fluentform / app / Services / Spout / Reader / XLSX / Helper / SharedStringsHelper.php

SharedStringsHelper.php in Fluent Forms – Customizable Contact Forms, Survey, Quiz, & Conversational Form Builder 6.2.5, at app/Services/Spout/Reader/XLSX/Helper/SharedStringsHelper.php

235 lines 8.7 KB

Raw Download Zip

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 $tempFolder Temporary folder where the temporary files to store shared strings will be stored
49	*/
50	public function __construct($filePath, $tempFolder = null)
51	{
52	$this->filePath = $filePath;
53	$this->tempFolder = $tempFolder;
54	}
55
56	/**
57	* Returns whether the XLSX file contains a shared strings XML file
58	*
59	* @return bool
60	*/
61	public function hasSharedStrings()
62	{
63	$hasSharedStrings = false;
64	$zip = new \ZipArchive();
65
66	if ($zip->open($this->filePath) === true) {
67	$hasSharedStrings = ($zip->locateName(self::SHARED_STRINGS_XML_FILE_PATH) !== false);
68	$zip->close();
69	}
70
71	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	*
84	* @return void
85	* @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml can't be read
86	*/
87	public function extractSharedStrings()
88	{
89	$xmlReader = new XMLReader();
90	$sharedStringIndex = 0;
91
92	if ($xmlReader->openFileInZip($this->filePath, self::SHARED_STRINGS_XML_FILE_PATH) === false) {
93	throw new IOException('Could not open "' . self::SHARED_STRINGS_XML_FILE_PATH . '".');
94	}
95
96	try {
97	$sharedStringsUniqueCount = $this->getSharedStringsUniqueCount($xmlReader);
98	$this->cachingStrategy = $this->getBestSharedStringsCachingStrategy($sharedStringsUniqueCount);
99
100	$xmlReader->readUntilNodeFound(self::XML_NODE_SI);
101
102	while ($xmlReader->getCurrentNodeName() === self::XML_NODE_SI) {
103	$this->processSharedStringsItem($xmlReader, $sharedStringIndex);
104	$sharedStringIndex++;
105
106	// jump to the next '<si>' tag
107	$xmlReader->next(self::XML_NODE_SI);
108	}
109
110	$this->cachingStrategy->closeCache();
111
112	} catch (XMLProcessingException $exception) {
113	throw new IOException("The sharedStrings.xml file is invalid and cannot be read. [{$exception->getMessage()}]");
114	}
115
116	$xmlReader->close();
117	}
118
119	/**
120	* Returns the shared strings unique count, as specified in <sst> tag.
121	*
122	* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader instance
123	* @return int\|null Number of unique shared strings in the sharedStrings.xml file
124	* @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml is invalid and can't be read
125	*/
126	protected function getSharedStringsUniqueCount($xmlReader)
127	{
128	$xmlReader->next(self::XML_NODE_SST);
129
130	// Iterate over the "sst" elements to get the actual "sst ELEMENT" (skips any DOCTYPE)
131	while ($xmlReader->getCurrentNodeName() === self::XML_NODE_SST && $xmlReader->nodeType !== XMLReader::ELEMENT) {
132	$xmlReader->read();
133	}
134
135	$uniqueCount = $xmlReader->getAttribute(self::XML_ATTRIBUTE_UNIQUE_COUNT);
136
137	// some software do not add the "uniqueCount" attribute but only use the "count" one
138	// @see https://github.com/box/spout/issues/254
139	if ($uniqueCount === null) {
140	$uniqueCount = $xmlReader->getAttribute(self::XML_ATTRIBUTE_COUNT);
141	}
142
143	return ($uniqueCount !== null) ? intval($uniqueCount) : null;
144	}
145
146	/**
147	* Returns the best shared strings caching strategy.
148	*
149	* @param int\|null $sharedStringsUniqueCount Number of unique shared strings (NULL if unknown)
150	* @return CachingStrategyInterface
151	*/
152	protected function getBestSharedStringsCachingStrategy($sharedStringsUniqueCount)
153	{
154	return CachingStrategyFactory::getInstance()
155	->getBestCachingStrategy($sharedStringsUniqueCount, $this->tempFolder);
156	}
157
158	/**
159	* Processes the shared strings item XML node which the given XML reader is positioned on.
160	*
161	* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XML Reader positioned on a "<si>" node
162	* @param int $sharedStringIndex Index of the processed shared strings item
163	* @return void
164	*/
165	protected function processSharedStringsItem($xmlReader, $sharedStringIndex)
166	{
167	$sharedStringValue = '';
168
169	// NOTE: expand() will automatically decode all XML entities of the child nodes
170	$siNode = $xmlReader->expand();
171	$textNodes = $siNode->getElementsByTagName(self::XML_NODE_T);
172
173	foreach ($textNodes as $textNode) {
174	if ($this->shouldExtractTextNodeValue($textNode)) {
175	$textNodeValue = $textNode->nodeValue;
176	$shouldPreserveWhitespace = $this->shouldPreserveWhitespace($textNode);
177
178	$sharedStringValue .= ($shouldPreserveWhitespace) ? $textNodeValue : trim($textNodeValue);
179	}
180	}
181
182	$this->cachingStrategy->addStringForIndex($sharedStringValue, $sharedStringIndex);
183	}
184
185	/**
186	* Not all text nodes' values must be extracted.
187	* Some text nodes are part of a node describing the pronunciation for instance.
188	* We'll only consider the nodes whose parents are "<si>" or "<r>".
189	*
190	* @param \DOMElement $textNode Text node to check
191	* @return bool Whether the given text node's value must be extracted
192	*/
193	protected function shouldExtractTextNodeValue($textNode)
194	{
195	$parentTagName = $textNode->parentNode->localName;
196	return ($parentTagName === self::XML_NODE_SI \|\| $parentTagName === self::XML_NODE_R);
197	}
198
199	/**
200	* If the text node has the attribute 'xml:space="preserve"', then preserve whitespace.
201	*
202	* @param \DOMElement $textNode The text node element (<t>) whose whitespace may be preserved
203	* @return bool Whether whitespace should be preserved
204	*/
205	protected function shouldPreserveWhitespace($textNode)
206	{
207	$spaceValue = $textNode->getAttribute(self::XML_ATTRIBUTE_XML_SPACE);
208	return ($spaceValue === self::XML_ATTRIBUTE_VALUE_PRESERVE);
209	}
210
211	/**
212	* Returns the shared string at the given index, using the previously chosen caching strategy.
213	*
214	* @param int $sharedStringIndex Index of the shared string in the sharedStrings.xml file
215	* @return string The shared string at the given index
216	* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If no shared string found for the given index
217	*/
218	public function getStringAtIndex($sharedStringIndex)
219	{
220	return $this->cachingStrategy->getStringAtIndex($sharedStringIndex);
221	}
222
223	/**
224	* Destroys the cache, freeing memory and removing any created artifacts
225	*
226	* @return void
227	*/
228	public function cleanup()
229	{
230	if ($this->cachingStrategy) {
231	$this->cachingStrategy->clearCache();
232	}
233	}
234	}
235