Changes for the OpenSpout Excel export (#356)

* Truncate the value when it is larger than the Excel cell limit
* Do not wrap text
* Decode HTML encoded characters in the value

src/Exporter/Excel/ExcelOpenSpoutExporter.php

@@ -13,6 +13,7 @@
 namespace Omines\DataTablesBundle\Exporter\Excel;
 
 use Omines\DataTablesBundle\Exporter\DataTableExporterInterface;
+use OpenSpout\Common\Entity\Cell;
 use OpenSpout\Common\Entity\Row;
 use OpenSpout\Common\Entity\Style\Style;
 use OpenSpout\Writer\AutoFilter;
@@ -28,28 +29,61 @@ public function export(array $columnNames, \Iterator $data): \SplFileInfo
     {
         $filePath = sys_get_temp_dir() . '/' . uniqid('dt') . '.xlsx';
 
-        // Header
-        $rows = [Row::fromValues($columnNames, (new Style())->setFontBold())];
+        // Style definitions
+        $noWrapTextStyle = (new Style())->setShouldWrapText(false);
+        $boldStyle = (new Style())->setFontBold();
 
-        // Data
-        foreach ($data as $row) {
-            // Remove HTML tags
-            $values = array_map('strip_tags', $row);
-            $rows[] = Row::fromValues($values);
-        }
-
-        // Write rows
         $writer = new Writer();
         $writer->openToFile($filePath);
-        $writer->addRows($rows);
+
+        // Add header
+        $writer->addRow(Row::fromValues($columnNames, $boldStyle));
+
+        $truncated = false;
+        $maxCharactersPerCell = 32767;  // E.g. https://support.microsoft.com/en-us/office/excel-specifications-and-limits-1672b34d-7043-467e-8e27-269d656771c3
+        $rowCount = 0;
+
+        foreach ($data as $rowValues) {
+            $row = new Row([]);
+            foreach ($rowValues as $value) {
+                // I assume that $value is always a string
+
+                // The data that we get may contain rich HTML. But OpenSpout does not support this.
+                // We just strip all HTML tags and unescape the remaining text.
+                $value = htmlspecialchars_decode(strip_tags($value), ENT_QUOTES | ENT_SUBSTITUTE);
+
+                // Excel has a limit of 32,767 characters per cell
+                if (mb_strlen($value) > $maxCharactersPerCell) {
+                    $truncated = true;
+                    $value = mb_substr($value, 0, $maxCharactersPerCell);
+                }
+
+                // Do not wrap text
+                $row->addCell(Cell::fromValue($value, $noWrapTextStyle));
+            }
+            $writer->addRow($row);
+            ++$rowCount;
+        }
 
         // Sheet configuration (AutoFilter, freeze row, better column width)
         $sheet = $writer->getCurrentSheet();
         $sheet->setAutoFilter(new AutoFilter(0, 1,
-            max(count($columnNames) - 1, 0), max(count($rows), 1)));
+            max(count($columnNames) - 1, 0), $rowCount + 1));
         $sheet->setSheetView((new SheetView())->setFreezeRow(2));
         $sheet->setColumnWidthForRange(24, 1, max(count($columnNames), 1));
 
+        if ($truncated) {
+            // Add a notice to the sheet if there is truncated data.
+            //
+            // TODO: when the user opens the XLSX, it will open at the first sheet, not at this notice sheet.
+            //  Thus the user won't see the notice immediately.
+            //  This needs to have a better solution.
+            $writer
+                ->addNewSheetAndMakeItCurrent()
+                ->setName('Notice');
+            $writer->addRow(Row::fromValues(['Some cell values were too long! They were truncated to fit the 32,767 character limit.'], $boldStyle));
+        }
+
         $writer->close();
 
         return new \SplFileInfo($filePath);

tests/Fixtures/AppBundle/Controller/ExporterController.php

@@ -13,6 +13,7 @@
 namespace Tests\Fixtures\AppBundle\Controller;
 
 use Doctrine\ORM\QueryBuilder;
+use Omines\DataTablesBundle\Adapter\ArrayAdapter;
 use Omines\DataTablesBundle\Adapter\Doctrine\ORMAdapter;
 use Omines\DataTablesBundle\Column\TextColumn;
 use Omines\DataTablesBundle\DataTableFactory;
@@ -100,4 +101,62 @@ public function exportEmptyDataTableAction(Request $request, DataTableFactory $d
             'datatable' => $table,
         ]);
     }
+
+    /**
+     * This route returns data which does not fit in an Excel cell (cells have a character limit of 32767).
+     */
+    public function exportLongText(Request $request, DataTableFactory $dataTableFactory): Response
+    {
+        $longText = str_repeat('a', 40000);
+
+        $table = $dataTableFactory
+            ->create()
+            ->add('longText', TextColumn::class)
+            ->createAdapter(ArrayAdapter::class, [
+                ['longText' => $longText],
+            ])
+            ->addEventListener(DataTableExporterEvents::PRE_RESPONSE, function (DataTableExporterResponseEvent $e) {
+                $response = $e->getResponse();
+                $response->deleteFileAfterSend(false);
+                $ext = $response->getFile()->getExtension();
+                $response->setContentDisposition(ResponseHeaderBag::DISPOSITION_ATTACHMENT, 'custom_filename.' . $ext);
+            })
+            ->handleRequest($request);
+
+        if ($table->isCallback()) {
+            return $table->getResponse();
+        }
+
+        return $this->render('@App/exporter.html.twig', [
+            'datatable' => $table,
+        ]);
+    }
+
+    /**
+     * This route returns data with HTML special characters.
+     */
+    public function exportSpecialChars(Request $request, DataTableFactory $dataTableFactory): Response
+    {
+        $table = $dataTableFactory
+            ->create()
+            ->add('specialChars', TextColumn::class)
+            ->createAdapter(ArrayAdapter::class, [
+                ['specialChars' => '<?xml version="1.0" encoding="UTF-8"?><hello>World</hello>'],
+            ])
+            ->addEventListener(DataTableExporterEvents::PRE_RESPONSE, function (DataTableExporterResponseEvent $e) {
+                $response = $e->getResponse();
+                $response->deleteFileAfterSend(false);
+                $ext = $response->getFile()->getExtension();
+                $response->setContentDisposition(ResponseHeaderBag::DISPOSITION_ATTACHMENT, 'custom_filename.' . $ext);
+            })
+            ->handleRequest($request);
+
+        if ($table->isCallback()) {
+            return $table->getResponse();
+        }
+
+        return $this->render('@App/exporter.html.twig', [
+            'datatable' => $table,
+        ]);
+    }
 }

tests/Fixtures/routing.yml

@@ -44,3 +44,11 @@ exporter:
 exporter_empty_datatable:
     path: /exporter-empty-datatable
     controller: Tests\Fixtures\AppBundle\Controller\ExporterController::exportEmptyDataTableAction
+
+exporter_long_text:
+    path: /exporter-long-text
+    controller: Tests\Fixtures\AppBundle\Controller\ExporterController::exportLongText
+
+exporter_special_chars:
+    path: /exporter-special-chars
+    controller: Tests\Fixtures\AppBundle\Controller\ExporterController::exportSpecialChars

tests/Functional/Exporter/Excel/ExcelOpenSpoutExporterTest.php

@@ -98,4 +98,36 @@ public function testWithSearch(): void
         static::assertEmpty($sheet->getCell('A3')->getFormattedValue());
         static::assertEmpty($sheet->getCell('B3')->getFormattedValue());
     }
+
+    public function testMaxCellLength(): void
+    {
+        $this->client->request('POST', '/exporter-long-text', [
+            '_dt' => 'dt',
+            '_exporter' => 'excel-openspout',
+        ]);
+
+        /** @var BinaryFileResponse $response */
+        $response = $this->client->getResponse();
+
+        $sheet = IOFactory::load($response->getFile()->getPathname())->getActiveSheet();
+
+        // Value should be truncated to 32767 characters
+        static::assertSame(str_repeat('a', 32767), $sheet->getCell('A2')->getFormattedValue());
+    }
+
+    public function testSpecialChars(): void
+    {
+        $this->client->request('POST', '/exporter-special-chars', [
+            '_dt' => 'dt',
+            '_exporter' => 'excel-openspout',
+        ]);
+
+        /** @var BinaryFileResponse $response */
+        $response = $this->client->getResponse();
+
+        $sheet = IOFactory::load($response->getFile()->getPathname())->getActiveSheet();
+
+        // Value should not contain HTML encoded characters
+        static::assertSame('<?xml version="1.0" encoding="UTF-8"?><hello>World</hello>', $sheet->getCell('A2')->getFormattedValue());
+    }
 }