Merge pull request #167 from box/fix_int_float_value_formatters

Fix CellValueFormatter for numeric values

composer.json

@@ -32,7 +32,7 @@
     },
     "extra": {
         "branch-alias": {
-            "dev-master": "2.4.x-dev"
+            "dev-master": "2.5.x-dev"
         }
     }
 

src/Spout/Reader/ODS/Helper/CellValueFormatter.php

@@ -47,7 +47,6 @@ class CellValueFormatter
 
     /**
      * Returns the (unescaped) correctly marshalled, cell value associated to the given XML node.
-     * @TODO Add other types !!
      * @see http://docs.oasis-open.org/office/v1.2/os/OpenDocument-v1.2-os-part1.html#refTable13
      *
      * @param \DOMNode $node
@@ -119,7 +118,8 @@ class CellValueFormatter
     protected function formatFloatCellValue($node)
     {
         $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_VALUE);
-        $cellValue = is_int($nodeValue) ? intval($nodeValue) : floatval($nodeValue);
+        $nodeIntValue = intval($nodeValue);
+        $cellValue = ($nodeIntValue == $nodeValue) ? $nodeIntValue : floatval($nodeValue);
         return $cellValue;
     }
 

src/Spout/Reader/ODS/Sheet.php

@@ -19,7 +19,7 @@ class Sheet implements SheetInterface
     /** @var int ID of the sheet */
     protected $id;
 
-    /** @var int Index of the sheet, based on order of creation (zero-based) */
+    /** @var int Index of the sheet, based on order in the workbook (zero-based) */
     protected $index;
 
     /** @var string Name of the sheet */
@@ -27,7 +27,7 @@ class Sheet implements SheetInterface
 
     /**
      * @param XMLReader $xmlReader XML Reader, positioned on the "<table:table>" element
-     * @param int $sheetIndex Index of the sheet, based on order of creation (zero-based)
+     * @param int $sheetIndex Index of the sheet, based on order in the workbook (zero-based)
      * @param string $sheetName Name of the sheet
      */
     public function __construct($xmlReader, $sheetIndex, $sheetName)
@@ -48,7 +48,7 @@ class Sheet implements SheetInterface
 
     /**
      * @api
-     * @return int Index of the sheet, based on order of creation (zero-based)
+     * @return int Index of the sheet, based on order in the workbook (zero-based)
      */
     public function getIndex()
     {

src/Spout/Reader/XLSX/Helper/CellValueFormatter.php

@@ -169,7 +169,8 @@ class CellValueFormatter
         if ($shouldFormatAsDate) {
             return $this->formatExcelTimestampValue(floatval($nodeValue));
         } else {
-            return is_int($nodeValue) ? intval($nodeValue) : floatval($nodeValue);
+            $nodeIntValue = intval($nodeValue);
+            return ($nodeIntValue == $nodeValue) ? $nodeIntValue : floatval($nodeValue);
         }
     }
 

src/Spout/Reader/XLSX/Helper/SheetHelper.php

@@ -13,9 +13,6 @@ use Box\Spout\Reader\XLSX\Sheet;
  */
 class SheetHelper
 {
-    /** Extension for XML files */
-    const XML_EXTENSION = '.xml';
-
     /** Paths of XML files relative to the XLSX file root */
     const CONTENT_TYPES_XML_FILE_PATH = '[Content_Types].xml';
     const WORKBOOK_XML_RELS_FILE_PATH = 'xl/_rels/workbook.xml.rels';
@@ -79,9 +76,15 @@ class SheetHelper
             $sheetNode = $sheetNodes[$i];
             $sheetDataXMLFilePath = $sheetNode->getAttribute('PartName');
 
-            $sheets[] = $this->getSheetFromXML($sheetDataXMLFilePath, $i);
+            $sheets[] = $this->getSheetFromXML($sheetDataXMLFilePath);
         }
 
+        // make sure the sheets are sorted by index
+        // (as the sheets are not necessarily in this order in the XML file)
+        usort($sheets, function ($sheet1, $sheet2) {
+            return ($sheet1->getIndex() - $sheet2->getIndex());
+        });
+
         return $sheets;
     }
 
@@ -91,60 +94,37 @@ class SheetHelper
      * Then we look at "xl/worbook.xml" to find the sheet entry associated to the found ID.
      * The entry contains the ID and name of the sheet.
      *
-     * If this piece of data can't be found by parsing the different XML files, the ID will default
-     * to the sheet index, based on order in [Content_Types].xml. Similarly, the sheet's name will
-     * default to the data sheet XML file name ("xl/worksheets/sheet2.xml" => "sheet2").
-     *
      * @param string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml
-     * @param int $sheetIndexZeroBased Index of the sheet, based on order in [Content_Types].xml (zero-based)
      * @return \Box\Spout\Reader\XLSX\Sheet Sheet instance
      */
-    protected function getSheetFromXML($sheetDataXMLFilePath, $sheetIndexZeroBased)
+    protected function getSheetFromXML($sheetDataXMLFilePath)
     {
-        $sheetName = $this->getDefaultSheetName($sheetDataXMLFilePath);
-
-        /*
-         * In [Content_Types].xml, the path is "/xl/worksheets/sheet1.xml"
-         * In workbook.xml.rels, it is only "worksheets/sheet1.xml"
-         */
+        // In [Content_Types].xml, the path is "/xl/worksheets/sheet1.xml"
+        // In workbook.xml.rels, it is only "worksheets/sheet1.xml"
         $sheetDataXMLFilePathInWorkbookXMLRels = ltrim($sheetDataXMLFilePath, '/xl/');
 
         // find the node associated to the given file path
         $workbookXMLResElement = $this->getWorkbookXMLRelsAsXMLElement();
         $relationshipNodes = $workbookXMLResElement->xpath('//ns:Relationship[@Target="' . $sheetDataXMLFilePathInWorkbookXMLRels . '"]');
+        $relationshipNode = $relationshipNodes[0];
 
-        if (count($relationshipNodes) === 1) {
-            $relationshipNode = $relationshipNodes[0];
-            $sheetId = $relationshipNode->getAttribute('Id');
+        $relationshipSheetId = $relationshipNode->getAttribute('Id');
 
-            $workbookXMLElement = $this->getWorkbookXMLAsXMLElement();
-            $sheetNodes = $workbookXMLElement->xpath('//ns:sheet[@r:id="' . $sheetId . '"]');
+        $workbookXMLElement = $this->getWorkbookXMLAsXMLElement();
+        $sheetNodes = $workbookXMLElement->xpath('//ns:sheet[@r:id="' . $relationshipSheetId . '"]');
+        $sheetNode = $sheetNodes[0];
 
-            if (count($sheetNodes) === 1) {
-                $sheetNode = $sheetNodes[0];
-                $escapedSheetName = $sheetNode->getAttribute('name');
+        $escapedSheetName = $sheetNode->getAttribute('name');
+        $sheetIdOneBased = $sheetNode->getAttribute('sheetId');
+        $sheetIndexZeroBased = $sheetIdOneBased - 1;
 
-                /** @noinspection PhpUnnecessaryFullyQualifiedNameInspection */
-                $escaper = new \Box\Spout\Common\Escaper\XLSX();
-                $sheetName = $escaper->unescape($escapedSheetName);
-            }
-        }
+        /** @noinspection PhpUnnecessaryFullyQualifiedNameInspection */
+        $escaper = new \Box\Spout\Common\Escaper\XLSX();
+        $sheetName = $escaper->unescape($escapedSheetName);
 
         return new Sheet($this->filePath, $sheetDataXMLFilePath, $this->sharedStringsHelper, $sheetIndexZeroBased, $sheetName);
     }
 
-    /**
-     * Returns the default name of the sheet whose data is located
-     * at the given path.
-     *
-     * @param string $sheetDataXMLFilePath Path of the sheet data XML file
-     * @return string The default sheet name
-     */
-    protected function getDefaultSheetName($sheetDataXMLFilePath)
-    {
-        return $this->globalFunctionsHelper->basename($sheetDataXMLFilePath, self::XML_EXTENSION);
-    }
-
     /**
      * Returns a representation of the workbook.xml.rels file, ready to be parsed.
      * The returned value is cached.

src/Spout/Reader/XLSX/Sheet.php

@@ -15,7 +15,7 @@ class Sheet implements SheetInterface
     /** @var \Box\Spout\Reader\XLSX\RowIterator To iterate over sheet's rows */
     protected $rowIterator;
 
-    /** @var int Index of the sheet, based on order of creation (zero-based) */
+    /** @var int Index of the sheet, based on order in the workbook (zero-based) */
     protected $index;
 
     /** @var string Name of the sheet */
@@ -25,7 +25,7 @@ class Sheet implements SheetInterface
      * @param string $filePath Path of the XLSX file being read
      * @param string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml
      * @param Helper\SharedStringsHelper Helper to work with shared strings
-     * @param int $sheetIndex Index of the sheet, based on order of creation (zero-based)
+     * @param int $sheetIndex Index of the sheet, based on order in the workbook (zero-based)
      * @param string $sheetName Name of the sheet
      */
     public function __construct($filePath, $sheetDataXMLFilePath, $sharedStringsHelper, $sheetIndex, $sheetName)
@@ -46,7 +46,7 @@ class Sheet implements SheetInterface
 
     /**
      * @api
-     * @return int Index of the sheet, based on order of creation (zero-based)
+     * @return int Index of the sheet, based on order in the workbook (zero-based)
      */
     public function getIndex()
     {

src/Spout/Writer/Common/Sheet.php

@@ -24,7 +24,7 @@ class Sheet
     /** @var array Associative array [SHEET_INDEX] => [SHEET_NAME] keeping track of sheets' name to enforce uniqueness */
     protected static $SHEETS_NAME_USED = [];
 
-    /** @var int Index of the sheet, based on order of creation (zero-based) */
+    /** @var int Index of the sheet, based on order in the workbook (zero-based) */
     protected $index;
 
     /** @var string Name of the sheet */
@@ -34,7 +34,7 @@ class Sheet
     protected $stringHelper;
 
     /**
-     * @param int $sheetIndex Index of the sheet, based on order of creation (zero-based)
+     * @param int $sheetIndex Index of the sheet, based on order in the workbook (zero-based)
      */
     public function __construct($sheetIndex)
     {
@@ -45,7 +45,7 @@ class Sheet
 
     /**
      * @api
-     * @return int Index of the sheet, based on order of creation (zero-based)
+     * @return int Index of the sheet, based on order in the workbook (zero-based)
      */
     public function getIndex()
     {

tests/Spout/Reader/XLSX/Helper/CellValueFormatterTest.php

@@ -12,23 +12,26 @@ class CellValueFormatterTest extends \PHPUnit_Framework_TestCase
     /**
      * @return array
      */
-    public function dataProviderForExcelDateTest()
+    public function dataProviderForTestExcelDate()
     {
         return [
-            [ CellValueFormatter::CELL_TYPE_NUMERIC, 42429, '2016-02-29 00:00:00' ],
-            [ CellValueFormatter::CELL_TYPE_NUMERIC, '146098', '2299-12-31 00:00:00' ],
-            [ CellValueFormatter::CELL_TYPE_NUMERIC, -700, null ],
-            [ CellValueFormatter::CELL_TYPE_NUMERIC, 0, null ],
-            [ CellValueFormatter::CELL_TYPE_NUMERIC, 0.5, null ],
-            [ CellValueFormatter::CELL_TYPE_NUMERIC, 1, '1900-01-01 00:00:00' ],
-            [ CellValueFormatter::CELL_TYPE_NUMERIC, 59.999988425926, '1900-02-28 23:59:59' ],
-            [ CellValueFormatter::CELL_TYPE_NUMERIC, 60.458333333333, '1900-02-28 11:00:00' ],
+            [CellValueFormatter::CELL_TYPE_NUMERIC, 42429, '2016-02-29 00:00:00'],
+            [CellValueFormatter::CELL_TYPE_NUMERIC, '146098', '2299-12-31 00:00:00'],
+            [CellValueFormatter::CELL_TYPE_NUMERIC, -700, null],
+            [CellValueFormatter::CELL_TYPE_NUMERIC, 0, null],
+            [CellValueFormatter::CELL_TYPE_NUMERIC, 0.5, null],
+            [CellValueFormatter::CELL_TYPE_NUMERIC, 1, '1900-01-01 00:00:00'],
+            [CellValueFormatter::CELL_TYPE_NUMERIC, 59.999988425926, '1900-02-28 23:59:59'],
+            [CellValueFormatter::CELL_TYPE_NUMERIC, 60.458333333333, '1900-02-28 11:00:00'],
         ];
     }
 
     /**
-     * @dataProvider dataProviderForExcelDateTest
+     * @dataProvider dataProviderForTestExcelDate
      *
+     * @param string $cellType
+     * @param int|float|string $nodeValue
+     * @param string|null $expectedDateAsString
      * @return void
      */
     public function testExcelDate($cellType, $nodeValue, $expectedDateAsString)
@@ -39,7 +42,7 @@ class CellValueFormatterTest extends \PHPUnit_Framework_TestCase
             ->expects($this->atLeastOnce())
             ->method('item')
             ->with(0)
-            ->will($this->returnValue((object)[ 'nodeValue' => $nodeValue ]));
+            ->will($this->returnValue((object)['nodeValue' => $nodeValue]));
 
         $nodeMock = $this->getMockBuilder('DOMElement')->disableOriginalConstructor()->getMock();
 
@@ -47,8 +50,8 @@ class CellValueFormatterTest extends \PHPUnit_Framework_TestCase
             ->expects($this->atLeastOnce())
             ->method('getAttribute')
             ->will($this->returnValueMap([
-                [ CellValueFormatter::XML_ATTRIBUTE_TYPE, $cellType ],
-                [ CellValueFormatter::XML_ATTRIBUTE_STYLE_ID, 123 ],
+                [CellValueFormatter::XML_ATTRIBUTE_TYPE, $cellType],
+                [CellValueFormatter::XML_ATTRIBUTE_STYLE_ID, 123],
             ]));
 
         $nodeMock
@@ -57,7 +60,7 @@ class CellValueFormatterTest extends \PHPUnit_Framework_TestCase
             ->with(CellValueFormatter::XML_NODE_VALUE)
             ->will($this->returnValue($nodeListMock));
 
-        $styleHelperMock = $this->getMockBuilder(__NAMESPACE__ . '\StyleHelper')->disableOriginalConstructor()->getMock();
+        $styleHelperMock = $this->getMockBuilder('Box\Spout\Reader\XLSX\Helper\StyleHelper')->disableOriginalConstructor()->getMock();
 
         $styleHelperMock
             ->expects($this->once())
@@ -65,9 +68,8 @@ class CellValueFormatterTest extends \PHPUnit_Framework_TestCase
             ->with(123)
             ->will($this->returnValue(true));
 
-        $instance = new CellValueFormatter(null, $styleHelperMock);
-
-        $result = $instance->extractAndFormatNodeValue($nodeMock);
+        $formatter = new CellValueFormatter(null, $styleHelperMock);
+        $result = $formatter->extractAndFormatNodeValue($nodeMock);
 
         if ($expectedDateAsString === null) {
             $this->assertNull($result);
@@ -77,4 +79,48 @@ class CellValueFormatterTest extends \PHPUnit_Framework_TestCase
         }
     }
 
+    /**
+     * @return array
+     */
+    public function dataProviderForTestFormatNumericCellValueWithNumbers()
+    {
+        return [
+            [42, 42, 'integer'],
+            [42.5, 42.5, 'double'],
+            [-42, -42, 'integer'],
+            [-42.5, -42.5, 'double'],
+            ['42', 42, 'integer'],
+            ['42.5', 42.5, 'double'],
+            [865640023012945, 865640023012945, 'integer'],
+            ['865640023012945', 865640023012945, 'integer'],
+            [865640023012945.5, 865640023012945.5, 'double'],
+            ['865640023012945.5', 865640023012945.5, 'double'],
+            [PHP_INT_MAX, PHP_INT_MAX, 'integer'],
+            [~PHP_INT_MAX + 1, ~PHP_INT_MAX + 1, 'integer'], // ~PHP_INT_MAX === PHP_INT_MIN, PHP_INT_MIN being PHP7+
+            [PHP_INT_MAX + 1, PHP_INT_MAX + 1, 'double'],
+        ];
+    }
+
+    /**
+     * @dataProvider dataProviderForTestFormatNumericCellValueWithNumbers
+     *
+     * @param int|float|string $value
+     * @param int|float $expectedFormattedValue
+     * @param string $expectedType
+     * @return void
+     */
+    public function testFormatNumericCellValueWithNumbers($value, $expectedFormattedValue, $expectedType)
+    {
+        $styleHelperMock = $this->getMockBuilder('Box\Spout\Reader\XLSX\Helper\StyleHelper')->disableOriginalConstructor()->getMock();
+        $styleHelperMock
+            ->expects($this->once())
+            ->method('shouldFormatNumericValueAsDate')
+            ->will($this->returnValue(false));
+
+        $formatter = new CellValueFormatter(null, $styleHelperMock);
+        $formattedValue = \ReflectionHelper::callMethodOnObject($formatter, 'formatNumericCellValue', $value, 0);
+
+        $this->assertEquals($expectedFormattedValue, $formattedValue);
+        $this->assertEquals($expectedType, gettype($formattedValue));
+    }
 }

tests/Spout/Reader/XLSX/ReaderTest.php

@@ -73,6 +73,28 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
         }
     }
 
+    /**
+     * @return void
+     */
+    public function testReadShouldSupportSheetsDefinitionInRandomOrder()
+    {
+        $allRows = $this->getAllRowsForFile('two_sheets_with_sheets_definition_in_reverse_order.xlsx');
+
+        $expectedRows = [
+            ['s1 - A1', 's1 - B1', 's1 - C1', 's1 - D1', 's1 - E1'],
+            ['s1 - A2', 's1 - B2', 's1 - C2', 's1 - D2', 's1 - E2'],
+            ['s1 - A3', 's1 - B3', 's1 - C3', 's1 - D3', 's1 - E3'],
+            ['s1 - A4', 's1 - B4', 's1 - C4', 's1 - D4', 's1 - E4'],
+            ['s1 - A5', 's1 - B5', 's1 - C5', 's1 - D5', 's1 - E5'],
+            ['s2 - A1', 's2 - B1', 's2 - C1', 's2 - D1', 's2 - E1'],
+            ['s2 - A2', 's2 - B2', 's2 - C2', 's2 - D2', 's2 - E2'],
+            ['s2 - A3', 's2 - B3', 's2 - C3', 's2 - D3', 's2 - E3'],
+            ['s2 - A4', 's2 - B4', 's2 - C4', 's2 - D4', 's2 - E4'],
+            ['s2 - A5', 's2 - B5', 's2 - C5', 's2 - D5', 's2 - E5'],
+        ];
+        $this->assertEquals($expectedRows, $allRows);
+    }
+
     /**
      * @return void
      */