From 5fbed42f76c324a313c09b5c92d6d6a9b0edc664 Mon Sep 17 00:00:00 2001
From: Mathieu De Keyzer <me@theus.be>
Date: Sat, 9 Sep 2023 00:51:28 +0200
Subject: [PATCH] feat: file encoding for csv reader

---
 EMS/common-bundle/src/Common/File/FileReader.php          | 8 +++++++-
 .../src/Contracts/File/FileReaderInterface.php            | 2 +-
 .../src/Command/FileReader/FileReaderImportCommand.php    | 6 +++++-
 3 files changed, 13 insertions(+), 3 deletions(-)

diff --git a/EMS/common-bundle/src/Common/File/FileReader.php b/EMS/common-bundle/src/Common/File/FileReader.php
index e2aa902f1..a5bd958dc 100644
--- a/EMS/common-bundle/src/Common/File/FileReader.php
+++ b/EMS/common-bundle/src/Common/File/FileReader.php
@@ -6,15 +6,21 @@
 
 use EMS\CommonBundle\Contracts\File\FileReaderInterface;
 use PhpOffice\PhpSpreadsheet\IOFactory;
+use PhpOffice\PhpSpreadsheet\Reader\Csv;
+use PhpOffice\PhpSpreadsheet\Reader\Html;
+use PhpOffice\PhpSpreadsheet\Reader\Slk;
 
 final class FileReader implements FileReaderInterface
 {
     /**
      * {@inheritDoc}
      */
-    public function getData(string $filename, bool $skipFirstRow = false): array
+    public function getData(string $filename, bool $skipFirstRow = false, string $encoding = null): array
     {
         $reader = IOFactory::createReaderForFile($filename);
+        if (($reader instanceof Csv || $reader instanceof Html || $reader instanceof Slk) && null !== $encoding) {
+            $reader->setInputEncoding($encoding);
+        }
 
         $data = $reader->load($filename)->getActiveSheet()->toArray();
 
diff --git a/EMS/common-bundle/src/Contracts/File/FileReaderInterface.php b/EMS/common-bundle/src/Contracts/File/FileReaderInterface.php
index fed2c0c47..8ee982038 100644
--- a/EMS/common-bundle/src/Contracts/File/FileReaderInterface.php
+++ b/EMS/common-bundle/src/Contracts/File/FileReaderInterface.php
@@ -9,5 +9,5 @@ interface FileReaderInterface
     /**
      * @return array<mixed>
      */
-    public function getData(string $filename, bool $skipFirstRow = false): array;
+    public function getData(string $filename, bool $skipFirstRow = false, string $encoding = null): array;
 }
diff --git a/elasticms-cli/src/Command/FileReader/FileReaderImportCommand.php b/elasticms-cli/src/Command/FileReader/FileReaderImportCommand.php
index f3f886887..80e5046f5 100644
--- a/elasticms-cli/src/Command/FileReader/FileReaderImportCommand.php
+++ b/elasticms-cli/src/Command/FileReader/FileReaderImportCommand.php
@@ -26,6 +26,7 @@ final class FileReaderImportCommand extends AbstractCommand
     private const OPTION_GENERATE_HASH = 'generate-hash';
     private const OPTION_DELETE_MISSING_DOCUMENTS = 'delete-missing-document';
     private const OPTION_HASH_FILE = 'hash-file';
+    private const OPTION_ENCODING = 'encoding';
     private string $ouuidExpression;
     private string $contentType;
     private string $file;
@@ -33,6 +34,7 @@ final class FileReaderImportCommand extends AbstractCommand
     private bool $hashOuuid;
     private bool $deleteMissingDocuments;
     private bool $hashFile;
+    private ?string $encoding;
 
     public function __construct(private readonly AdminHelper $adminHelper, private readonly FileReaderInterface $fileReader)
     {
@@ -50,6 +52,7 @@ protected function configure(): void
             ->addOption(self::OPTION_DELETE_MISSING_DOCUMENTS, null, InputOption::VALUE_NONE, 'The command will delete content type document that are missing in the import file')
             ->addOption(self::OPTION_OUUID_EXPRESSION, null, InputOption::VALUE_OPTIONAL, 'Expression language apply to excel rows in order to identify the document by its ouuid. If equal to null new document will be created', "row['ouuid']")
             ->addOption(self::OPTION_HASH_FILE, null, InputOption::VALUE_NONE, 'Specify that the file argument is a file hash not a file path.')
+            ->addOption(self::OPTION_ENCODING, null, InputOption::VALUE_OPTIONAL, 'Specify the file\'s encoding for csv, html and Slk file')
         ;
     }
 
@@ -63,6 +66,7 @@ protected function initialize(InputInterface $input, OutputInterface $output): v
         $this->hashOuuid = $this->getOptionBool(self::OPTION_GENERATE_HASH);
         $this->deleteMissingDocuments = $this->getOptionBool(self::OPTION_DELETE_MISSING_DOCUMENTS);
         $this->hashFile = $this->getOptionBool(self::OPTION_HASH_FILE);
+        $this->encoding = $this->getOptionStringNull(self::OPTION_ENCODING);
     }
 
     protected function execute(InputInterface $input, OutputInterface $output): int
@@ -79,7 +83,7 @@ protected function execute(InputInterface $input, OutputInterface $output): int
         $file = $this->hashFile ? $this->getFileByHash($this->file) : $this->file;
 
         $expressionLanguage = new ExpressionLanguage();
-        $rows = $this->fileReader->getData($file);
+        $rows = $this->fileReader->getData($file, false, $this->encoding);
         $header = $rows[0] ?? [];
 
         $ouuids = [];