firefly-import-preprocessor/tests/CsvReaderTest.php

<?php

namespace UbsCsvTransformer\Tests;

use PHPUnit\Framework\TestCase;
use UbsCsvTransformer\CsvReader;

class CsvReaderTest extends TestCase
{
    private string $tempFile = '';

    protected function setUp(): void
    {
        $this->tempFile = tempnam(sys_get_temp_dir(), 'csvreader_test_');
    }

    protected function tearDown(): void
    {
        if (file_exists($this->tempFile)) {
            unlink($this->tempFile);
        }
    }

    private function write(string $content): void
    {
        file_put_contents($this->tempFile, $content);
    }

    private function reader(int $headerLine = 1, string $delimiter = ';', bool $hasBom = false): CsvReader
    {
        return new CsvReader(
            $this->tempFile,
            ['inputDelimiter' => $delimiter, 'headerLine' => $headerLine, 'hasBom' => $hasBom]
        );
    }

    // -------------------------------------------------------------------------
    // readCsvData – basic parsing
    // -------------------------------------------------------------------------

    public function testReadCsvDataSimple(): void
    {
        $this->write("Name;Age\nAlice;30\nBob;25\n");
        $data = $this->reader()->readCsvData();
        $this->assertCount(2, $data);
        $this->assertSame('Alice', $data[0]['Name']);
        $this->assertSame('30', $data[0]['Age']);
        $this->assertSame('Bob', $data[1]['Name']);
    }

    public function testReadCsvDataTrimsWhitespace(): void
    {
        $this->write("Name ; Age\n Alice ; 30 \n");
        $data = $this->reader()->readCsvData();
        $this->assertSame('Alice', $data[0]['Name']);
        $this->assertSame('30', $data[0]['Age']);
    }

    // -------------------------------------------------------------------------
    // headerLine offset
    // -------------------------------------------------------------------------

    public function testHeaderLineOffset(): void
    {
        // Lines 1+2 are metadata, header is line 3
        $this->write("Meta1\nMeta2\nColA;ColB\nval1;val2\n");
        $data = $this->reader(3)->readCsvData();
        $this->assertCount(1, $data);
        $this->assertSame('val1', $data[0]['ColA']);
        $this->assertSame('val2', $data[0]['ColB']);
    }

    // -------------------------------------------------------------------------
    // readMetadataLines
    // -------------------------------------------------------------------------

    public function testReadMetadataLines(): void
    {
        $this->write("Line1\nLine2\nHeader;Col\nData;Row\n");
        $meta = $this->reader(3)->readMetadataLines();
        $this->assertCount(2, $meta);
        $this->assertSame('Line1', $meta[0]);
        $this->assertSame('Line2', $meta[1]);
    }

    public function testReadMetadataLinesHeaderOnLine1IsEmpty(): void
    {
        $this->write("Name;Age\nAlice;30\n");
        $meta = $this->reader(1)->readMetadataLines();
        $this->assertSame([], $meta);
    }

    public function testReadMetadataLinesHeaderOnLine2ReturnsOneLine(): void
    {
        $this->write("MetaInfo\nName;Age\nAlice;30\n");
        $meta = $this->reader(2)->readMetadataLines();
        $this->assertCount(1, $meta);
        $this->assertSame('MetaInfo', $meta[0]);
    }

    // -------------------------------------------------------------------------
    // maxDataRows limit
    // -------------------------------------------------------------------------

    public function testMaxDataRowsLimit(): void
    {
        $this->write("Name;Age\nAlice;30\nBob;25\nCarol;20\n");
        $data = $this->reader()->readCsvData(2);
        $this->assertCount(2, $data);
        $this->assertSame('Alice', $data[0]['Name']);
        $this->assertSame('Bob', $data[1]['Name']);
    }

    public function testMaxDataRowsZeroMeansAll(): void
    {
        $this->write("Name;Age\nAlice;30\nBob;25\nCarol;20\n");
        $data = $this->reader()->readCsvData(0);
        $this->assertCount(3, $data);
    }

    // -------------------------------------------------------------------------
    // Empty line skipping
    // -------------------------------------------------------------------------

    public function testEmptyLinesAreSkipped(): void
    {
        $this->write("Name;Age\nAlice;30\n\nBob;25\n\n");
        $data = $this->reader()->readCsvData();
        $this->assertCount(2, $data);
    }

    // -------------------------------------------------------------------------
    // BOM removal
    // -------------------------------------------------------------------------

    public function testBomIsRemovedFromFirstColumnName(): void
    {
        // UTF-8 BOM followed immediately by the header
        $this->write("\xEF\xBB\xBFName;Age\nAlice;30\n");
        $data = $this->reader(1, ';', true)->readCsvData();
        $this->assertCount(1, $data);
        // The column must be 'Name', not the BOM-prefixed version
        $this->assertArrayHasKey('Name', $data[0]);
        $this->assertSame('Alice', $data[0]['Name']);
    }

    public function testNoBomFlagLeavesHeaderIntact(): void
    {
        // If hasBom is false, BOM bytes stay and the key will be mangled — we only
        // assert that the clean path (hasBom=true) works, tested above.
        // Here we just verify normal CSV without BOM also works when hasBom=false.
        $this->write("Name;Age\nAlice;30\n");
        $data = $this->reader(1, ';', false)->readCsvData();
        $this->assertArrayHasKey('Name', $data[0]);
    }

    // -------------------------------------------------------------------------
    // getHeaders
    // -------------------------------------------------------------------------

    public function testGetHeaders(): void
    {
        $this->write("Col1;Col2;Col3\nA;B;C\n");
        $headers = $this->reader()->getHeaders();
        $this->assertSame(['Col1', 'Col2', 'Col3'], $headers);
    }

    // -------------------------------------------------------------------------
    // Row with fewer columns than headers
    // -------------------------------------------------------------------------

    public function testShortRowPaddedWithEmpty(): void
    {
        $this->write("A;B;C\n1;2\n");
        $data = $this->reader()->readCsvData();
        $this->assertCount(1, $data);
        $this->assertSame('1', $data[0]['A']);
        $this->assertSame('2', $data[0]['B']);
        $this->assertSame('', $data[0]['C']);
    }

    // -------------------------------------------------------------------------
    // Error cases
    // -------------------------------------------------------------------------

    public function testFileNotFoundThrowsRuntimeException(): void
    {
        $this->expectException(\RuntimeException::class);
        $reader = new CsvReader('/nonexistent/path/file.csv', ['inputDelimiter' => ';', 'headerLine' => 1]);
        $reader->readLines();
    }

    public function testHeaderLineBeyondFileLengthThrows(): void
    {
        $this->write("Name;Age\nAlice;30\n");
        $this->expectException(\RuntimeException::class);
        $this->reader(99)->readCsvData();
    }
}