From d7da839c5f7545490712647eed55067fc6a68a70 Mon Sep 17 00:00:00 2001
From: Beno!t POLASZEK <bpolaszek@users.noreply.github.com>
Date: Thu, 21 Dec 2023 12:04:27 +0100
Subject: [PATCH] Feat(CSV): Add a `skipFirstRow` option (#56)

---
 src/Iterator/CSVIterator.php            | 31 +++++++++++++++++++++----
 tests/Unit/Iterator/CSVIteratorTest.php | 30 ++++++++++++++++++++++++
 2 files changed, 56 insertions(+), 5 deletions(-)

diff --git a/src/Iterator/CSVIterator.php b/src/Iterator/CSVIterator.php
index 8b2dd09..d0ec0df 100644
--- a/src/Iterator/CSVIterator.php
+++ b/src/Iterator/CSVIterator.php
@@ -28,13 +28,27 @@
 final readonly class CSVIterator implements IteratorAggregate
 {
     /**
-     * @var array{delimiter: string, enclosure: string, escapeString: string, columns: 'auto'|string[]|null, normalizers: ValueNormalizerInterface[]}
+     * @var array{
+     *     delimiter: string,
+     *     enclosure: string,
+     *     escapeString: string,
+     *     columns: 'auto'|string[]|null,
+     *     normalizers: ValueNormalizerInterface[],
+     *     skipFirstRow: bool,
+     * }
      */
     private array $options;
 
     /**
-     * @param Traversable<string>                                                                                                                            $text
-     * @param array{delimiter?: string, enclosure?: string, escapeString?: string, columns?: 'auto'|string[]|null, normalizers?: ValueNormalizerInterface[]} $options
+     * @param Traversable<string> $text
+     * @param array{
+     *     delimiter?: string,
+     *     enclosure?: string,
+     *     escapeString?: string,
+     *     columns?: 'auto'|string[]|null,
+     *     normalizers?: ValueNormalizerInterface[],
+     *     skipFirstRow?: bool,
+     * } $options
      */
     public function __construct(
         private Traversable $text,
@@ -50,6 +64,7 @@ public function __construct(
                 new NumericStringToNumberNormalizer(),
                 new EmptyStringToNullNormalizer(),
             ],
+            'skipFirstRow' => false,
         ]);
         $resolver->setAllowedTypes('delimiter', 'string');
         $resolver->setAllowedTypes('enclosure', 'string');
@@ -59,6 +74,7 @@ public function __construct(
         $resolver->setAllowedValues('columns', function (array|string|null $value) {
             return 'auto' === $value || null === $value || is_array($value);
         });
+        $resolver->setAllowedTypes('skipFirstRow', 'bool');
         $this->options = $resolver->resolve($options);
     }
 
@@ -92,6 +108,11 @@ public function getIterator(): Traversable
         return $this->iterateFromContent($this->text);
     }
 
+    private function shouldSkipFirstRow(): bool
+    {
+        return $this->options['skipFirstRow'] || 'auto' === $this->options['columns'];
+    }
+
     /**
      * @return Traversable<mixed>
      */
@@ -112,7 +133,7 @@ private function iterateFromFile(SplFileObject $file): Traversable
             if ([null] === $fields) {
                 continue;
             }
-            if ('auto' === $this->options['columns'] && 0 === $file->key()) {
+            if (0 === $file->key() && $this->shouldSkipFirstRow()) {
                 $columns ??= $fields;
                 continue;
             }
@@ -139,7 +160,7 @@ private function iterateFromContent(Traversable $content): Traversable
                 $this->options['enclosure'],
                 $this->options['escapeString'],
             );
-            if ('auto' === $this->options['columns'] && 0 === $r) {
+            if (0 === $r && $this->shouldSkipFirstRow()) {
                 $columns ??= $fields;
                 continue;
             }
diff --git a/tests/Unit/Iterator/CSVIteratorTest.php b/tests/Unit/Iterator/CSVIteratorTest.php
index 50c53f7..a1a6358 100644
--- a/tests/Unit/Iterator/CSVIteratorTest.php
+++ b/tests/Unit/Iterator/CSVIteratorTest.php
@@ -90,6 +90,36 @@
     yield 'file' => new CSVIterator(new SplFileObject($filename), ['columns' => $columns]);
 });
 
+it('skips the 1st row when asked to', function (CSVIterator $iterator) {
+    $rows = [...$iterator];
+
+    expect($rows[0])->toBe([
+        'cityEnglishName' => 'New York',
+        'cityLocalName' => 'New York',
+        'countryIsoCode' => 'US',
+        'continent' => 'North America',
+        'population' => 8537673,
+    ])
+        ->and($rows[2])->toBe([
+            'cityEnglishName' => 'Tokyo',
+            'cityLocalName' => '東京',
+            'countryIsoCode' => 'JP',
+            'continent' => 'Asia',
+            'population' => 13929286,
+        ]);
+})->with(function () {
+    $columns = [
+        'cityEnglishName',
+        'cityLocalName',
+        'countryIsoCode',
+        'continent',
+        'population',
+    ];
+    $filename = dirname(__DIR__, 2).'/Data/10-biggest-cities.csv';
+    yield 'string content' => new CSVIterator(new StrTokIterator(file_get_contents($filename)), ['columns' => $columns, 'skipFirstRow' => true]);
+    yield 'file' => new CSVIterator(new SplFileObject($filename), ['columns' => $columns, 'skipFirstRow' => true]);
+});
+
 it('adds fields when the row has not enough columns', function (CSVIterator $iterator) {
     $rows = [...$iterator];