diff --git a/core/modules/file/config/install/file.settings.yml b/core/modules/file/config/install/file.settings.yml
index 9191a0b8004e61f6174aa7f935f79d7007470530..a3809aaa9d2f99b52b9c4d8299057009e1f27449 100644
--- a/core/modules/file/config/install/file.settings.yml
+++ b/core/modules/file/config/install/file.settings.yml
@@ -4,3 +4,10 @@ description:
icon:
directory: 'core/modules/file/icons'
make_unused_managed_files_temporary: false
+filename_sanitization:
+ transliterate: false
+ replace_whitespace: false
+ replace_non_alphanumeric: false
+ deduplicate_separators: false
+ lowercase: false
+ replacement_character: '-'
diff --git a/core/modules/file/config/schema/file.schema.yml b/core/modules/file/config/schema/file.schema.yml
index f94fbec950f14c28d23bd71a84058f985ae759aa..aaa74af750c4fa9a7a4abc49d3b615d585fd8637 100644
--- a/core/modules/file/config/schema/file.schema.yml
+++ b/core/modules/file/config/schema/file.schema.yml
@@ -24,6 +24,28 @@ file.settings:
make_unused_managed_files_temporary:
type: boolean
label: 'Controls if unused files should be marked temporary'
+ filename_sanitization:
+ type: mapping
+ label: 'Uploaded filename sanitization options'
+ mapping:
+ transliterate:
+ type: boolean
+ label: 'Transliterate'
+ replace_whitespace:
+ type: boolean
+ label: 'Replace whitespace'
+ replace_non_alphanumeric:
+ type: boolean
+ label: 'Replace non-alphanumeric characters except dot, underscore and dash'
+ deduplicate_separators:
+ type: boolean
+ label: 'Replace sequences of dots, underscores and/or dashes with the replacement character'
+ lowercase:
+ type: boolean
+ label: 'Convert to lowercase'
+ replacement_character:
+ type: string
+ label: 'Character to use in replacements'
field.storage_settings.file:
type: base_entity_reference_field_settings
diff --git a/core/modules/file/file.module b/core/modules/file/file.module
index 34b00de5c68027bdd7a342b17b839a6b0a8e79ed..840850806999ad811a051cf52c0e90f958c9fb53 100644
--- a/core/modules/file/file.module
+++ b/core/modules/file/file.module
@@ -1552,3 +1552,80 @@ function file_field_find_file_reference_column(FieldDefinitionInterface $field)
}
return FALSE;
}
+
+/**
+ * Implements hook_form_FORM_ID_alter().
+ *
+ * Injects the file sanitization options into /admin/config/media/file-system.
+ *
+ * These settings are enforced during upload by the FileEventSubscriber that
+ * listens to the FileUploadSanitizeNameEvent event.
+ *
+ * @see \Drupal\system\Form\FileSystemForm
+ * @see \Drupal\Core\File\Event\FileUploadSanitizeNameEvent
+ * @see \Drupal\file\EventSubscriber\FileEventSubscriber
+ */
+function file_form_system_file_system_settings_alter(array &$form, FormStateInterface $form_state) {
+ $config = \Drupal::config('file.settings');
+ $form['filename_sanitization'] = [
+ '#type' => 'details',
+ '#title' => t('Sanitize filenames'),
+ '#description' => t('These settings only apply to new files as they are uploaded. Changes here do not affect existing file names.'),
+ '#open' => TRUE,
+ '#tree' => TRUE,
+ ];
+
+ $form['filename_sanitization']['replacement_character'] = [
+ '#type' => 'select',
+ '#title' => t('Replacement character'),
+ '#default_value' => $config->get('filename_sanitization.replacement_character'),
+ '#options' => [
+ '-' => t('Dash (-)'),
+ '_' => t('Underscore (_)'),
+ ],
+ '#description' => t('Used when replacing whitespace, replacing non-alphanumeric characters or transliterating unknown characters.'),
+ ];
+
+ $form['filename_sanitization']['transliterate'] = [
+ '#type' => 'checkbox',
+ '#title' => t('Transliterate'),
+ '#default_value' => $config->get('filename_sanitization.transliterate'),
+ '#description' => t('Transliteration replaces any characters that are not alphanumeric, underscores, periods or hyphens with the replacement character. It ensures filenames only contain ASCII characters. It is recommended to keep transliteration enabled.'),
+ ];
+
+ $form['filename_sanitization']['replace_whitespace'] = [
+ '#type' => 'checkbox',
+ '#title' => t('Replace whitespace with the replacement character'),
+ '#default_value' => $config->get('filename_sanitization.replace_whitespace'),
+ ];
+
+ $form['filename_sanitization']['replace_non_alphanumeric'] = [
+ '#type' => 'checkbox',
+ '#title' => t('Replace non-alphanumeric characters with the replacement character'),
+ '#default_value' => $config->get('filename_sanitization.replace_non_alphanumeric'),
+ '#description' => t('Alphanumeric characters, dots <span aria-hidden="true">(.)</span>, underscores <span aria-hidden="true">(_)</span> and dashes <span aria-hidden="true">(-)</span> are preserved.'),
+ ];
+
+ $form['filename_sanitization']['deduplicate_separators'] = [
+ '#type' => 'checkbox',
+ '#title' => t('Replace sequences of dots, underscores and/or dashes with the replacement character'),
+ '#default_value' => $config->get('filename_sanitization.deduplicate_separators'),
+ ];
+
+ $form['filename_sanitization']['lowercase'] = [
+ '#type' => 'checkbox',
+ '#title' => t('Convert to lowercase'),
+ '#default_value' => $config->get('filename_sanitization.lowercase'),
+ ];
+
+ $form['#submit'][] = 'file_system_settings_submit';
+}
+
+/**
+ * Form submission handler for file system settings form.
+ */
+function file_system_settings_submit(array &$form, FormStateInterface $form_state) {
+ $config = \Drupal::configFactory()->getEditable('file.settings')
+ ->set('filename_sanitization', $form_state->getValue('filename_sanitization'));
+ $config->save();
+}
diff --git a/core/modules/file/file.post_update.php b/core/modules/file/file.post_update.php
index 22bbdd512ed42bf6ecde0cb839fc40fb1624045e..73c815b1cc0233c03d950cf278ff024527b06d79 100644
--- a/core/modules/file/file.post_update.php
+++ b/core/modules/file/file.post_update.php
@@ -29,3 +29,17 @@ function file_post_update_add_permissions_to_roles(?array &$sandbox = NULL): voi
return TRUE;
});
}
+
+/**
+ * Add default filename sanitization configuration.
+ */
+function file_post_update_add_default_filename_sanitization_configuration() {
+ $config = \Drupal::configFactory()->getEditable('file.settings');
+ $config->set('filename_sanitization.transliterate', FALSE);
+ $config->set('filename_sanitization.replace_whitespace', FALSE);
+ $config->set('filename_sanitization.replace_non_alphanumeric', FALSE);
+ $config->set('filename_sanitization.deduplicate_separators', FALSE);
+ $config->set('filename_sanitization.lowercase', FALSE);
+ $config->set('filename_sanitization.replacement_character', '-');
+ $config->save();
+}
diff --git a/core/modules/file/file.services.yml b/core/modules/file/file.services.yml
index 64c612cfbbca3fae76b8b1a035e3a6e7a98e1073..370972c5ff0b754b8db32717c180b808258c6675 100644
--- a/core/modules/file/file.services.yml
+++ b/core/modules/file/file.services.yml
@@ -1,4 +1,9 @@
services:
+ file.event.subscriber:
+ class: Drupal\file\EventSubscriber\FileEventSubscriber
+ arguments: ['@config.factory', '@transliteration', '@language_manager']
+ tags:
+ - { name: event_subscriber }
file.usage:
class: Drupal\file\FileUsage\DatabaseFileUsageBackend
arguments: ['@config.factory', '@database', 'file_usage']
diff --git a/core/modules/file/src/EventSubscriber/FileEventSubscriber.php b/core/modules/file/src/EventSubscriber/FileEventSubscriber.php
new file mode 100644
index 0000000000000000000000000000000000000000..17608c583fb48589fa5824c8c66da3fbe9fb9cb4
--- /dev/null
+++ b/core/modules/file/src/EventSubscriber/FileEventSubscriber.php
@@ -0,0 +1,110 @@
+<?php
+
+namespace Drupal\file\EventSubscriber;
+
+use Drupal\Component\Transliteration\TransliterationInterface;
+use Drupal\Core\Config\ConfigFactoryInterface;
+use Drupal\Core\Language\LanguageInterface;
+use Drupal\Core\Language\LanguageManagerInterface;
+use Drupal\Core\File\Event\FileUploadSanitizeNameEvent;
+use Symfony\Component\EventDispatcher\EventSubscriberInterface;
+
+/**
+ * Sanitizes uploaded filenames.
+ *
+ * @package Drupal\file\EventSubscriber
+ */
+class FileEventSubscriber implements EventSubscriberInterface {
+
+ /**
+ * Constructs a new file event listener.
+ *
+ * @param \Drupal\Core\Config\ConfigFactoryInterface $configFactory
+ * The config factory.
+ * @param \Drupal\Component\Transliteration\TransliterationInterface $transliteration
+ * The transliteration service.
+ * @param \Drupal\Core\Language\LanguageManagerInterface $languageManager
+ * The language manager.
+ */
+ public function __construct(
+ protected ConfigFactoryInterface $configFactory,
+ protected TransliterationInterface $transliteration,
+ protected LanguageManagerInterface $languageManager,
+ ) {}
+
+ /**
+ * {@inheritdoc}
+ */
+ public static function getSubscribedEvents(): array {
+ return [
+ FileUploadSanitizeNameEvent::class => 'sanitizeFilename',
+ ];
+ }
+
+ /**
+ * Sanitizes the filename of a file being uploaded.
+ *
+ * @param \Drupal\Core\File\Event\FileUploadSanitizeNameEvent $event
+ * File upload sanitize name event.
+ *
+ * @see file_form_system_file_system_settings_alter()
+ */
+ public function sanitizeFilename(FileUploadSanitizeNameEvent $event) {
+ $fileSettings = $this->configFactory->get('file.settings');
+ $transliterate = $fileSettings->get('filename_sanitization.transliterate');
+
+ $filename = $event->getFilename();
+ $extension = pathinfo($filename, PATHINFO_EXTENSION);
+ if ($extension !== '') {
+ $extension = '.' . $extension;
+ $filename = pathinfo($filename, PATHINFO_FILENAME);
+ }
+
+ // Sanitize the filename according to configuration.
+ $alphanumeric = $fileSettings->get('filename_sanitization.replace_non_alphanumeric');
+ $replacement = $fileSettings->get('filename_sanitization.replacement_character');
+ if ($transliterate) {
+ $transliterated_filename = $this->transliteration->transliterate(
+ $filename,
+ $this->languageManager->getCurrentLanguage(LanguageInterface::TYPE_CONTENT)->getId(),
+ $replacement
+ );
+ if (mb_strlen($transliterated_filename) > 0) {
+ $filename = $transliterated_filename;
+ }
+ else {
+ // If transliteration has resulted in a zero length string enable the
+ // 'replace_non_alphanumeric' option and ignore the result of
+ // transliteration.
+ $alphanumeric = TRUE;
+ }
+ }
+ if ($fileSettings->get('filename_sanitization.replace_whitespace')) {
+ $filename = preg_replace('/\s/u', $replacement, trim($filename));
+ }
+ // Only honor replace_non_alphanumeric if transliterate is enabled.
+ if ($transliterate && $alphanumeric) {
+ $filename = preg_replace('/[^0-9A-Za-z_.-]/u', $replacement, $filename);
+ }
+ if ($fileSettings->get('filename_sanitization.deduplicate_separators')) {
+ $filename = preg_replace('/(_)_+|(\.)\.+|(-)-+/u', $replacement, $filename);
+ // Replace multiple separators with single one.
+ $filename = preg_replace('/(_|\.|\-)[(_|\.|\-)]+/u', $replacement, $filename);
+ $filename = preg_replace('/' . preg_quote($replacement) . '[' . preg_quote($replacement) . ']*/u', $replacement, $filename);
+ // Remove replacement character from the end of the filename.
+ $filename = rtrim($filename, $replacement);
+
+ // If there is an extension remove dots from the end of the filename to
+ // prevent duplicate dots.
+ if (!empty($extension)) {
+ $filename = rtrim($filename, '.');
+ }
+ }
+ if ($fileSettings->get('filename_sanitization.lowercase')) {
+ // Force lowercase to prevent issues on case-insensitive file systems.
+ $filename = mb_strtolower($filename);
+ }
+ $event->setFilename($filename . $extension);
+ }
+
+}
diff --git a/core/modules/file/tests/src/Functional/SaveUploadTest.php b/core/modules/file/tests/src/Functional/SaveUploadTest.php
index d7ffe5a252e7ed062f25c0e46e1a71a738b076b5..863ea5ebec3683e98df0e8b888ae2d9d3268d217 100644
--- a/core/modules/file/tests/src/Functional/SaveUploadTest.php
+++ b/core/modules/file/tests/src/Functional/SaveUploadTest.php
@@ -8,6 +8,8 @@
use Drupal\file\Entity\File;
use Drupal\Tests\TestFileCreationTrait;
+// cSpell:ignore TÉXT Pácê
+
/**
* Tests the file_save_upload() function.
*
@@ -59,13 +61,20 @@ class SaveUploadTest extends FileManagedTestBase {
*/
protected $imageExtension;
+ /**
+ * The user used by the test.
+ *
+ * @var \Drupal\user\Entity\User
+ */
+ protected $account;
+
/**
* {@inheritdoc}
*/
protected function setUp(): void {
parent::setUp();
- $account = $this->drupalCreateUser(['access site reports']);
- $this->drupalLogin($account);
+ $this->account = $this->drupalCreateUser(['access site reports']);
+ $this->drupalLogin($this->account);
$image_files = $this->drupalGetTestFiles('image');
$this->image = File::create((array) current($image_files));
@@ -756,4 +765,135 @@ public function testRequired() {
$this->assertSession()->responseContains('You WIN!');
}
+ /**
+ * Tests filename sanitization.
+ */
+ public function testSanitization() {
+ $file = $this->generateFile('TÉXT-œ', 64, 5, 'text');
+
+ $this->drupalGet('file-test/upload');
+ // Upload a file with a name with uppercase and unicode characters.
+ $edit = [
+ 'files[file_test_upload]' => \Drupal::service('file_system')->realpath($file),
+ 'extensions' => 'txt',
+ 'is_image_file' => FALSE,
+ ];
+ $this->submitForm($edit, 'Submit');
+ $this->assertSession()->statusCodeEquals(200);
+ // Test that the file name has not been sanitized.
+ $this->assertSession()->responseContains('File name is TÉXT-œ.txt.');
+
+ // Enable sanitization via the UI.
+ $admin = $this->createUser(['administer site configuration']);
+ $this->drupalLogin($admin);
+
+ // For now, just transliterate, with no other transformations.
+ $options = [
+ 'filename_sanitization[transliterate]' => TRUE,
+ 'filename_sanitization[replace_whitespace]' => FALSE,
+ 'filename_sanitization[replace_non_alphanumeric]' => FALSE,
+ 'filename_sanitization[deduplicate_separators]' => FALSE,
+ 'filename_sanitization[lowercase]' => FALSE,
+ 'filename_sanitization[replacement_character]' => '-',
+ ];
+ $this->drupalGet('admin/config/media/file-system');
+ $this->submitForm($options, 'Save configuration');
+
+ $this->drupalLogin($this->account);
+
+ // Upload a file with a name with uppercase and unicode characters.
+ $this->drupalGet('file-test/upload');
+ $this->submitForm($edit, 'Submit');
+ $this->assertSession()->statusCodeEquals(200);
+ // Test that the file name has been transliterated.
+ $this->assertSession()->responseContains('File name is TEXT-oe.txt.');
+ // Make sure we got a message about the rename.
+ $message = 'Your upload has been renamed to <em class="placeholder">TEXT-oe.txt</em>';
+ $this->assertSession()->responseContains($message);
+
+ // Generate another file with a name with All The Things(tm) we care about.
+ $file = $this->generateFile('S Pácê--táb# #--🙈', 64, 5, 'text');
+ $edit = [
+ 'files[file_test_upload]' => \Drupal::service('file_system')->realpath($file),
+ 'extensions' => 'txt',
+ 'is_image_file' => FALSE,
+ ];
+ $this->submitForm($edit, 'Submit');
+ $this->assertSession()->statusCodeEquals(200);
+ // Test that the file name has only been transliterated.
+ $this->assertSession()->responseContains('File name is S Pace--tab# #---.txt.');
+
+ // Leave transliteration on and enable whitespace replacement.
+ $this->drupalLogin($admin);
+ $options['filename_sanitization[replace_whitespace]'] = TRUE;
+ $this->drupalGet('admin/config/media/file-system');
+ $this->submitForm($options, 'Save configuration');
+ $this->drupalLogin($this->account);
+
+ // Try again with the monster filename.
+ $this->drupalGet('file-test/upload');
+ $this->submitForm($edit, 'Submit');
+ $this->assertSession()->statusCodeEquals(200);
+ // Test that the file name has been transliterated and whitespace replaced.
+ $this->assertSession()->responseContains('File name is S--Pace--tab#-#---.txt.');
+
+ // Leave transliteration and whitespace replacement on, replace non-alpha.
+ $this->drupalLogin($admin);
+ $options['filename_sanitization[replace_non_alphanumeric]'] = TRUE;
+ $options['filename_sanitization[replacement_character]'] = '_';
+ $this->drupalGet('admin/config/media/file-system');
+ $this->submitForm($options, 'Save configuration');
+ $this->drupalLogin($this->account);
+
+ // Try again with the monster filename.
+ $this->drupalGet('file-test/upload');
+ $this->submitForm($edit, 'Submit');
+ $this->assertSession()->statusCodeEquals(200);
+
+ // Test that the file name has been transliterated, whitespace replaced with
+ // '_', and non-alphanumeric characters replaced with '_'.
+ $this->assertSession()->responseContains('File name is S__Pace--tab___--_.txt.');
+
+ // Now turn on the setting to remove duplicate separators.
+ $this->drupalLogin($admin);
+ $options['filename_sanitization[deduplicate_separators]'] = TRUE;
+ $options['filename_sanitization[replacement_character]'] = '-';
+ $this->drupalGet('admin/config/media/file-system');
+ $this->submitForm($options, 'Save configuration');
+ $this->drupalLogin($this->account);
+
+ // Try again with the monster filename.
+ $this->drupalGet('file-test/upload');
+ $this->submitForm($edit, 'Submit');
+ $this->assertSession()->statusCodeEquals(200);
+
+ // Test that the file name has been transliterated, whitespace replaced,
+ // non-alphanumeric characters replaced, and duplicate separators removed.
+ $this->assertSession()->responseContains('File name is S-Pace-tab.txt.');
+
+ // Finally, check the lowercase setting.
+ $this->drupalLogin($admin);
+ $options['filename_sanitization[lowercase]'] = TRUE;
+ $this->drupalGet('admin/config/media/file-system');
+ $this->submitForm($options, 'Save configuration');
+ $this->drupalLogin($this->account);
+
+ // Generate another file since we're going to start getting collisions with
+ // previously uploaded and renamed copies.
+ $file = $this->generateFile('S Pácê--táb# #--🙈-2', 64, 5, 'text');
+ $edit = [
+ 'files[file_test_upload]' => \Drupal::service('file_system')->realpath($file),
+ 'extensions' => 'txt',
+ 'is_image_file' => FALSE,
+ ];
+ $this->drupalGet('file-test/upload');
+ $this->submitForm($edit, 'Submit');
+ $this->assertSession()->statusCodeEquals(200);
+ // Make sure all the sanitization options work as intended.
+ $this->assertSession()->responseContains('File name is s-pace-tab-2.txt.');
+ // Make sure we got a message about the rename.
+ $message = 'Your upload has been renamed to <em class="placeholder">s-pace-tab-2.txt</em>';
+ $this->assertSession()->responseContains($message);
+ }
+
}
diff --git a/core/modules/file/tests/src/Unit/SanitizeNameTest.php b/core/modules/file/tests/src/Unit/SanitizeNameTest.php
new file mode 100644
index 0000000000000000000000000000000000000000..dfe3b4675c2162511bc617694726660555662bc9
--- /dev/null
+++ b/core/modules/file/tests/src/Unit/SanitizeNameTest.php
@@ -0,0 +1,247 @@
+<?php
+
+namespace Drupal\Tests\file\Unit;
+
+use Drupal\Component\Transliteration\PhpTransliteration;
+use Drupal\Core\File\Event\FileUploadSanitizeNameEvent;
+use Drupal\Core\Language\Language;
+use Drupal\Core\Language\LanguageInterface;
+use Drupal\Core\Language\LanguageManagerInterface;
+use Drupal\file\EventSubscriber\FileEventSubscriber;
+use Drupal\Tests\UnitTestCase;
+
+// cSpell:ignore TÉXT äöüåøhello aouaohello aeoeueaohello Pácê
+
+/**
+ * Filename sanitization tests.
+ *
+ * @group file
+ */
+class SanitizeNameTest extends UnitTestCase {
+
+ /**
+ * Test file name sanitization.
+ *
+ * @param string $original
+ * The original filename.
+ * @param string $expected
+ * The expected filename.
+ * @param array $options
+ * Array of filename sanitization options, in this order:
+ * 0: boolean Transliterate.
+ * 1: string Replace whitespace.
+ * 2: string Replace non-alphanumeric characters.
+ * 3: boolean De-duplicate separators.
+ * 4: boolean Convert to lowercase.
+ * @param string $language_id
+ * Optional language code for transliteration. Defaults to 'en'.
+ *
+ * @dataProvider provideFilenames
+ *
+ * @covers \Drupal\file\EventSubscriber\FileEventSubscriber::sanitizeFilename
+ * @covers \Drupal\Core\File\Event\FileUploadSanitizeNameEvent::__construct
+ */
+ public function testFileNameTransliteration($original, $expected, array $options, $language_id = 'en') {
+ $sanitization_options = [
+ 'transliterate' => $options[0],
+ 'replacement_character' => $options[1],
+ 'replace_whitespace' => $options[2],
+ 'replace_non_alphanumeric' => $options[3],
+ 'deduplicate_separators' => $options[4],
+ 'lowercase' => $options[5],
+ ];
+ $config_factory = $this->getConfigFactoryStub([
+ 'file.settings' => [
+ 'filename_sanitization' => $sanitization_options,
+ ],
+ ]);
+
+ $language = new Language(['id' => $language_id]);
+ $language_manager = $this->prophesize(LanguageManagerInterface::class);
+ $language_manager->getCurrentLanguage(LanguageInterface::TYPE_CONTENT)->willReturn($language);
+
+ $event = new FileUploadSanitizeNameEvent($original, $language_id);
+ $subscriber = new FileEventSubscriber($config_factory, new PhpTransliteration(), $language_manager->reveal());
+ $subscriber->sanitizeFilename($event);
+
+ // Check the results of the configured sanitization.
+ $this->assertEquals($expected, $event->getFilename());
+ }
+
+ /**
+ * Provides data for testFileNameTransliteration().
+ *
+ * @return array
+ * Arrays with original name, expected name, and sanitization options.
+ */
+ public function provideFilenames() {
+ return [
+ 'Test default options' => [
+ 'TÉXT-œ.txt',
+ 'TÉXT-œ.txt',
+ [FALSE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Test raw file without extension' => [
+ 'TÉXT-œ',
+ 'TÉXT-œ',
+ [FALSE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Test only transliteration: simple' => [
+ 'Á-TÉXT-œ.txt',
+ 'A-TEXT-oe.txt',
+ [TRUE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Test only transliteration: raw file without extension' => [
+ 'Á-TÉXT-œ',
+ 'A-TEXT-oe',
+ [TRUE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Test only transliteration: complex and replace (-)' => [
+ 'S Pácê--táb# #--🙈.jpg',
+ 'S Pace--tab# #---.jpg',
+ [TRUE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Test only transliteration: complex and replace (_)' => [
+ 'S Pácê--táb# #--🙈.jpg',
+ 'S Pace--tab# #--_.jpg',
+ [TRUE, '_', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Test transliteration, replace (-) and replace whitespace (trim front)' => [
+ ' S Pácê--táb# #--🙈.png',
+ 'S--Pace--tab#-#---.png',
+ [TRUE, '-', TRUE, FALSE, FALSE, FALSE],
+ ],
+ 'Test transliteration, replace (-) and replace whitespace (trim both sides)' => [
+ ' S Pácê--táb# #--🙈 .jpg',
+ 'S--Pace--tab#-#---.jpg',
+ [TRUE, '-', TRUE, FALSE, FALSE, FALSE],
+ ],
+ 'Test transliteration, replace (_) and replace whitespace (trim both sides)' => [
+ ' S Pácê--táb# #--🙈 .jpg',
+ 'S__Pace--tab#_#--_.jpg',
+ [TRUE, '_', TRUE, FALSE, FALSE, FALSE],
+ ],
+ 'Test transliteration, replace (_), replace whitespace and replace non-alphanumeric' => [
+ ' S Pácê--táb# #--🙈.txt',
+ 'S__Pace--tab___--_.txt',
+ [TRUE, '_', TRUE, TRUE, FALSE, FALSE],
+ ],
+ 'Test transliteration, replace (-), replace whitespace and replace non-alphanumeric' => [
+ ' S Pácê--táb# #--🙈.txt',
+ 'S--Pace--tab------.txt',
+ [TRUE, '-', TRUE, TRUE, FALSE, FALSE],
+ ],
+ 'Test transliteration, replace (-), replace whitespace, replace non-alphanumeric and removing duplicate separators' => [
+ 'S Pácê--táb# #--🙈.txt',
+ 'S-Pace-tab.txt',
+ [TRUE, '-', TRUE, TRUE, TRUE, FALSE],
+ ],
+ 'Test transliteration, replace (-), replace whitespace and deduplicate separators' => [
+ ' S Pácê--táb# #--🙈.txt',
+ 'S-Pace-tab#-#.txt',
+ [TRUE, '-', TRUE, FALSE, TRUE, FALSE],
+ ],
+ 'Test transliteration, replace (_), replace whitespace, replace non-alphanumeric and deduplicate separators' => [
+ 'S Pácê--táb# #--🙈.txt',
+ 'S_Pace_tab.txt',
+ [TRUE, '_', TRUE, TRUE, TRUE, FALSE],
+ ],
+ 'Test transliteration, replace (-), replace whitespace, replace non-alphanumeric, deduplicate separators and lowercase conversion' => [
+ 'S Pácê--táb# #--🙈.jpg',
+ 's-pace-tab.jpg',
+ [TRUE, '-', TRUE, TRUE, TRUE, TRUE],
+ ],
+ 'Test transliteration, replace (_), replace whitespace, replace non-alphanumeric, deduplicate separators and lowercase conversion' => [
+ 'S Pácê--táb# #--🙈.txt',
+ 's_pace_tab.txt',
+ [TRUE, '_', TRUE, TRUE, TRUE, TRUE],
+ ],
+ 'Ignore non-alphanumeric replacement if transliteration is not set, but still replace whitespace, deduplicate separators, and lowercase' => [
+ ' 2S Pácê--táb# #--🙈.txt',
+ '2s-pácê-táb#-#-🙈.txt',
+ [FALSE, '-', TRUE, TRUE, TRUE, TRUE],
+ ],
+ 'Only lowercase, simple' => [
+ 'TEXT.txt',
+ 'text.txt',
+ [TRUE, '-', FALSE, FALSE, FALSE, TRUE],
+ ],
+ 'Only lowercase, with unicode' => [
+ 'TÉXT.txt',
+ 'text.txt',
+ [TRUE, '-', FALSE, FALSE, FALSE, TRUE],
+ ],
+ 'No transformations' => [
+ 'Ä Ö Ü Å Ø äöüåøhello.txt',
+ 'Ä Ö Ü Å Ø äöüåøhello.txt',
+ [FALSE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Transliterate via en (not de), no other transformations' => [
+ 'Ä Ö Ü Å Ø äöüåøhello.txt',
+ 'A O U A O aouaohello.txt',
+ [TRUE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Transliterate via de (not en), no other transformations' => [
+ 'Ä Ö Ü Å Ø äöüåøhello.txt',
+ 'Ae Oe Ue A O aeoeueaohello.txt',
+ [TRUE, '-', FALSE, FALSE, FALSE, FALSE], 'de',
+ ],
+ 'Transliterate via de not en, plus whitespace + lowercase' => [
+ 'Ä Ö Ü Å Ø äöüåøhello.txt',
+ 'ae-oe-ue-a-o-aeoeueaohello.txt',
+ [TRUE, '-', TRUE, FALSE, FALSE, TRUE], 'de',
+ ],
+ 'Remove duplicate separators with falsey extension' => [
+ 'foo.....0',
+ 'foo.0',
+ [TRUE, '-', FALSE, FALSE, TRUE, FALSE],
+ ],
+ 'Remove duplicate separators with extension and ending in dot' => [
+ 'foo.....txt',
+ 'foo.txt',
+ [TRUE, '-', FALSE, FALSE, TRUE, FALSE],
+ ],
+ 'Remove duplicate separators without extension and ending in dot' => [
+ 'foo.....',
+ 'foo',
+ [TRUE, '-', FALSE, FALSE, TRUE, FALSE],
+ ],
+ 'All unknown unicode' => [
+ '🙈🙈🙈.txt',
+ '---.txt',
+ [TRUE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ '✓ unicode' => [
+ '✓.txt',
+ '-.txt',
+ [TRUE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Multiple ✓ unicode' => [
+ '✓✓✓.txt',
+ '---.txt',
+ [TRUE, '-', FALSE, FALSE, FALSE, FALSE],
+ ],
+ 'Test transliteration, replace (-), replace whitespace and removing multiple duplicate separators #1' => [
+ 'Test_-_file.png',
+ 'test-file.png',
+ [TRUE, '-', TRUE, TRUE, TRUE, TRUE],
+ ],
+ 'Test transliteration, replace (-), replace whitespace and removing multiple duplicate separators #2' => [
+ 'Test .. File.png',
+ 'test-file.png',
+ [TRUE, '-', TRUE, TRUE, TRUE, TRUE],
+ ],
+ 'Test transliteration, replace (-), replace whitespace and removing multiple duplicate separators #3' => [
+ 'Test -..__-- file.png',
+ 'test-file.png',
+ [TRUE, '-', TRUE, TRUE, TRUE, TRUE],
+ ],
+ 'Test transliteration, replace (-), replace sequences of dots, underscores and/or dashes with the replacement character' => [
+ 'abc. --_._-- .abc.jpeg',
+ 'abc. - .abc.jpeg',
+ [TRUE, '-', FALSE, FALSE, TRUE, FALSE],
+ ],
+ ];
+ }
+
+}