mcndave/spout
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Refactor SharedStringsHelper::extractSharedStrings

Browse Source
This commit is contained in:
Adrien Loison 2016-10-17 23:54:45 -07:00
parent 2fa01cd838
commit 441c33100f
1 changed files with 51 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

79
src/Spout/Reader/XLSX/Helper/SharedStringsHelper.php
View File

@ -94,44 +94,19 @@ class SharedStringsHelper
$xmlReader->readUntilNodeFound('si');
while ($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);
$this->processSharedStringsItem($xmlReader, $sharedStringIndex, $escaper);
$sharedStringIndex++;
// jump to the next 'si' tag
$xmlReader->next('si');
}
$this->cachingStrategy->closeCache();
} catch (XMLProcessingException $exception) {
throw new IOException("The sharedStrings.xml file is invalid and cannot be read. [{$exception->getMessage()}]");
}
$this->cachingStrategy->closeCache();
$xmlReader->close();
}
@ -182,6 +157,31 @@ class SharedStringsHelper
->getBestCachingStrategy($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
* @param int $sharedStringIndex Index of the processed shared strings item
* @param \Box\Spout\Common\Escaper\XLSX $escaper Helper to escape values
* @return void
*/
protected function processSharedStringsItem($xmlReader, $sharedStringIndex, $escaper)
{
$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 = $this->extractTextValueForNodes($textNodes);
$unescapedTextValue = $escaper->unescape($textValue);
$this->cachingStrategy->addStringForIndex($unescapedTextValue, $sharedStringIndex);
}
/**
* Returns a SimpleXMLElement node from the current node in the given XMLReader instance.
* This is to simplify the parsing of the subtree.
@ -225,6 +225,29 @@ class SharedStringsHelper
return $parentNode;
}
/**
* @param array $textNodes Text XML nodes ("<t>")
* @return string The value associated with the given text node(s)
*/
protected function extractTextValueForNodes($textNodes)
{
$textValue = '';
foreach ($textNodes as $nodeIndex => $textNode) {
if ($nodeIndex !== 0) {
// add a space between each "t" node
$textValue .= ' ';
}
$textNodeAsString = $textNode->__toString();
$shouldPreserveWhitespace = $this->shouldPreserveWhitespace($textNode);
$textValue .= ($shouldPreserveWhitespace) ? $textNodeAsString : trim($textNodeAsString);
}
return $textValue;
}
/**
* If the text node has the attribute 'xml:space="preserve"', then preserve whitespace.
*