Merge e7529ba into 6080327

src/Extract/CsvFileExtractor.php

@@ -0,0 +1,169 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 15/08/18
+ * Time: 7:08 PM
+ */
+
+namespace MilesAsylum\Slurp\Extract;
+
+
+use MilesAsylum\Slurp\Extract\Exception\MalformedCsvException;
+
+class CsvFileExtractor implements ExtractorInterface
+{
+    /**
+     * @var string
+     */
+    protected $filePath;
+
+    /**
+     * @var bool|resource
+     */
+    protected $fileHandle;
+
+    protected $columnNames = [];
+
+    protected $firstRowValueCount;
+
+    /**
+     * @var int
+     */
+    private $lineLength;
+
+    /**
+     * @var string
+     */
+    private $delimiter;
+
+    /**
+     * @var string
+     */
+    private $enclosure;
+
+    /**
+     * @var string
+     */
+    private $escape;
+
+    /**
+     * @var array
+     */
+    protected $currentLine;
+
+    protected $currentLineNo = null;
+
+    /**
+     * CsvFileExtractor constructor.
+     * @param $filePath
+     * @param $hasHeader
+     * @param int $lineLength
+     * @param string $delimiter
+     * @param string $enclosure
+     * @param string $escape
+     * @throws MalformedCsvException
+     */
+    public function __construct(
+        $filePath,
+        $hasHeader,
+        $lineLength = 0,
+        $delimiter = ',',
+        $enclosure = '"',
+        $escape = '\\'
+    ) {
+        $this->filePath = $filePath;
+        $this->fileHandle = fopen($this->filePath, "r");
+        $this->lineLength = $lineLength;
+        $this->delimiter = $delimiter;
+        $this->enclosure = $enclosure;
+        $this->escape = $escape;
+
+        if ($hasHeader) {
+            $this->columnNames = $this->loadCsvLine();
+
+            if (count($this->columnNames) !== count(array_unique($this->columnNames))) {
+                throw new MalformedCsvException(
+                    "The loaded file {$this->filePath} contains duplicate column names."
+                );
+            }
+        }
+
+        $this->currentLine = $this->loadCsvLine();
+        $this->firstRowValueCount = count($this->currentLine);
+    }
+
+    public function getColumnNames()
+    {
+        return $this->columnNames;
+    }
+
+    /**
+     * @return array|false|null
+     * @throws MalformedCsvException
+     */
+    public function current()
+    {
+        if (!empty($this->columnNames)) {
+            if (count($this->columnNames) !== count($this->currentLine)) {
+                throw new MalformedCsvException(
+                    sprintf(
+                        'Line %s in %s has %s values where we expected %s.',
+                        $this->currentLineNo,
+                        $this->filePath,
+                        count($this->currentLine),
+                        count($this->columnNames)
+                    )
+                );
+            }
+            return array_combine($this->columnNames, $this->currentLine);
+        } elseif ($this->firstRowValueCount != count($this->currentLine)) {
+            throw new MalformedCsvException(
+                sprintf(
+                    'Line %s in %s has %s values where previous rows had %s.',
+                    $this->currentLineNo,
+                    $this->filePath,
+                    count($this->currentLine),
+                    $this->firstRowValueCount
+                )
+            );
+        }
+
+        return $this->currentLine;
+    }
+
+    public function next()
+    {
+        $this->currentLine = $this->loadCsvLine();
+    }
+
+    public function key()
+    {
+        return $this->currentLineNo;
+    }
+
+    public function valid()
+    {
+        return $this->currentLine !== false;
+    }
+
+    public function rewind()
+    {
+        if ($this->currentLineNo > 1) {
+            $this->currentLineNo = null;
+            rewind($this->fileHandle);
+            $this->loadCsvLine(); // Skip the line containing headers.
+            $this->currentLine = $this->loadCsvLine();
+        }
+    }
+
+    protected function loadCsvLine()
+    {
+        if ($this->currentLineNo === null) {
+            $this->currentLineNo = 0;
+        } else {
+            $this->currentLineNo++;
+        }
+
+        return fgetcsv($this->fileHandle, $this->lineLength, $this->delimiter, $this->enclosure, $this->escape);
+    }
+}

src/Extract/Exception/ExceptionInterface.php

@@ -0,0 +1,13 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 15/08/18
+ * Time: 11:30 PM
+ */
+
+namespace MilesAsylum\Slurp\Extract\Exception;
+
+interface ExceptionInterface
+{
+
+}

src/Extract/Exception/MalformedCsvException.php

@@ -0,0 +1,13 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 15/08/18
+ * Time: 11:30 PM
+ */
+
+namespace MilesAsylum\Slurp\Extract\Exception;
+
+class MalformedCsvException extends \Exception implements ExceptionInterface
+{
+
+}

src/Extract/ExtractorInterface.php

@@ -14,5 +14,5 @@ interface ExtractorInterface extends \Iterator
      * Get the defined columns for the source.
      * @return array
      */
-    public function getColumns();
+    public function getColumnNames();
 }

tests/Slurp/Extract/CsvFileExtractorTest.php

@@ -0,0 +1,126 @@
+<?php
+/**
+ * Author: Courtney Miles
+ * Date: 15/08/18
+ * Time: 10:15 PM
+ */
+
+namespace MilesAsylum\Slurp\Tests\Slurp\Extract;
+
+use MilesAsylum\Slurp\Extract\CsvFileExtractor;
+use MilesAsylum\Slurp\Extract\Exception\MalformedCsvException;
+use PHPUnit\Framework\TestCase;
+
+class CsvFileExtractorTest extends TestCase
+{
+    public function testGetColumnNamesWithHeaders()
+    {
+        $csv = new CsvFileExtractor(__DIR__ . '/_fixture/with_headers.csv', true);
+
+        $this->assertSame(
+            ['user','date','class','value'],
+            $csv->getColumnNames()
+        );
+    }
+
+    public function testGetColumnNamesWithoutHeaders()
+    {
+        $csv = new CsvFileExtractor(__DIR__ . '/_fixture/without_headers.csv', false);
+
+        $this->assertSame(
+            [],
+            $csv->getColumnNames()
+        );
+    }
+
+    public function testIterateFileWithHeaders()
+    {
+        $csv = new CsvFileExtractor(__DIR__ . '/_fixture/with_headers.csv', true);
+
+        $this->assertSame(
+            [
+                1 => [
+                    'user' => 'user123',
+                    'date' => '2018-01-01',
+                    'class' => 'foo',
+                    'value' => '123.45'
+                ],
+                2 => [
+                    'user' => 'user456',
+                    'date' => '2018-02-01',
+                    'class' => 'bar',
+                    'value' => '678.90'
+                ]
+            ],
+            $this->iterateCsv($csv)
+        );
+    }
+
+    /**
+     * @depends testIterateFileWithHeaders
+     */
+    public function testRewindIterator()
+    {
+        $csv = new CsvFileExtractor(__DIR__ . '/_fixture/with_headers.csv', true);
+
+        $this->assertSame($this->iterateCsv($csv), $this->iterateCsv($csv));
+    }
+
+    public function testIterateFileWithoutHeaders()
+    {
+        $csv = new CsvFileExtractor(__DIR__ . '/_fixture/without_headers.csv', false);
+
+        $this->assertSame(
+            [
+                0 => ['user123', '2018-01-01', 'foo', '123.45'],
+                1 => ['user456', '2018-02-01', 'bar', '678.90']
+            ],
+            $this->iterateCsv($csv)
+        );
+    }
+
+    public function testIterateFileWithMismatchColumnCount()
+    {
+        $this->expectException(MalformedCsvException::class);
+        $this->expectExceptionMessage(
+            'Line 1 in ' . __DIR__ . '/_fixture/mismatch_column_count.csv has 5 values where we expected 4.'
+        );
+
+        $csv = new CsvFileExtractor(__DIR__ . '/_fixture/mismatch_column_count.csv', true);
+
+        $this->iterateCsv($csv);
+    }
+
+    public function testIterateFileWithVaryingColumnCount()
+    {
+        $this->expectException(MalformedCsvException::class);
+        $this->expectExceptionMessage(
+            'Line 1 in ' . __DIR__ . '/_fixture/varying_column_count.csv has 3 values where previous rows had 4.'
+        );
+
+        $csv = new CsvFileExtractor(__DIR__ . '/_fixture/varying_column_count.csv', false);
+
+        $this->iterateCsv($csv);
+    }
+
+    public function testConstructFileWithDuplicateColumnNames()
+    {
+        $this->expectException(MalformedCsvException::class);
+        $this->expectExceptionMessage(
+            'The loaded file ' . __DIR__ . '/_fixture/duplicate_headers.csv contains duplicate column names.'
+        );
+
+        new CsvFileExtractor(__DIR__ . '/_fixture/duplicate_headers.csv', true);
+    }
+
+    protected function iterateCsv(CsvFileExtractor $csv)
+    {
+        $results = [];
+
+        foreach ($csv as $linoNo => $arr) {
+            $results[$linoNo] = $arr;
+        }
+
+        return $results;
+    }
+}

tests/Slurp/Extract/_fixture/duplicate_headers.csv

@@ -0,0 +1,3 @@
+user,user,date,class,value
+user123,user123,2018-01-01,foo,123.45
+user456,user456,2018-02-01,bar,678.90

tests/Slurp/Extract/_fixture/mismatch_column_count.csv

@@ -0,0 +1,3 @@
+user,date,class,value
+user123,2018-01-01,foo,123.45,111
+user456,2018-02-01,bar,678.90,222

tests/Slurp/Extract/_fixture/varying_column_count.csv

@@ -0,0 +1,2 @@
+user123,2018-01-01,foo,123.45
+user456,2018-02-01,bar

tests/Slurp/Extract/_fixture/with_headers.csv

@@ -0,0 +1,3 @@
+user,date,class,value
+user123,2018-01-01,foo,123.45
+user456,2018-02-01,bar,678.90

tests/Slurp/Extract/_fixture/without_headers.csv

@@ -0,0 +1,2 @@
+user123,2018-01-01,foo,123.45
+user456,2018-02-01,bar,678.90