Merge pull request #113 from box/all_ods_cell_types

Support all ODS cell types

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

@@ -12,15 +12,25 @@ class CellValueFormatter
 {
     /** Definition of all possible cell types */
     const CELL_TYPE_STRING = 'string';
-    const CELL_TYPE_BOOLEAN = 'boolean';
     const CELL_TYPE_FLOAT = 'float';
+    const CELL_TYPE_BOOLEAN = 'boolean';
+    const CELL_TYPE_DATE = 'date';
+    const CELL_TYPE_TIME = 'time';
+    const CELL_TYPE_CURRENCY = 'currency';
+    const CELL_TYPE_PERCENTAGE = 'percentage';
+    const CELL_TYPE_VOID = 'void';
 
     /** Definition of XML nodes names used to parse data */
     const XML_NODE_P = 'p';
     const XML_NODE_S = 'text:s';
 
     /** Definition of XML attribute used to parse data */
     const XML_ATTRIBUTE_TYPE = 'office:value-type';
+    const XML_ATTRIBUTE_VALUE = 'office:value';
+    const XML_ATTRIBUTE_BOOLEAN_VALUE = 'office:boolean-value';
+    const XML_ATTRIBUTE_DATE_VALUE = 'office:date-value';
+    const XML_ATTRIBUTE_TIME_VALUE = 'office:time-value';
+    const XML_ATTRIBUTE_CURRENCY = 'office:currency';
     const XML_ATTRIBUTE_C = 'text:c';
 
     /** @var \Box\Spout\Common\Escaper\ODS Used to unescape XML data */
@@ -38,45 +48,36 @@ public function __construct()
     /**
      * 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 $node
-     * @return string|int|float|bool The value associated with the cell (or empty string if cell's type is undefined)
+     * @return string|int|float|bool|\DateTime|\DateInterval|null The value associated with the cell, empty string if cell's type is void/undefined, null on error
      */
     public function extractAndFormatNodeValue($node)
     {
         $cellType = $node->getAttribute(self::XML_ATTRIBUTE_TYPE);
-        $pNodeValue = $this->getFirstPNodeValue($node);
 
         switch ($cellType) {
             case self::CELL_TYPE_STRING:
                 return $this->formatStringCellValue($node);
             case self::CELL_TYPE_FLOAT:
-                return $this->formatFloatCellValue($pNodeValue);
+                return $this->formatFloatCellValue($node);
             case self::CELL_TYPE_BOOLEAN:
-                return $this->formatBooleanCellValue($pNodeValue);
+                return $this->formatBooleanCellValue($node);
+            case self::CELL_TYPE_DATE:
+                return $this->formatDateCellValue($node);
+            case self::CELL_TYPE_TIME:
+                return $this->formatTimeCellValue($node);
+            case self::CELL_TYPE_CURRENCY:
+                return $this->formatCurrencyCellValue($node);
+            case self::CELL_TYPE_PERCENTAGE:
+                return $this->formatPercentageCellValue($node);
+            case self::CELL_TYPE_VOID:
             default:
                 return '';
         }
     }
 
-    /**
-     * Returns the value of the first "<text:p>" node within the given node.
-     *
-     * @param \DOMNode $node
-     * @return string Value for the first "<text:p>" node or empty string if no "<text:p>" found
-     */
-    protected function getFirstPNodeValue($node)
-    {
-        $nodeValue = '';
-        $pNodes = $node->getElementsByTagName(self::XML_NODE_P);
-
-        if ($pNodes->length > 0) {
-            $nodeValue = $pNodes->item(0)->nodeValue;
-        }
-
-        return $nodeValue;
-    }
-
     /**
      * Returns the cell String value.
      *
@@ -110,27 +111,87 @@ protected function formatStringCellValue($node)
     }
 
     /**
-     * Returns the cell Numeric value from string of nodeValue.
+     * Returns the cell Numeric value from the given node.
      *
-     * @param string $pNodeValue
+     * @param \DOMNode $node
      * @return int|float The value associated with the cell
      */
-    protected function formatFloatCellValue($pNodeValue)
+    protected function formatFloatCellValue($node)
     {
-        $cellValue = is_int($pNodeValue) ? intval($pNodeValue) : floatval($pNodeValue);
+        $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_VALUE);
+        $cellValue = is_int($nodeValue) ? intval($nodeValue) : floatval($nodeValue);
         return $cellValue;
     }
 
     /**
-     * Returns the cell Boolean value from a specific node's Value.
+     * Returns the cell Boolean value from the given node.
      *
-     * @param string $pNodeValue
+     * @param \DOMNode $node
      * @return bool The value associated with the cell
      */
-    protected function formatBooleanCellValue($pNodeValue)
+    protected function formatBooleanCellValue($node)
     {
+        $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_BOOLEAN_VALUE);
         // !! is similar to boolval()
-        $cellValue = !!$pNodeValue;
+        $cellValue = !!$nodeValue;
         return $cellValue;
     }
+
+    /**
+     * Returns the cell Date value from the given node.
+     *
+     * @param \DOMNode $node
+     * @return \DateTime|null The value associated with the cell or NULL if invalid date value
+     */
+    protected function formatDateCellValue($node)
+    {
+        try {
+            $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_DATE_VALUE);
+            return new \DateTime($nodeValue);
+        } catch (\Exception $e) {
+            return null;
+        }
+    }
+
+    /**
+     * Returns the cell Time value from the given node.
+     *
+     * @param \DOMNode $node
+     * @return \DateInterval|null The value associated with the cell or NULL if invalid time value
+     */
+    protected function formatTimeCellValue($node)
+    {
+        try {
+            $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_TIME_VALUE);
+            return new \DateInterval($nodeValue);
+        } catch (\Exception $e) {
+            return null;
+        }
+    }
+
+    /**
+     * Returns the cell Currency value from the given node.
+     *
+     * @param \DOMNode $node
+     * @return string The value associated with the cell (e.g. "100 USD" or "9.99 EUR")
+     */
+    protected function formatCurrencyCellValue($node)
+    {
+        $value = $node->getAttribute(self::XML_ATTRIBUTE_VALUE);
+        $currency = $node->getAttribute(self::XML_ATTRIBUTE_CURRENCY);
+
+        return "$value $currency";
+    }
+
+    /**
+     * Returns the cell Percentage value from the given node.
+     *
+     * @param \DOMNode $node
+     * @return int|float The value associated with the cell
+     */
+    protected function formatPercentageCellValue($node)
+    {
+        // percentages are formatted like floats
+        return $this->formatFloatCellValue($node);
+    }
 }

src/Spout/Reader/ODS/RowIterator.php

@@ -174,7 +174,7 @@ protected function getNumColumnsRepeatedForCurrentNode()
      * Returns the (unescaped) correctly marshalled, cell value associated to the given XML node.
      *
      * @param \DOMNode $node
-     * @return string|int|float|bool The value associated with the cell (or empty string if cell's type is undefined)
+     * @return string|int|float|bool|\DateTime|\DateInterval|null The value associated with the cell, empty string if cell's type is void/undefined, null on error
      */
     protected function getCellValue($node)
     {

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

@@ -165,7 +165,7 @@ protected function formatBooleanCellValue($nodeValue)
      * Returns a cell's PHP Date value, associated to the given stored nodeValue.
      *
      * @param string $nodeValue
-     * @return \DateTime|null The value associated with the cell (null when the cell has an error)
+     * @return \DateTime|null The value associated with the cell or NULL if invalid date value
      */
     protected function formatDateCellValue($nodeValue)
     {

src/Spout/Writer/ODS/Internal/Worksheet.php

@@ -194,7 +194,7 @@ protected function getCellContent($cellValue, $styleIndex, $numTimesValueRepeate
 
             $data .= '</table:table-cell>';
         } else if (CellHelper::isBoolean($cellValue)) {
-            $data .= ' office:value-type="boolean" calcext:value-type="boolean" office:value="' . $cellValue . '">';
+            $data .= ' office:value-type="boolean" calcext:value-type="boolean" office:boolean-value="' . $cellValue . '">';
             $data .= '<text:p>' . $cellValue . '</text:p>';
             $data .= '</table:table-cell>';
         } else if (CellHelper::isNumeric($cellValue)) {

tests/Spout/Reader/ODS/ReaderTest.php

@@ -143,13 +143,21 @@ public function testReadWithFilesGeneratedByExternalSoftwares($fileName, $skipLa
      */
     public function testReadShouldSupportAllCellTypes()
     {
+        $utcTz = new \DateTimeZone('UTC');
+        $honoluluTz = new \DateTimeZone('Pacific/Honolulu'); // UTC-10
+
         $allRows = $this->getAllRowsForFile('sheet_with_all_cell_types.ods');
 
         $expectedRows = [
             [
                 'ods--11', 'ods--12',
                 true, false,
                 0, 10.43,
+                new \DateTime('1987-11-29T00:00:00', $utcTz), new \DateTime('1987-11-29T13:37:00', $utcTz),
+                new \DateTime('1987-11-29T13:37:00', $utcTz), new \DateTime('1987-11-29T13:37:00', $honoluluTz),
+                new \DateInterval('PT13H37M00S'),
+                0, 0.42,
+                '42 USD', '9.99 EUR',
                 '',
             ],
         ];
@@ -165,6 +173,15 @@ public function testReadShouldReturnEmptyStringOnUndefinedCellType()
         $this->assertEquals([['ods--11', '', 'ods--13']], $allRows);
     }
 
+    /**
+     * @return void
+     */
+    public function testReadShouldReturnNullOnInvalidDateOrTime()
+    {
+        $allRows = $this->getAllRowsForFile('sheet_with_invalid_date_time.ods');
+        $this->assertEquals([[null, null]], $allRows);
+    }
+
     /**
      * @return void
      */