SharedStringsHelper::shouldExtractTextNodeValue() - Code Metrics - Inspection of "Fix shared strings XML Entities auto decode (#411)" - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 1eb01a...048105 )

by Adrien

created 2017-04-28 00:27 UTC

SharedStringsHelper::shouldExtractTextNodeValue() A

↳ Parent: SharedStringsHelper

Complexity

Conditions	2
Paths	2

Size

Total Lines	5
Code Lines	3

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	3
CRAP Score	2

Importance

Changes

Metric	Value
dl	0
loc	5
ccs	3
cts	3
cp	1
rs	9.4285
c	0
b	0
f	0
cc	2
eloc	3
nc	2
nop	1
crap	2

<?php

namespace Box\Spout\Reader\XLSX\Helper;

use Box\Spout\Common\Exception\IOException;
use Box\Spout\Reader\Exception\XMLProcessingException;
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';

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

    /**
     * @param string $filePath Path of the XLSX file being read
     * @param string|null|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;

        $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(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();
    }

    /**
     * @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 $xmlReader XMLReader instance
     * @return int|null Number of unique shared strings in the sharedStrings.xml file
     * @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml is invalid and can't be read
     */
    protected function getSharedStringsUniqueCount($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 CachingStrategyFactory::getInstance()
                ->getBestCachingStrategy($sharedStringsUniqueCount, $this->tempFolder);
    }

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

box / spout

Push — master ( 1eb01a...048105 )

SharedStringsHelper::shouldExtractTextNodeValue() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like