Catch array_combine errors and skip items with warning (cherry pick 16f6ca8) (#42)

Author: yann-eugone

src/batch-box-spout/src/FlatFileReader.php

@@ -14,6 +14,7 @@
 use Yokai\Batch\Job\Item\ItemReaderInterface;
 use Yokai\Batch\Job\JobExecutionAwareInterface;
 use Yokai\Batch\Job\JobExecutionAwareTrait;
+use Yokai\Batch\Warning;
 
 final class FlatFileReader implements
     ItemReaderInterface,
@@ -110,7 +111,35 @@ public function read(): iterable
                 }
 
                 if (is_array($headers)) {
-                    $row = array_combine($headers, $row);
+                    try {
+                        /** @var array<string, mixed>|false $combined */
+                        $combined = @array_combine($headers, $row);
+                        if ($combined === false) {
+                            // @codeCoverageIgnoreStart
+                            // Prior to PHP 8.0 array_combine only trigger a warning
+                            // Now it is throwing a ValueError
+                            throw new \ValueError(
+                                'array_combine(): Argument #1 ($keys) and argument #2 ($values) ' .
+                                'must have the same number of elements'
+                            );
+                            // @codeCoverageIgnoreEnd
+                        }
+                    } catch (\ValueError $exception) {
+                        $this->jobExecution->addWarning(
+                            new Warning(
+                                'Expecting row {row} to have exactly {expected} columns(s), but got {actual}.',
+                                [
+                                    '{row}' => (string)$rowIndex,
+                                    '{expected}' => (string)count($headers),
+                                    '{actual}' => (string)count($row),
+                                ],
+                                ['headers' => $headers, 'row' => $row]
+                            )
+                        );
+                        continue;
+                    }
+
+                    $row = $combined;
                 }
 
                 yield $row;

src/batch-box-spout/tests/FlatFileReaderTest.php

@@ -57,6 +57,48 @@ public function testMissingFileToRead(string $type)
         iterator_to_array($reader->read());
     }
 
+    public function testReadWrongLineSize(): void
+    {
+        $file = __DIR__ . '/fixtures/wrong-line-size.csv';
+        $jobExecution = JobExecution::createRoot('123456789', 'parent');
+        $reader = new FlatFileReader(
+            'csv',
+            [],
+            FlatFileReader::HEADERS_MODE_COMBINE,
+            null,
+            $file
+        );
+        $reader->setJobExecution($jobExecution);
+
+        /** @var \Iterator $result */
+        $result = $reader->read();
+        self::assertInstanceOf(\Iterator::class, $result);
+        self::assertSame(
+            [
+                ['firstName' => 'John', 'lastName' => 'Doe'],
+                ['firstName' => 'Jack', 'lastName' => 'Doe'],
+            ],
+            iterator_to_array($result)
+        );
+
+        self::assertSame(
+            'Expecting row {row} to have exactly {expected} columns(s), but got {actual}.',
+            $jobExecution->getWarnings()[0]->getMessage()
+        );
+        self::assertSame(
+            [
+                '{row}' => '3',
+                '{expected}' => '2',
+                '{actual}' => '3',
+            ],
+            $jobExecution->getWarnings()[0]->getParameters()
+        );
+        self::assertSame(
+            ['headers' => ['firstName', 'lastName'], 'row' => ['Jane', 'Doe', 'too much data']],
+            $jobExecution->getWarnings()[0]->getContext()
+        );
+    }
+
     public function types()
     {
         foreach ([Type::CSV, Type::XLSX, Type::ODS] as $type) {

src/batch-box-spout/tests/fixtures/wrong-line-size.csv

@@ -0,0 +1,4 @@
+firstName,lastName
+John,Doe
+Jane,Doe,too much data
+Jack,Doe