SharedStringsHelper::getBestSharedStringsCachingStrategy() - Code Metrics - Inspection of "XLSX Reader should add a space between text nodes" - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#229)

by Adrien

created 2016-05-23 21:13 UTC

getBestSharedStringsCachingStrategy() A

↳ Parent: SharedStringsHelper

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	1
Bugs	0	Features	0

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

<?php

namespace Box\Spout\Reader\XLSX\Helper;

use Box\Spout\Common\Exception\IOException;
use Box\Spout\Reader\Exception\XMLProcessingException;
use Box\Spout\Reader\Wrapper\SimpleXMLElement;
use Box\Spout\Reader\Wrapper\XMLReader;
use Box\Spout\Reader\XLSX\Helper\SharedStringsCaching\CachingStrategyFactory;
use Box\Spout\Reader\XLSX\Helper\SharedStringsCaching\CachingStrategyInterface;

/**
 * Class SharedStringsHelper
 * This class provides helper functions for reading sharedStrings XML file
 *
 * @package Box\Spout\Reader\XLSX\Helper
 */
class SharedStringsHelper
{
    /** 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';

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

    /**
     * @param string $filePath Path of the XLSX file being read
     * @param string|void $tempFolder Temporary folder where the temporary files to store shared strings will be stored
     */
    public function __construct($filePath, $tempFolder = null)
    {
        $this->filePath = $filePath;
        $this->tempFolder = $tempFolder;
    }

    /**
     * Returns whether the XLSX file contains a shared strings XML file
     *
     * @return bool
     */
    public function hasSharedStrings()
    {
        $hasSharedStrings = false;
        $zip = new \ZipArchive();

        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.
     * Please note that SimpleXML does not provide such a functionality but since it is faster
     * and more handy to parse few XML nodes, it is used in combination with XMLReader for that purpose.
     *
     * @return void
     * @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml can't be read
     */
    public function extractSharedStrings()
    {
        $xmlReader = new XMLReader();
        $sharedStringIndex = 0;
        /** @noinspection PhpUnnecessaryFullyQualifiedNameInspection */
        $escaper = new \Box\Spout\Common\Escaper\XLSX();

        $sharedStringsFilePath = $this->getSharedStringsFilePath();
        if ($xmlReader->open($sharedStringsFilePath) === 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('si');

            while ($xmlReader->name === 'si') {
                $node = $this->getSimpleXmlElementNodeFromXMLReader($xmlReader);
                $node->registerXPathNamespace('ns', self::MAIN_NAMESPACE_FOR_SHARED_STRINGS_XML);

                // removes nodes that should not be read, like the pronunciation of the Kanji characters
                $cleanNode = $this->removeSuperfluousTextNodes($node);

                // find all text nodes "t"; there can be multiple if the cell contains formatting
                $textNodes = $cleanNode->xpath('//ns:t');

                $textValue = '';
                foreach ($textNodes as $nodeIndex => $textNode) {
                    if ($nodeIndex !== 0) {
                        // add a space between each "t" node
                        $textValue .= ' ';
                    }

                    if ($this->shouldPreserveWhitespace($textNode)) {
                        $textValue .= $textNode->__toString();
                    } else {
                        $textValue .= trim($textNode->__toString());
                    }
                }

                $unescapedTextValue = $escaper->unescape($textValue);
                $this->cachingStrategy->addStringForIndex($unescapedTextValue, $sharedStringIndex);

                $sharedStringIndex++;

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

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

        $this->cachingStrategy->closeCache();

        $xmlReader->close();
    }

    /**
     * @return string The path to the shared strings XML file
     */
    protected function getSharedStringsFilePath()
    {
        return 'zip://' . $this->filePath . '#' . self::SHARED_STRINGS_XML_FILE_PATH;
    }

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

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

        return intval($xmlReader->getAttribute('uniqueCount'));
    }

    /**
     * Returns the best shared strings caching strategy.
     *
     * @param int $sharedStringsUniqueCount
     * @return CachingStrategyInterface
     */
    protected function getBestSharedStringsCachingStrategy($sharedStringsUniqueCount)
    {
        return CachingStrategyFactory::getInstance()
                ->getBestCachingStrategy($sharedStringsUniqueCount, $this->tempFolder);
    }

    /**
     * Returns a SimpleXMLElement node from the current node in the given XMLReader instance.
     * This is to simplify the parsing of the subtree.
     *
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader
     * @return \Box\Spout\Reader\Wrapper\SimpleXMLElement
     * @throws \Box\Spout\Common\Exception\IOException If the current node cannot be read
     */
    protected function getSimpleXmlElementNodeFromXMLReader($xmlReader)
    {
        $node = null;
        try {
            $node = new SimpleXMLElement($xmlReader->readOuterXml());
        } catch (XMLProcessingException $exception) {
            throw new IOException("The sharedStrings.xml file contains unreadable data [{$exception->getMessage()}].");
        }

        return $node;
    }

    /**
     * Removes nodes that should not be read, like the pronunciation of the Kanji characters.
     * By keeping them, their text content would be added to the read string.
     *
     * @param \Box\Spout\Reader\Wrapper\SimpleXMLElement $parentNode Parent node that may contain nodes to remove
     * @return \Box\Spout\Reader\Wrapper\SimpleXMLElement Cleaned parent node
     */
    protected function removeSuperfluousTextNodes($parentNode)
    {
        $tagsToRemove = [
            'rPh', // Pronunciation of the text
            'pPr', // Paragraph Properties / Previous Paragraph Properties
            'rPr', // Run Properties for the Paragraph Mark / Previous Run Properties for the Paragraph Mark
        ];

        foreach ($tagsToRemove as $tagToRemove) {
            $xpath = '//ns:' . $tagToRemove;
            $parentNode->removeNodesMatchingXPath($xpath);
        }

        return $parentNode;
    }

    /**
     * If the text node has the attribute 'xml:space="preserve"', then preserve whitespace.
     *
     * @param \Box\Spout\Reader\Wrapper\SimpleXMLElement $textNode The text node element (<t>) whitespace may be preserved
     * @return bool Whether whitespace should be preserved
     */
    protected function shouldPreserveWhitespace($textNode)
    {
        $spaceValue = $textNode->getAttribute('space', 'xml');
        return ($spaceValue === '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\Helper;
4
5		use Box\Spout\Common\Exception\IOException;
6		use Box\Spout\Reader\Exception\XMLProcessingException;
7		use Box\Spout\Reader\Wrapper\SimpleXMLElement;
8		use Box\Spout\Reader\Wrapper\XMLReader;
9		use Box\Spout\Reader\XLSX\Helper\SharedStringsCaching\CachingStrategyFactory;
10		use Box\Spout\Reader\XLSX\Helper\SharedStringsCaching\CachingStrategyInterface;
11
12		/**
13		* Class SharedStringsHelper
14		* This class provides helper functions for reading sharedStrings XML file
15		*
16		* @package Box\Spout\Reader\XLSX\Helper
17		*/
18		class SharedStringsHelper
19		{
20		/** Path of sharedStrings XML file inside the XLSX file */
21		const SHARED_STRINGS_XML_FILE_PATH = 'xl/sharedStrings.xml';
22
23		/** Main namespace for the sharedStrings.xml file */
24		const MAIN_NAMESPACE_FOR_SHARED_STRINGS_XML = 'http://schemas.openxmlformats.org/spreadsheetml/2006/main';
25
26		/** @var string Path of the XLSX file being read */
27		protected $filePath;
28
29		/** @var string Temporary folder where the temporary files to store shared strings will be stored */
30		protected $tempFolder;
31
32		/** @var CachingStrategyInterface The best caching strategy for storing shared strings */
33		protected $cachingStrategy;
34
35		/**
36		* @param string $filePath Path of the XLSX file being read
37		* @param string\|void $tempFolder Temporary folder where the temporary files to store shared strings will be stored
38		*/
39	102	public function __construct($filePath, $tempFolder = null)
40		{
41	102	$this->filePath = $filePath;
42	102	$this->tempFolder = $tempFolder;
43	102	}
44
45		/**
46		* Returns whether the XLSX file contains a shared strings XML file
47		*
48		* @return bool
49		*/
50	89	public function hasSharedStrings()
51		{
52	87	$hasSharedStrings = false;
53	87	$zip = new \ZipArchive();
54
55	87	if ($zip->open($this->filePath) === true) {
56	87	$hasSharedStrings = ($zip->locateName(self::SHARED_STRINGS_XML_FILE_PATH) !== false);
57	87	$zip->close();
58	87	}
59
60	89	return $hasSharedStrings;
61		}
62
63		/**
64		* Builds an in-memory array containing all the shared strings of the sheet.
65		* All the strings are stored in a XML file, located at 'xl/sharedStrings.xml'.
66		* It is then accessed by the sheet data, via the string index in the built table.
67		*
68		* More documentation available here: http://msdn.microsoft.com/en-us/library/office/gg278314.aspx
69		*
70		* The XML file can be really big with sheets containing a lot of data. That is why
71		* we need to use a XML reader that provides streaming like the XMLReader library.
72		* Please note that SimpleXML does not provide such a functionality but since it is faster
73		* and more handy to parse few XML nodes, it is used in combination with XMLReader for that purpose.
74		*
75		* @return void
76		* @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml can't be read
77		*/
78	90	public function extractSharedStrings()
79		{
80	90	$xmlReader = new XMLReader();
81	90	$sharedStringIndex = 0;
82		/** @noinspection PhpUnnecessaryFullyQualifiedNameInspection */
83	90	$escaper = new \Box\Spout\Common\Escaper\XLSX();
84
85	90	$sharedStringsFilePath = $this->getSharedStringsFilePath();
86	90	if ($xmlReader->open($sharedStringsFilePath) === false) {
87		throw new IOException('Could not open "' . self::SHARED_STRINGS_XML_FILE_PATH . '".');
88		}
89
90		try {
91	90	$sharedStringsUniqueCount = $this->getSharedStringsUniqueCount($xmlReader);
92	87	$this->cachingStrategy = $this->getBestSharedStringsCachingStrategy($sharedStringsUniqueCount);
93
94	87	$xmlReader->readUntilNodeFound('si');
95
96	87	while ($xmlReader->name === 'si') {
97	57	$node = $this->getSimpleXmlElementNodeFromXMLReader($xmlReader);
98	54	$node->registerXPathNamespace('ns', self::MAIN_NAMESPACE_FOR_SHARED_STRINGS_XML);
99
100		// removes nodes that should not be read, like the pronunciation of the Kanji characters
101	54	$cleanNode = $this->removeSuperfluousTextNodes($node);
102
103		// find all text nodes "t"; there can be multiple if the cell contains formatting
104	54	$textNodes = $cleanNode->xpath('//ns:t');
105
106	54	$textValue = '';
107	54	foreach ($textNodes as $nodeIndex => $textNode) {
108	54	if ($nodeIndex !== 0) {
109		// add a space between each "t" node
110	3	$textValue .= ' ';
111	3	}
112
113	54	if ($this->shouldPreserveWhitespace($textNode)) {
114	6	$textValue .= $textNode->__toString();
115	6	} else {
116	48	$textValue .= trim($textNode->__toString());
117		}
118	54	}
119
120	54	$unescapedTextValue = $escaper->unescape($textValue);
121	54	$this->cachingStrategy->addStringForIndex($unescapedTextValue, $sharedStringIndex);
122
123	54	$sharedStringIndex++;
124
125		// jump to the next 'si' tag
126	54	$xmlReader->next('si');
127	54	}
128
129	90	} catch (XMLProcessingException $exception) {
130	3	throw new IOException("The sharedStrings.xml file is invalid and cannot be read. [{$exception->getMessage()}]");
131		}
132
133	84	$this->cachingStrategy->closeCache();
134
135	84	$xmlReader->close();
136	84	}
137
138		/**
139		* @return string The path to the shared strings XML file
140		*/
141	90	protected function getSharedStringsFilePath()
142		{
143	90	return 'zip://' . $this->filePath . '#' . self::SHARED_STRINGS_XML_FILE_PATH;
144		}
145
146		/**
147		* Returns the shared strings unique count, as specified in <sst> tag.
148		*
149		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader instance
150		* @return int Number of unique shared strings in the sharedStrings.xml file
151		* @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml is invalid and can't be read
152		*/
153	90	protected function getSharedStringsUniqueCount($xmlReader)
154		{
155	90	$xmlReader->next('sst');
156
157		// Iterate over the "sst" elements to get the actual "sst ELEMENT" (skips any DOCTYPE)
158	87	while ($xmlReader->name === 'sst' && $xmlReader->nodeType !== XMLReader::ELEMENT) {
159	3	$xmlReader->read();
160	3	}
161
162	87	return intval($xmlReader->getAttribute('uniqueCount'));
163		}
164
165		/**
166		* Returns the best shared strings caching strategy.
167		*
168		* @param int $sharedStringsUniqueCount
169		* @return CachingStrategyInterface
170		*/
171	87	protected function getBestSharedStringsCachingStrategy($sharedStringsUniqueCount)
172		{
173	87	return CachingStrategyFactory::getInstance()
174	87	->getBestCachingStrategy($sharedStringsUniqueCount, $this->tempFolder);
175		}
176
177		/**
178		* Returns a SimpleXMLElement node from the current node in the given XMLReader instance.
179		* This is to simplify the parsing of the subtree.
180		*
181		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader
182		* @return \Box\Spout\Reader\Wrapper\SimpleXMLElement
183		* @throws \Box\Spout\Common\Exception\IOException If the current node cannot be read
184		*/
185	57	protected function getSimpleXmlElementNodeFromXMLReader($xmlReader)
186		{
187	57	$node = null;
188		try {
189	57	$node = new SimpleXMLElement($xmlReader->readOuterXml());
190	57	} catch (XMLProcessingException $exception) {
191	3	throw new IOException("The sharedStrings.xml file contains unreadable data [{$exception->getMessage()}].");
192		}
193
194	54	return $node;
195		}
196
197		/**
198		* Removes nodes that should not be read, like the pronunciation of the Kanji characters.
199		* By keeping them, their text content would be added to the read string.
200		*
201		* @param \Box\Spout\Reader\Wrapper\SimpleXMLElement $parentNode Parent node that may contain nodes to remove
202		* @return \Box\Spout\Reader\Wrapper\SimpleXMLElement Cleaned parent node
203		*/
204	54	protected function removeSuperfluousTextNodes($parentNode)
205		{
206		$tagsToRemove = [
207	54	'rPh', // Pronunciation of the text
208	54	'pPr', // Paragraph Properties / Previous Paragraph Properties
209	54	'rPr', // Run Properties for the Paragraph Mark / Previous Run Properties for the Paragraph Mark
210	54	];
211
212	54	foreach ($tagsToRemove as $tagToRemove) {
213	54	$xpath = '//ns:' . $tagToRemove;
214	54	$parentNode->removeNodesMatchingXPath($xpath);
215	54	}
216
217	54	return $parentNode;
218		}
219
220		/**
221		* If the text node has the attribute 'xml:space="preserve"', then preserve whitespace.
222		*
223		* @param \Box\Spout\Reader\Wrapper\SimpleXMLElement $textNode The text node element (<t>) whitespace may be preserved
224		* @return bool Whether whitespace should be preserved
225		*/
226	54	protected function shouldPreserveWhitespace($textNode)
227		{
228	54	$spaceValue = $textNode->getAttribute('space', 'xml');
229	54	return ($spaceValue === 'preserve');
230		}
231
232		/**
233		* Returns the shared string at the given index, using the previously chosen caching strategy.
234		*
235		* @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
236		* @return string The shared string at the given index
237		* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If no shared string found for the given index
238		*/
239	54	public function getStringAtIndex($sharedStringIndex)
240		{
241	54	return $this->cachingStrategy->getStringAtIndex($sharedStringIndex);
242		}
243
244		/**
245		* Destroys the cache, freeing memory and removing any created artifacts
246		*
247		* @return void
248		*/
249	90	public function cleanup()
250		{
251	90	if ($this->cachingStrategy) {
252	78	$this->cachingStrategy->clearCache();
253	78	}
254	90	}
255		}
256

box / spout

Pull Request — master (#229)

getBestSharedStringsCachingStrategy() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like