From a93b81bda36e616a855b33b54f1c78d47cf93d75 Mon Sep 17 00:00:00 2001
From: Luka Trovic <luka@nextcloud.com>
Date: Thu, 22 Feb 2024 14:02:58 +0100
Subject: [PATCH] feat: autodetect data type during import

---
 lib/Service/ColumnService.php | 25 +++++++++-
 lib/Service/ImportService.php | 87 +++++++++++++++++++++++++++++++----
 2 files changed, 100 insertions(+), 12 deletions(-)

diff --git a/lib/Service/ColumnService.php b/lib/Service/ColumnService.php
index 7bd41eaec..b65c3e532 100644
--- a/lib/Service/ColumnService.php
+++ b/lib/Service/ColumnService.php
@@ -523,8 +523,29 @@ public function findOrCreateColumnsByTitleForTableAsArray(?int $tableId, ?int $v
 			// if column was not found
 			if($result[$i] === '' && $createUnknownColumns) {
 				$description = $this->l->t('This column was automatically created by the import service.');
-				$subType = $dataTypes[$i] === 'text' ? 'line' : '';
-				$result[$i] = $this->create($userId, $tableId, $viewId, $dataTypes[$i], $subType, $title, false, $description, null, null, null, null, null, null, null, null, null, null, null, null, []);
+				$result[$i] = $this->create(
+					$userId,
+					$tableId,
+					$viewId,
+					$dataTypes[$i]['type'],
+					$dataTypes[$i]['subtype'] ?? '',
+					$title,
+					false,
+					$description,
+					null,
+					null,
+					null,
+					$dataTypes[$i]['number_prefix'] ?? null,
+					$dataTypes[$i]['number_suffix'] ?? null,
+					null,
+					null,
+					null,
+					$dataTypes[$i]['number_decimals'] ?? null,
+					null,
+					$dataTypes[$i]['selection_default'] ?? null,
+					null,
+					[]
+				);
 				$countCreatedColumns++;
 			}
 		}
diff --git a/lib/Service/ImportService.php b/lib/Service/ImportService.php
index adc22461b..a2e8efd5c 100644
--- a/lib/Service/ImportService.php
+++ b/lib/Service/ImportService.php
@@ -17,6 +17,8 @@
 use OCP\Files\NotPermittedException;
 use OCP\IUserManager;
 use OCP\Server;
+use PhpOffice\PhpSpreadsheet\Cell\Cell;
+use PhpOffice\PhpSpreadsheet\Cell\DataType;
 use PhpOffice\PhpSpreadsheet\IOFactory;
 use PhpOffice\PhpSpreadsheet\Shared\Date;
 use PhpOffice\PhpSpreadsheet\Worksheet\Row;
@@ -201,10 +203,13 @@ private function createRow(Row $row): void {
 				continue;
 			}
 
+			/** @var Column $column */
+			$column = $this->columns[$i];
+
 			// if cell is empty
 			if(!$cell || $cell->getValue() === null) {
 				$this->logger->info('Cell is empty while fetching rows data for importing.');
-				if($this->columns[$i]->getMandatory()) {
+				if($column->getMandatory()) {
 					$this->logger->warning('Mandatory column was not set');
 					$this->countErrors++;
 					return;
@@ -213,8 +218,12 @@ private function createRow(Row $row): void {
 			}
 
 			$value = $cell->getValue();
-			if ($this->columns[$i]->getType() === 'datetime') {
+			if ($column->getType() === 'datetime') {
 				$value = Date::excelToDateTimeObject($value)->format('Y-m-d H:i');
+			} elseif ($column->getType() === 'number' && $column->getNumberSuffix() === '%') {
+				$value = $value * 100;
+			} elseif ($column->getType() === 'selection' && $column->getSubtype() === 'check') {
+				$value = $cell->getFormattedValue() === 'TRUE' ? 'true' : 'false';
 			}
 
 			$data[] = [
@@ -255,16 +264,9 @@ private function getColumns(Row $firstRow, Row $secondRow): void {
 		foreach ($cellIterator as $cell) {
 			if ($cell && $cell->getValue() !== null && $cell->getValue() !== '') {
 				$titles[] = $cell->getValue();
-				$originDataType = $secondRowCellIterator->current()->getDataType();
 
 				// Convert data type to our data type
-				if (Date::isDateTime($secondRowCellIterator->current()) || $originDataType === 'd') {
-					$dataTypes[] = 'datetime';
-				} elseif ($originDataType === 'n') {
-					$dataTypes[] = 'number';
-				} else {
-					$dataTypes[] = 'text';
-				}
+				$dataTypes[] = $this->parseColumnDataType($secondRowCellIterator->current());
 			} else {
 				$this->logger->debug('No cell given or cellValue is empty while loading columns for importing');
 				$this->countErrors++;
@@ -278,4 +280,69 @@ private function getColumns(Row $firstRow, Row $secondRow): void {
 		}
 	}
 
+	private function parseColumnDataType(Cell $cell): array {
+		$originDataType = $cell->getDataType();
+		$value = $cell->getValue();
+		$formattedValue = $cell->getFormattedValue();
+		$dataType = [
+			'type' => 'text',
+			'subtype' => 'line',
+		];
+
+		if (Date::isDateTime($cell) || $originDataType === DataType::TYPE_ISO_DATE) {
+			$dataType = [
+				'type' => 'datetime',
+			];
+		} elseif ($originDataType === DataType::TYPE_NUMERIC) {
+			if (str_contains($formattedValue, '%')) {
+				$dataType = [
+					'type' => 'number',
+					'number_decimals' => 2,
+					'number_suffix' => '%',
+				];
+			} elseif (str_contains($formattedValue, '€')) {
+				$dataType = [
+					'type' => 'number',
+					'number_decimals' => 2,
+					'number_suffix' => '€',
+				];
+			} elseif (str_contains($formattedValue, 'EUR')) {
+				$dataType = [
+					'type' => 'number',
+					'number_decimals' => 2,
+					'number_suffix' => 'EUR',
+				];
+			} elseif (str_contains($formattedValue, '$')) {
+				$dataType = [
+					'type' => 'number',
+					'number_decimals' => 2,
+					'number_prefix' => '$',
+				];
+			} elseif (str_contains($formattedValue, 'USD')) {
+				$dataType = [
+					'type' => 'number',
+					'number_decimals' => 2,
+					'number_suffix' => 'USD',
+				];
+			} elseif (is_float($value)) {
+				$decimals = strlen(substr(strrchr($value, "."), 1));
+				$dataType = [
+					'type' => 'number',
+					'number_decimals' => $decimals,
+				];
+			} else {
+				$dataType = [
+					'type' => 'number',
+				];
+			}
+		} elseif ($originDataType === DataType::TYPE_BOOL) {
+			$dataType = [
+				'type' => 'selection',
+				'subtype' => 'check',
+				'selection_default' => 'false',
+			];
+		}
+
+		return $dataType;
+	}
 }