| 1 |
<?php |
| 2 |
|
| 3 |
namespace OpenSpout\Reader\XLSX\Helper; |
| 4 |
|
| 5 |
use OpenSpout\Reader\Exception\InvalidValueException; |
| 6 |
use OpenSpout\Reader\XLSX\Manager\SharedStringsManager; |
| 7 |
use OpenSpout\Reader\XLSX\Manager\StyleManager; |
| 8 |
|
| 9 |
/** |
| 10 |
* This class provides helper functions to format cell values. |
| 11 |
*/ |
| 12 |
class CellValueFormatter |
| 13 |
{ |
| 14 |
/** Definition of all possible cell types */ |
| 15 |
public const CELL_TYPE_INLINE_STRING = 'inlineStr'; |
| 16 |
public const CELL_TYPE_STR = 'str'; |
| 17 |
public const CELL_TYPE_SHARED_STRING = 's'; |
| 18 |
public const CELL_TYPE_BOOLEAN = 'b'; |
| 19 |
public const CELL_TYPE_NUMERIC = 'n'; |
| 20 |
public const CELL_TYPE_DATE = 'd'; |
| 21 |
public const CELL_TYPE_ERROR = 'e'; |
| 22 |
|
| 23 |
/** Definition of XML nodes names used to parse data */ |
| 24 |
public const XML_NODE_VALUE = 'v'; |
| 25 |
public const XML_NODE_INLINE_STRING_VALUE = 't'; |
| 26 |
|
| 27 |
/** Definition of XML attributes used to parse data */ |
| 28 |
public const XML_ATTRIBUTE_TYPE = 't'; |
| 29 |
public const XML_ATTRIBUTE_STYLE_ID = 's'; |
| 30 |
|
| 31 |
/** Constants used for date formatting */ |
| 32 |
public const NUM_SECONDS_IN_ONE_DAY = 86400; |
| 33 |
|
| 34 |
/** @var SharedStringsManager Manages shared strings */ |
| 35 |
protected $sharedStringsManager; |
| 36 |
|
| 37 |
/** @var StyleManager Manages styles */ |
| 38 |
protected $styleManager; |
| 39 |
|
| 40 |
/** @var bool Whether date/time values should be returned as PHP objects or be formatted as strings */ |
| 41 |
protected $shouldFormatDates; |
| 42 |
|
| 43 |
/** @var bool Whether date/time values should use a calendar starting in 1904 instead of 1900 */ |
| 44 |
protected $shouldUse1904Dates; |
| 45 |
|
| 46 |
/** @var \OpenSpout\Common\Helper\Escaper\XLSX Used to unescape XML data */ |
| 47 |
protected $escaper; |
| 48 |
|
| 49 |
/** |
| 50 |
* @param SharedStringsManager $sharedStringsManager Manages shared strings |
| 51 |
* @param StyleManager $styleManager Manages styles |
| 52 |
* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings |
| 53 |
* @param bool $shouldUse1904Dates Whether date/time values should use a calendar starting in 1904 instead of 1900 |
| 54 |
* @param \OpenSpout\Common\Helper\Escaper\XLSX $escaper Used to unescape XML data |
| 55 |
*/ |
| 56 |
public function __construct($sharedStringsManager, $styleManager, $shouldFormatDates, $shouldUse1904Dates, $escaper) |
| 57 |
{ |
| 58 |
$this->sharedStringsManager = $sharedStringsManager; |
| 59 |
$this->styleManager = $styleManager; |
| 60 |
$this->shouldFormatDates = $shouldFormatDates; |
| 61 |
$this->shouldUse1904Dates = $shouldUse1904Dates; |
| 62 |
$this->escaper = $escaper; |
| 63 |
} |
| 64 |
|
| 65 |
/** |
| 66 |
* Returns the (unescaped) correctly marshalled, cell value associated to the given XML node. |
| 67 |
* |
| 68 |
* @param \DOMElement $node |
| 69 |
* |
| 70 |
* @throws InvalidValueException If the value is not valid |
| 71 |
* |
| 72 |
* @return bool|\DateTime|float|int|string The value associated with the cell |
| 73 |
*/ |
| 74 |
public function extractAndFormatNodeValue($node) |
| 75 |
{ |
| 76 |
// Default cell type is "n" |
| 77 |
$cellType = $node->getAttribute(self::XML_ATTRIBUTE_TYPE) ?: self::CELL_TYPE_NUMERIC; |
| 78 |
$cellStyleId = (int) $node->getAttribute(self::XML_ATTRIBUTE_STYLE_ID); |
| 79 |
$vNodeValue = $this->getVNodeValue($node); |
| 80 |
|
| 81 |
if (('' === $vNodeValue) && (self::CELL_TYPE_INLINE_STRING !== $cellType)) { |
| 82 |
return $vNodeValue; |
| 83 |
} |
| 84 |
|
| 85 |
switch ($cellType) { |
| 86 |
case self::CELL_TYPE_INLINE_STRING: |
| 87 |
return $this->formatInlineStringCellValue($node); |
| 88 |
|
| 89 |
case self::CELL_TYPE_SHARED_STRING: |
| 90 |
return $this->formatSharedStringCellValue($vNodeValue); |
| 91 |
|
| 92 |
case self::CELL_TYPE_STR: |
| 93 |
return $this->formatStrCellValue($vNodeValue); |
| 94 |
|
| 95 |
case self::CELL_TYPE_BOOLEAN: |
| 96 |
return $this->formatBooleanCellValue($vNodeValue); |
| 97 |
|
| 98 |
case self::CELL_TYPE_NUMERIC: |
| 99 |
return $this->formatNumericCellValue($vNodeValue, $cellStyleId); |
| 100 |
|
| 101 |
case self::CELL_TYPE_DATE: |
| 102 |
return $this->formatDateCellValue($vNodeValue); |
| 103 |
|
| 104 |
default: |
| 105 |
throw new InvalidValueException($vNodeValue); |
| 106 |
} |
| 107 |
} |
| 108 |
|
| 109 |
/** |
| 110 |
* Returns the cell's string value from a node's nested value node. |
| 111 |
* |
| 112 |
* @param \DOMElement $node |
| 113 |
* |
| 114 |
* @return string The value associated with the cell |
| 115 |
*/ |
| 116 |
protected function getVNodeValue($node) |
| 117 |
{ |
| 118 |
// for cell types having a "v" tag containing the value. |
| 119 |
// if not, the returned value should be empty string. |
| 120 |
$vNode = $node->getElementsByTagName(self::XML_NODE_VALUE)->item(0); |
| 121 |
|
| 122 |
return (null !== $vNode) ? $vNode->nodeValue : ''; |
| 123 |
} |
| 124 |
|
| 125 |
/** |
| 126 |
* Returns the cell String value where string is inline. |
| 127 |
* |
| 128 |
* @param \DOMElement $node |
| 129 |
* |
| 130 |
* @return string The value associated with the cell |
| 131 |
*/ |
| 132 |
protected function formatInlineStringCellValue($node) |
| 133 |
{ |
| 134 |
// inline strings are formatted this way (they can contain any number of <t> nodes): |
| 135 |
// <c r="A1" t="inlineStr"><is><t>[INLINE_STRING]</t><t>[INLINE_STRING_2]</t></is></c> |
| 136 |
$tNodes = $node->getElementsByTagName(self::XML_NODE_INLINE_STRING_VALUE); |
| 137 |
|
| 138 |
$cellValue = ''; |
| 139 |
for ($i = 0; $i < $tNodes->count(); ++$i) { |
| 140 |
$tNode = $tNodes->item($i); |
| 141 |
$cellValue .= $this->escaper->unescape($tNode->nodeValue); |
| 142 |
} |
| 143 |
|
| 144 |
return $cellValue; |
| 145 |
} |
| 146 |
|
| 147 |
/** |
| 148 |
* Returns the cell String value from shared-strings file using nodeValue index. |
| 149 |
* |
| 150 |
* @param string $nodeValue |
| 151 |
* |
| 152 |
* @return string The value associated with the cell |
| 153 |
*/ |
| 154 |
protected function formatSharedStringCellValue($nodeValue) |
| 155 |
{ |
| 156 |
// shared strings are formatted this way: |
| 157 |
// <c r="A1" t="s"><v>[SHARED_STRING_INDEX]</v></c> |
| 158 |
$sharedStringIndex = (int) $nodeValue; |
| 159 |
$escapedCellValue = $this->sharedStringsManager->getStringAtIndex($sharedStringIndex); |
| 160 |
|
| 161 |
return $this->escaper->unescape($escapedCellValue); |
| 162 |
} |
| 163 |
|
| 164 |
/** |
| 165 |
* Returns the cell String value, where string is stored in value node. |
| 166 |
* |
| 167 |
* @param string $nodeValue |
| 168 |
* |
| 169 |
* @return string The value associated with the cell |
| 170 |
*/ |
| 171 |
protected function formatStrCellValue($nodeValue) |
| 172 |
{ |
| 173 |
$escapedCellValue = trim($nodeValue); |
| 174 |
|
| 175 |
return $this->escaper->unescape($escapedCellValue); |
| 176 |
} |
| 177 |
|
| 178 |
/** |
| 179 |
* Returns the cell Numeric value from string of nodeValue. |
| 180 |
* The value can also represent a timestamp and a DateTime will be returned. |
| 181 |
* |
| 182 |
* @param string $nodeValue |
| 183 |
* @param int $cellStyleId 0 being the default style |
| 184 |
* |
| 185 |
* @return \DateTime|float|int The value associated with the cell |
| 186 |
*/ |
| 187 |
protected function formatNumericCellValue($nodeValue, $cellStyleId) |
| 188 |
{ |
| 189 |
// Numeric values can represent numbers as well as timestamps. |
| 190 |
// We need to look at the style of the cell to determine whether it is one or the other. |
| 191 |
$shouldFormatAsDate = $this->styleManager->shouldFormatNumericValueAsDate($cellStyleId); |
| 192 |
|
| 193 |
if ($shouldFormatAsDate) { |
| 194 |
$cellValue = $this->formatExcelTimestampValue((float) $nodeValue, $cellStyleId); |
| 195 |
} else { |
| 196 |
$nodeIntValue = (int) $nodeValue; |
| 197 |
$nodeFloatValue = (float) $nodeValue; |
| 198 |
$cellValue = ((float) $nodeIntValue === $nodeFloatValue) ? $nodeIntValue : $nodeFloatValue; |
| 199 |
} |
| 200 |
|
| 201 |
return $cellValue; |
| 202 |
} |
| 203 |
|
| 204 |
/** |
| 205 |
* Returns a cell's PHP Date value, associated to the given timestamp. |
| 206 |
* NOTE: The timestamp is a float representing the number of days since the base Excel date: |
| 207 |
* Dec 30th 1899, 1900 or Jan 1st, 1904, depending on the Workbook setting. |
| 208 |
* NOTE: The timestamp can also represent a time, if it is a value between 0 and 1. |
| 209 |
* |
| 210 |
* @see ECMA-376 Part 1 - §18.17.4 |
| 211 |
* |
| 212 |
* @param float $nodeValue |
| 213 |
* @param int $cellStyleId 0 being the default style |
| 214 |
* |
| 215 |
* @throws InvalidValueException If the value is not a valid timestamp |
| 216 |
* |
| 217 |
* @return \DateTime The value associated with the cell |
| 218 |
*/ |
| 219 |
protected function formatExcelTimestampValue($nodeValue, $cellStyleId) |
| 220 |
{ |
| 221 |
if ($this->isValidTimestampValue($nodeValue)) { |
| 222 |
$cellValue = $this->formatExcelTimestampValueAsDateTimeValue($nodeValue, $cellStyleId); |
| 223 |
} else { |
| 224 |
throw new InvalidValueException($nodeValue); |
| 225 |
} |
| 226 |
|
| 227 |
return $cellValue; |
| 228 |
} |
| 229 |
|
| 230 |
/** |
| 231 |
* Returns whether the given timestamp is supported by SpreadsheetML. |
| 232 |
* |
| 233 |
* @see ECMA-376 Part 1 - §18.17.4 - this specifies the timestamp boundaries. |
| 234 |
* |
| 235 |
* @param float $timestampValue |
| 236 |
* |
| 237 |
* @return bool |
| 238 |
*/ |
| 239 |
protected function isValidTimestampValue($timestampValue) |
| 240 |
{ |
| 241 |
// @NOTE: some versions of Excel don't support negative dates (e.g. Excel for Mac 2011) |
| 242 |
return |
| 243 |
$this->shouldUse1904Dates && $timestampValue >= -695055 && $timestampValue <= 2957003.9999884 |
| 244 |
|| !$this->shouldUse1904Dates && $timestampValue >= -693593 && $timestampValue <= 2958465.9999884 |
| 245 |
; |
| 246 |
} |
| 247 |
|
| 248 |
/** |
| 249 |
* Returns a cell's PHP DateTime value, associated to the given timestamp. |
| 250 |
* Only the time value matters. The date part is set to the base Excel date: |
| 251 |
* Dec 30th 1899, 1900 or Jan 1st, 1904, depending on the Workbook setting. |
| 252 |
* |
| 253 |
* @param float $nodeValue |
| 254 |
* @param int $cellStyleId 0 being the default style |
| 255 |
* |
| 256 |
* @return \DateTime|string The value associated with the cell |
| 257 |
*/ |
| 258 |
protected function formatExcelTimestampValueAsDateTimeValue($nodeValue, $cellStyleId) |
| 259 |
{ |
| 260 |
$baseDate = $this->shouldUse1904Dates ? '1904-01-01' : '1899-12-30'; |
| 261 |
|
| 262 |
$daysSinceBaseDate = (int) $nodeValue; |
| 263 |
$timeRemainder = fmod($nodeValue, 1); |
| 264 |
$secondsRemainder = round($timeRemainder * self::NUM_SECONDS_IN_ONE_DAY, 0); |
| 265 |
|
| 266 |
$dateObj = \DateTime::createFromFormat('|Y-m-d', $baseDate); |
| 267 |
$dateObj->modify('+'.$daysSinceBaseDate.'days'); |
| 268 |
$dateObj->modify('+'.$secondsRemainder.'seconds'); |
| 269 |
|
| 270 |
if ($this->shouldFormatDates) { |
| 271 |
$styleNumberFormatCode = $this->styleManager->getNumberFormatCode($cellStyleId); |
| 272 |
$phpDateFormat = DateFormatHelper::toPHPDateFormat($styleNumberFormatCode); |
| 273 |
$cellValue = $dateObj->format($phpDateFormat); |
| 274 |
} else { |
| 275 |
$cellValue = $dateObj; |
| 276 |
} |
| 277 |
|
| 278 |
return $cellValue; |
| 279 |
} |
| 280 |
|
| 281 |
/** |
| 282 |
* Returns the cell Boolean value from a specific node's Value. |
| 283 |
* |
| 284 |
* @param string $nodeValue |
| 285 |
* |
| 286 |
* @return bool The value associated with the cell |
| 287 |
*/ |
| 288 |
protected function formatBooleanCellValue($nodeValue) |
| 289 |
{ |
| 290 |
return (bool) $nodeValue; |
| 291 |
} |
| 292 |
|
| 293 |
/** |
| 294 |
* Returns a cell's PHP Date value, associated to the given stored nodeValue. |
| 295 |
* |
| 296 |
* @see ECMA-376 Part 1 - §18.17.4 |
| 297 |
* |
| 298 |
* @param string $nodeValue ISO 8601 Date string |
| 299 |
* |
| 300 |
* @throws InvalidValueException If the value is not a valid date |
| 301 |
* |
| 302 |
* @return \DateTime|string The value associated with the cell |
| 303 |
*/ |
| 304 |
protected function formatDateCellValue($nodeValue) |
| 305 |
{ |
| 306 |
// Mitigate thrown Exception on invalid date-time format (http://php.net/manual/en/datetime.construct.php) |
| 307 |
try { |
| 308 |
$cellValue = ($this->shouldFormatDates) ? $nodeValue : new \DateTime($nodeValue); |
| 309 |
} catch (\Exception $e) { |
| 310 |
throw new InvalidValueException($nodeValue); |
| 311 |
} |
| 312 |
|
| 313 |
return $cellValue; |
| 314 |
} |
| 315 |
} |
| 316 |
|
