improved uri parsing

2026-02-27 08:31:19 +01:00
parent d24675094d
commit d25229cb95
19 changed files with 1132 additions and 95 deletions
@@ -18,12 +18,12 @@
 * along with this program. If not, see <http://www.gnu.org/licenses/>.
 */

-const _supportedSchemes = {'https', 'http', 'ftp', 'file', 'content', 'about'};
+import 'package:weblibre/utils/uri_policy.dart';

 extension UriX on Uri {
  Uri get base => Uri.parse('$scheme://$authority');

-  bool get hasSupportedScheme => _supportedSchemes.contains(scheme);
+  bool get hasSupportedScheme => allSupportedSchemes.any((s) => s.name == scheme);

  bool get isHttp => isScheme('http');
  bool get isHttps => isScheme('https');
@@ -88,7 +88,14 @@ class EditBangScreen extends HookConsumerWidget {
                  return;
                }

-                final uri = Uri.parse(urlTextController.text);
+                final uri = parseValidatedUrl(
+                  urlTextController.text,
+                  eagerParsing: false,
+                  onlyHttpProtocol: true,
+                );
+                if (uri == null) {
+                  return;
+                }

                final bang = Bang(
                  group: BangGroup.user,
@@ -28,7 +28,7 @@ import 'package:weblibre/features/geckoview/features/bookmarks/domain/repositori
 import 'package:weblibre/features/geckoview/features/bookmarks/presentation/dialogs/delete_bookmark_dialog.dart';
 import 'package:weblibre/features/geckoview/features/bookmarks/presentation/widgets/folder_tree_picker.dart';
 import 'package:weblibre/utils/form_validators.dart';
-import 'package:weblibre/utils/uri_parser.dart' as uri_parser;
+import 'package:weblibre/utils/uri_input_parser.dart';

 class BookmarkEntryEditScreen extends HookConsumerWidget {
  final BookmarkInfo? initialInfo;
@@ -66,10 +66,17 @@ class BookmarkEntryEditScreen extends HookConsumerWidget {
          IconButton(
            onPressed: () async {
              if (formKey.currentState?.validate() ?? false) {
-                final newUrl = uri_parser.tryParseUrl(
+                var newUrl = parseValidatedUrl(
                  urlTextController.text,
                  eagerParsing: true,
-                )!;
+                  onlyHttpProtocol: true,
+                );
+
+                if (newUrl == null) {
+                  return;
+                }
+
+                newUrl = redactUriCredentials(newUrl);

                if (exisitingEntry != null) {
                  await ref
@@ -23,6 +23,7 @@
 import 'dart:convert';
 import 'package:flutter_mozilla_components/flutter_mozilla_components.dart';
 import 'package:weblibre/core/logger.dart';
+import 'package:weblibre/utils/uri_input_parser.dart';

 class BookmarkJSONUtils {
  final GeckoBookmarksService _service;
@@ -156,7 +157,11 @@ class BookmarkJSONUtils {
              await _service.addItem(parentGuid, uri, title, i);
              count++;
            } else {
-              logger.w('Skipping invalid URL: $url');
+              final parsed = Uri.tryParse(url);
+              final redacted = parsed != null
+                  ? redactUriCredentials(parsed)
+                  : url;
+              logger.w('Skipping invalid URL: $redacted');
            }
          } catch (e) {
            logger.e('Failed to import bookmark "$title": $e');
@@ -74,7 +74,10 @@ class OpenSharedContent extends HookConsumerWidget {
      return null;
    }, [currentUrl]);

-    final parsedDebouncedUrl = Uri.tryParse(debouncedUrl.value);
+    final parsedDebouncedUrl = parseValidatedUrl(
+      debouncedUrl.value,
+      eagerParsing: false,
+    );
    final hasExternalApp = useCachedFuture(
      // ignore: discarded_futures useFuture
      () => parsedDebouncedUrl != null
@@ -163,10 +166,18 @@ class OpenSharedContent extends HookConsumerWidget {

    Future<void> openTab(TabMode tabMode) async {
      if (formKey.currentState?.validate() == true) {
+        final parsedUrl = parseValidatedUrl(
+          textController.text,
+          eagerParsing: false,
+        );
+        if (parsedUrl == null) {
+          return;
+        }
+
        await ref
            .read(tabRepositoryProvider.notifier)
            .addTab(
-              url: Uri.parse(textController.text),
+              url: parsedUrl,
              tabMode: tabMode,
              containerSelection: selectedContainer.value == null
                  ? const TabContainerSelection.unassigned()
@@ -183,8 +194,16 @@ class OpenSharedContent extends HookConsumerWidget {

    Future<void> openCustomTab(bool isPrivate) async {
      if (formKey.currentState?.validate() == true) {
+        final parsedUrl = parseValidatedUrl(
+          textController.text,
+          eagerParsing: false,
+        );
+        if (parsedUrl == null) {
+          return;
+        }
+
        await GeckoBrowserService().openInCustomTab(
-          url: Uri.parse(textController.text),
+          url: parsedUrl,
          private: isPrivate,
          contextId: selectedContainer.value?.id,
        );
@@ -197,7 +216,7 @@ class OpenSharedContent extends HookConsumerWidget {

    Future<void> openInApp() async {
      if (formKey.currentState?.validate() == true) {
-        final uri = Uri.tryParse(textController.text);
+        final uri = parseValidatedUrl(textController.text, eagerParsing: false);
        if (uri == null) return;

        final success = await _appLinksService.openAppLink(uri);
@@ -20,14 +20,27 @@
 import 'package:flutter/material.dart';
 import 'package:flutter_mozilla_components/flutter_mozilla_components.dart'
    show GeckoBrowserService;
+
 import 'package:weblibre/utils/ui_helper.dart';
+import 'package:weblibre/utils/uri_input_parser.dart';
+import 'package:weblibre/utils/uri_policy.dart';

 Future<void> openInPrivateCustomTab(BuildContext context, String url) async {
  try {
-    await GeckoBrowserService().openInCustomTab(
-      url: Uri.parse(url),
-      private: true,
+    final parsedUrl = parseUserInputUrl(
+      url,
+      policy: SchemePolicy.internalIntent,
+      allowSchemelessHosts: true,
+      enforceMaxInputLength: true,
    );
+    if (parsedUrl == null) {
+      if (context.mounted) {
+        showErrorMessage(context, 'Could not open link: $url');
+      }
+      return;
+    }
+
+    await GeckoBrowserService().openInCustomTab(url: parsedUrl, private: true);
  } catch (e) {
    if (context.mounted) {
      showErrorMessage(context, 'Could not open link: $url');
@@ -47,8 +47,9 @@ import 'package:weblibre/features/geckoview/features/tabs/presentation/widgets/c
 import 'package:weblibre/features/user/domain/repositories/general_settings.dart';
 import 'package:weblibre/presentation/hooks/on_listenable_change_selector.dart';
 import 'package:weblibre/presentation/hooks/sampled_value_notifier.dart';
+import 'package:weblibre/utils/input_classification.dart';
 import 'package:weblibre/utils/text_field_line_count.dart';
-import 'package:weblibre/utils/uri_parser.dart' as uri_parser;
+import 'package:weblibre/utils/ui_helper.dart' as ui_helper;

 class SearchScreen extends HookConsumerWidget {
  final String? initialSearchText;
@@ -358,12 +359,26 @@ class SearchScreen extends HookConsumerWidget {
                      unfocusOnTapOutside: false,
                      onSubmitted: (value) async {
                        if (value.isNotEmpty) {
-                          var newUrl = uri_parser.tryParseUrl(
-                            value,
-                            eagerParsing: true,
-                          );
+                          final classification = classifyAddressBarInput(value);
+                          Uri? newUrl;
+                          String? searchQuery;

-                          if (newUrl == null) {
+                          switch (classification) {
+                            case NavigateInputClassification(:final uri):
+                              newUrl = uri;
+                            case SearchInputClassification(:final query):
+                              searchQuery = query;
+                            case InvalidInputClassification():
+                              if (context.mounted) {
+                                ui_helper.showErrorMessage(
+                                  context,
+                                  'Invalid address',
+                                );
+                              }
+                              return;
+                          }
+
+                          if (newUrl == null && searchQuery != null) {
                            // Read from both providers - use site if set, otherwise global
                            final siteBang = isEditMode
                                ? ref.read(
@@ -383,12 +398,12 @@ class SearchScreen extends HookConsumerWidget {
                                );

                            if (bang != null) {
-                              newUrl = bang.getTemplateUrl(value);
+                              newUrl = bang.getTemplateUrl(searchQuery);

                              if (!privateTabMode) {
                                await ref
                                    .read(bangSearchProvider.notifier)
-                                    .triggerBangSearch(bang, value);
+                                    .triggerBangSearch(bang, searchQuery);
                              }
                            }
                          }
@@ -26,7 +26,6 @@ import 'package:weblibre/features/geckoview/features/tabs/domain/repositories/co
 import 'package:weblibre/presentation/widgets/url_icon.dart';
 import 'package:weblibre/utils/form_validators.dart';
 import 'package:weblibre/utils/ui_helper.dart' as ui_helper;
-import 'package:weblibre/utils/uri_parser.dart' as uri_parser;

 class ContainerSitesScreen extends HookConsumerWidget {
  final Set<Uri> initialSites;
@@ -91,11 +90,16 @@ class ContainerSitesScreen extends HookConsumerWidget {
                        return uriValid;
                      }

-                      final origin = Uri.parse(
-                        uri_parser
-                            .tryParseUrl(value, eagerParsing: true)!
-                            .origin,
+                      final parsedUrl = parseValidatedUrl(
+                        value,
+                        eagerParsing: true,
+                        onlyHttpProtocol: true,
                      );
+                      if (parsedUrl == null) {
+                        return 'Invalid URL';
+                      }
+
+                      final origin = Uri.parse(parsedUrl.origin);

                      if (sites.value.contains(origin)) {
                        return 'This site has been already assigned';
@@ -104,11 +108,16 @@ class ContainerSitesScreen extends HookConsumerWidget {
                      return null;
                    },
                    onSaved: (newValue) async {
-                      final origin = Uri.parse(
-                        uri_parser
-                            .tryParseUrl(newValue, eagerParsing: true)!
-                            .origin,
+                      final parsedUrl = parseValidatedUrl(
+                        newValue,
+                        eagerParsing: true,
+                        onlyHttpProtocol: true,
                      );
+                      if (parsedUrl == null) {
+                        return;
+                      }
+
+                      final origin = Uri.parse(parsedUrl.origin);

                      final isAssigned = await ref
                          .read(containerRepositoryProvider.notifier)
@@ -18,14 +18,13 @@
 * along with this program. If not, see <http://www.gnu.org/licenses/>.
 */
 import 'package:fast_equatable/fast_equatable.dart';
-import 'package:weblibre/utils/uri_parser.dart' as uri_parser;
+import 'package:weblibre/utils/input_classification.dart';

 sealed class SharedContent with FastEquatable {
  SharedContent();

  factory SharedContent.parse(String content) {
-    if (uri_parser.tryParseUrl(content, eagerParsing: true)
-        case final Uri uri) {
+    if (parseSharedIntentUrl(content) case final Uri uri) {
      return SharedUrl(uri);
    } else {
      return SharedText(content);
@@ -24,7 +24,6 @@ import 'package:hooks_riverpod/hooks_riverpod.dart';
 import 'package:weblibre/core/routing/routes.dart';
 import 'package:weblibre/features/web_feed/domain/providers/add_dialog_blocking.dart';
 import 'package:weblibre/utils/form_validators.dart';
-import 'package:weblibre/utils/uri_parser.dart' as uri_parser;

 class AddFeedDialog extends HookConsumerWidget {
  final Uri? initialUri;
@@ -83,9 +82,16 @@ class AddFeedDialog extends HookConsumerWidget {
        TextButton(
          onPressed: () {
            if (formKey.currentState?.validate() == true) {
-              FeedCreateRoute(
-                feedId: uri_parser.tryParseUrl(textController.text)!,
-              ).pushReplacement(context);
+              final feedId = parseValidatedUrl(
+                textController.text,
+                eagerParsing: false,
+                onlyHttpProtocol: true,
+              );
+              if (feedId == null) {
+                return;
+              }
+
+              FeedCreateRoute(feedId: feedId).pushReplacement(context);
            }
          },
          child: const Text('Add'),
@@ -34,7 +34,6 @@ import 'package:weblibre/features/web_feed/presentation/widgets/tag_field.dart';
 import 'package:weblibre/presentation/widgets/failure_widget.dart';
 import 'package:weblibre/presentation/widgets/url_icon.dart';
 import 'package:weblibre/utils/form_validators.dart';
-import 'package:weblibre/utils/uri_parser.dart' as uri_parser;

 enum _DialogMode { create, edit }

@@ -156,19 +155,22 @@ class _FeedEditContent extends HookConsumerWidget {
            onPressed: () async {
              if (formKey.currentState?.validate() ?? false) {
                final feedData = FeedData(
-                  url: uri_parser.tryParseUrl(
+                  url: parseValidatedUrl(
                    urlTextController.text,
-                    eagerParsing: true,
+                    eagerParsing: false,
+                    onlyHttpProtocol: true,
                  )!,
                  authors: initialFeed.authors,
                  description: descriptionTextController.text.whenNotEmpty,
-                  icon: uri_parser.tryParseUrl(
+                  icon: parseValidatedUrl(
                    iconUrlTextController.text,
-                    eagerParsing: true,
+                    eagerParsing: false,
+                    onlyHttpProtocol: true,
                  ),
-                  siteLink: uri_parser.tryParseUrl(
+                  siteLink: parseValidatedUrl(
                    siteLinkTextController.text,
-                    eagerParsing: true,
+                    eagerParsing: false,
+                    onlyHttpProtocol: true,
                  ),
                  tags: tags.value.map((tag) => FeedCategory(id: tag)).toList(),
                  title: titleTextController.text.whenNotEmpty,
@@ -21,8 +21,25 @@ import 'dart:io';

 import 'package:nullability/nullability.dart';
 import 'package:path/path.dart' as p;
-import 'package:weblibre/extensions/uri.dart';
-import 'package:weblibre/utils/uri_parser.dart' as uri_parser;
+import 'package:weblibre/utils/uri_input_parser.dart';
+import 'package:weblibre/utils/uri_policy.dart';
+
+Uri? parseValidatedUrl(
+  String? value, {
+  required bool eagerParsing,
+  bool onlyHttpProtocol = false,
+}) {
+  final policy = onlyHttpProtocol
+      ? SchemePolicy.strictHttpOnly
+      : SchemePolicy.internalIntent;
+
+  return parseUserInputUrl(
+    value,
+    policy: policy,
+    allowSchemelessHosts: eagerParsing,
+    enforceMaxInputLength: true,
+  );
+}

 String? validateUrl(
  String? value, {
@@ -39,16 +56,18 @@ String? validateUrl(
    }
  }

-  if (uri_parser.tryParseUrl(value, eagerParsing: eagerParsing)
+  if (parseValidatedUrl(
+        value,
+        eagerParsing: eagerParsing,
+        onlyHttpProtocol: onlyHttpProtocol,
+      )
      case final Uri url) {
    if (!requireAuthority || url.authority.isNotEmpty) {
-      if (!onlyHttpProtocol || url.isHttpOrHttps) {
-        return null;
-      }
+      return null;
    }
  }

-  return 'Inavlid URL';
+  return 'Invalid URL';
 }

 String? validateRequired(String? value, {String message = 'Value required'}) {
@@ -0,0 +1,198 @@
+/*
+ * Copyright (c) 2024-2026 Fabian Freund.
+ *
+ * This file is part of WebLibre
+ * (see https://weblibre.eu).
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU Affero General Public License as
+ * published by the Free Software Foundation, either version 3 of the
+ * License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Affero General Public License for more details.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+import 'dart:io';
+
+import 'package:weblibre/utils/uri_input_parser.dart';
+import 'package:weblibre/utils/uri_policy.dart';
+
+enum NavigationReason {
+  explicitScheme,
+  schemelessHost,
+  localhost,
+  ipLiteral,
+  aboutLike,
+}
+
+enum SearchReason {
+  containsWhitespace,
+  notHostCandidate,
+  invalidHostChars,
+  fallback,
+}
+
+enum InvalidReason {
+  unsupportedScheme,
+  malformedUri,
+  containsControlChars,
+  emptyInput,
+  tooLong,
+}
+
+sealed class InputClassification {
+  const InputClassification();
+
+  const factory InputClassification.navigate(Uri uri, NavigationReason reason) =
+      NavigateInputClassification;
+  const factory InputClassification.search(String query, SearchReason reason) =
+      SearchInputClassification;
+  const factory InputClassification.invalid(String raw, InvalidReason reason) =
+      InvalidInputClassification;
+}
+
+final class NavigateInputClassification extends InputClassification {
+  final Uri uri;
+  final NavigationReason reason;
+
+  const NavigateInputClassification(this.uri, this.reason);
+}
+
+final class SearchInputClassification extends InputClassification {
+  final String query;
+  final SearchReason reason;
+
+  const SearchInputClassification(this.query, this.reason);
+}
+
+final class InvalidInputClassification extends InputClassification {
+  final String raw;
+  final InvalidReason reason;
+
+  const InvalidInputClassification(this.raw, this.reason);
+}
+
+InputClassification classifyAddressBarInput(
+  String input, {
+  SchemePolicy policy = SchemePolicy.addressBarTyped,
+}) {
+  final normalizedInput = normalizeInput(input);
+
+  if (normalizedInput.isEmpty) {
+    return InputClassification.invalid(
+      normalizedInput,
+      InvalidReason.emptyInput,
+    );
+  }
+
+  if (exceedsMaxInputLength(normalizedInput)) {
+    return InputClassification.invalid(normalizedInput, InvalidReason.tooLong);
+  }
+
+  if (containsControlChars(normalizedInput)) {
+    return InputClassification.invalid(
+      normalizedInput,
+      InvalidReason.containsControlChars,
+    );
+  }
+
+  if (hasExplicitScheme(normalizedInput)) {
+    final explicitUri = parseExplicitUri(normalizedInput, policy: policy);
+    if (explicitUri != null) {
+      final reason = explicitUri.isScheme('about')
+          ? NavigationReason.aboutLike
+          : NavigationReason.explicitScheme;
+
+      return InputClassification.navigate(explicitUri, reason);
+    }
+
+    if (hasAllowedScheme(normalizedInput, policy.allowedSchemes)) {
+      return InputClassification.invalid(
+        normalizedInput,
+        InvalidReason.malformedUri,
+      );
+    }
+
+    return InputClassification.invalid(
+      normalizedInput,
+      InvalidReason.unsupportedScheme,
+    );
+  }
+
+  if (normalizedInput.contains(whitespaceRegex)) {
+    return InputClassification.search(
+      normalizedInput,
+      SearchReason.containsWhitespace,
+    );
+  }
+
+  final schemelessUri = parseSchemelessWebHost(
+    normalizedInput,
+    allowedSchemes: policy.allowedSchemes,
+  );
+
+  if (schemelessUri != null) {
+    if (schemelessUri.host.toLowerCase() == 'localhost') {
+      return InputClassification.navigate(
+        schemelessUri,
+        NavigationReason.localhost,
+      );
+    }
+
+    if (InternetAddress.tryParse(schemelessUri.host) != null) {
+      return InputClassification.navigate(
+        schemelessUri,
+        NavigationReason.ipLiteral,
+      );
+    }
+
+    return InputClassification.navigate(
+      schemelessUri,
+      NavigationReason.schemelessHost,
+    );
+  }
+
+  if (looksLikeHostExpression(normalizedInput) &&
+      hasInvalidHostLikeChars(normalizedInput)) {
+    return InputClassification.search(
+      normalizedInput,
+      SearchReason.invalidHostChars,
+    );
+  }
+
+  return InputClassification.search(
+    normalizedInput,
+    SearchReason.notHostCandidate,
+  );
+}
+
+Uri? parseSharedIntentUrl(
+  String input, {
+  SchemePolicy policy = SchemePolicy.sharedIntent,
+}) {
+  final normalizedInput = normalizeInput(input);
+  if (normalizedInput.isEmpty ||
+      exceedsMaxInputLength(normalizedInput) ||
+      containsControlChars(normalizedInput)) {
+    return null;
+  }
+
+  if (hasExplicitScheme(normalizedInput)) {
+    return parseExplicitUri(normalizedInput, policy: policy);
+  }
+
+  if (normalizedInput.contains(whitespaceRegex)) {
+    return null;
+  }
+
+  return parseSchemelessWebHost(
+    normalizedInput,
+    allowedSchemes: policy.allowedSchemes,
+  );
+}
@@ -0,0 +1,245 @@
+/*
+ * Copyright (c) 2024-2026 Fabian Freund.
+ *
+ * This file is part of WebLibre
+ * (see https://weblibre.eu).
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU Affero General Public License as
+ * published by the Free Software Foundation, either version 3 of the
+ * License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Affero General Public License for more details.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+import 'dart:io';
+
+import 'package:weblibre/utils/uri_policy.dart';
+
+const maxUriInputLength = 4096;
+
+final _controlCharacterRegex = RegExp(r'[\x00-\x1F\u007F-\u009F]');
+final _explicitSchemeRegex = RegExp(r'^([a-zA-Z][a-zA-Z0-9+\-.]*):');
+final _hostPortSuffixRegex = RegExp(r'^\d{1,5}([/?#].*)?$');
+final _hostLabelRegex = RegExp(r'^[a-zA-Z0-9-]{1,63}$');
+final _topLevelDomainRegex = RegExp(r'^[a-zA-Z]{2,63}$');
+final _possibleHostCharsRegex = RegExp(
+  r"^[a-zA-Z0-9\-._~:/?#\[\]@!$&'()*+,;=%]+$",
+);
+final whitespaceRegex = RegExp(r'\s');
+
+String normalizeInput(String input) {
+  return input.trim();
+}
+
+bool containsControlChars(String input) {
+  return _controlCharacterRegex.hasMatch(input);
+}
+
+bool hasAllowedScheme(String input, Set<String> allowedSchemes) {
+  final scheme = _extractExplicitScheme(input);
+  if (scheme == null) {
+    return false;
+  }
+
+  return allowedSchemes.contains(scheme);
+}
+
+bool hasExplicitScheme(String input) {
+  final match = _explicitSchemeRegex.firstMatch(input);
+  if (match == null) {
+    return false;
+  }
+
+  final scheme = match.group(1)?.toLowerCase();
+  if (scheme == null) {
+    return false;
+  }
+
+  final suffix = input.substring(match.end);
+  if (suffix.startsWith('//')) {
+    return true;
+  }
+
+  // Keep host:port shorthand treated as schemeless input.
+  if ((scheme == 'localhost' || scheme.contains('.')) &&
+      _hostPortSuffixRegex.hasMatch(suffix)) {
+    return false;
+  }
+
+  return true;
+}
+
+bool exceedsMaxInputLength(String input, {int limit = maxUriInputLength}) {
+  return input.length > limit;
+}
+
+bool isValidHostCandidate(String hostCandidate) {
+  final host = hostCandidate.trim().toLowerCase();
+  if (host.isEmpty) {
+    return false;
+  }
+
+  if (host == 'localhost') {
+    return true;
+  }
+
+  if (InternetAddress.tryParse(host) != null) {
+    return true;
+  }
+
+  final labels = host.split('.');
+  if (labels.length < 2) {
+    return false;
+  }
+
+  if (!_topLevelDomainRegex.hasMatch(labels.last)) {
+    return false;
+  }
+
+  for (final label in labels) {
+    if (!_hostLabelRegex.hasMatch(label) ||
+        label.startsWith('-') ||
+        label.endsWith('-')) {
+      return false;
+    }
+  }
+
+  return true;
+}
+
+bool looksLikeHostExpression(String input) {
+  return input.contains('.') ||
+      input.contains(':') ||
+      input.startsWith('[') ||
+      input.contains('/');
+}
+
+bool hasInvalidHostLikeChars(String input) {
+  return !_possibleHostCharsRegex.hasMatch(input);
+}
+
+Uri? parseExplicitUri(
+  String input, {
+  required SchemePolicy policy,
+  bool enforceMaxInputLength = false,
+}) {
+  final normalizedInput = normalizeInput(input);
+  if (normalizedInput.isEmpty ||
+      (enforceMaxInputLength && exceedsMaxInputLength(normalizedInput)) ||
+      containsControlChars(normalizedInput) ||
+      !hasExplicitScheme(normalizedInput) ||
+      !hasAllowedScheme(normalizedInput, policy.allowedSchemes)) {
+    return null;
+  }
+
+  final uri = Uri.tryParse(normalizedInput);
+  if (uri == null) {
+    return null;
+  }
+
+  final scheme = uri.scheme.toLowerCase();
+  if (!policy.allows(scheme)) {
+    return null;
+  }
+
+  if (schemeRequiresAuthority(scheme) && uri.authority.isEmpty) {
+    return null;
+  }
+
+  return uri.replace(scheme: scheme);
+}
+
+Uri? parseSchemelessWebHost(
+  String input, {
+  Set<String>? allowedSchemes,
+  bool enforceMaxInputLength = false,
+}) {
+  final normalizedInput = normalizeInput(input);
+  if (normalizedInput.isEmpty ||
+      (enforceMaxInputLength && exceedsMaxInputLength(normalizedInput)) ||
+      containsControlChars(normalizedInput) ||
+      hasExplicitScheme(normalizedInput) ||
+      normalizedInput.contains(whitespaceRegex)) {
+    return null;
+  }
+
+  final probeUri = Uri.tryParse('https://$normalizedInput');
+  if (probeUri == null || probeUri.host.isEmpty) {
+    return null;
+  }
+
+  if (!isValidHostCandidate(probeUri.host)) {
+    return null;
+  }
+
+  if (probeUri.hasPort && (probeUri.port < 1 || probeUri.port > 65535)) {
+    return null;
+  }
+
+  final scheme = probeUri.host.toLowerCase() == 'localhost' ? 'http' : 'https';
+  if (allowedSchemes != null && !allowedSchemes.contains(scheme)) {
+    return null;
+  }
+
+  return probeUri.replace(scheme: scheme);
+}
+
+Uri? parseUserInputUrl(
+  String? input, {
+  required SchemePolicy policy,
+  bool allowSchemelessHosts = false,
+  bool enforceMaxInputLength = false,
+}) {
+  if (input == null) {
+    return null;
+  }
+
+  final normalizedInput = normalizeInput(input);
+  if (normalizedInput.isEmpty ||
+      (enforceMaxInputLength && exceedsMaxInputLength(normalizedInput)) ||
+      containsControlChars(normalizedInput)) {
+    return null;
+  }
+
+  if (hasExplicitScheme(normalizedInput)) {
+    return parseExplicitUri(
+      normalizedInput,
+      policy: policy,
+      enforceMaxInputLength: enforceMaxInputLength,
+    );
+  }
+
+  if (!allowSchemelessHosts) {
+    return null;
+  }
+
+  return parseSchemelessWebHost(
+    normalizedInput,
+    allowedSchemes: policy.allowedSchemes,
+    enforceMaxInputLength: enforceMaxInputLength,
+  );
+}
+
+Uri redactUriCredentials(Uri uri) {
+  if (uri.userInfo.isEmpty) {
+    return uri;
+  }
+
+  return uri.replace(userInfo: '');
+}
+
+String? _extractExplicitScheme(String input) {
+  final match = _explicitSchemeRegex.firstMatch(input);
+  if (match == null) {
+    return null;
+  }
+
+  return match.group(1)?.toLowerCase();
+}
@@ -18,49 +18,13 @@
 * along with this program. If not, see <http://www.gnu.org/licenses/>.
 */

-import 'dart:io';
-
-import 'package:weblibre/extensions/uri.dart';
-
-final _domainRegex = RegExp(r'\.[a-zA-Z]{2,}$');
+import 'package:weblibre/utils/uri_input_parser.dart';
+import 'package:weblibre/utils/uri_policy.dart';

 Uri? tryParseUrl(String? input, {bool eagerParsing = false}) {
-  if (input != null) {
-    var uri = Uri.tryParse(input);
-    if (uri != null) {
-      if (uri.authority.isEmpty && eagerParsing) {
-        if (uri.pathSegments.isNotEmpty) {
-          int? port;
-          var firstSegment = uri.pathSegments.first;
-
-          //When there is no scheme while aprsing, the port becomse the first segment because : is treated as delimeter
-          if (int.tryParse(firstSegment) case final int segmentPort) {
-            port = segmentPort;
-            firstSegment = uri.scheme;
-          }
-
-          if (_domainRegex.hasMatch(firstSegment)) {
-            uri = Uri.tryParse('https://$input')?.replace(port: port);
-          } else if (firstSegment == 'localhost') {
-            uri = Uri.tryParse('http://$input')?.replace(port: port);
-          } else if (InternetAddress.tryParse(firstSegment) != null) {
-            uri = Uri.tryParse('https://$input')?.replace(port: port);
-          }
-        }
-      }
-
-      if (uri != null) {
-        if (uri.isScheme('about')) {
-          return uri;
-        }
-
-        if (uri.authority.isNotEmpty) {
-          if (uri.hasSupportedScheme || !uri.hasScheme) {
-            return uri;
-          }
-        }
-      }
-    }
-  }
-  return null;
+  return parseUserInputUrl(
+    input,
+    policy: SchemePolicy.internalIntent,
+    allowSchemelessHosts: eagerParsing,
+  );
 }
@@ -0,0 +1,71 @@
+/*
+ * Copyright (c) 2024-2026 Fabian Freund.
+ *
+ * This file is part of WebLibre
+ * (see https://weblibre.eu).
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU Affero General Public License as
+ * published by the Free Software Foundation, either version 3 of the
+ * License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Affero General Public License for more details.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+class Scheme {
+  final String name;
+  final bool requiresAuthority;
+  const Scheme(this.name, {this.requiresAuthority = false});
+}
+
+const kSchemeHttp = Scheme('http', requiresAuthority: true);
+const kSchemeHttps = Scheme('https', requiresAuthority: true);
+const kSchemeFtp = Scheme('ftp', requiresAuthority: true);
+const kSchemeFile = Scheme('file');
+const kSchemeContent = Scheme('content', requiresAuthority: true);
+const kSchemeAbout = Scheme('about');
+const kSchemeMozExtension = Scheme('moz-extension', requiresAuthority: true);
+
+const allSupportedSchemes = [
+  kSchemeHttp,
+  kSchemeHttps,
+  kSchemeFtp,
+  kSchemeFile,
+  kSchemeContent,
+  kSchemeAbout,
+  kSchemeMozExtension,
+];
+
+const httpOnlySchemes = [kSchemeHttp, kSchemeHttps];
+
+bool schemeRequiresAuthority(String scheme) {
+  return allSupportedSchemes.any((s) => s.name == scheme && s.requiresAuthority);
+}
+
+enum SchemePolicy {
+  addressBarTyped,
+  strictHttpOnly,
+  sharedIntent,
+  internalIntent,
+}
+
+extension SchemePolicyX on SchemePolicy {
+  List<Scheme> get _schemes => switch (this) {
+    SchemePolicy.addressBarTyped => allSupportedSchemes,
+    SchemePolicy.strictHttpOnly => httpOnlySchemes,
+    SchemePolicy.sharedIntent => httpOnlySchemes,
+    SchemePolicy.internalIntent => allSupportedSchemes,
+  };
+
+  Set<String> get allowedSchemes => _schemes.map((s) => s.name).toSet();
+
+  bool allows(String scheme) {
+    return allowedSchemes.contains(scheme.toLowerCase());
+  }
+}
@@ -0,0 +1,45 @@
+/*
+ * Copyright (c) 2024-2026 Fabian Freund.
+ *
+ * This file is part of WebLibre
+ * (see https://weblibre.eu).
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU Affero General Public License as
+ * published by the Free Software Foundation, either version 3 of the
+ * License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Affero General Public License for more details.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+import 'package:flutter_test/flutter_test.dart';
+import 'package:weblibre/features/share_intent/domain/entities/shared_content.dart';
+
+void main() {
+  group('SharedContent.parse', () {
+    test('returns SharedUrl for explicit https uri', () {
+      final parsed = SharedContent.parse('https://weblibre.eu/path');
+
+      expect(parsed, isA<SharedUrl>());
+      expect((parsed as SharedUrl).url.host, 'weblibre.eu');
+    });
+
+    test('returns SharedText for explicit non-http scheme', () {
+      final parsed = SharedContent.parse('moz-extension://abc/index.html');
+
+      expect(parsed, isA<SharedText>());
+    });
+
+    test('returns SharedText for plain sentence', () {
+      final parsed = SharedContent.parse('WebLibre README.md');
+
+      expect(parsed, isA<SharedText>());
+    });
+  });
+}
@@ -0,0 +1,242 @@
+/*
+ * Copyright (c) 2024-2026 Fabian Freund.
+ *
+ * This file is part of WebLibre
+ * (see https://weblibre.eu).
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU Affero General Public License as
+ * published by the Free Software Foundation, either version 3 of the
+ * License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Affero General Public License for more details.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+import 'package:flutter_test/flutter_test.dart';
+import 'package:weblibre/utils/input_classification.dart';
+
+void main() {
+  group('classifyAddressBarInput', () {
+    test('searches free text containing whitespace', () {
+      final result = classifyAddressBarInput('WebLibre README.md');
+
+      expect(result, isA<SearchInputClassification>());
+      final search = result as SearchInputClassification;
+      expect(search.reason, SearchReason.containsWhitespace);
+      expect(search.query, 'WebLibre README.md');
+    });
+
+    test('navigates schemeless domain', () {
+      final result = classifyAddressBarInput('weblibre.eu');
+
+      expect(result, isA<NavigateInputClassification>());
+      final navigation = result as NavigateInputClassification;
+      expect(navigation.reason, NavigationReason.schemelessHost);
+      expect(navigation.uri.toString(), 'https://weblibre.eu');
+    });
+
+    test('navigates localhost over http', () {
+      final result = classifyAddressBarInput('localhost:8080');
+
+      expect(result, isA<NavigateInputClassification>());
+      final navigation = result as NavigateInputClassification;
+      expect(navigation.reason, NavigationReason.localhost);
+      expect(navigation.uri.toString(), 'http://localhost:8080');
+    });
+
+    test('navigates explicit moz-extension uri', () {
+      final result = classifyAddressBarInput('moz-extension://abc/index.html');
+
+      expect(result, isA<NavigateInputClassification>());
+      final navigation = result as NavigateInputClassification;
+      expect(navigation.reason, NavigationReason.explicitScheme);
+      expect(navigation.uri.scheme, 'moz-extension');
+    });
+
+    test('rejects explicit unsupported scheme as invalid', () {
+      final result = classifyAddressBarInput('myapp://callback?token=abc');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.unsupportedScheme);
+    });
+
+    test('rejects dotted explicit unsupported scheme as invalid', () {
+      final result = classifyAddressBarInput('my.app://callback');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.unsupportedScheme);
+    });
+
+    test('rejects javascript scheme as invalid', () {
+      final result = classifyAddressBarInput('JAVASCRIPT:alert(1)');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.unsupportedScheme);
+    });
+
+    test('does not decode encoded scheme before classification', () {
+      final result = classifyAddressBarInput('%6aavascript:alert(1)');
+
+      expect(result, isA<SearchInputClassification>());
+    });
+
+    test('rejects control chars as invalid', () {
+      final result = classifyAddressBarInput('example.com\x00path');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.containsControlChars);
+    });
+
+    test('invalid schemeless port falls back to search', () {
+      final result = classifyAddressBarInput('weblibre.eu:99999');
+
+      expect(result, isA<SearchInputClassification>());
+    });
+
+    test('explicit disallowed scheme with whitespace remains invalid', () {
+      final result = classifyAddressBarInput('javascript: alert(1)');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.unsupportedScheme);
+    });
+
+    test('unknown explicit scheme with whitespace remains invalid', () {
+      final result = classifyAddressBarInput('myapp: secret token');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.unsupportedScheme);
+    });
+
+    test('colon-prefixed query with whitespace is treated as invalid', () {
+      final result = classifyAddressBarInput('site:weblibre.eu privacy');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.unsupportedScheme);
+    });
+
+    test('symbolic colon query with whitespace is treated as invalid', () {
+      final result = classifyAddressBarInput('c++: tutorial');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.unsupportedScheme);
+    });
+
+    test('navigates schemeless domain with valid port', () {
+      final result = classifyAddressBarInput('weblibre.eu:8080');
+
+      expect(result, isA<NavigateInputClassification>());
+      final navigation = result as NavigateInputClassification;
+      expect(navigation.reason, NavigationReason.schemelessHost);
+      expect(navigation.uri.toString(), 'https://weblibre.eu:8080');
+    });
+
+    test('navigates IP literal', () {
+      final result = classifyAddressBarInput('192.168.1.1');
+
+      expect(result, isA<NavigateInputClassification>());
+      final navigation = result as NavigateInputClassification;
+      expect(navigation.reason, NavigationReason.ipLiteral);
+      expect(navigation.uri.toString(), 'https://192.168.1.1');
+    });
+
+    test('navigates explicit scheme with space in path', () {
+      final result = classifyAddressBarInput('https://example.com/a b');
+
+      expect(result, isA<NavigateInputClassification>());
+      final navigation = result as NavigateInputClassification;
+      expect(navigation.reason, NavigationReason.explicitScheme);
+      expect(navigation.uri.host, 'example.com');
+    });
+
+    test('rejects data scheme as invalid', () {
+      final result = classifyAddressBarInput(
+        'data:text/html,<script>alert(1)</script>',
+      );
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.unsupportedScheme);
+    });
+
+    test('treats missing scheme prefix as search', () {
+      final result = classifyAddressBarInput('://example.com');
+
+      expect(result, isA<SearchInputClassification>());
+    });
+
+    test('navigates URL with credentials', () {
+      final result = classifyAddressBarInput('https://user:pass@example.com');
+
+      expect(result, isA<NavigateInputClassification>());
+      final navigation = result as NavigateInputClassification;
+      expect(navigation.reason, NavigationReason.explicitScheme);
+      expect(navigation.uri.host, 'example.com');
+    });
+
+    test('rejects very long input as tooLong', () {
+      final longInput = 'a' * 5000;
+      final result = classifyAddressBarInput(longInput);
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.tooLong);
+    });
+
+    test('rejects empty input as emptyInput', () {
+      final result = classifyAddressBarInput('');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.emptyInput);
+    });
+
+    test('rejects whitespace-only input as emptyInput', () {
+      final result = classifyAddressBarInput('   ');
+
+      expect(result, isA<InvalidInputClassification>());
+      final invalid = result as InvalidInputClassification;
+      expect(invalid.reason, InvalidReason.emptyInput);
+    });
+  });
+
+  group('parseSharedIntentUrl', () {
+    test('parses explicit https', () {
+      final uri = parseSharedIntentUrl('https://weblibre.eu');
+
+      expect(uri, isNotNull);
+      expect(uri!.scheme, 'https');
+    });
+
+    test('treats explicit non-http scheme as non-url', () {
+      final uri = parseSharedIntentUrl('moz-extension://abc/index.html');
+
+      expect(uri, isNull);
+    });
+
+    test('treats explicit dotted non-http scheme as non-url', () {
+      final uri = parseSharedIntentUrl('my.app://callback');
+
+      expect(uri, isNull);
+    });
+
+    test('treats plain sentence as non-url', () {
+      final uri = parseSharedIntentUrl('This is WebLibre README.md');
+
+      expect(uri, isNull);
+    });
+  });
+}
@@ -0,0 +1,171 @@
+/*
+ * Copyright (c) 2024-2026 Fabian Freund.
+ *
+ * This file is part of WebLibre
+ * (see https://weblibre.eu).
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU Affero General Public License as
+ * published by the Free Software Foundation, either version 3 of the
+ * License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Affero General Public License for more details.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+import 'package:flutter_test/flutter_test.dart';
+import 'package:weblibre/utils/uri_input_parser.dart';
+import 'package:weblibre/utils/uri_policy.dart';
+
+void main() {
+  group('parseExplicitUri', () {
+    test('parses supported explicit scheme', () {
+      final uri = parseExplicitUri(
+        'https://weblibre.eu/path',
+        policy: SchemePolicy.addressBarTyped,
+      );
+
+      expect(uri, isNotNull);
+      expect(uri!.scheme, 'https');
+      expect(uri.host, 'weblibre.eu');
+    });
+
+    test('rejects unsupported explicit scheme', () {
+      final uri = parseExplicitUri(
+        'javascript:alert(1)',
+        policy: SchemePolicy.addressBarTyped,
+      );
+
+      expect(uri, isNull);
+    });
+
+    test('rejects malformed explicit uri with required authority', () {
+      final uri = parseExplicitUri(
+        'https:///path',
+        policy: SchemePolicy.addressBarTyped,
+      );
+
+      expect(uri, isNull);
+    });
+  });
+
+  group('parseSchemelessWebHost', () {
+    test('upgrades domain to https', () {
+      final uri = parseSchemelessWebHost('weblibre.eu');
+
+      expect(uri, isNotNull);
+      expect(uri!.toString(), 'https://weblibre.eu');
+    });
+
+    test('upgrades localhost to http', () {
+      final uri = parseSchemelessWebHost('localhost:8080');
+
+      expect(uri, isNotNull);
+      expect(uri!.toString(), 'http://localhost:8080');
+    });
+
+    test('rejects spaces in host candidate', () {
+      expect(parseSchemelessWebHost('foo bar.com'), isNull);
+    });
+
+    test('rejects invalid port range', () {
+      expect(parseSchemelessWebHost('weblibre.eu:99999'), isNull);
+    });
+  });
+
+  group('parseUserInputUrl', () {
+    test('parses schemeless host only when enabled', () {
+      expect(
+        parseUserInputUrl('weblibre.eu', policy: SchemePolicy.addressBarTyped),
+        isNull,
+      );
+
+      expect(
+        parseUserInputUrl(
+          'weblibre.eu',
+          policy: SchemePolicy.addressBarTyped,
+          allowSchemelessHosts: true,
+        ),
+        isNotNull,
+      );
+    });
+
+    test('rejects control chars', () {
+      expect(
+        parseUserInputUrl(
+          'example.com\x00path',
+          policy: SchemePolicy.addressBarTyped,
+          allowSchemelessHosts: true,
+        ),
+        isNull,
+      );
+    });
+
+    test('allows long persisted urls when max length is not enforced', () {
+      final longPath = 'a' * 5000;
+      final uri = parseUserInputUrl(
+        'https://weblibre.eu/$longPath',
+        policy: SchemePolicy.addressBarTyped,
+      );
+
+      expect(uri, isNotNull);
+    });
+
+    test('rejects long user input urls when max length is enforced', () {
+      final longPath = 'a' * 5000;
+      final uri = parseUserInputUrl(
+        'https://weblibre.eu/$longPath',
+        policy: SchemePolicy.addressBarTyped,
+        enforceMaxInputLength: true,
+      );
+
+      expect(uri, isNull);
+    });
+  });
+
+  group('redactUriCredentials', () {
+    test('strips userInfo from URI', () {
+      final uri = Uri.parse('https://user:pass@example.com/path');
+      final redacted = redactUriCredentials(uri);
+
+      expect(redacted.userInfo, isEmpty);
+      expect(redacted.host, 'example.com');
+      expect(redacted.path, '/path');
+      expect(redacted.toString(), 'https://example.com/path');
+    });
+
+    test('is no-op for URIs without credentials', () {
+      final uri = Uri.parse('https://example.com/path');
+      final redacted = redactUriCredentials(uri);
+
+      expect(identical(redacted, uri), isTrue);
+    });
+  });
+
+  group('isValidHostCandidate', () {
+    test('rejects single-label hosts', () {
+      expect(isValidHostCandidate('example'), isFalse);
+      expect(isValidHostCandidate('intranet'), isFalse);
+    });
+  });
+
+  group('containsControlChars', () {
+    test('detects null bytes', () {
+      expect(containsControlChars('example\x00.com'), isTrue);
+    });
+
+    test('detects C1 control chars', () {
+      expect(containsControlChars('example\u0080.com'), isTrue);
+      expect(containsControlChars('example\u009F.com'), isTrue);
+    });
+
+    test('allows normal text', () {
+      expect(containsControlChars('example.com'), isFalse);
+    });
+  });
+}