| 1 |
<?php |
| 2 |
|
| 3 |
namespace Box\Spout\Reader\XLSX\Helper; |
| 4 |
|
| 5 |
use Box\Spout\Reader\Wrapper\XMLReader; |
| 6 |
|
| 7 |
/** |
| 8 |
* Class StyleHelper |
| 9 |
* This class provides helper functions related to XLSX styles |
| 10 |
* |
| 11 |
* @package Box\Spout\Reader\XLSX\Helper |
| 12 |
*/ |
| 13 |
class StyleHelper |
| 14 |
{ |
| 15 |
/** Paths of XML files relative to the XLSX file root */ |
| 16 |
const STYLES_XML_FILE_PATH = 'xl/styles.xml'; |
| 17 |
|
| 18 |
/** Nodes used to find relevant information in the styles XML file */ |
| 19 |
const XML_NODE_NUM_FMTS = 'numFmts'; |
| 20 |
const XML_NODE_NUM_FMT = 'numFmt'; |
| 21 |
const XML_NODE_CELL_XFS = 'cellXfs'; |
| 22 |
const XML_NODE_XF = 'xf'; |
| 23 |
|
| 24 |
/** Attributes used to find relevant information in the styles XML file */ |
| 25 |
const XML_ATTRIBUTE_NUM_FMT_ID = 'numFmtId'; |
| 26 |
const XML_ATTRIBUTE_FORMAT_CODE = 'formatCode'; |
| 27 |
const XML_ATTRIBUTE_APPLY_NUMBER_FORMAT = 'applyNumberFormat'; |
| 28 |
|
| 29 |
/** By convention, default style ID is 0 */ |
| 30 |
const DEFAULT_STYLE_ID = 0; |
| 31 |
|
| 32 |
const NUMBER_FORMAT_GENERAL = 'General'; |
| 33 |
|
| 34 |
/** |
| 35 |
* @see https://msdn.microsoft.com/en-us/library/ff529597(v=office.12).aspx |
| 36 |
* @var array Mapping between built-in numFmtId and the associated format - for dates only |
| 37 |
*/ |
| 38 |
protected static $builtinNumFmtIdToNumFormatMapping = [ |
| 39 |
14 => 'm/d/yyyy', // @NOTE: ECMA spec is 'mm-dd-yy' |
| 40 |
15 => 'd-mmm-yy', |
| 41 |
16 => 'd-mmm', |
| 42 |
17 => 'mmm-yy', |
| 43 |
18 => 'h:mm AM/PM', |
| 44 |
19 => 'h:mm:ss AM/PM', |
| 45 |
20 => 'h:mm', |
| 46 |
21 => 'h:mm:ss', |
| 47 |
22 => 'm/d/yyyy h:mm', // @NOTE: ECMA spec is 'm/d/yy h:mm', |
| 48 |
45 => 'mm:ss', |
| 49 |
46 => '[h]:mm:ss', |
| 50 |
47 => 'mm:ss.0', // @NOTE: ECMA spec is 'mmss.0', |
| 51 |
]; |
| 52 |
|
| 53 |
/** @var string Path of the XLSX file being read */ |
| 54 |
protected $filePath; |
| 55 |
|
| 56 |
/** @var array Array containing the IDs of built-in number formats indicating a date */ |
| 57 |
protected $builtinNumFmtIdIndicatingDates; |
| 58 |
|
| 59 |
/** @var array Array containing a mapping NUM_FMT_ID => FORMAT_CODE */ |
| 60 |
protected $customNumberFormats; |
| 61 |
|
| 62 |
/** @var array Array containing a mapping STYLE_ID => [STYLE_ATTRIBUTES] */ |
| 63 |
protected $stylesAttributes; |
| 64 |
|
| 65 |
/** @var array Cache containing a mapping NUM_FMT_ID => IS_DATE_FORMAT. Used to avoid lots of recalculations */ |
| 66 |
protected $numFmtIdToIsDateFormatCache = []; |
| 67 |
|
| 68 |
/** |
| 69 |
* @param string $filePath Path of the XLSX file being read |
| 70 |
*/ |
| 71 |
public function __construct($filePath) |
| 72 |
{ |
| 73 |
$this->filePath = $filePath; |
| 74 |
$this->builtinNumFmtIdIndicatingDates = array_keys(self::$builtinNumFmtIdToNumFormatMapping); |
| 75 |
} |
| 76 |
|
| 77 |
/** |
| 78 |
* Returns whether the style with the given ID should consider |
| 79 |
* numeric values as timestamps and format the cell as a date. |
| 80 |
* |
| 81 |
* @param int $styleId Zero-based style ID |
| 82 |
* @return bool Whether the cell with the given cell should display a date instead of a numeric value |
| 83 |
*/ |
| 84 |
public function shouldFormatNumericValueAsDate($styleId) |
| 85 |
{ |
| 86 |
$stylesAttributes = $this->getStylesAttributes(); |
| 87 |
|
| 88 |
// Default style (0) does not format numeric values as timestamps. Only custom styles do. |
| 89 |
// Also if the style ID does not exist in the styles.xml file, format as numeric value. |
| 90 |
// Using isset here because it is way faster than array_key_exists... |
| 91 |
if ($styleId === self::DEFAULT_STYLE_ID || !isset($stylesAttributes[$styleId])) { |
| 92 |
return false; |
| 93 |
} |
| 94 |
|
| 95 |
$styleAttributes = $stylesAttributes[$styleId]; |
| 96 |
|
| 97 |
return $this->doesStyleIndicateDate($styleAttributes); |
| 98 |
} |
| 99 |
|
| 100 |
/** |
| 101 |
* Reads the styles.xml file and extract the relevant information from the file. |
| 102 |
* |
| 103 |
* @return void |
| 104 |
*/ |
| 105 |
protected function extractRelevantInfo() |
| 106 |
{ |
| 107 |
$this->customNumberFormats = []; |
| 108 |
$this->stylesAttributes = []; |
| 109 |
|
| 110 |
$xmlReader = new XMLReader(); |
| 111 |
|
| 112 |
if ($xmlReader->openFileInZip($this->filePath, self::STYLES_XML_FILE_PATH)) { |
| 113 |
while ($xmlReader->read()) { |
| 114 |
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_NUM_FMTS)) { |
| 115 |
$this->extractNumberFormats($xmlReader); |
| 116 |
|
| 117 |
} else if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_CELL_XFS)) { |
| 118 |
$this->extractStyleAttributes($xmlReader); |
| 119 |
} |
| 120 |
} |
| 121 |
|
| 122 |
$xmlReader->close(); |
| 123 |
} |
| 124 |
} |
| 125 |
|
| 126 |
/** |
| 127 |
* Extracts number formats from the "numFmt" nodes. |
| 128 |
* For simplicity, the styles attributes are kept in memory. This is possible thanks |
| 129 |
* to the reuse of formats. So 1 million cells should not use 1 million formats. |
| 130 |
* |
| 131 |
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XML Reader positioned on the "numFmts" node |
| 132 |
* @return void |
| 133 |
*/ |
| 134 |
protected function extractNumberFormats($xmlReader) |
| 135 |
{ |
| 136 |
while ($xmlReader->read()) { |
| 137 |
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_NUM_FMT)) { |
| 138 |
$numFmtId = intval($xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_FMT_ID)); |
| 139 |
$formatCode = $xmlReader->getAttribute(self::XML_ATTRIBUTE_FORMAT_CODE); |
| 140 |
$this->customNumberFormats[$numFmtId] = $formatCode; |
| 141 |
} else if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_NUM_FMTS)) { |
| 142 |
// Once done reading "numFmts" node's children |
| 143 |
break; |
| 144 |
} |
| 145 |
} |
| 146 |
} |
| 147 |
|
| 148 |
/** |
| 149 |
* Extracts style attributes from the "xf" nodes, inside the "cellXfs" section. |
| 150 |
* For simplicity, the styles attributes are kept in memory. This is possible thanks |
| 151 |
* to the reuse of styles. So 1 million cells should not use 1 million styles. |
| 152 |
* |
| 153 |
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XML Reader positioned on the "cellXfs" node |
| 154 |
* @return void |
| 155 |
*/ |
| 156 |
protected function extractStyleAttributes($xmlReader) |
| 157 |
{ |
| 158 |
while ($xmlReader->read()) { |
| 159 |
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_XF)) { |
| 160 |
$numFmtId = $xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_FMT_ID); |
| 161 |
$normalizedNumFmtId = ($numFmtId !== null) ? intval($numFmtId) : null; |
| 162 |
|
| 163 |
$applyNumberFormat = $xmlReader->getAttribute(self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT); |
| 164 |
$normalizedApplyNumberFormat = ($applyNumberFormat !== null) ? !!$applyNumberFormat : null; |
| 165 |
|
| 166 |
$this->stylesAttributes[] = [ |
| 167 |
self::XML_ATTRIBUTE_NUM_FMT_ID => $normalizedNumFmtId, |
| 168 |
self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT => $normalizedApplyNumberFormat, |
| 169 |
]; |
| 170 |
} else if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_CELL_XFS)) { |
| 171 |
// Once done reading "cellXfs" node's children |
| 172 |
break; |
| 173 |
} |
| 174 |
} |
| 175 |
} |
| 176 |
|
| 177 |
/** |
| 178 |
* @return array The custom number formats |
| 179 |
*/ |
| 180 |
protected function getCustomNumberFormats() |
| 181 |
{ |
| 182 |
if (!isset($this->customNumberFormats)) { |
| 183 |
$this->extractRelevantInfo(); |
| 184 |
} |
| 185 |
|
| 186 |
return $this->customNumberFormats; |
| 187 |
} |
| 188 |
|
| 189 |
/** |
| 190 |
* @return array The styles attributes |
| 191 |
*/ |
| 192 |
protected function getStylesAttributes() |
| 193 |
{ |
| 194 |
if (!isset($this->stylesAttributes)) { |
| 195 |
$this->extractRelevantInfo(); |
| 196 |
} |
| 197 |
|
| 198 |
return $this->stylesAttributes; |
| 199 |
} |
| 200 |
|
| 201 |
/** |
| 202 |
* @param array $styleAttributes Array containing the style attributes (2 keys: "applyNumberFormat" and "numFmtId") |
| 203 |
* @return bool Whether the style with the given attributes indicates that the number is a date |
| 204 |
*/ |
| 205 |
protected function doesStyleIndicateDate($styleAttributes) |
| 206 |
{ |
| 207 |
$applyNumberFormat = $styleAttributes[self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT]; |
| 208 |
$numFmtId = $styleAttributes[self::XML_ATTRIBUTE_NUM_FMT_ID]; |
| 209 |
|
| 210 |
// A style may apply a date format if it has: |
| 211 |
// - "applyNumberFormat" attribute not set to "false" |
| 212 |
// - "numFmtId" attribute set |
| 213 |
// This is a preliminary check, as having "numFmtId" set just means the style should apply a specific number format, |
| 214 |
// but this is not necessarily a date. |
| 215 |
if ($applyNumberFormat === false || $numFmtId === null) { |
| 216 |
return false; |
| 217 |
} |
| 218 |
|
| 219 |
return $this->doesNumFmtIdIndicateDate($numFmtId); |
| 220 |
} |
| 221 |
|
| 222 |
/** |
| 223 |
* Returns whether the number format ID indicates that the number is a date. |
| 224 |
* The result is cached to avoid recomputing the same thing over and over, as |
| 225 |
* "numFmtId" attributes can be shared between multiple styles. |
| 226 |
* |
| 227 |
* @param int $numFmtId |
| 228 |
* @return bool Whether the number format ID indicates that the number is a date |
| 229 |
*/ |
| 230 |
protected function doesNumFmtIdIndicateDate($numFmtId) |
| 231 |
{ |
| 232 |
if (!isset($this->numFmtIdToIsDateFormatCache[$numFmtId])) { |
| 233 |
$formatCode = $this->getFormatCodeForNumFmtId($numFmtId); |
| 234 |
|
| 235 |
$this->numFmtIdToIsDateFormatCache[$numFmtId] = ( |
| 236 |
$this->isNumFmtIdBuiltInDateFormat($numFmtId) || |
| 237 |
$this->isFormatCodeCustomDateFormat($formatCode) |
| 238 |
); |
| 239 |
} |
| 240 |
|
| 241 |
return $this->numFmtIdToIsDateFormatCache[$numFmtId]; |
| 242 |
} |
| 243 |
|
| 244 |
/** |
| 245 |
* @param int $numFmtId |
| 246 |
* @return string|null The custom number format or NULL if none defined for the given numFmtId |
| 247 |
*/ |
| 248 |
protected function getFormatCodeForNumFmtId($numFmtId) |
| 249 |
{ |
| 250 |
$customNumberFormats = $this->getCustomNumberFormats(); |
| 251 |
|
| 252 |
// Using isset here because it is way faster than array_key_exists... |
| 253 |
return (isset($customNumberFormats[$numFmtId])) ? $customNumberFormats[$numFmtId] : null; |
| 254 |
} |
| 255 |
|
| 256 |
/** |
| 257 |
* @param int $numFmtId |
| 258 |
* @return bool Whether the number format ID indicates that the number is a date |
| 259 |
*/ |
| 260 |
protected function isNumFmtIdBuiltInDateFormat($numFmtId) |
| 261 |
{ |
| 262 |
return in_array($numFmtId, $this->builtinNumFmtIdIndicatingDates); |
| 263 |
} |
| 264 |
|
| 265 |
/** |
| 266 |
* @param string|null $formatCode |
| 267 |
* @return bool Whether the given format code indicates that the number is a date |
| 268 |
*/ |
| 269 |
protected function isFormatCodeCustomDateFormat($formatCode) |
| 270 |
{ |
| 271 |
// if no associated format code or if using the default "General" format |
| 272 |
if ($formatCode === null || strcasecmp($formatCode, self::NUMBER_FORMAT_GENERAL) === 0) { |
| 273 |
return false; |
| 274 |
} |
| 275 |
|
| 276 |
return $this->isFormatCodeMatchingDateFormatPattern($formatCode); |
| 277 |
} |
| 278 |
|
| 279 |
/** |
| 280 |
* @param string $formatCode |
| 281 |
* @return bool Whether the given format code matches a date format pattern |
| 282 |
*/ |
| 283 |
protected function isFormatCodeMatchingDateFormatPattern($formatCode) |
| 284 |
{ |
| 285 |
// Remove extra formatting (what's between [ ], the brackets should not be preceded by a "\") |
| 286 |
$pattern = '((?<!\\\)\[.+?(?<!\\\)\])'; |
| 287 |
$formatCode = preg_replace($pattern, '', $formatCode); |
| 288 |
|
| 289 |
// custom date formats contain specific characters to represent the date: |
| 290 |
// e - yy - m - d - h - s |
| 291 |
// and all of their variants (yyyy - mm - dd...) |
| 292 |
$dateFormatCharacters = ['e', 'yy', 'm', 'd', 'h', 's']; |
| 293 |
|
| 294 |
$hasFoundDateFormatCharacter = false; |
| 295 |
foreach ($dateFormatCharacters as $dateFormatCharacter) { |
| 296 |
// character not preceded by "\" (case insensitive) |
| 297 |
$pattern = '/(?<!\\\)' . $dateFormatCharacter . '/i'; |
| 298 |
|
| 299 |
if (preg_match($pattern, $formatCode)) { |
| 300 |
$hasFoundDateFormatCharacter = true; |
| 301 |
break; |
| 302 |
} |
| 303 |
} |
| 304 |
|
| 305 |
return $hasFoundDateFormatCharacter; |
| 306 |
} |
| 307 |
|
| 308 |
/** |
| 309 |
* Returns the format as defined in "styles.xml" of the given style. |
| 310 |
* NOTE: It is assumed that the style DOES have a number format associated to it. |
| 311 |
* |
| 312 |
* @param int $styleId Zero-based style ID |
| 313 |
* @return string The number format code associated with the given style |
| 314 |
*/ |
| 315 |
public function getNumberFormatCode($styleId) |
| 316 |
{ |
| 317 |
$stylesAttributes = $this->getStylesAttributes(); |
| 318 |
$styleAttributes = $stylesAttributes[$styleId]; |
| 319 |
$numFmtId = $styleAttributes[self::XML_ATTRIBUTE_NUM_FMT_ID]; |
| 320 |
|
| 321 |
if ($this->isNumFmtIdBuiltInDateFormat($numFmtId)) { |
| 322 |
$numberFormatCode = self::$builtinNumFmtIdToNumFormatMapping[$numFmtId]; |
| 323 |
} else { |
| 324 |
$customNumberFormats = $this->getCustomNumberFormats(); |
| 325 |
$numberFormatCode = $customNumberFormats[$numFmtId]; |
| 326 |
} |
| 327 |
|
| 328 |
return $numberFormatCode; |
| 329 |
} |
| 330 |
} |
| 331 |
|
