[GoogleSheet] Prevent fatal error when extracted columns don't match headers amount

composer.json

@@ -307,7 +307,8 @@
             "tools/phpunit/vendor/bin/phpunit --testsuite=adapter-elasticsearch-integration --log-junit ./var/phpunit/logs/adapter-elasticsearch-integration.junit.xml --coverage-clover=./var/phpunit/coverage/clover/adapter-elasticsearch-integration.coverage.xml --coverage-html=./var/phpunit/coverage/html/adapter-elasticsearch-integration"
         ],
         "test:adapter:google-sheet": [
-            "tools/phpunit/vendor/bin/phpunit --testsuite=adapter-google-sheet-unit --log-junit ./var/phpunit/logs/adapter-google-sheet-unit.junit.xml --coverage-clover=./var/phpunit/coverage/clover/adapter-google-sheet-unit.coverage.xml --coverage-html=./var/phpunit/coverage/html/adapter-google-sheet-unit"
+            "tools/phpunit/vendor/bin/phpunit --testsuite=adapter-google-sheet-unit --log-junit ./var/phpunit/logs/adapter-google-sheet-unit.junit.xml --coverage-clover=./var/phpunit/coverage/clover/adapter-google-sheet-unit.coverage.xml --coverage-html=./var/phpunit/coverage/html/adapter-google-sheet-unit",
+            "tools/phpunit/vendor/bin/phpunit --testsuite=adapter-google-sheet-integration --log-junit ./var/phpunit/coverage/clover/adapter-google-sheet-integration.junit.xml --coverage-clover=./var/phpunit/coverage/clover/adapter-google-sheet-integration.coverage.xml --coverage-html=./var/phpunit/coverage/html/adapter-google-sheet-integration"
         ],
         "test:adapter:http": [
             "tools/phpunit/vendor/bin/phpunit --testsuite=adapter-http-unit --log-junit ./var/phpunit/logs/adapter-http-unit.junit.xml --coverage-clover=./var/phpunit/coverage/clover/adapter-http-unit.coverage.xml --coverage-html=./var/phpunit/coverage/html/adapter-http-unit",

phpunit.xml.dist

@@ -120,6 +120,9 @@
     <testsuite name="adapter-google-sheet-unit">
       <directory>src/adapter/etl-adapter-google-sheet/tests/Flow/ETL/Adapter/GoogleSheet/Tests/Unit</directory>
     </testsuite>
+    <testsuite name="adapter-google-sheet-integration">
+      <directory>src/adapter/etl-adapter-google-sheet/tests/Flow/ETL/Adapter/GoogleSheet/Tests/Integration</directory>
+    </testsuite>
     <testsuite name="adapter-http-unit">
       <directory>src/adapter/etl-adapter-http/tests/Flow/ETL/Adapter/HTTP/Tests/Unit</directory>
     </testsuite>

src/adapter/etl-adapter-google-sheet/src/Flow/ETL/Adapter/GoogleSheet/GoogleSheetExtractor.php

@@ -14,6 +14,8 @@
 {
     use Limitable;
 
+    private bool $dropExtraColumns = true;
+
     /**
      * @param array{dateTimeRenderOption?: string, majorDimension?: string, valueRenderOption?: string} $options
      */
@@ -57,24 +59,31 @@
             $totalRows = 0;
         }
 
+        $headersCount = \count($headers);
+
         $shouldPutInputIntoRows = $context->config->shouldPutInputIntoRows();
 
         while (\count($values)) {
             $rows = \array_map(
-                function (array $rowData) use ($headers, $shouldPutInputIntoRows) {
-                    if (\count($headers) > \count($rowData)) {
+                function (array $rowData) use ($headers, $headersCount, $shouldPutInputIntoRows) {
+                    $rowDataCount = \count($rowData);
+
+                    if ($headersCount > $rowDataCount) {
                         \array_push(
                             $rowData,
                             ...\array_map(
                                 static fn (int $i) => null,
-                                \range(1, \count($headers) - \count($rowData))
+                                \range(1, $headersCount - $rowDataCount)
                             )
                         );
                     }
 
-                    if (\count($rowData) > \count($headers)) {
-                        /** @phpstan-ignore-next-line */
-                        $rowData = \array_chunk($rowData, \count($headers));
+                    if ($rowDataCount > $headersCount) {
+                        if (!$this->dropExtraColumns) {
+                            throw InvalidArgumentException::because('Row has more columns (%d) than headers (%d)', $rowDataCount, $headersCount);
+                        }
+
+                        $rowData = \array_slice($rowData, 0, $headersCount);
                     }
 
                     $row = \array_combine($headers, $rowData);
@@ -105,7 +114,7 @@
             }
 
             $cellsRange = $cellsRange->nextRows($this->rowsPerPage);
-            /** @var Sheets\ValueRange $response */
+
             $response = $this->service->spreadsheets_values->get($this->spreadsheetId, $cellsRange->toString(), $this->options);
             /**
              * @var array<array> $values
@@ -114,6 +123,13 @@
         }
     }
 
+    public function withDropExtraColumns(bool $dropExtraColumns) : self
+    {
+        $this->dropExtraColumns = $dropExtraColumns;
+
+        return $this;
+    }
+
     public function withHeader(bool $withHeader) : self
     {
         $this->withHeader = $withHeader;

src/adapter/etl-adapter-google-sheet/src/Flow/ETL/Adapter/GoogleSheet/functions.php

@@ -5,7 +5,6 @@
 namespace Flow\ETL\Adapter\GoogleSheet;
 
 use Flow\ETL\Attribute\{DocumentationDSL, Module, Type};
-use Flow\ETL\Extractor;
 use Google\Client;
 use Google\Service\Sheets;
 
@@ -25,7 +24,7 @@ function from_google_sheet(
     bool $with_header = true,
     int $rows_per_page = 1000,
     array $options = [],
-) : Extractor {
+) : GoogleSheetExtractor {
     if ($auth_config instanceof Sheets) {
         $sheets = $auth_config;
     } else {
@@ -64,7 +63,7 @@ function from_google_sheet_columns(
     bool $with_header = true,
     int $rows_per_page = 1000,
     array $options = [],
-) : Extractor {
+) : GoogleSheetExtractor {
     if ($auth_config instanceof Sheets) {
         $sheets = $auth_config;
     } else {

src/adapter/etl-adapter-google-sheet/tests/Flow/ETL/Adapter/GoogleSheet/Tests/Fixtures/extra-columns.json

@@ -0,0 +1,17 @@
+{
+  "range": "Sheet!A1:D11",
+  "majorDimension": "ROWS",
+  "values": [
+    ["Header 1", "Header 2", "Header 3"],
+    ["A2", "B2", "C2"],
+    ["A3", "B3", "C3"],
+    ["A4", "B4", "C4"],
+    ["A5", "B5", "C5"],
+    ["A6", "B6", "C6"],
+    ["A7", "B7", "C7"],
+    ["A8", "B8", "C8"],
+    ["A9", "B9", "C9"],
+    ["A10", "B10", "C10"],
+    ["A11", "B11", "C11", "D11"]
+  ]
+}

src/adapter/etl-adapter-google-sheet/tests/Flow/ETL/Adapter/GoogleSheet/Tests/GoogleSheetsContext.php

@@ -0,0 +1,47 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flow\ETL\Adapter\GoogleSheet\Tests;
+
+use Google\Client as GoogleClient;
+use Google\Service\Sheets;
+use GuzzleHttp\{Client as HttpClient, HandlerStack};
+use GuzzleHttp\Handler\MockHandler;
+use GuzzleHttp\Psr7\Response;
+
+final readonly class GoogleSheetsContext
+{
+    private GoogleClient $client;
+
+    public function __construct()
+    {
+        $this->client = new GoogleClient();
+    }
+
+    public function sheets(string $fixtureFile) : Sheets
+    {
+        $this->client->setHttpClient($this->createHttpClient($fixtureFile));
+
+        return new Sheets($this->client);
+    }
+
+    private function createHttpClient(string $fixtureFile) : HttpClient
+    {
+        return new HttpClient(
+            [
+                'handler' => HandlerStack::create(
+                    new MockHandler(
+                        [
+                            new Response(
+                                200,
+                                ['Content-Type' => 'application/json'],
+                                file_get_contents($fixtureFile) ?: throw new \RuntimeException('Failed to read file: ' . $fixtureFile)
+                            ),
+                        ]
+                    )
+                ),
+            ]
+        );
+    }
+}

src/adapter/etl-adapter-google-sheet/tests/Flow/ETL/Adapter/GoogleSheet/Tests/Integration/GoogleSheetExtractorTest.php

@@ -0,0 +1,54 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flow\ETL\Adapter\GoogleSheet\Tests\Integration;
+
+use function Flow\ETL\DSL\{config, flow_context};
+use Flow\ETL\Adapter\GoogleSheet\{Columns, GoogleSheetExtractor, Tests\GoogleSheetsContext};
+use Flow\ETL\Exception\InvalidArgumentException;
+use Flow\ETL\Tests\FlowTestCase;
+
+final class GoogleSheetExtractorTest extends FlowTestCase
+{
+    private GoogleSheetsContext $context;
+
+    protected function setUp() : void
+    {
+        $this->context = new GoogleSheetsContext();
+    }
+
+    public function test_extract_with_cut_extra_columns() : void
+    {
+        $extractor = new GoogleSheetExtractor(
+            $this->context->sheets(__DIR__ . '/../Fixtures/extra-columns.json'),
+            '1234567890',
+            new Columns('Sheet', 'A', 'Z'),
+        );
+
+        $rows = $extractor->extract(flow_context(config()));
+
+        foreach ($rows as $row) {
+            self::assertNotNull($row);
+        }
+    }
+
+    public function test_extract_without_cut_extra_columns() : void
+    {
+        $extractor = new GoogleSheetExtractor(
+            $this->context->sheets(__DIR__ . '/../Fixtures/extra-columns.json'),
+            '1234567890',
+            new Columns('Sheet', 'A', 'Z'),
+        );
+        $extractor->withDropExtraColumns(false);
+
+        $rows = $extractor->extract(flow_context(config()));
+
+        $this->expectException(InvalidArgumentException::class);
+        $this->expectExceptionMessage('Row has more columns (4) than headers (3)');
+
+        foreach ($rows as $row) {
+            self::assertNotNull($row);
+        }
+    }
+}

src/adapter/etl-adapter-google-sheet/tests/Flow/ETL/Adapter/GoogleSheet/Tests/Unit/GoogleSheetExtractorTest.php

@@ -23,9 +23,8 @@ public function test_its_stop_fetching_data_if_processed_row_count_is_less_then_
             $sheetName = 'sheet',
             'A',
             'B',
-            true,
-            2,
-        );
+        )->withHeader(true)
+            ->withRowsPerPage(2);
         $spreadSheetIdEntry = string_entry('_spread_sheet_id', $spreadSheetId);
         $sheetNameEntry = string_entry('_sheet_name', $sheetName);
         $firstValueRangeMock = $this->createMock(ValueRange::class);
@@ -63,9 +62,7 @@ public function test_rows_in_batch_must_be_positive_integer() : void
             'sheet',
             'A',
             'B',
-            true,
-            0
-        );
+        )->withRowsPerPage(0);
     }
 
     public function test_works_for_no_data() : void
@@ -76,9 +73,8 @@ public function test_works_for_no_data() : void
             'sheet',
             'A',
             'B',
-            true,
-            20
-        );
+        )->withHeader(true)
+            ->withRowsPerPage(20);
         $ValueRangeMock = $this->createMock(ValueRange::class);
         $ValueRangeMock->method('getValues')->willReturn(null);