fix: spool parquet archives before upload

app/Console/Commands/BackfillBusinessDataColumnar.php

@@ -13,6 +13,7 @@ use Modules\EcommerceFinance\Services\BusinessDataPlacement;
 use Modules\EcommerceFinance\Services\BusinessMirrorDatasetRegistry;
 use Modules\EcommerceFinance\Services\ClickHouseClient;
 use Modules\EcommerceFinance\Services\ClickHouseSchemaManager;
+use Psr\Http\Message\StreamInterface;
 use RuntimeException;
 use Throwable;
 
@@ -224,7 +225,7 @@ class BackfillBusinessDataColumnar extends Command
                 continue;
             }
             $stream = $columnar->parquetPartition($dataset, $generation, $partition);
-            $resource = StreamWrapper::getResource($stream);
+            $resource = $this->seekableResource($stream);
             try {
                 $disk->writeStream($objectKey, $resource);
             } finally {
@@ -254,4 +255,28 @@ class BackfillBusinessDataColumnar extends Command
 
         $this->info($datasetKey . '：32 个 Parquet 哈希分区已写入对象存储并登记清单；MariaDB 原表仍保留。');
     }
+
+    /** @return resource */
+    private function seekableResource(StreamInterface $stream): mixed
+    {
+        $source = StreamWrapper::getResource($stream);
+        $target = tmpfile();
+        if ($target === false) {
+            fclose($source);
+            throw new RuntimeException('business_data_archive_temp_file_unavailable');
+        }
+
+        try {
+            if (stream_copy_to_stream($source, $target) === false || ! rewind($target)) {
+                throw new RuntimeException('business_data_archive_spool_failed');
+            }
+        } catch (Throwable $exception) {
+            fclose($target);
+            throw $exception;
+        } finally {
+            fclose($source);
+        }
+
+        return $target;
+    }
 }

tests/Unit/Console/BusinessDataArchiveStreamTest.php

@@ -0,0 +1,30 @@
+<?php
+
+namespace Tests\Unit\Console;
+
+use App\Console\Commands\BackfillBusinessDataColumnar;
+use GuzzleHttp\Psr7\PumpStream;
+use PHPUnit\Framework\TestCase;
+use ReflectionMethod;
+
+class BusinessDataArchiveStreamTest extends TestCase
+{
+    public function test_non_seekable_parquet_stream_is_spooled_to_a_seekable_resource(): void
+    {
+        $chunks = ['PAR1-', 'fixture'];
+        $stream = new PumpStream(static function () use (&$chunks): string|false {
+            return array_shift($chunks) ?? false;
+        });
+        $this->assertFalse($stream->isSeekable());
+
+        $method = new ReflectionMethod(BackfillBusinessDataColumnar::class, 'seekableResource');
+        $resource = $method->invoke(new BackfillBusinessDataColumnar(), $stream);
+
+        try {
+            $this->assertTrue(stream_get_meta_data($resource)['seekable']);
+            $this->assertSame('PAR1-fixture', stream_get_contents($resource));
+        } finally {
+            fclose($resource);
+        }
+    }
+}