Simplify Excel & GoogleSheet extractor cell expanding (#1661)

Author: stloyd

src/adapter/etl-adapter-excel/src/Flow/ETL/Adapter/Excel/ExcelExtractor.php

@@ -120,28 +120,17 @@ public function withSheetName(string $sheetName) : self
         return $this;
     }
 
-    private function createRowsFromCells(Row $row) : array
+    private function createRowsFromCells(Row $row, int $previousRowDataCount = 0) : array
     {
-        return \array_map(
-            fn (Cell $cell) =>
-                // Convert empty values to nullables if allowed
-                $this->convertEmptyToNull && '' === $cell->getValue() ? null : $cell->getValue(),
+        $rowData = \array_map(
+            // Convert empty values to nullables if allowed
+            fn (Cell $cell) => $this->convertEmptyToNull && '' === $cell->getValue() ? null : $cell->getValue(),
             $row->getCells()
         );
-    }
 
-    private function extendRowData(int $headersCount, array $rowData) : array
-    {
-        $rowDataCount = \count($rowData);
-
-        if ($headersCount > $rowDataCount) {
-            \array_push(
-                $rowData,
-                ...\array_map(
-                    static fn (int $i) => null,
-                    \range(1, $headersCount - $rowDataCount)
-                )
-            );
+        // Expand columns to the size of the previous row
+        for ($i = \count($rowData); $i < $previousRowDataCount; $i++) {
+            $rowData[$i] = null;
         }
 
         return $rowData;
@@ -172,7 +161,7 @@ private function extractRows(SourceStream $stream, array $headers, int $offset)
                 }
 
                 // ODS format reader skips empty cells when reading rows
-                $rowData = $this->extendRowData($previousRowDataCount, $this->createRowsFromCells($row));
+                $rowData = $this->createRowsFromCells($row, $previousRowDataCount);
                 $previousRowDataCount = \count($rowData);
 
                 if ($this->withHeader) {

src/adapter/etl-adapter-excel/tests/Flow/ETL/Adapter/Excel/Tests/Fixtures/nullable_fixture.ods

@@ -20,6 +20,12 @@ public static function provide_fixtures() : iterable
         yield 'xlsx' => [__DIR__ . '/../Fixtures/fixture.xlsx'];
     }
 
+    public static function provide_nullable_fixtures() : iterable
+    {
+        yield 'ods' => [__DIR__ . '/../Fixtures/nullable_fixture.ods'];
+        yield 'xlsx' => [__DIR__ . '/../Fixtures/nullable_fixture.xlsx'];
+    }
+
     #[DataProvider('provide_fixtures')]
     public function test_extract_excel_file(string $fixtureName) : void
     {
@@ -165,6 +171,24 @@ public function test_extract_excel_file_without_header(string $fixtureName) : vo
         self::assertSame(10, $total);
     }
 
+    #[DataProvider('provide_nullable_fixtures')]
+    public function test_extract_excel_nullable_file(string $fixtureName) : void
+    {
+        $extractor = from_excel($fixtureName);
+
+        $total = 0;
+
+        foreach ($extractor->extract(flow_context(config())) as $rows) {
+            $rows->each(function (Row $row) : void {
+                $this->assertSame(['id', 'name', 'email'], \array_keys($row->toArray()));
+                $this->assertCount(3, $row->toArray());
+            });
+            $total += $rows->count();
+        }
+
+        self::assertSame(5, $total);
+    }
+
     public function test_extract_with_wrongly_selected_reader() : void
     {
         $extractor = from_excel(__DIR__ . '/../Fixtures/fixture.xlsx');

src/adapter/etl-adapter-excel/tests/Flow/ETL/Adapter/Excel/Tests/Fixtures/nullable_fixture.xlsx

@@ -68,14 +68,9 @@ public function extract(FlowContext $context) : \Generator
                 function (array $rowData) use ($headers, $headersCount, $shouldPutInputIntoRows) {
                     $rowDataCount = \count($rowData);
 
-                    if ($headersCount > $rowDataCount) {
-                        \array_push(
-                            $rowData,
-                            ...\array_map(
-                                static fn (int $i) => null,
-                                \range(1, $headersCount - $rowDataCount)
-                            )
-                        );
+                    // Expand columns to the size of the previous row
+                    for ($i = $rowDataCount; $i < $headersCount; $i++) {
+                        $rowData[$i] = null;
                     }
 
                     if ($rowDataCount > $headersCount) {