->parentNode->localName;
213	24	return ($parentTagName === self::XML_NODE_SI \|\| $parentTagName === self::XML_NODE_R);
214		}
215
216		/**
217		* If the text node has the attribute 'xml:space="preserve"', then preserve whitespace.
218		*
219		* @param \DOMElement $textNode The text node element (<t>) whose whitespace may be preserved
220		* @return bool Whether whitespace should be preserved
221		*/
222	24	protected function shouldPreserveWhitespace($textNode)
223		{
224	24	$spaceValue = $textNode->getAttribute(self::XML_ATTRIBUTE_XML_SPACE);
225	24	return ($spaceValue === self::XML_ATTRIBUTE_VALUE_PRESERVE);
226		}
227
228		/**
229		* Returns the shared string at the given index, using the previously chosen caching strategy.
230		*
231		* @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
232		* @return string The shared string at the given index
233		* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If no shared string found for the given index
234		*/
235	24	public function getStringAtIndex($sharedStringIndex)
236		{
237	24	return $this->cachingStrategy->getStringAtIndex($sharedStringIndex);
238		}
239
240		/**
241		* Destroys the cache, freeing memory and removing any created artifacts
242		*
243		* @return void
244		*/
245	38	public function cleanup()
246		{
247	38	if ($this->cachingStrategy) {
248	32	$this->cachingStrategy->clearCache();
249		}
250	38	}
251		}
252

box / spout

Pull Request — develop_3.0 (#460)

SharedStringsManager::shouldPreserveWhitespace() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like