Merge pull request #19 from courtney-miles/enhance/13/EnsureConsisten…

…tRecordValues #13 Ensure CSV rows have consistent number of fields.
courtney-miles · Sep 3, 2018 · bda34ad · bda34ad
2 parents 2e30671 + 3982ba6
commit bda34ad
Show file tree

Hide file tree

Showing 8 changed files with 272 additions and 15 deletions.
diff --git a/src/Extract/CsvFileExtractor/CsvFileExtractor.php b/src/Extract/CsvFileExtractor/CsvFileExtractor.php
@@ -7,6 +7,7 @@
 
 namespace MilesAsylum\Slurp\Extract\CsvFileExtractor;
 
+use CallbackFilterIterator;
 use League\Csv\Reader;
 use MilesAsylum\Slurp\Extract\ExtractorInterface;
 
@@ -19,18 +20,20 @@ class CsvFileExtractor implements ExtractorInterface
 
     private $headers = [];
 
+    private $headerOffset;
+
     public function __construct(Reader $csvReader)
     {
         $this->csvReader = $csvReader;
     }
 
     /**
      * Loads the first row in the CSV file as the headers.
-     * @throws \League\Csv\Exception
      */
     public function loadHeadersFromFile() : void
     {
-        $this->csvReader->setHeaderOffset(0);
+        $this->headers = $this->csvReader->fetchOne();
+        $this->headerOffset = 0;
     }
 
     public function setHeaders(array $headers) : void
@@ -40,6 +43,24 @@ public function setHeaders(array $headers) : void
 
     public function getIterator() : \Iterator
     {
-        return $this->csvReader->getRecords($this->headers);
+        return $this->prepareRecords($this->csvReader->getRecords(), $this->headers);
+    }
+
+    protected function prepareRecords(\Iterator $records, array $headers): \Iterator
+    {
+        if ($this->headerOffset !== null) {
+            $records = new CallbackFilterIterator($records, function (array $record, int $offset): bool {
+                return $offset !== $this->headerOffset;
+            });
+        }
+
+        $valueCount = !empty($headers) ? count($headers) : count($this->csvReader->fetchOne());
+        $records = new VerifyValueCountIterator($records, $valueCount);
+
+        if (!empty($headers)) {
+            $records = new MapIterator($records, $headers);
+        }
+
+        return $records;
     }
 }
diff --git a/src/Extract/CsvFileExtractor/MapIterator.php b/src/Extract/CsvFileExtractor/MapIterator.php
@@ -0,0 +1,31 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 27/08/18
+ * Time: 6:32 PM
+ */
+
+namespace MilesAsylum\Slurp\Extract\CsvFileExtractor;
+
+use Traversable;
+
+class MapIterator extends \IteratorIterator
+{
+    /**
+     * @var array
+     */
+    private $headers;
+
+    public function __construct(Traversable $iterator, array $headers)
+    {
+        parent::__construct($iterator);
+        $this->headers = $headers;
+    }
+
+    public function current()
+    {
+        $record = parent::current();
+
+        return array_combine($this->headers, $record);
+    }
+}
diff --git a/src/Extract/CsvFileExtractor/VerifyValueCountIterator.php b/src/Extract/CsvFileExtractor/VerifyValueCountIterator.php
@@ -0,0 +1,45 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 27/08/18
+ * Time: 6:05 PM
+ */
+
+namespace MilesAsylum\Slurp\Extract\CsvFileExtractor;
+
+use MilesAsylum\Slurp\Extract\Exception\ValueCountMismatchException;
+use Traversable;
+
+class VerifyValueCountIterator extends \IteratorIterator
+{
+    private $expectedValueCount;
+
+    public function __construct(Traversable $iterator, int $expectedValueCount)
+    {
+        parent::__construct($iterator);
+        $this->expectedValueCount = $expectedValueCount;
+    }
+
+    /**
+     * {@inheritdoc}
+     * @throws ValueCountMismatchException
+     */
+    public function current()
+    {
+        $record = parent::current();
+
+        if (count($record) !== $this->expectedValueCount) {
+            $recordId = $this->key();
+            throw new ValueCountMismatchException(
+                sprintf(
+                    'Record %s contained %s values where we expected %s.',
+                    $recordId,
+                    count($record),
+                    $this->expectedValueCount
+                )
+            );
+        }
+
+        return $record;
+    }
+}
diff --git a/src/Extract/Exception/ExceptionInterface.php b/src/Extract/Exception/ExceptionInterface.php
@@ -0,0 +1,12 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 27/08/18
+ * Time: 6:10 PM
+ */
+
+namespace MilesAsylum\Slurp\Extract\Exception;
+
+interface ExceptionInterface
+{
+}
diff --git a/src/Extract/Exception/ValueCountMismatchException.php b/src/Extract/Exception/ValueCountMismatchException.php
@@ -0,0 +1,13 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 27/08/18
+ * Time: 6:10 PM
+ */
+
+namespace MilesAsylum\Slurp\Extract\Exception;
+
+class ValueCountMismatchException extends \Exception implements ExceptionInterface
+{
+
+}
diff --git a/tests/Slurp/Extract/CsvFileExtractor/CsvFileExtractorTest.php b/tests/Slurp/Extract/CsvFileExtractor/CsvFileExtractorTest.php
@@ -9,10 +9,15 @@
 
 use League\Csv\Reader;
 use MilesAsylum\Slurp\Extract\CsvFileExtractor\CsvFileExtractor;
+use MilesAsylum\Slurp\Extract\Exception\ValueCountMismatchException;
 use Mockery\Adapter\Phpunit\MockeryPHPUnitIntegration;
 use Mockery\MockInterface;
 use PHPUnit\Framework\TestCase;
 
+
+/**
+ * @covers \MilesAsylum\Slurp\Extract\CsvFileExtractor\CsvFileExtractor
+ */
 class CsvFileExtractorTest extends TestCase
 {
     use MockeryPHPUnitIntegration;
@@ -35,25 +40,78 @@ public function setUp()
         $this->csvExtractor = new CsvFileExtractor($this->mockReader);
     }
 
-    public function testLoadHeaders()
+    public function testNoHeaders()
     {
-        $this->mockReader->shouldReceive('setHeaderOffset')
-            ->with(0)
-            ->once();
+        $csvRows = [[123, 234]];
+        $this->setUpMockReader($this->mockReader, $csvRows);
 
-        $this->csvExtractor->loadHeadersFromFile();
+        foreach ($this->csvExtractor as $rowId => $row) {
+            $this->assertArrayHasKey($rowId, $csvRows);
+            $this->assertSame($csvRows[$rowId], $row);
+        }
     }
 
-    public function testUseSuppliedHeaders()
+    public function testUseHeadersFromFile()
     {
-        $headers = ['col 1', 'col 2'];
-        $mockIterator = \Mockery::mock(\Iterator::class);
-        $this->mockReader->shouldReceive('getRecords')
-            ->with($headers)
-            ->andReturn($mockIterator);
+        $csvRows = [
+            ['col_1', 'col_2'],
+            [123, 234]
+        ];
+        $this->setUpMockReader($this->mockReader, $csvRows);
+        $this->csvExtractor->loadHeadersFromFile();
 
+        foreach ($this->csvExtractor as $rowId => $row) {
+            $this->assertArrayHasKey($rowId, $csvRows);
+            $this->assertSame(array_combine($csvRows[0], $csvRows[$rowId]), $row);
+        }
+    }
+
+    public function testSetHeaders()
+    {
+        $headers = ['col_1', 'col_2'];
+        $csvRows = [[123, 234]];
+        $this->setUpMockReader($this->mockReader, $csvRows);
         $this->csvExtractor->setHeaders($headers);
 
-        $this->assertSame($mockIterator, $this->csvExtractor->getIterator());
+        foreach ($this->csvExtractor as $rowId => $row) {
+            $this->assertArrayHasKey($rowId, $csvRows);
+            $this->assertSame(array_combine($headers, $csvRows[$rowId]), $row);
+        }
+    }
+
+    public function testExceptionOnInconsistentNumberOfRowValuesWithoutHeaders()
+    {
+        $this->expectException(ValueCountMismatchException::class);
+
+        $csvRows = [[123, 234],[345,456,567]];
+        $this->setUpMockReader($this->mockReader, $csvRows);
+
+        foreach ($this->csvExtractor as $rowId => $row) {
+            // Do nothing.
+        }
+    }
+
+    public function testExceptionOnInconsistentNumberOfRowValuesWithHeaders()
+    {
+        $this->expectException(ValueCountMismatchException::class);
+
+        $csvRows = [[123, 234, 345]];
+        $this->setUpMockReader($this->mockReader, $csvRows);
+        $this->csvExtractor->setHeaders(['col_1', 'col_2']);
+
+        foreach ($this->csvExtractor as $rowId => $row) {
+            // Do nothing.
+        }
+    }
+
+    public function setUpMockReader(MockInterface $mockReader, array $rows)
+    {
+        $mockReader->shouldReceive('fetchOne')
+            ->withNoArgs()
+            ->andReturn($rows[0]);
+
+        $mockReader->shouldReceive('getRecords')
+            ->withNoArgs()
+            ->andReturn(new \ArrayIterator($rows));
     }
 }
diff --git a/tests/Slurp/Extract/CsvFileExtractor/MapIteratorTest.php b/tests/Slurp/Extract/CsvFileExtractor/MapIteratorTest.php
@@ -0,0 +1,29 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 27/08/18
+ * Time: 6:47 PM
+ */
+
+namespace MilesAsylum\Slurp\Tests\Slurp\Extract\CsvFileExtractor;
+
+use MilesAsylum\Slurp\Extract\CsvFileExtractor\MapIterator;
+use PHPUnit\Framework\TestCase;
+
+class MapIteratorTest extends TestCase
+{
+    public function testMapHeaders()
+    {
+        $mi = new MapIterator(
+            new \ArrayIterator([[123, 234]]),
+            ['val_1', 'val_2']
+        );
+
+        $mi->rewind();
+
+        $this->assertSame(
+            ['val_1' => 123, 'val_2' => 234],
+            $mi->current()
+        );
+    }
+}
diff --git a/tests/Slurp/Extract/CsvFileExtractor/VerifyValueCountIteratorTest.php b/tests/Slurp/Extract/CsvFileExtractor/VerifyValueCountIteratorTest.php
@@ -0,0 +1,48 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 27/08/18
+ * Time: 6:53 PM
+ */
+
+namespace MilesAsylum\Slurp\Tests\Slurp\Extract\CsvFileExtractor;
+
+use MilesAsylum\Slurp\Extract\CsvFileExtractor\VerifyValueCountIterator;
+use MilesAsylum\Slurp\Extract\Exception\ValueCountMismatchException;
+use PHPUnit\Framework\TestCase;
+
+class VerifyValueCountIteratorTest extends TestCase
+{
+    /**
+     * @dataProvider getValueCountMismatchTestData
+     */
+    public function testValueCountMismatch($values, $expectedCount)
+    {
+        $this->expectException(ValueCountMismatchException::class);
+        $this->expectExceptionMessage(
+            sprintf(
+                'Record 0 contained %s values where we expected %.',
+                count($values),
+                $expectedCount
+            )
+        );
+
+        $iterator = new VerifyValueCountIterator(
+            new \ArrayIterator(
+                [$values]
+            ),
+            $expectedCount
+        );
+
+        $iterator->rewind();
+        $iterator->current();
+    }
+
+    public function getValueCountMismatchTestData()
+    {
+        return [
+            [[123,234], 1],
+            [[123,234], 3],
+        ];
+    }
+}