| 1 |
<?php |
| 2 |
|
| 3 |
namespace Box\Spout\Reader\XLSX\Helper; |
| 4 |
|
| 5 |
use Box\Spout\Reader\Wrapper\XMLReader; |
| 6 |
use Box\Spout\Reader\XLSX\Sheet; |
| 7 |
|
| 8 |
/** |
| 9 |
* Class SheetHelper |
| 10 |
* This class provides helper functions related to XLSX sheets |
| 11 |
* |
| 12 |
* @package Box\Spout\Reader\XLSX\Helper |
| 13 |
*/ |
| 14 |
class SheetHelper |
| 15 |
{ |
| 16 |
/** Paths of XML files relative to the XLSX file root */ |
| 17 |
const WORKBOOK_XML_RELS_FILE_PATH = 'xl/_rels/workbook.xml.rels'; |
| 18 |
const WORKBOOK_XML_FILE_PATH = 'xl/workbook.xml'; |
| 19 |
|
| 20 |
/** Definition of XML node names used to parse data */ |
| 21 |
const XML_NODE_WORKBOOK_VIEW = 'workbookView'; |
| 22 |
const XML_NODE_SHEET = 'sheet'; |
| 23 |
const XML_NODE_SHEETS = 'sheets'; |
| 24 |
const XML_NODE_RELATIONSHIP = 'Relationship'; |
| 25 |
|
| 26 |
/** Definition of XML attributes used to parse data */ |
| 27 |
const XML_ATTRIBUTE_ACTIVE_TAB = 'activeTab'; |
| 28 |
const XML_ATTRIBUTE_R_ID = 'r:id'; |
| 29 |
const XML_ATTRIBUTE_NAME = 'name'; |
| 30 |
const XML_ATTRIBUTE_ID = 'Id'; |
| 31 |
const XML_ATTRIBUTE_TARGET = 'Target'; |
| 32 |
|
| 33 |
/** @var string Path of the XLSX file being read */ |
| 34 |
protected $filePath; |
| 35 |
|
| 36 |
/** @var \Box\Spout\Reader\XLSX\ReaderOptions Reader's current options */ |
| 37 |
protected $options; |
| 38 |
|
| 39 |
/** @var \Box\Spout\Reader\XLSX\Helper\SharedStringsHelper Helper to work with shared strings */ |
| 40 |
protected $sharedStringsHelper; |
| 41 |
|
| 42 |
/** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper Helper to work with global functions */ |
| 43 |
protected $globalFunctionsHelper; |
| 44 |
|
| 45 |
/** |
| 46 |
* @param string $filePath Path of the XLSX file being read |
| 47 |
* @param \Box\Spout\Reader\XLSX\ReaderOptions $options Reader's current options |
| 48 |
* @param \Box\Spout\Reader\XLSX\Helper\SharedStringsHelper Helper to work with shared strings |
| 49 |
* @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper |
| 50 |
*/ |
| 51 |
public function __construct($filePath, $options, $sharedStringsHelper, $globalFunctionsHelper) |
| 52 |
{ |
| 53 |
$this->filePath = $filePath; |
| 54 |
$this->options = $options; |
| 55 |
$this->sharedStringsHelper = $sharedStringsHelper; |
| 56 |
$this->globalFunctionsHelper = $globalFunctionsHelper; |
| 57 |
} |
| 58 |
|
| 59 |
/** |
| 60 |
* Returns the sheets metadata of the file located at the previously given file path. |
| 61 |
* The paths to the sheets' data are read from the [Content_Types].xml file. |
| 62 |
* |
| 63 |
* @return Sheet[] Sheets within the XLSX file |
| 64 |
*/ |
| 65 |
public function getSheets() |
| 66 |
{ |
| 67 |
$sheets = []; |
| 68 |
$sheetIndex = 0; |
| 69 |
$activeSheetIndex = 0; // By default, the first sheet is active |
| 70 |
|
| 71 |
$xmlReader = new XMLReader(); |
| 72 |
if ($xmlReader->openFileInZip($this->filePath, self::WORKBOOK_XML_FILE_PATH)) { |
| 73 |
while ($xmlReader->read()) { |
| 74 |
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_WORKBOOK_VIEW)) { |
| 75 |
// The "workbookView" node is located before "sheet" nodes, ensuring that |
| 76 |
// the active sheet is known before parsing sheets data. |
| 77 |
$activeSheetIndex = (int) $xmlReader->getAttribute(self::XML_ATTRIBUTE_ACTIVE_TAB); |
| 78 |
} else if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_SHEET)) { |
| 79 |
$isSheetActive = ($sheetIndex === $activeSheetIndex); |
| 80 |
$sheets[] = $this->getSheetFromSheetXMLNode($xmlReader, $sheetIndex, $isSheetActive); |
| 81 |
$sheetIndex++; |
| 82 |
} else if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_SHEETS)) { |
| 83 |
// stop reading once all sheets have been read |
| 84 |
break; |
| 85 |
} |
| 86 |
} |
| 87 |
|
| 88 |
$xmlReader->close(); |
| 89 |
} |
| 90 |
|
| 91 |
return $sheets; |
| 92 |
} |
| 93 |
|
| 94 |
/** |
| 95 |
* Returns an instance of a sheet, given the XML node describing the sheet - from "workbook.xml". |
| 96 |
* We can find the XML file path describing the sheet inside "workbook.xml.res", by mapping with the sheet ID |
| 97 |
* ("r:id" in "workbook.xml", "Id" in "workbook.xml.res"). |
| 98 |
* |
| 99 |
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReaderOnSheetNode XML Reader instance, pointing on the node describing the sheet, as defined in "workbook.xml" |
| 100 |
* @param int $sheetIndexZeroBased Index of the sheet, based on order of appearance in the workbook (zero-based) |
| 101 |
* @param bool $isSheetActive Whether this sheet was defined as active |
| 102 |
* @return \Box\Spout\Reader\XLSX\Sheet Sheet instance |
| 103 |
*/ |
| 104 |
protected function getSheetFromSheetXMLNode($xmlReaderOnSheetNode, $sheetIndexZeroBased, $isSheetActive) |
| 105 |
{ |
| 106 |
$sheetId = $xmlReaderOnSheetNode->getAttribute(self::XML_ATTRIBUTE_R_ID); |
| 107 |
$escapedSheetName = $xmlReaderOnSheetNode->getAttribute(self::XML_ATTRIBUTE_NAME); |
| 108 |
|
| 109 |
/** @noinspection PhpUnnecessaryFullyQualifiedNameInspection */ |
| 110 |
$escaper = \Box\Spout\Common\Escaper\XLSX::getInstance(); |
| 111 |
$sheetName = $escaper->unescape($escapedSheetName); |
| 112 |
|
| 113 |
$sheetDataXMLFilePath = $this->getSheetDataXMLFilePathForSheetId($sheetId); |
| 114 |
|
| 115 |
return new Sheet( |
| 116 |
$this->filePath, $sheetDataXMLFilePath, |
| 117 |
$sheetIndexZeroBased, $sheetName, $isSheetActive, |
| 118 |
$this->options, $this->sharedStringsHelper |
| 119 |
); |
| 120 |
} |
| 121 |
|
| 122 |
/** |
| 123 |
* @param string $sheetId The sheet ID, as defined in "workbook.xml" |
| 124 |
* @return string The XML file path describing the sheet inside "workbook.xml.res", for the given sheet ID |
| 125 |
*/ |
| 126 |
protected function getSheetDataXMLFilePathForSheetId($sheetId) |
| 127 |
{ |
| 128 |
$sheetDataXMLFilePath = ''; |
| 129 |
|
| 130 |
// find the file path of the sheet, by looking at the "workbook.xml.res" file |
| 131 |
$xmlReader = new XMLReader(); |
| 132 |
if ($xmlReader->openFileInZip($this->filePath, self::WORKBOOK_XML_RELS_FILE_PATH)) { |
| 133 |
while ($xmlReader->read()) { |
| 134 |
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_RELATIONSHIP)) { |
| 135 |
$relationshipSheetId = $xmlReader->getAttribute(self::XML_ATTRIBUTE_ID); |
| 136 |
|
| 137 |
if ($relationshipSheetId === $sheetId) { |
| 138 |
// In workbook.xml.rels, it is only "worksheets/sheet1.xml" |
| 139 |
// In [Content_Types].xml, the path is "/xl/worksheets/sheet1.xml" |
| 140 |
$sheetDataXMLFilePath = $xmlReader->getAttribute(self::XML_ATTRIBUTE_TARGET); |
| 141 |
|
| 142 |
// sometimes, the sheet data file path already contains "/xl/"... |
| 143 |
if (strpos($sheetDataXMLFilePath, '/xl/') !== 0) { |
| 144 |
$sheetDataXMLFilePath = '/xl/' . $sheetDataXMLFilePath; |
| 145 |
break; |
| 146 |
} |
| 147 |
} |
| 148 |
} |
| 149 |
} |
| 150 |
|
| 151 |
$xmlReader->close(); |
| 152 |
} |
| 153 |
|
| 154 |
return $sheetDataXMLFilePath; |
| 155 |
} |
| 156 |
} |
| 157 |
|