Merge pull request #173 from sfichera/master

Support for variable EOL for CSV

Author: adrilo

README.md

@@ -110,6 +110,7 @@ use Box\Spout\Common\Type;
 $reader = ReaderFactory::create(Type::CSV);
 $reader->setFieldDelimiter('|');
 $reader->setFieldEnclosure('@');
+$reader->setEndOfLineCharacter("\r");
 ```
 
 Additionally, if you need to read non UTF-8 files, you can specify the encoding of your file this way:

src/Spout/Reader/CSV/Reader.php

@@ -29,6 +29,9 @@ class Reader extends AbstractReader
     /** @var string Encoding of the CSV file to be read */
     protected $encoding = EncodingHelper::ENCODING_UTF8;
 
+    /** @var string Defines the End of line */
+    protected $endOfLineCharacter = "\n";    
+
     /**
      * Sets the field delimiter for the CSV.
      * Needs to be called before opening the reader.
@@ -68,6 +71,19 @@ public function setEncoding($encoding)
         return $this;
     }
 
+    /**
+     * Sets the EOL for the CSV.
+     * Needs to be called before opening the reader.
+     *
+     * @param string $endOfLineCharacter used to properly get lines from the CSV file.
+     * @return Reader
+     */
+    public function setEndOfLineCharacter($endOfLineCharacter)
+    {
+        $this->endOfLineCharacter = $endOfLineCharacter;
+        return $this;
+    }  
+
     /**
      * Opens the file at the given path to make it ready to be read.
      * If setEncoding() was not called, it assumes that the file is encoded in UTF-8.
@@ -88,6 +104,7 @@ protected function openReader($filePath)
             $this->fieldDelimiter,
             $this->fieldEnclosure,
             $this->encoding,
+            $this->endOfLineCharacter,
             $this->globalFunctionsHelper
         );
     }

src/Spout/Reader/CSV/RowIterator.php

@@ -49,19 +49,23 @@ class RowIterator implements IteratorInterface
     /** @var string End of line delimiter, encoded using the same encoding as the CSV */
     protected $encodedEOLDelimiter;
 
+    /** @var string End of line delimiter, given by the user as input. */
+    protected $inputEOLDelimiter;
+
     /**
      * @param resource $filePointer Pointer to the CSV file to read
      * @param string $fieldDelimiter Character that delimits fields
      * @param string $fieldEnclosure Character that enclose fields
      * @param string $encoding Encoding of the CSV file to be read
      * @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
      */
-    public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $globalFunctionsHelper)
+    public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineDelimiter, $globalFunctionsHelper)
     {
         $this->filePointer = $filePointer;
         $this->fieldDelimiter = $fieldDelimiter;
         $this->fieldEnclosure = $fieldEnclosure;
         $this->encoding = $encoding;
+        $this->inputEOLDelimiter = $endOfLineDelimiter;
         $this->globalFunctionsHelper = $globalFunctionsHelper;
 
         $this->encodingHelper = new EncodingHelper($globalFunctionsHelper);
@@ -172,7 +176,7 @@ protected function getNextUTF8EncodedLine()
     protected function getEncodedEOLDelimiter()
     {
         if (!isset($this->encodedEOLDelimiter)) {
-            $this->encodedEOLDelimiter = $this->encodingHelper->attemptConversionFromUTF8("\n", $this->encoding);
+            $this->encodedEOLDelimiter = $this->encodingHelper->attemptConversionFromUTF8($this->inputEOLDelimiter, $this->encoding);
         }
 
         return $this->encodedEOLDelimiter;

src/Spout/Reader/CSV/Sheet.php

@@ -21,9 +21,9 @@ class Sheet implements SheetInterface
      * @param string $encoding Encoding of the CSV file to be read
      * @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
      */
-    public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $globalFunctionsHelper)
+    public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineCharacter, $globalFunctionsHelper)
     {
-        $this->rowIterator = new RowIterator($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $globalFunctionsHelper);
+        $this->rowIterator = new RowIterator($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineCharacter, $globalFunctionsHelper);
     }
 
     /**

src/Spout/Reader/CSV/SheetIterator.php

@@ -25,9 +25,9 @@ class SheetIterator implements IteratorInterface
      * @param string $encoding Encoding of the CSV file to be read
      * @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
      */
-    public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $globalFunctionsHelper)
+    public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineCharacter, $globalFunctionsHelper)
     {
-        $this->sheet = new Sheet($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $globalFunctionsHelper);
+        $this->sheet = new Sheet($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineCharacter, $globalFunctionsHelper);
     }
 
     /**

tests/Spout/Reader/CSV/ReaderTest.php

@@ -377,4 +377,50 @@ private function getAllRowsForFile(
 
         return $allRows;
     }
+
+    /**
+     * @return array
+     */
+    public function dataProviderForTestReadCustomEOL()
+    {
+        return [
+            ['csv_with_CR_EOL.csv', "\r"],
+            ['csv_standard.csv', "\n"],
+        ];
+    }
+
+    /**
+     * @dataProvider dataProviderForTestReadCustomEOL
+     *
+     * @param string $fileName
+     * @param string $customEOL
+     * @return void
+     */
+    public function testReadCustomEOLs($fileName, $customEOL)
+    {
+        $allRows = [];
+        $resourcePath = $this->getResourcePath($fileName);
+
+        /** @var \Box\Spout\Reader\CSV\Reader $reader */
+        $reader = ReaderFactory::create(Type::CSV);
+        $reader
+            ->setEndOfLineCharacter($customEOL)
+            ->open($resourcePath);
+
+        foreach ($reader->getSheetIterator() as $sheet) {
+            foreach ($sheet->getRowIterator() as $row) {
+                $allRows[] = $row;
+            }
+        }
+
+        $reader->close();
+
+        $expectedRows = [
+            ['csv--11', 'csv--12', 'csv--13'],
+            ['csv--21', 'csv--22', 'csv--23'],
+            ['csv--31', 'csv--32', 'csv--33'],
+        ];
+        $this->assertEquals($expectedRows, $allRows);
+    }
+
 }

tests/resources/csv/csv_with_CR_EOL.csv

@@ -0,0 +1 @@
+csv--11,csv--12,csv--13csv--21,csv--22,csv--23csv--31,csv--32,csv--33