Csv Reader Allow Use of mimetype=text/html Files Without Extension

CHANGELOG.md

@@ -29,6 +29,7 @@ and this project adheres to [Semantic Versioning](https://semver.org).
 - POWER Null/Bool Args. [PR #4031](https://github.com/PHPOffice/PhpSpreadsheet/pull/4031)
 - Do Not Output Alignment and Protection for Conditional Format. [Issue #4025](https://github.com/PHPOffice/PhpSpreadsheet/issues/4025) [PR #4027](https://github.com/PHPOffice/PhpSpreadsheet/pull/4027)
 - Xls Conditional Format Improvements. [PR #4030](https://github.com/PHPOffice/PhpSpreadsheet/pull/4030)
+- Csv Reader allow use of html mimetype. [Issue #4036](https://github.com/PHPOffice/PhpSpreadsheet/issues/4036) [PR #4049](https://github.com/PHPOffice/PhpSpreadsheet/pull/4040)
 
 ## 2024-05-11 - 2.1.0
 

src/PhpSpreadsheet/Reader/Csv.php

@@ -566,6 +566,7 @@ public function canRead(string $filename): bool
             'text/csv',
             'text/plain',
             'inode/x-empty',
+            'text/html',
         ];
 
         return in_array($type, $supportedTypes, true);

src/PhpSpreadsheet/Writer/ZipStream3.php

@@ -2,7 +2,6 @@
 
 namespace PhpOffice\PhpSpreadsheet\Writer;
 
-use ZipStream\Option\Archive;
 use ZipStream\ZipStream;
 
 class ZipStream3

tests/PhpSpreadsheetTests/IOFactoryTest.php

@@ -103,6 +103,8 @@ public static function providerIdentify(): array
             //['samples/templates/Excel2003XMLTest.xml', 'Xml', Reader\Xml::class],
             ['samples/templates/46readHtml.html', 'Html', Reader\Html::class],
             ['tests/data/Reader/CSV/encoding.utf8bom.csv', 'Csv', Reader\Csv::class],
+            ['tests/data/Reader/HTML/charset.UTF-16.lebom.html', 'Html', Reader\Html::class],
+            ['tests/data/Reader/HTML/charset.UTF-8.bom.html', 'Html', Reader\Html::class],
         ];
     }
 

tests/PhpSpreadsheetTests/Reader/Csv/NotHtmlTest.php

@@ -0,0 +1,83 @@
+<?php
+
+declare(strict_types=1);
+
+namespace PhpOffice\PhpSpreadsheetTests\Reader\Csv;
+
+use PhpOffice\PhpSpreadsheet\IOFactory;
+use PhpOffice\PhpSpreadsheet\Reader\Csv as CsvReader;
+use PhpOffice\PhpSpreadsheet\Shared\File;
+use PHPUnit\Framework\TestCase;
+
+class NotHtmlTest extends TestCase
+{
+    private string $tempFile = '';
+
+    protected function tearDown(): void
+    {
+        if ($this->tempFile !== '') {
+            unlink($this->tempFile);
+            $this->tempFile = '';
+        }
+    }
+
+    public function testHtmlCantRead(): void
+    {
+        // This test has a file which IOFactory will identify as Csv.
+        // So file can be read using either Csv Reader or IOFactory.
+        $this->tempFile = $filename = File::temporaryFilename();
+        $cells = [
+            ['1', '<a href="http://example.com">example</a>', '3'],
+            ['4', '5', '6'],
+        ];
+        $handle = fopen($filename, 'wb');
+        self::assertNotFalse($handle);
+        foreach ($cells as $row) {
+            fwrite($handle, "{$row[0]},{$row[1]},{$row[2]}\n");
+        }
+        fclose($handle);
+        // Php8.3- identify file as text/html.
+        // Php8.4+ identify file as text/csv, and this type of change
+        //    has been known to be retrofitted to prior versions.
+        $mime = mime_content_type($filename);
+        if ($mime !== 'text/csv') {
+            self::assertSame('text/html', $mime);
+        }
+        self::assertSame('Csv', IOFactory::identify($filename));
+        $reader = new CsvReader();
+        $spreadsheet = $reader->load($filename);
+        $sheet = $spreadsheet->getActiveSheet();
+        self::assertSame($cells, $sheet->toArray());
+        $spreadsheet->disconnectWorksheets();
+    }
+
+    public function testHtmlCanRead(): void
+    {
+        // This test has a file which IOFactory will identify as Html.
+        // So file has to be read using Csv Reader, not IOFactory.
+        $this->tempFile = $filename = File::temporaryFilename();
+        $cells = [
+            ['<a href="http://example.com">example</a>', '<div>hello', '3'],
+            ['4', '5', '</div>'],
+        ];
+        $handle = fopen($filename, 'wb');
+        self::assertNotFalse($handle);
+        foreach ($cells as $row) {
+            fwrite($handle, "{$row[0]},{$row[1]},{$row[2]}\n");
+        }
+        fclose($handle);
+        // Php8.3- identify file as text/html.
+        // Php8.4+ identify file as text/csv, and this type of change
+        //    has been known to be retrofitted to prior versions.
+        $mime = mime_content_type($filename);
+        if ($mime !== 'text/csv') {
+            self::assertSame('text/html', $mime);
+        }
+        self::assertSame('Html', IOFactory::identify($filename));
+        $reader = new CsvReader();
+        $spreadsheet = $reader->load($filename);
+        $sheet = $spreadsheet->getActiveSheet();
+        self::assertSame($cells, $sheet->toArray());
+        $spreadsheet->disconnectWorksheets();
+    }
+}