Update privacy policy anchor hash algorithm. (#1146)

We now replace umlaute and diacritics. | Input | Old | New | |--------|--------|--------| | 1. Einführung | 1-einfhrung | 1-einfuehrung | | abc pokémon | abc-pokmon | abc-pokemon | Fixes #1145
SharezoneApp · Nov 5, 2023 · e961f6b · e961f6b
1 parent 2870d8f
commit e961f6b
Show file tree

Hide file tree

Showing 4 changed files with 98 additions and 26 deletions.
diff --git a/app/lib/privacy_policy/src/ui/sharezone_markdown_extensions.dart b/app/lib/privacy_policy/src/ui/sharezone_markdown_extensions.dart
@@ -7,12 +7,11 @@
 // SPDX-License-Identifier: EUPL-1.2
 
 import 'package:markdown/markdown.dart';
+import 'package:diacritic/diacritic.dart' as pkg;
 
 /// We define our own classes for automatic ID generation as we don't want any
 /// chance of breakage if `package:markdown` changes/updates its algorithm.
 ///
-/// It's the exact same algorithm (for now).
-///
 /// It might break if we e.g. would use one updated algorithm in our backend to
 /// generate the anchor ids for the table of contents section list, but our
 /// client uses an old algorithm to generate/search for the anchor ids inside
@@ -86,5 +85,28 @@ String _generateAnchorHashFromElement(Element element) =>
 String generateAnchorHash(String text) => text
     .toLowerCase()
     .trim()
+    .replaceAllMapped(regexp, (char) => _replacementMap[char.group(0)]!)
+    .removeDiacritics()
     .replaceAll(RegExp(r'[^a-z0-9 _-]'), '')
     .replaceAll(RegExp(r'\s'), '-');
+
+// Match every character in [_replacementMap.keys].
+RegExp get regexp => RegExp(_replacementMap.keys.join('|'));
+
+const _replacementMap = {
+  'Ä': 'ae',
+  'ä': 'ae',
+  'Ö': 'oe',
+  'ö': 'oe',
+  'Ü': 'ue',
+  'ü': 'ue',
+  'ẞ': 'ss',
+  'ß': 'ss',
+};
+
+extension on String {
+  /// E.g. `é` -> `e`
+  String removeDiacritics() {
+    return pkg.removeDiacritics(this);
+  }
+}
diff --git a/app/pubspec.lock b/app/pubspec.lock
diff --git a/app/pubspec.yaml b/app/pubspec.yaml
@@ -58,6 +58,7 @@ dependencies:
     path: ../lib/date
   design:
     path: ../lib/design
+  diacritic: ^0.1.4
   dio: ^5.3.2
   dynamic_links:
     path: ../lib/dynamic_links

diff --git a/app/test/privacy_policy/generate_anchor_hash_test.dart b/app/test/privacy_policy/generate_anchor_hash_test.dart
@@ -12,29 +12,70 @@ import 'package:sharezone/privacy_policy/src/privacy_policy_src.dart';
 void main() {
   /// We are testing this since changing the anchor hash generation algorithm
   /// might break stuff. See comments in [generateAnchorHash] file.
-  test('Generates correct anchor hashes', () {
-    final sections = [
-      '1. Einführung',
-      '2. Kontaktinformationen',
-      '3. Wichtige Begriffe, die du kennen solltest',
-      '4. Welche Informationen erfassen wir grundsätzlich?',
-      '5. An wen geben wir deine Daten weiter?',
-      '6. Wie lange speichern wir deine Daten?',
-      '7. Welche Rechte hast du?',
-      'Glückwunsch, du hast es geschafft'
-    ];
-
-    final anchorHashes = sections.map(generateAnchorHash).toList();
-
-    expect(anchorHashes, [
-      '1-einfhrung',
-      '2-kontaktinformationen',
-      '3-wichtige-begriffe-die-du-kennen-solltest',
-      '4-welche-informationen-erfassen-wir-grundstzlich',
-      '5-an-wen-geben-wir-deine-daten-weiter',
-      '6-wie-lange-speichern-wir-deine-daten',
-      '7-welche-rechte-hast-du',
-      'glckwunsch-du-hast-es-geschafft',
-    ]);
+  group('Generates correct anchor hashes', () {
+    test('real life examples', () {
+      final sections = [
+        '1. Einführung',
+        '2. Kontaktinformationen',
+        '3. Wichtige Begriffe, die du kennen solltest',
+        '4. Welche Informationen erfassen wir grundsätzlich?',
+        '5. An wen geben wir deine Daten weiter?',
+        '6. Wie lange speichern wir deine Daten?',
+        '7. Welche Rechte hast du?',
+        'Glückwunsch, du hast es geschafft',
+      ];
+
+      final anchorHashes = sections.map(generateAnchorHash).toList();
+
+      expect(anchorHashes, [
+        '1-einfuehrung',
+        '2-kontaktinformationen',
+        '3-wichtige-begriffe-die-du-kennen-solltest',
+        '4-welche-informationen-erfassen-wir-grundsaetzlich',
+        '5-an-wen-geben-wir-deine-daten-weiter',
+        '6-wie-lange-speichern-wir-deine-daten',
+        '7-welche-rechte-hast-du',
+        'glueckwunsch-du-hast-es-geschafft',
+      ]);
+    });
+    test('replaces special chars', () {
+      final sections = [
+        r'1-?!=""§$pokémon~+',
+        'Ist das eine Frage???',
+        'Wort-mit-Bindestrichen'
+      ];
+
+      final anchorHashes = sections.map(generateAnchorHash).toList();
+
+      expect(anchorHashes, [
+        '1-pokemon',
+        'ist-das-eine-frage',
+        'wort-mit-bindestrichen',
+      ]);
+    });
+
+    test('replaces umlaute', () {
+      final umlaute = [
+        'ä',
+        'Ä',
+        'ö',
+        'Ö',
+        'ü',
+        'Ü',
+        'ß',
+      ];
+
+      final anchorHashes = umlaute.map(generateAnchorHash).toList();
+
+      expect(anchorHashes, [
+        'ae',
+        'ae',
+        'oe',
+        'oe',
+        'ue',
+        'ue',
+        'ss',
+      ]);
+    });
   });
 }