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

Add support for more cell types

Browse Source 
Added proper support for booleans, dates, numbers, errors.
Added unescaping of the read string.
Fixed a bug when cells did not have any values => now returns empty string.
This commit is contained in:
Adrien Loison 2015-06-03 10:54:53 -07:00
parent d617a50b85
commit 8bac924d48
3 changed files with 78 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

77
src/Spout/Reader/XLSX.php
View File

@ -177,6 +177,7 @@ class XLSX extends AbstractReader
throw new BadUsageException('You must call nextSheet() before calling hasNextRow() or nextRow()');
}
$escaper = new \Box\Spout\Common\Escaper\XLSX();
$isInsideRowTag = false;
$rowData = [];
@ -188,6 +189,7 @@ class XLSX extends AbstractReader
$lastCellIndex = $matches[1];
$this->numberOfColumns = CellHelper::getColumnIndexFromCellIndex($lastCellIndex) + 1;
}
} else if ($this->xmlReader->nodeType == \XMLReader::ELEMENT && $this->xmlReader->name === 'row') {
// Start of the row description
$isInsideRowTag = true;
@ -200,32 +202,15 @@ class XLSX extends AbstractReader
$numberOfColumnsForRow = intval($numberOfColumnsForRow);
}
$rowData = ($numberOfColumnsForRow !== 0) ? array_fill(0, $numberOfColumnsForRow, '') : [];
} else if ($isInsideRowTag && $this->xmlReader->nodeType == \XMLReader::ELEMENT && $this->xmlReader->name === 'c') {
// Start of a cell description
$currentCellIndex = $this->xmlReader->getAttribute('r');
$currentColumnIndex = CellHelper::getColumnIndexFromCellIndex($currentCellIndex);
$node = $this->xmlReader->expand();
$rowData[$currentColumnIndex] = $this->getCellValue($node, $escaper);
$hasInlineString = ($this->xmlReader->getAttribute('t') === 'inlineStr');
$hasSharedString = ($this->xmlReader->getAttribute('t') === 's');
if ($hasInlineString) {
// inline strings are formatted this way:
// <c r="A1" t="inlineStr"><is><t>[INLINE_STRING]</t></is></c>
$tNode = $node->getElementsByTagName('t')->item(0);
$rowData[$currentColumnIndex] = trim($tNode->nodeValue);
} else if ($hasSharedString) {
// shared strings are formatted this way:
// <c r="A1" t="s"><v>[SHARED_STRING_INDEX]</v></c>
$vNode = $node->getElementsByTagName('v')->item(0);
$sharedStringIndex = intval($vNode->nodeValue);
$rowData[$currentColumnIndex] = $this->sharedStringsHelper->getStringAtIndex($sharedStringIndex);
} else {
// other values are formatted this way:
// <c r="A1"><v>[VALUE]</v></c>
$vNode = $node->getElementsByTagName('v')->item(0);
$rowData[$currentColumnIndex] = intval($vNode->nodeValue);
}
} else if ($this->xmlReader->nodeType == \XMLReader::END_ELEMENT && $this->xmlReader->name === 'row') {
// End of the row description
// If needed, we fill the empty cells
@ -238,6 +223,58 @@ class XLSX extends AbstractReader
return ($rowData !== []) ? $rowData : null;
}
/**
* Returns the (unescaped) cell value associated to the given XML node.
*
* @param \DOMNode $node
* @param \Box\Spout\Common\Escaper\XLSX $escaper
* @return string|int|float|bool|null The value associated with the cell (null when the cell has an error)
*/
protected function getCellValue($node, $escaper)
{
$cellValue = '';
// Default cell type is "n"
$cellType = $node->getAttribute('t') ?: 'n';
if ($cellType === 'inlineStr') {
// inline strings are formatted this way:
// <c r="A1" t="inlineStr"><is><t>[INLINE_STRING]</t></is></c>
$tNode = $node->getElementsByTagName('t')->item(0);
$escapedCellValue = trim($tNode->nodeValue);
$cellValue = $escaper->unescape($escapedCellValue);
} else {
// all other cell types should have a "v" tag containing the value.
// if not, the returned value should be empty string.
$vNode = $node->getElementsByTagName('v')->item(0);
if ($vNode !== null) {
if ($cellType === 's') {
// shared strings are formatted this way:
// <c r="A1" t="s"><v>[SHARED_STRING_INDEX]</v></c>
$sharedStringIndex = intval($vNode->nodeValue);
$escapedCellValue = $this->sharedStringsHelper->getStringAtIndex($sharedStringIndex);
$cellValue = $escaper->unescape($escapedCellValue);
} else if ($cellType === 'b') {
// !! is similar to boolval()
$cellValue = !!$vNode->nodeValue;
} else if ($cellType === 'n') {
$nodeValue = $vNode->nodeValue;
$cellValue = is_int($nodeValue) ? intval($nodeValue) : floatval($nodeValue);
} else if ($cellType === 'd') {
$cellValue = new \DateTime($vNode->nodeValue);
} else if ($cellType === 'e') {
$cellValue = null;
} else if ($cellType === 'str') {
$escapedCellValue = trim($vNode->nodeValue);
$cellValue = $escaper->unescape($escapedCellValue);
}
}
}
return $cellValue;
}
/**
* Closes the reader. To be used after reading the file.
*

21
tests/Spout/Reader/XLSXTest.php
View File

@ -114,6 +114,27 @@ class XLSXTest extends \PHPUnit_Framework_TestCase
$this->assertEquals($expectedRows, $allRows);
}
/**
* @return void
*/
public function testReadShouldSupportAllCellTypes()
{
$allRows = $this->getAllRowsForFile('sheet_with_all_cell_types.xlsx');
$expectedRows = [
[
's1--A1', 's1--A2',
false, true,
\DateTime::createFromFormat('Y-m-d H:i:s', '2015-06-03 13:21:58'),
\DateTime::createFromFormat('Y-m-d H:i:s', '2015-06-01 00:00:00'),
10, 10.43,
null,
],
['', '', '', '', '', '', '', '', ''],
];
$this->assertEquals($expectedRows, $allRows);
}
/**
* @return void
*/

BIN
tests/resources/xlsx/sheet_with_all_cell_types.xlsx Normal file
View File

Binary file not shown.