- dropped SKIP_EMPTY and READ_AHEAD due to strange behavior when using c...

lib/EasyCSV/AbstractBase.php

@@ -14,7 +14,7 @@ public function __construct($path, $mode = 'r+')
             touch($path);
         }
         $this->handle = new \SplFileObject($path, $mode);
-        $this->handle->setFlags(\SplFileObject::SKIP_EMPTY | \SplFileObject::DROP_NEW_LINE | \SplFileObject::READ_AHEAD);
+        $this->handle->setFlags(\SplFileObject::DROP_NEW_LINE);
     }
 
     public function __destruct()

lib/EasyCSV/Reader.php

@@ -2,41 +2,88 @@
 
 namespace EasyCSV;
 
+/**
+ * Class Reader
+ *
+ * @package EasyCSV
+ */
 class Reader extends AbstractBase
 {
+    /**
+     * @var bool
+     */
     private $headersInFirstRow = true;
-    private $headers;
+    /**
+     * @var bool
+     */
+    private $headers = false;
+    /**
+     * @var int
+     */
     private $line;
+
+    /**
+     * @var
+     */
     private $init;
 
+    /**
+     * @var bool
+     */
+    private $headerLine = false;
+    /**
+     * @var bool
+     */
+    private $lastLine = false;
+
+    /**
+     * @param $path
+     * @param string $mode
+     * @param bool $headersInFirstRow
+     */
     public function __construct($path, $mode = 'r+', $headersInFirstRow = true)
     {
         parent::__construct($path, $mode);
         $this->headersInFirstRow = $headersInFirstRow;
         $this->line = 0;
     }
 
+    /**
+     * @return bool
+     */
     public function getHeaders()
     {
         $this->init();
         return $this->headers;
     }
-
+
+    /**
+     * @return array|bool
+     */
     public function getRow()
     {
         $this->init();
         if ($this->handle->eof()) {
             return false;
         }
 
-        if (($row = $this->handle->fgetcsv($this->delimiter, $this->enclosure)) !== false && $row != null) {
+        $row = $this->handle->fgetcsv($this->delimiter, $this->enclosure);
+        $isEmpty = $this->rowIsEmpty( $row );
+
+        if ($row !== false && $row != null && $isEmpty === FALSE ) {
             $this->line++;
             return $this->headers ? array_combine($this->headers, $row) : $row;
+        } elseif( $isEmpty ){
+            // empty row, transparently try the next row
+            return $this->getRow();
         } else {
             return false;
         }
     }
 
+    /**
+     * @return array
+     */
     public function getAll()
     {
         $data = array();
@@ -46,22 +93,107 @@ public function getAll()
         return $data;
     }
 
+    /**
+     * @return int zero-based index
+     */
     public function getLineNumber()
     {
-        return $this->line;
+        return $this->handle->key();
+    }
+
+    /**
+     * @return int zero-based index
+     */
+    public function getLastLineNumber()
+    {
+        if( $this->lastLine !== FALSE ) {
+            return $this->lastLine;
+        }
+
+        $this->handle->seek($this->handle->getSize());
+        $lastLine = $this->handle->key();
+
+        $this->handle->rewind();
+
+        return $this->lastLine = $lastLine;
+    }
+
+    /**
+     * @return array
+     */
+    public function getCurrentRow()
+    {
+        return str_getcsv( $this->handle->current(), $this->delimiter, $this->enclosure );
+    }
+
+    /**
+     * @param $lineNumber zero-based index
+     */
+    public function advanceTo( $lineNumber )
+    {
+        if( $this->headerLine > $lineNumber){
+            throw new \LogicException("Line Number $lineNumber is before the header line that was set");
+        } elseif( $this->headerLine === $lineNumber ){
+            throw new \LogicException("Line Number $lineNumber is equal to the header line that was set");
+        }
+
+        $this->line = $lineNumber;
+
+        $this->handle->seek( $lineNumber );
+    }
+
+    /**
+     * @param $lineNumber zero-based index
+     */
+    public function setHeaderLine( $lineNumber )
+    {
+        if( $lineNumber !== 0 ) {
+            $this->headersInFirstRow = false;
+        } else {
+            return false;
+        }
+
+        $this->headerLine = $lineNumber;
+
+        // seek to line before headers
+        $this->handle->seek( $lineNumber );
+
+        // get headers
+        $this->headers = $this->getCurrentRow();
     }
-    
+
     protected function init()
     {
         if (true === $this->init) {
             return;
         }
-        $this->init    = true;
-        $this->headers = $this->headersInFirstRow === true ? $this->getRow() : false;
+        $this->init = true;
+
+        if( $this->headersInFirstRow === true) {
+            $this->handle->rewind();
+
+            $this->headers = $this->getRow();
+        }
     }
 
-    protected function incrementLine()
+    /**
+     * @param $row
+     * @return bool
+     */
+    protected function rowIsEmpty( $row )
     {
-        $this->line++;
+        $emptyRow = ( $row === array( NULL ) );
+        $emptyRowWithDelimiters = ( array_filter( $row ) === array() );
+        $isEmpty = FALSE;
+
+        if ( $emptyRow ) {
+            $isEmpty = TRUE;
+            return $isEmpty;
+        } elseif ( $emptyRowWithDelimiters ) {
+            $isEmpty = TRUE;
+            return $isEmpty;
+        }
+
+        return $isEmpty;
     }
 }

tests/EasyCSV/Tests/ReaderTest.php

@@ -6,6 +6,40 @@
 
 class ReaderTest extends \PHPUnit_Framework_TestCase
 {
+    protected $headerValues = array( "column1", "column2", "column3" );
+    protected $expectedRows = array (
+        0 =>
+            array (
+                'column1' => '1column2value',
+                'column2' => '1column3value',
+                'column3' => '1column4value',
+            ),
+        1 =>
+            array (
+                'column1' => '2column2value',
+                'column2' => '2column3value',
+                'column3' => '2column4value',
+            ),
+        2 =>
+            array (
+                'column1' => '3column2value',
+                'column2' => '3column3value',
+                'column3' => '3column4value',
+            ),
+        3 =>
+            array (
+                'column1' => '4column2value',
+                'column2' => '4column3value',
+                'column3' => '4column4value',
+            ),
+        4 =>
+            array (
+                'column1' => '5column2value',
+                'column2' => '5column3value',
+                'column3' => '5column4value',
+            ),
+    );
+
     /**
      * @dataProvider getReaders
      */
@@ -30,15 +64,125 @@ public function testGetAll(Reader $reader)
      */
     public function testGetHeaders(Reader $reader)
     {
-        $this->assertEquals(array("column1", "column2", "column3"), $reader->getHeaders());
+        $this->assertEquals( $this->headerValues, $reader->getHeaders());
     }
 
+    /**
+     * @dataProvider getReaders
+     */
+    public function testAdvanceto(Reader $reader)
+    {
+        $reader->advanceTo( 3 );
+
+        $this->assertEquals( 3, $reader->getLineNumber() );
+
+        $reader->advanceTo( 0 );
+
+        $row = array
+        (
+            'column1' => '1column2value',
+            'column2' => '1column3value',
+            'column3' => '1column4value',
+        );
+
+        $actualRow = $reader->getRow();
+        $this->assertEquals( $row, $actualRow );
+
+        $reader->advanceTo( 3 );
+
+        $row = array
+        (
+            'column1' => '4column2value',
+            'column2' => '4column3value',
+            'column3' => '4column4value',
+        );
+
+        $this->assertEquals( $row, $reader->getRow() );
+    }
+
+    /**
+     * @dataProvider getReadersNoHeadersFirstRow
+     */
+    public function testAdvanceToNoHeadersFirstRow(Reader $reader)
+    {
+        $row = array (
+            0 => 'Some Meta Data',
+            1 => '',
+            2 => '',
+        );
+
+        $actualRow = $reader->getRow();
+        $this->assertEquals( $row, $actualRow );
+
+        // give it the ol' one-two-switcharoo
+        $reader->advanceTo(3);
+        $reader->getRow();
+        $reader->advanceTo(0);
+
+        $this->assertEquals( $row, $reader->getRow() );
+    }
+
+    /**
+     * @dataProvider getReaders
+     */
+    public function testSetHeaderLine( Reader $reader )
+    {
+        $headers = $this->headerValues;
+
+        $this->assertEquals( $headers, $reader->getHeaders() );
+
+        $reader->setHeaderLine(0);
+
+        $this->assertEquals( $headers, $reader->getHeaders() );
+    }
+
+    /**
+     * @dataProvider getReadersNoHeadersFirstRow
+     */
+    public function testSetHeaderLineNoHeadersFirstRow( Reader $reader  ){
+        // set headers
+        $reader->setHeaderLine( 3 );
+
+        $this->assertEquals( $this->headerValues, $reader->getHeaders() );
+
+        $rows = $reader->getAll();
+
+        $this->assertCount(5, $rows);
+        $this->assertEquals($this->expectedRows, $rows);
+    }
+
+    /**
+     * @dataProvider getReaders
+     */
+    public function testGetLastLineNumber( Reader $reader )
+    {
+        $this->assertEquals( 5, $reader->getLastLineNumber() );
+    }
+
+    /**
+     * @dataProvider getReadersNoHeadersFirstRow
+     */
+    public function testGetLastLineNumberNoHeadersFirstRow( Reader $reader )
+    {
+        $this->assertEquals( 10, $reader->getLastLineNumber() );
+    }
+
     public function getReaders()
     {
-        $readerSemiColon = new \EasyCSV\Reader(__DIR__ . '/read_sc.csv');
+        $readerSemiColon = new Reader(__DIR__ . '/read_sc.csv');
+        $readerSemiColon->setDelimiter(';');
+        return array(
+            array(new Reader(__DIR__ . '/read.csv')),
+            array($readerSemiColon),
+        );
+    }
+
+    public function getReadersNoHeadersFirstRow()
+    {
+        $readerSemiColon = new Reader(__DIR__ . '/read_header_line_sc.csv', 'r+', FALSE );
         $readerSemiColon->setDelimiter(';');
         return array(
-            array(new \EasyCSV\Reader(__DIR__ . '/read.csv')),
+            array(new Reader(__DIR__ . '/read_header_line.csv', 'r+', FALSE )),
             array($readerSemiColon),
         );
     }

tests/EasyCSV/Tests/read_header_line.csv

@@ -0,0 +1,11 @@
+"Some Meta Data",,
+"Field: Value",,
+,,
+"column1","column2","column3"
+"1column2value","1column3value","1column4value"
+"2column2value","2column3value","2column4value"
+"3column2value","3column3value","3column4value"
+
+,,
+"4column2value","4column3value","4column4value"
+"5column2value","5column3value","5column4value"