Merge pull request #196 from madflow/trim-inline-strings

Fix #195

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

@@ -118,8 +118,7 @@ class CellValueFormatter
         // inline strings are formatted this way:
         // <c r="A1" t="inlineStr"><is><t>[INLINE_STRING]</t></is></c>
         $tNode = $node->getElementsByTagName(self::XML_NODE_INLINE_STRING_VALUE)->item(0);
-        $escapedCellValue = trim($tNode->nodeValue);
+        $cellValue = $this->escaper->unescape($tNode->nodeValue);
-        $cellValue = $this->escaper->unescape($escapedCellValue);
         return $cellValue;
     }
 

tests/Spout/Reader/CSV/ReaderTest.php

@@ -384,6 +384,23 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
         $this->assertEquals($expectedRows, $allRows, 'There should be 3 rows, with equal length');
     }
 
+    /**
+     * https://github.com/box/spout/issues/195
+     * @return void
+     */
+    public function testReaderShouldNotTrimCellValues()
+    {
+        $allRows = $this->getAllRowsForFile('sheet_with_untrimmed_strings.csv');
+
+        $expectedRows = [
+            ['A'],
+            [' A '],
+            ["\n\tA\n\t"],
+        ];
+
+        $this->assertEquals($expectedRows, $allRows, 'Cell values should not be trimmed');
+    }
+
     /**
      * @param string $fileName
      * @param string|void $fieldDelimiter

tests/Spout/Reader/ODS/ReaderTest.php

@@ -419,6 +419,23 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
         $this->assertEquals($expectedRows, $allRows, 'There should be 3 rows, with equal length');
     }
 
+    /**
+     * https://github.com/box/spout/issues/195
+     * @return void
+     */
+    public function testReaderShouldNotTrimCellValues()
+    {
+        $allRows = $this->getAllRowsForFile('sheet_with_untrimmed_strings.ods');
+
+        $expectedRows = [
+            ['A'],
+            [' A '],
+            ["\n\tA\n\t"],
+        ];
+
+        $this->assertEquals($expectedRows, $allRows, 'Cell values should not be trimmed');
+    }
+
     /**
      * @param string $fileName
      * @return array All the read rows the given file

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

@@ -123,4 +123,46 @@ class CellValueFormatterTest extends \PHPUnit_Framework_TestCase
         $this->assertEquals($expectedFormattedValue, $formattedValue);
         $this->assertEquals($expectedType, gettype($formattedValue));
     }
+
+    /**
+     * @return array
+     */
+    public function dataProviderForTestFormatStringCellValue()
+    {
+        return [
+            ['A', 'A'],
+            [' A ', ' A '],
+            ["\n\tA\n\t", "\n\tA\n\t"],
+            [' ', ' '],
+        ];
+    }
+
+    /**
+     * @dataProvider dataProviderForTestFormatStringCellValue
+     *
+     * @param string $value
+     * @param string $expectedFormattedValue
+     * @return void
+     */
+    public function testFormatInlineStringCellValue($value, $expectedFormattedValue)
+    {
+        $nodeListMock = $this->getMockBuilder('DOMNodeList')->disableOriginalConstructor()->getMock();
+        $nodeListMock
+            ->expects($this->atLeastOnce())
+            ->method('item')
+            ->with(0)
+            ->will($this->returnValue((object)['nodeValue' => $value]));
+
+        $nodeMock = $this->getMockBuilder('DOMElement')->disableOriginalConstructor()->getMock();
+        $nodeMock
+            ->expects($this->atLeastOnce())
+            ->method('getElementsByTagName')
+            ->with(CellValueFormatter::XML_NODE_INLINE_STRING_VALUE)
+            ->will($this->returnValue($nodeListMock));
+
+        $formatter = new CellValueFormatter(null, null);
+        $formattedValue = \ReflectionHelper::callMethodOnObject($formatter, 'formatInlineStringCellValue', $nodeMock);
+
+        $this->assertEquals($expectedFormattedValue, $formattedValue);
+    }
 }

tests/Spout/Reader/XLSX/ReaderTest.php

@@ -461,6 +461,24 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
     }
 
 
+    /**
+     * https://github.com/box/spout/issues/195
+     * @return void
+     */
+    public function testReaderShouldNotTrimCellValues()
+    {
+        $allRows = $this->getAllRowsForFile('sheet_with_untrimmed_inline_strings.xlsx');
+
+        $expectedRows = [
+            ['A'],
+            [' A '],
+            ["\n\tA\n\t"],
+        ];
+
+        $this->assertEquals($expectedRows, $allRows, 'Cell values should not be trimmed');
+    }
+
+
     /**
      * @param string $fileName
      * @return array All the read rows the given file

tests/resources/csv/sheet_with_untrimmed_strings.csv

@@ -0,0 +1,5 @@
+A
+" A "
+"
+	A
+	"