Merge pull request #187 from skeleton/issue-183

Fix line breaks on CSV reader

src/Spout/Common/Helper/GlobalFunctionsHelper.php

@@ -291,35 +291,6 @@ public function stream_get_wrappers()
         return stream_get_wrappers();
     }
 
-    /**
-     * Wrapper around global function stream_get_line()
-     * @see stream_get_line()
-     *
-     * @param resource $handle
-     * @param int $length
-     * @param string|void $ending
-     * @return string|bool
-     */
-    public function stream_get_line($handle, $length, $ending = null)
-    {
-        return stream_get_line($handle, $length, $ending);
-    }
-
-    /**
-     * Wrapper around global function str_getcsv()
-     * @see str_getcsv()
-     *
-     * @param string $input
-     * @param string|void $delimiter
-     * @param string|void $enclosure
-     * @param string|void $escape
-     * @return array
-     */
-    public function str_getcsv($input, $delimiter = null, $enclosure = null, $escape = null)
-    {
-        return str_getcsv($input, $delimiter, $enclosure, $escape);
-    }
-
     /**
      * Wrapper around global function function_exists()
      * @see function_exists()

src/Spout/Reader/CSV/Reader.php

@@ -32,6 +32,9 @@ class Reader extends AbstractReader
     /** @var string Defines the End of line */
     protected $endOfLineCharacter = "\n";
 
+    /** @var string */
+    protected $autoDetectLineEndings;
+
     /**
      * Sets the field delimiter for the CSV.
      * Needs to be called before opening the reader.
@@ -104,6 +107,9 @@ protected function doesSupportStreamWrapper()
      */
     protected function openReader($filePath)
     {
+        $this->autoDetectLineEndings = ini_get('auto_detect_line_endings');
+        ini_set('auto_detect_line_endings', '1');
+
         $this->filePointer = $this->globalFunctionsHelper->fopen($filePath, 'r');
         if (!$this->filePointer) {
             throw new IOException("Could not open file $filePath for reading.");
@@ -140,5 +146,7 @@ protected function closeReader()
         if ($this->filePointer) {
             $this->globalFunctionsHelper->fclose($this->filePointer);
         }
+
+        ini_set('auto_detect_line_endings', $this->autoDetectLineEndings);
     }
 }

src/Spout/Reader/CSV/RowIterator.php

@@ -14,7 +14,7 @@
 class RowIterator implements IteratorInterface
 {
     /**
-     * If no value is given to stream_get_line(), it defaults to 8192 (which may be too low).
+     * If no value is given to fgetcsv(), it defaults to 8192 (which may be too low).
      * Alignement with other functions like fgets() is discussed here: https://bugs.php.net/bug.php?id=48421
      */
     const MAX_READ_BYTES_PER_LINE = 32768;
@@ -128,16 +128,12 @@ public function next()
         }
 
         do {
-            $lineData = false;
-            $utf8EncodedLineData = $this->getNextUTF8EncodedLine();
-            if ($utf8EncodedLineData !== false) {
-                $lineData = $this->globalFunctionsHelper->str_getcsv($utf8EncodedLineData, $this->fieldDelimiter, $this->fieldEnclosure);
-            }
+            $rowData = $this->getNextUTF8EncodedRow();
             $hasNowReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);
-        } while (($lineData === false && !$hasNowReachedEndOfFile) || $this->isEmptyLine($lineData));
+        } while (($rowData === false && !$hasNowReachedEndOfFile) || $this->isEmptyLine($rowData));
 
-        if ($lineData !== false) {
-            $this->rowDataBuffer = $lineData;
+        if ($rowData !== false) {
+            $this->rowDataBuffer = $rowData;
             $this->numReadRows++;
         } else {
             // If we reach this point, it means end of file was reached.
@@ -147,24 +143,39 @@ public function next()
     }
 
     /**
-     * Returns the next line, converted if necessary to UTF-8.
-     * Neither fgets nor fgetcsv don't work with non UTF-8 data... so we need to do some things manually.
+     * Returns the next row, converted if necessary to UTF-8.
+     * As fgetcsv() does not manage correctly encoding for non UTF-8 data,
+     * we remove manually whitespace with ltrim or rtrim (depending on the order of the bytes)
      *
-     * @return string|false The next line for the current file pointer, encoded in UTF-8 or FALSE if nothing to read
+     * @return array|false The row for the current file pointer, encoded in UTF-8 or FALSE if nothing to read
      * @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
      */
-    protected function getNextUTF8EncodedLine()
+    protected function getNextUTF8EncodedRow()
     {
-        // Read until the EOL delimiter or EOF is reached. The delimiter's encoding needs to match the CSV's encoding.
-        $encodedEOLDelimiter = $this->getEncodedEOLDelimiter();
-        $encodedLineData = $this->globalFunctionsHelper->stream_get_line($this->filePointer, self::MAX_READ_BYTES_PER_LINE, $encodedEOLDelimiter);
+        $encodedRowData = fgetcsv($this->filePointer, self::MAX_READ_BYTES_PER_LINE, $this->fieldDelimiter, $this->fieldEnclosure);
+        if (false === $encodedRowData) {
+            return false;
+        }
 
-        // If the line could have been read, it can be converted to UTF-8
-        $utf8EncodedLineData = ($encodedLineData !== false) ?
-            $this->encodingHelper->attemptConversionToUTF8($encodedLineData, $this->encoding) :
-            false;
+        foreach ($encodedRowData as $cellIndex => $cellValue) {
+            switch($this->encoding) {
+                case EncodingHelper::ENCODING_UTF16_LE:
+                case EncodingHelper::ENCODING_UTF32_LE:
+                    // remove whitespace from the beginning of a string as fgetcsv() add extra whitespace when it try to explode non UTF-8 data
+                    $cellValue = ltrim($cellValue);
+                    break;
+
+                case EncodingHelper::ENCODING_UTF16_BE:
+                case EncodingHelper::ENCODING_UTF32_BE:
+                    // remove whitespace from the end of a string as fgetcsv() add extra whitespace when it try to explode non UTF-8 data
+                    $cellValue = rtrim($cellValue);
+                    break;
+            }
+
+            $encodedRowData[$cellIndex] = $this->encodingHelper->attemptConversionToUTF8($cellValue, $this->encoding);
+        }
 
-        return $utf8EncodedLineData;
+        return $encodedRowData;
     }
 
     /**

tests/Spout/Reader/CSV/ReaderTest.php

@@ -204,6 +204,15 @@ public function testReadShouldSupportCustomFieldEnclosure()
         $this->assertEquals('This is, a comma', $allRows[0][0]);
     }
 
+    /**
+     * @return void
+     */
+    public function testReadShouldNotTruncateLineBreak()
+    {
+        $allRows = $this->getAllRowsForFile('csv_with_line_breaks.csv', ',');
+        $this->assertEquals("This is,\na comma", $allRows[0][0]);
+    }
+
     /**
      * @return array
      */

tests/resources/csv/csv_with_line_breaks.csv

@@ -0,0 +1,2 @@
+"This is,
+a comma",csv--12