pkg/analysis_server/lib/src/legacy_analysis_server.dart

// Copyright (c) 2014, the Dart project authors. Please see the AUTHORS file
// for details. All rights reserved. Use of this source code is governed by a
// BSD-style license that can be found in the LICENSE file.

/// @docImport 'package:analysis_server_plugin/src/plugin_server.dart';
library;

import 'dart:async';
import 'dart:io' as io;
import 'dart:math' show max;

import 'package:analysis_server/lsp_protocol/protocol.dart' as lsp;
import 'package:analysis_server/protocol/protocol.dart';
import 'package:analysis_server/protocol/protocol_constants.dart';
import 'package:analysis_server/protocol/protocol_generated.dart'
    hide AnalysisOptions, MessageType;
import 'package:analysis_server/src/analysis_server.dart';
import 'package:analysis_server/src/analytics/analytics_manager.dart';
import 'package:analysis_server/src/channel/channel.dart';
import 'package:analysis_server/src/computer/computer_highlights.dart';
import 'package:analysis_server/src/domains/analysis/occurrences.dart';
import 'package:analysis_server/src/domains/analysis/occurrences_dart.dart';
import 'package:analysis_server/src/flutter/flutter_notifications.dart';
import 'package:analysis_server/src/handler/legacy/analysis_get_errors.dart';
import 'package:analysis_server/src/handler/legacy/analysis_get_hover.dart';
import 'package:analysis_server/src/handler/legacy/analysis_get_imported_elements.dart';
import 'package:analysis_server/src/handler/legacy/analysis_get_navigation.dart';
import 'package:analysis_server/src/handler/legacy/analysis_get_signature.dart';
import 'package:analysis_server/src/handler/legacy/analysis_reanalyze.dart';
import 'package:analysis_server/src/handler/legacy/analysis_set_analysis_roots.dart';
import 'package:analysis_server/src/handler/legacy/analysis_set_general_subscriptions.dart';
import 'package:analysis_server/src/handler/legacy/analysis_set_priority_files.dart';
import 'package:analysis_server/src/handler/legacy/analysis_set_subscriptions.dart';
import 'package:analysis_server/src/handler/legacy/analysis_update_content.dart';
import 'package:analysis_server/src/handler/legacy/analysis_update_options.dart';
import 'package:analysis_server/src/handler/legacy/analytics_enable.dart';
import 'package:analysis_server/src/handler/legacy/analytics_is_enabled.dart';
import 'package:analysis_server/src/handler/legacy/analytics_send_event.dart';
import 'package:analysis_server/src/handler/legacy/analytics_send_timing.dart';
import 'package:analysis_server/src/handler/legacy/completion_get_suggestion_details2.dart';
import 'package:analysis_server/src/handler/legacy/completion_get_suggestions2.dart';
import 'package:analysis_server/src/handler/legacy/diagnostic_get_diagnostics.dart';
import 'package:analysis_server/src/handler/legacy/diagnostic_get_server_port.dart';
import 'package:analysis_server/src/handler/legacy/edit_bulk_fixes.dart';
import 'package:analysis_server/src/handler/legacy/edit_format.dart';
import 'package:analysis_server/src/handler/legacy/edit_format_if_enabled.dart';
import 'package:analysis_server/src/handler/legacy/edit_get_assists.dart';
import 'package:analysis_server/src/handler/legacy/edit_get_available_refactorings.dart';
import 'package:analysis_server/src/handler/legacy/edit_get_fixes.dart';
import 'package:analysis_server/src/handler/legacy/edit_get_postfix_completion.dart';
import 'package:analysis_server/src/handler/legacy/edit_get_refactoring.dart';
import 'package:analysis_server/src/handler/legacy/edit_get_statement_completion.dart';
import 'package:analysis_server/src/handler/legacy/edit_import_elements.dart';
import 'package:analysis_server/src/handler/legacy/edit_is_postfix_completion_applicable.dart';
import 'package:analysis_server/src/handler/legacy/edit_list_postfix_completion_templates.dart';
import 'package:analysis_server/src/handler/legacy/edit_organize_directives.dart';
import 'package:analysis_server/src/handler/legacy/edit_sort_members.dart';
import 'package:analysis_server/src/handler/legacy/execution_create_context.dart';
import 'package:analysis_server/src/handler/legacy/execution_delete_context.dart';
import 'package:analysis_server/src/handler/legacy/execution_get_suggestions.dart';
import 'package:analysis_server/src/handler/legacy/execution_map_uri.dart';
import 'package:analysis_server/src/handler/legacy/execution_set_subscriptions.dart';
import 'package:analysis_server/src/handler/legacy/flutter_get_widget_description.dart';
import 'package:analysis_server/src/handler/legacy/flutter_set_subscriptions.dart';
import 'package:analysis_server/src/handler/legacy/flutter_set_widget_property_value.dart';
import 'package:analysis_server/src/handler/legacy/legacy_handler.dart';
import 'package:analysis_server/src/handler/legacy/lsp_notification_over_legacy_handler.dart';
import 'package:analysis_server/src/handler/legacy/lsp_over_legacy_handler.dart';
import 'package:analysis_server/src/handler/legacy/search_find_element_references.dart';
import 'package:analysis_server/src/handler/legacy/search_find_member_declarations.dart';
import 'package:analysis_server/src/handler/legacy/search_find_member_references.dart';
import 'package:analysis_server/src/handler/legacy/search_find_top_level_declarations.dart';
import 'package:analysis_server/src/handler/legacy/search_get_element_declarations.dart';
import 'package:analysis_server/src/handler/legacy/search_get_type_hierarchy.dart';
import 'package:analysis_server/src/handler/legacy/server_cancel_request.dart';
import 'package:analysis_server/src/handler/legacy/server_get_version.dart';
import 'package:analysis_server/src/handler/legacy/server_set_client_capabilities.dart';
import 'package:analysis_server/src/handler/legacy/server_set_subscriptions.dart';
import 'package:analysis_server/src/handler/legacy/server_shutdown.dart';
import 'package:analysis_server/src/handler/legacy/unsupported_request.dart';
import 'package:analysis_server/src/lsp/client_capabilities.dart' as lsp;
import 'package:analysis_server/src/lsp/client_configuration.dart' as lsp;
import 'package:analysis_server/src/lsp/constants.dart' as lsp;
import 'package:analysis_server/src/lsp/handlers/handler_states.dart';
import 'package:analysis_server/src/lsp/lsp_analysis_server.dart';
import 'package:analysis_server/src/operation/operation_analysis.dart';
import 'package:analysis_server/src/plugin/notification_manager.dart';
import 'package:analysis_server/src/protocol_server.dart' as server;
import 'package:analysis_server/src/scheduler/scheduled_message.dart';
import 'package:analysis_server/src/server/crash_reporting_attachments.dart';
import 'package:analysis_server/src/server/debounce_requests.dart';
import 'package:analysis_server/src/server/detachable_filesystem_manager.dart';
import 'package:analysis_server/src/server/diagnostic_server.dart';
import 'package:analysis_server/src/server/error_notifier.dart';
import 'package:analysis_server/src/server/features.dart';
import 'package:analysis_server/src/server/sdk_configuration.dart';
import 'package:analysis_server/src/services/completion/completion_state.dart';
import 'package:analysis_server/src/services/execution/execution_context.dart';
import 'package:analysis_server/src/services/flutter/widget_descriptions.dart';
import 'package:analysis_server/src/services/refactoring/legacy/refactoring_manager.dart';
import 'package:analysis_server/src/session_logger/session_logger.dart';
import 'package:analysis_server/src/status/performance_logger.dart';
import 'package:analysis_server/src/utilities/extensions/resource_provider.dart';
import 'package:analysis_server/src/utilities/process.dart';
import 'package:analysis_server_plugin/src/correction/performance.dart';
import 'package:analyzer/dart/analysis/results.dart';
import 'package:analyzer/dart/analysis/session.dart';
import 'package:analyzer/dart/ast/ast.dart';
import 'package:analyzer/exception/exception.dart';
import 'package:analyzer/file_system/file_system.dart';
import 'package:analyzer/instrumentation/instrumentation.dart';
import 'package:analyzer/src/analysis_options/analysis_options.dart';
import 'package:analyzer/src/dart/analysis/status.dart' as analysis;
import 'package:analyzer/src/generated/sdk.dart';
import 'package:analyzer/src/util/file_paths.dart' as file_paths;
import 'package:analyzer/src/util/performance/operation_performance.dart';
import 'package:analyzer/src/utilities/cancellation.dart';
import 'package:analyzer_plugin/protocol/protocol_common.dart' hide Element;
import 'package:analyzer_plugin/src/utilities/client_uri_converter.dart';
import 'package:analyzer_plugin/src/utilities/navigation/navigation.dart';
import 'package:analyzer_plugin/src/utilities/navigation/navigation_dart.dart';
import 'package:http/http.dart' as http;
import 'package:language_server_protocol/json_parsing.dart' as lsp;
import 'package:meta/meta.dart';
import 'package:telemetry/crash_reporting.dart';
import 'package:watcher/watcher.dart';

typedef AnalysisOptionsBuilderUpdater = void Function(
  AnalysisOptionsBuilder analysisOptionsBuilder,
);

/// A function that can be executed to create a handler for a request.
typedef HandlerGenerator = LegacyHandler Function(
  LegacyAnalysisServer,
  Request,
  CancellationToken,
  OperationPerformanceImpl,
);

/// Various IDE options.
class AnalysisServerOptions {
  String? newAnalysisDriverLog;

  String? clientId;
  String? clientVersion;

  /// Base path where to cache data.
  String? cacheFolder;

  /// If `true`, don't create a persistent byte store.
  ///
  /// Intended for benchmarking as it makes "instructions:u" in `perf stat` more
  /// stable, although it then doesn't benchmark that code.
  bool? disableFileByteStore;

  /// Whether to use the experimental single-file persistent byte store.
  /// Ignored when [disableFileByteStore] is `true` or a byte store is supplied.
  bool useSingleFileByteStore = false;

  /// The path to the package config file override.
  /// If `null`, then the default discovery mechanism is used.
  String? packagesFile;

  /// The crash report sender instance; note, this object can be `null`, and
  /// should be accessed via a null-aware operator.
  CrashReportSender? crashReportSender;

  /// An optional set of configuration overrides specified by the SDK.
  ///
  /// These overrides can provide new values for configuration settings, and are
  /// generally used in specific SDKs (like the internal google3 one).
  SdkConfiguration? configurationOverrides;

  /// Whether to use the Language Server Protocol.
  bool useLanguageServerProtocol = false;

  /// The set of enabled features.
  FeatureSet featureSet = FeatureSet();

  /// If set, this string will be reported as the protocol version.
  String? reportProtocolVersion;

  /// Experiments which have been enabled (or disabled) via the
  /// `--enable-experiment` command-line option.
  List<String> enabledExperiments = [];

  /// Whether to enable fine-grained dependencies.
  bool withFineDependencies = false;

  bool usePlugins = true;
}

/// Instances of the class [LegacyAnalysisServer] implement a server that
/// listens on a [ServerCommunicationChannel] for analysis requests and
/// processes them.
class LegacyAnalysisServer extends AnalysisServer {
  /// A map from the name of a request to a function used to create a request
  /// handler.
  ///
  /// Requests that don't match anything in this map will be passed to
  /// [LspOverLegacyHandler] for possible handling before returning an error.
  static final Map<String, HandlerGenerator> requestHandlerGenerators = {
    analysisRequestGetErrors: AnalysisGetErrorsHandler.new,
    analysisRequestGetHover: AnalysisGetHoverHandler.new,
    analysisRequestGetImportedElements: AnalysisGetImportedElementsHandler.new,
    analysisRequestGetLibraryDependencies: UnsupportedRequestHandler.new,
    analysisRequestGetNavigation: AnalysisGetNavigationHandler.new,
    analysisRequestGetReachableSources: UnsupportedRequestHandler.new,
    analysisRequestGetSignature: AnalysisGetSignatureHandler.new,
    analysisRequestReanalyze: AnalysisReanalyzeHandler.new,
    analysisRequestSetAnalysisRoots: AnalysisSetAnalysisRootsHandler.new,
    analysisRequestSetGeneralSubscriptions:
        AnalysisSetGeneralSubscriptionsHandler.new,
    analysisRequestSetPriorityFiles: AnalysisSetPriorityFilesHandler.new,
    analysisRequestSetSubscriptions: AnalysisSetSubscriptionsHandler.new,
    analysisRequestUpdateContent: AnalysisUpdateContentHandler.new,
    analysisRequestUpdateOptions: AnalysisUpdateOptionsHandler.new,

    analyticsRequestIsEnabled: AnalyticsIsEnabledHandler.new,
    analyticsRequestEnable: AnalyticsEnableHandler.new,
    analyticsRequestSendEvent: AnalyticsSendEventHandler.new,
    analyticsRequestSendTiming: AnalyticsSendTimingHandler.new,

    completionRequestGetSuggestionDetails2:
        CompletionGetSuggestionDetails2Handler.new,
    completionRequestGetSuggestions2: CompletionGetSuggestions2Handler.new,

    diagnosticRequestGetDiagnostics: DiagnosticGetDiagnosticsHandler.new,
    diagnosticRequestGetServerPort: DiagnosticGetServerPortHandler.new,

    editRequestFormat: EditFormatHandler.new,
    editRequestFormatIfEnabled: EditFormatIfEnabledHandler.new,
    editRequestGetAssists: EditGetAssistsHandler.new,
    editRequestGetAvailableRefactorings:
        EditGetAvailableRefactoringsHandler.new,
    editRequestBulkFixes: EditBulkFixes.new,
    editRequestGetFixes: EditGetFixesHandler.new,
    editRequestGetRefactoring: EditGetRefactoringHandler.new,
    editRequestImportElements: EditImportElementsHandler.new,
    editRequestOrganizeDirectives: EditOrganizeDirectivesHandler.new,
    editRequestSortMembers: EditSortMembersHandler.new,
    editRequestGetStatementCompletion: EditGetStatementCompletionHandler.new,
    editRequestIsPostfixCompletionApplicable:
        EditIsPostfixCompletionApplicableHandler.new,
    editRequestGetPostfixCompletion: EditGetPostfixCompletionHandler.new,
    editRequestListPostfixCompletionTemplates:
        EditListPostfixCompletionTemplatesHandler.new,

    executionRequestCreateContext: ExecutionCreateContextHandler.new,
    executionRequestDeleteContext: ExecutionDeleteContextHandler.new,
    executionRequestGetSuggestions: ExecutionGetSuggestionsHandler.new,
    executionRequestMapUri: ExecutionMapUriHandler.new,
    executionRequestSetSubscriptions: ExecutionSetSubscriptionsHandler.new,

    flutterRequestGetWidgetDescription: FlutterGetWidgetDescriptionHandler.new,
    flutterRequestSetWidgetPropertyValue:
        FlutterSetWidgetPropertyValueHandler.new,
    flutterRequestSetSubscriptions: FlutterSetSubscriptionsHandler.new,

    searchRequestFindElementReferences: SearchFindElementReferencesHandler.new,
    searchRequestFindMemberDeclarations:
        SearchFindMemberDeclarationsHandler.new,
    searchRequestFindMemberReferences: SearchFindMemberReferencesHandler.new,
    searchRequestFindTopLevelDeclarations:
        SearchFindTopLevelDeclarationsHandler.new,
    searchRequestGetElementDeclarations:
        SearchGetElementDeclarationsHandler.new,
    searchRequestGetTypeHierarchy: SearchGetTypeHierarchyHandler.new,

    serverRequestCancelRequest: ServerCancelRequestHandler.new,
    serverRequestGetVersion: ServerGetVersionHandler.new,
    serverRequestSetClientCapabilities: ServerSetClientCapabilitiesHandler.new,
    serverRequestSetSubscriptions: ServerSetSubscriptionsHandler.new,
    serverRequestShutdown: ServerShutdownHandler.new,

    lspRequestHandle: LspOverLegacyHandler.new,
  };

  /// The channel from which requests are received and to which responses should
  /// be sent.
  final ServerCommunicationChannel channel;

  /// The handler that is used for LSP messages once LSP initialization is
  /// complete (see [completeLspInitialization]).
  late final InitializedStateMessageHandler lspMessageHandler =
      InitializedStateMessageHandler(this);

  /// Whether the client has sent `server.setClientCapabilities`.
  ///
  /// A client that never sends it uses the default capabilities and
  /// configuration, so LSP initialization does not need to wait for anything
  /// (see [ensureLspInitializedForClientWithoutCapabilities]).
  bool _hasReceivedClientCapabilities = false;

  /// Whether either the last status message sent to the client or the last
  /// status message sent from any [PluginServer] indicated `isWorking: true`.
  @visibleForTesting
  bool statusAnalyzing = false;

  /// Whether the analysis server is currently analyzing (not including any
  /// plugins).
  bool serverStatusAnalyzing = false;

  /// A set of the [ServerService]s to send notifications for.
  Set<ServerService> serverServices = {};

  /// A table mapping request ids to cancellation tokens that allow cancelling
  /// the request.
  ///
  /// Tokens are removed once a request completes and should not be assumed to
  /// exist in this table just because cancellation was requested.
  Map<String, CancelableToken> cancellationTokens = {};

  /// A set of the [GeneralAnalysisService]s to send notifications for.
  Set<GeneralAnalysisService> generalAnalysisServices = {};

  /// A table mapping [AnalysisService]s to the file paths for which these
  /// notifications should be sent.
  Map<AnalysisService, Set<String>> analysisServices = {};

  /// The most recently registered set of client capabilities. The default is to
  /// have no registered requests and no additional capabilities.
  ///
  /// Must be modified through the [clientCapabilities] setter.
  ServerSetClientCapabilitiesParams _clientCapabilities =
      ServerSetClientCapabilitiesParams([]);

  /// See [editorClientCapabilities].
  var _editorClientCapabilities = lsp.fixedBasicLspClientCapabilities;

  @override
  final lsp.LspClientConfiguration lspClientConfiguration;

  /// A table mapping [FlutterService]s to the file paths for which these
  /// notifications should be sent.
  Map<FlutterService, Set<String>> flutterServices = {};

  /// The support for Flutter properties.
  WidgetDescriptions flutterWidgetDescriptions = WidgetDescriptions();

  /// The state used by the completion domain handlers.
  final CompletionState completionState = CompletionState();

  /// The object used to manage uncompleted refactorings.
  late RefactoringManager? _refactoringManager;

  /// The context used by the execution domain handlers.
  final ExecutionContext executionContext = ExecutionContext();

  /// The next search response id.
  int nextSearchId = 0;

  /// The [Completer] that completes when analysis is complete.
  ///
  /// This Completer is not used for communicating to the client whether we are
  /// analyzing; it is only used by some 'search_find' handlers, and in some tests.
  Completer<void>? _onAnalysisCompleteCompleter;

  /// The controller that is notified when analysis is started.
  final StreamController<bool> _onAnalysisStartedController =
      StreamController.broadcast();

  /// If the "analysis.analyzedFiles" notification is currently being subscribed
  /// to (see [generalAnalysisServices]), and at least one such notification has
  /// been sent since the subscription was enabled, the set of analyzed files
  /// that was delivered in the most recently sent notification.  Otherwise
  /// `null`.
  Set<String>? prevAnalyzedFiles;

  /// The controller for [onAnalysisSetChanged].
  final StreamController<void> _onAnalysisSetChangedController =
      StreamController.broadcast(sync: true);

  /// An optional manager to handle file systems which may not always be
  /// available.
  final DetachableFileSystemManager? detachableFileSystemManager;

  /// The broadcast stream of requests that were discarded because there
  /// was another request that made this one irrelevant.
  @visibleForTesting
  final StreamController<Request> discardedRequests =
      StreamController.broadcast(sync: true);

  /// The index of the next request from the server to the client.
  int nextServerRequestId = 0;

  /// A table mapping the ids of requests sent from the server to the client
  /// that have not yet received a response, to the completer used to return the
  /// response when it has been received.
  Map<String, Completer<Response>> pendingServerRequests = {};

  /// Whether the server should send LSP notifications.
  ///
  /// This is set once the client sends any LSP request or client capability
  /// that depends on LSP functionality.
  bool sendLspNotifications = false;

  /// Initialize a newly created server to receive requests from and send
  /// responses to the given [channel].
  new(
    this.channel,
    ResourceProvider baseResourceProvider,
    AnalysisServerOptions options,
    DartSdkManager sdkManager,
    AnalyticsManager analyticsManager,
    CrashReportingAttachmentsBuilder crashReportingAttachmentsBuilder,
    InstrumentationService instrumentationService,
    SessionLogger sessionLogger, {
    http.Client? httpClient,
    ProcessRunner? processRunner,
    super.requestStatistics,
    DiagnosticServer? diagnosticServer,
    this.detachableFileSystemManager,
    super.enableBlazeWatcher,
    super.dartFixPromptManager,
    super.providedByteStore,
    super.pluginManager,
    super.messageSchedulerListener,
    PerformanceLogger? performanceLogger,
    super.environment,
  }) : lspClientConfiguration = lsp.LspClientConfiguration(
         baseResourceProvider.pathContext,
       ),
       super(
         options,
         sdkManager,
         diagnosticServer,
         analyticsManager,
         crashReportingAttachmentsBuilder,
         baseResourceProvider,
         instrumentationService,
         sessionLogger,
         httpClient,
         processRunner,
         NotificationManager(channel, baseResourceProvider.pathContext),
         usePlugins: options.usePlugins,
       ) {
    notificationManager.analysisServer = this;
    var contextManagerCallbacks = ServerContextManagerCallbacks(
      this,
      resourceProvider,
    );
    contextManager.callbacks = contextManagerCallbacks;

    analysisDriverSchedulerEventsSubscription = analysisDriverScheduler.events
        .listen(handleAnalysisEvent);
    analysisDriverScheduler.start();

    onAnalysisStarted.first.then((_) {
      onAnalysisComplete.then((_) {
        performance = performanceAfterStartup = ServerPerformance();
      });
    });
    channel.sendNotification(
      ServerConnectedParams(
        options.reportProtocolVersion ?? PROTOCOL_VERSION,
        io.pid,
      ).toNotification(clientUriConverter: uriConverter),
    );
    debounceRequests(
      channel,
      discardedRequests,
    ).listen(handleClientMessage, onDone: done, onError: error);
    _newRefactoringManager();

    pluginManager.initializedCompleter.future.then((_) {
      // Perform "on idle" tasks in case the `pluginManger` determines that no
      // plugins should be run, _after_ the analysis server has reported its
      // final `isAnalyzing: false` status.
      _performOnIdleActions(
        // Use the existing plugin analyzing status.
        isPluginAnalyzing: notificationManager.pluginStatusAnalyzing,
      );
    });

    notificationManager.pluginAnalysisStatusChanges.listen((
      pluginStatusAnalyzing,
    ) {
      if (!pluginManager.initializedCompleter.isCompleted) {
        // Without `this.`, some portion of the analyzer believes we are
        // accessing the super parameter, instead of the field in the super
        // class.  See https://github.com/dart-lang/sdk/issues/59996.
        // ignore: unnecessary_this
        this.pluginManager.initializedCompleter.complete();
      } else {
        _performOnIdleActions(isPluginAnalyzing: pluginStatusAnalyzing);
      }
    });
  }

  /// The most recently registered set of client capabilities. The default is to
  /// have no registered requests and no additional capabilities.
  ServerSetClientCapabilitiesParams get clientCapabilities =>
      _clientCapabilities;

  /// Updates the current set of client capabilities.
  set clientCapabilities(ServerSetClientCapabilitiesParams capabilities) {
    _clientCapabilities = capabilities;
    _hasReceivedClientCapabilities = true;

    if (capabilities.supportsUris ?? false) {
      // URI support implies LSP, as that's the only way to access (and get
      // change notifications for) custom-scheme files.
      uriConverter = ClientUriConverter.withVirtualFileSupport(
        resourceProvider.pathContext,
      );
      // supportsUris implies LSP-over-Legacy support.
      initializeLspOverLegacy();
    } else {
      uriConverter = ClientUriConverter.noop(resourceProvider.pathContext);
    }

    if (capabilities.lspCapabilities
        case Map<Object?, Object?> lspCapabilities) {
      initializeLspOverLegacy();

      // First validate the capabilities so we can get a better message if it's
      // invalid.
      var reporter = lsp.LspJsonReporter();
      if (!lsp.ClientCapabilities.canParse(lspCapabilities, reporter)) {
        throw RequestError(
          RequestErrorCode.INVALID_PARAMETER,
          "The 'lspCapabilities' parameter was invalid: ${reporter.errors.join(', ')}",
        );
      }

      _editorClientCapabilities = lsp.LspClientCapabilities(
        lsp.ClientCapabilities.fromJson(lspCapabilities.cast<String, Object>()),
      );
    }
  }

  @override
  lsp.LspClientCapabilities get editorClientCapabilities =>
      _editorClientCapabilities;

  @override
  // Currently, we don't support this for LSP-over-Legacy.
  LspInitializationOptions? get initializationOptions => null;

  @override
  NotificationManager get notificationManager =>
      super.notificationManager as NotificationManager;

  /// The [Future] that completes when analysis is complete.
  ///
  /// This Future is not used for communicating to the client whether we are
  /// analyzing; it is only used by 'search_find' handlers, tests, and for
  /// performance calculations.
  Future<void> get onAnalysisComplete {
    if (_isAnalysisComplete) {
      return Future.value();
    }
    var completer = _onAnalysisCompleteCompleter ??= Completer<void>();
    return completer.future;
  }

  /// The stream that is notified when the analysis set is changed - this might
  /// be a change to a file, external via a watch event, or internal via
  /// overlay. This means that the resolved world might have changed.
  ///
  /// The type of produced elements is not specified and should not be used.
  Stream<void> get onAnalysisSetChanged =>
      _onAnalysisSetChangedController.stream;

  /// The stream that is notified with `true` when analysis is started.
  Stream<bool> get onAnalysisStarted {
    return _onAnalysisStartedController.stream;
  }

  @override
  OpenUriNotificationSender? get openUriNotificationSender {
    if (!clientCapabilities.requests.contains('openUrlRequest')) {
      return null;
    }

    return (Uri uri) async {
      var requestId = '${nextServerRequestId++}';
      await sendRequest(
        ServerOpenUrlRequestParams('$uri')
            .toRequest(requestId, clientUriConverter: uriConverter),
      );
    };
  }

  RefactoringManager? get refactoringManager {
    var refactoringManager = _refactoringManager;
    if (refactoringManager == null) {
      return null;
    }
    if (refactoringManager.hasPendingRequest) {
      refactoringManager.cancel();
      _newRefactoringManager();
    }
    return _refactoringManager;
  }

  String get sdkPath {
    return sdkManager.defaultSdkDirectory;
  }

  @override
  @protected
  bool get supportsShowMessageRequest =>
      clientCapabilities.requests.contains('showMessageRequest');

  // TODO(srawlins): Do we need to alter this to account for plugin status?
  bool get _isAnalysisComplete => !analysisDriverScheduler.isWorking;

  void cancelRequest(String id) {
    cancellationTokens[id]?.cancel();
  }

  /// Completes LSP initialization unless that has already happened.
  ///
  /// Unlike the LSP server, which completes initialization exactly once from
  /// the `initialized` notification, the legacy server has several triggers and
  /// some of them can run more than once: `server.setClientCapabilities` (which
  /// a client could send multiple times, but is not recommended) completes it
  /// either immediately or once the configuration fetch it started has
  /// finished, and every LSP-over-Legacy message from a client that has not
  /// sent capabilities completes it as well. Whichever runs first completes
  /// the initialization, the others must be no-ops.
  void completeLspInitializationIfNeeded() {
    if (!isLspInitialized) {
      completeLspInitialization(lspMessageHandler);
    }
  }

  /// The socket from which requests are being read has been closed.
  void done() {}

  /// Completes LSP initialization for a client that will never send
  /// `server.setClientCapabilities`.
  ///
  /// Such a client gets the default capabilities and configuration, so there is
  /// nothing to wait for. A client that does send capabilities completes the
  /// initialization itself once it has provided its configuration (see
  /// [ServerSetClientCapabilitiesHandler]).
  ///
  /// This is invoked for every LSP-over-Legacy message, so it is a no-op after
  /// the first one.
  void ensureLspInitializedForClientWithoutCapabilities() {
    if (!_hasReceivedClientCapabilities) {
      completeLspInitializationIfNeeded();
    }
  }

  /// There was an error related to the socket from which requests are being
  /// read.
  void error(Object argument) {}

  /// Return the cached analysis result for the file with the given [path].
  /// If there is no cached result, return `null`.
  ResolvedUnitResult? getCachedResolvedUnit(String path) {
    if (!file_paths.isDart(resourceProvider.pathContext, path)) {
      return null;
    }

    var driver = getAnalysisDriver(path);
    return driver?.getCachedResolvedUnit(path);
  }

  @override
  FutureOr<void> handleAnalysisStatusChange(analysis.AnalysisStatus status) {
    super.handleAnalysisStatusChange(status);
    sendStatusNotificationNew(status);
  }

  /// Handle a [message] that was read from the communication channel.
  void handleClientMessage(ClientMessage message) {
    switch (message) {
      case Request():
        var cancellationToken = CancelableToken();
        cancellationTokens[message.id] = cancellationToken;
        messageScheduler.add(
          LegacyMessage(request: message, cancellationToken: cancellationToken),
        );
      case Response():
        handleResponse(message);
      case Notification():
        messageScheduler.add(LegacyNotificationMessage(notification: message));
    }
  }

  /// Handle a [notification] that was read from the communication channel. The
  /// completer is used to indicate when the notification handling is done.
  void handleNotification(
    Notification notification,
    Completer<void> completer,
  ) {
    // Because we don't `await` the execution of the handler, we wrap the
    // execution in order to have one central place to handle exceptions.
    runZonedGuarded(
      () async {
        if (notification.event == lspNotificationNotification) {
          await LspNotificationOverLegacyHandler(this, notification).handle();
        } else {
          // There is no response to carry an error back to the client, so an
          // unknown notification can only be logged.
          instrumentationService.logError(
            'Unknown notification ${notification.event}',
          );
          sessionLogger.logException(
            exception: 'Unknown notification ${notification.event}',
          );
        }
        completer.complete();
      },
      (exception, stackTrace) {
        instrumentationService.logException(
          FatalException(
            'Failed to handle notification: ${notification.event}',
            exception,
            stackTrace,
          ),
          null,
          crashReportingAttachmentsBuilder.forException(exception),
        );
        completer.complete();
      },
    );
  }

  /// Handle a [request] that was read from the communication channel. The completer
  /// is used to indicate when the request handling is done.
  void handleRequest(
    Request request,
    Completer<void> completer,
    CancelableToken? cancellationToken,
  ) {
    var startTime = DateTime.now();
    performance.logRequestTiming(request.clientRequestTime);

    // Because we don't `await` the execution of the handlers, we wrap the
    // execution in order to have one central place to handle exceptions.
    runZonedGuarded(
      () async {
        // Record performance information for the request.
        var rootPerformance = OperationPerformanceImpl('<root>');
        RequestPerformance? requestPerformance;
        await rootPerformance.runAsync('request', (performance) async {
          requestPerformance = RequestPerformance(
            operation: request.method,
            performance: performance,
            requestLatency: request.timeSinceRequest,
            startTime: startTime,
          );
          recentPerformance.requests.add(requestPerformance!);

          var token = cancellationToken ??= CancelableToken();
          cancellationTokens[request.id] = token;
          var generator = requestHandlerGenerators[request.method];
          if (generator != null) {
            var handler = generator(this, request, token, performance);
            if (!handler.recordsOwnAnalytics) {
              analyticsManager.startedRequest(
                request: request,
                startTime: startTime,
              );
            }
            await handler.handle();
          } else {
            analyticsManager.startedRequest(
              request: request,
              startTime: startTime,
            );
            sendResponse(Response.unknownRequest(request));
          }
        });
        if (requestPerformance != null &&
            requestPerformance!.performance.elapsed >
                ServerRecentPerformance.slowRequestsThreshold) {
          recentPerformance.slowRequests.add(requestPerformance!);
        }
        completer.complete();
      },
      (exception, stackTrace) {
        if (exception is InconsistentAnalysisException) {
          sendResponse(Response.contentModified(request));
        } else if (exception is RequestFailure) {
          sendResponse(exception.response);
        } else {
          // Log the exception.
          var fatalException = FatalException(
            'Failed to handle request: ${request.method}',
            exception,
            stackTrace,
          );
          var attachments = crashReportingAttachmentsBuilder.forException(
            exception,
          );
          instrumentationService.logException(
            fatalException,
            null,
            attachments,
          );
          sessionLogger.logException(
            exception: fatalException,
            attachments: attachments,
          );
          // Then return an error response to the client.
          var error = RequestError(
            RequestErrorCode.SERVER_ERROR,
            exception.toString(),
          );
          error.stackTrace = stackTrace.toString();
          var response = Response(request.id, error: error);
          sendResponse(response);
        }
        completer.complete();
      },
      zoneValues: {
        // A container for capturing additional timings (such as analysis) stored
        // in zoneValues so it can be accessed anywhere in the processing of this
        // request.
        RequestPerformanceAdditionalTimings.zoneValueKey:
            RequestPerformanceAdditionalTimings(),
      },
    );
  }

  /// Handle a [response] that was read from the communication channel.
  void handleResponse(Response response) {
    var completer = pendingServerRequests.remove(response.id);
    if (completer != null) {
      completer.complete(response);
    }
  }

  /// Initializes LSP support over the legacy server.
  ///
  /// This method is called when the client sends an LSP request, or indicates
  /// that it will use LSP-overy-Legacy via client capabilities.
  ///
  /// This only applies to LSP over the legacy protocol and not DTD, since we
  /// do not want a DTD-LSP client to trigger LSP notifications going to the
  /// legacy protocol client, only the legacy protocol client should do that.
  ///
  /// For convenience, this method can be called multiple times from different
  /// code paths.
  void initializeLspOverLegacy() {
    sendLspNotifications = true;
  }

  @override
  void notifyFlutterWidgetDescriptions(String path) {
    flutterWidgetDescriptions.flush();
  }

  /// Send the given LSP [notification] to the client.
  @override
  void sendLspNotification(lsp.NotificationMessage notification) {
    if (!sendLspNotifications) {
      return;
    }

    channel.sendNotification(
      LspNotificationParams(notification)
          .toNotification(clientUriConverter: uriConverter),
    );
  }

  /// Sends an LSP request to the server (wrapped in 'lsp.handle') and unwraps
  /// the LSP response from the result of the legacy response.
  @override
  Future<lsp.ResponseMessage> sendLspRequest(
    lsp.Method method,
    Object params,
  ) async {
    var id = nextServerRequestId++;

    // Build the complete LSP RequestMessage to send.
    var lspMessage = lsp.RequestMessage(
      id: lsp.Either2<int, String>.t1(id),
      jsonrpc: lsp.jsonRpcVersion,
      method: method,
      params: params,
    );

    // Wrap the LSP message inside a call to lsp.handle.
    var response = await sendRequest(
      Request(
        id.toString(),
        lspRequestHandle,
        LspHandleParams(lspMessage).toJson(clientUriConverter: uriConverter),
      ),
    );

    // Unwrap the LSP response from the legacy response.
    var result = LspHandleResult.fromResponse(
      response,
      clientUriConverter: uriConverter,
    );
    var lspResponse = result.lspResponse;

    return lspResponse is Map<String, Object?>
        ? lsp.ResponseMessage.fromJson(lspResponse)
        : lsp.ResponseMessage(
            jsonrpc: lsp.jsonRpcVersion,
            error: lsp.ResponseError(
              code: lsp.ServerErrorCodes.unhandledError,
              message:
                  "The client responded to a '$method' LSP request but"
                  ' did not include a valid response in the lspResponse field',
            ),
          );
  }

  /// Send the given [notification] to the client.
  void sendNotification(Notification notification) {
    channel.sendNotification(notification);
  }

  /// Send the given [request] to the client.
  Future<Response> sendRequest(Request request) {
    var completer = Completer<Response>();
    pendingServerRequests[request.id] = completer;
    channel.sendRequest(request);
    return completer.future;
  }

  /// Send the given [response] to the client.
  void sendResponse(Response response) {
    channel.sendResponse(response);
    analyticsManager.sentResponse(response: response);
    cancellationTokens.remove(response.id);
  }

  /// If the [path] is not a valid file path, that is absolute and normalized,
  /// send an error response, and return `true`. If OK then return `false`.
  bool sendResponseErrorIfInvalidFilePath(Request request, String path) {
    if (!resourceProvider.isAbsoluteAndNormalized(path)) {
      sendResponse(Response.invalidFilePathFormat(request, path));
      return true;
    }
    return false;
  }

  /// Sends a `server.error` notification.
  @override
  void sendServerErrorNotification(
    String message,
    Object exception,
    StackTrace? stackTrace, {
    bool fatal = false,
  }) {
    var msg = '$message: $exception';
    if (stackTrace == null && exception is! CaughtException) {
      stackTrace = StackTrace.current;
    }

    // send the notification
    channel.sendNotification(
      ServerErrorParams(
        fatal,
        msg,
        '$stackTrace',
      ).toNotification(clientUriConverter: uriConverter),
    );

    // remember the last few exceptions
    if (exception is CaughtException) {
      stackTrace ??= exception.stackTrace;
    }

    exceptions.add(
      ServerException(
        message,
        exception,
        stackTrace is StackTrace ? stackTrace : StackTrace.current,
        fatal,
      ),
    );
  }

  /// Send status notification to the client. The state of analysis is given by
  /// the [status] information.
  void sendStatusNotificationNew(analysis.AnalysisStatus status) {
    var isServerAnalyzing = status.isWorking;
    if (isServerAnalyzing) {
      _onAnalysisStartedController.add(true);
    }
    var onAnalysisCompleteCompleter = _onAnalysisCompleteCompleter;
    if (onAnalysisCompleteCompleter != null && !isServerAnalyzing) {
      onAnalysisCompleteCompleter.complete();
      _onAnalysisCompleteCompleter = null;
    }
    serverStatusAnalyzing = isServerAnalyzing;

    _performOnIdleActions(
      // Use the existing plugin analyzing status.
      isPluginAnalyzing: notificationManager.pluginStatusAnalyzing,
    );
  }

  /// Implementation for `analysis.setAnalysisRoots`.
  ///
  // TODO(scheglov): implement complete projects/contexts semantics.
  //
  // The current implementation is intentionally simplified and expected
  // that only folders are given each given folder corresponds to the exactly
  // one context.
  //
  // So, we can start working in parallel on adding services and improving
  // projects/contexts support.
  Future<void> setAnalysisRoots(
    String requestId,
    List<String> includedPaths,
    List<String> excludedPaths,
  ) async {
    var completer = analysisContextRebuildCompleter = Completer();
    try {
      notificationManager.setAnalysisRoots(includedPaths, excludedPaths);
      try {
        await contextManager.setRoots(includedPaths, excludedPaths);
      } on UnimplementedError catch (e) {
        throw RequestFailure(
          Response.unsupportedFeature(
            requestId,
            e.message ?? 'Unsupported feature.',
          ),
        );
      }
    } finally {
      completer.complete();
    }
  }

  /// Implementation for `analysis.setSubscriptions`.
  void setAnalysisSubscriptions(
    Map<AnalysisService, Set<String>> subscriptions,
  ) {
    notificationManager.setSubscriptions(subscriptions);
    analysisServices = subscriptions;
    _sendSubscriptions(analysis: true);
  }

  /// Implementation for `flutter.setSubscriptions`.
  void setFlutterSubscriptions(Map<FlutterService, Set<String>> subscriptions) {
    flutterServices = subscriptions;
    _sendSubscriptions(flutter: true);
  }

  /// Implementation for `analysis.setGeneralSubscriptions`.
  void setGeneralAnalysisSubscriptions(
    List<GeneralAnalysisService> subscriptions,
  ) {
    var newServices = subscriptions.toSet();
    var newServicesContainsAnalyzedFiles = newServices.contains(
      GeneralAnalysisService.ANALYZED_FILES,
    );
    var generalServicesContainsAnalyzedFiles = generalAnalysisServices.contains(
      GeneralAnalysisService.ANALYZED_FILES,
    );
    if (newServicesContainsAnalyzedFiles &&
        !generalServicesContainsAnalyzedFiles &&
        _isAnalysisComplete) {
      sendAnalysisNotificationAnalyzedFiles(this);
    } else if (!newServicesContainsAnalyzedFiles &&
        generalServicesContainsAnalyzedFiles) {
      prevAnalyzedFiles = null;
    }
    generalAnalysisServices = newServices;
  }

  /// Set the priority files to the given [files].
  void setPriorityFiles(String requestId, List<String> files) {
    bool isPubspec(String filePath) =>
        file_paths.isPubspecYaml(resourceProvider.pathContext, filePath);

    // When pubspecs are opened, trigger pre-loading of pub package names and
    // versions.
    var pubspecs = files.where(isPubspec).toList();
    if (pubspecs.isNotEmpty) {
      pubPackageService.beginCachePreloads(pubspecs);
    }

    priorityFiles.clear();
    priorityFiles.addAll(files);
    // Set priority files in drivers.
    for (var driver in driverMap.values) {
      driver.priorityFiles = files;
    }
  }

  @override
  @visibleForOverriding
  Future<String?> showUserPrompt(
    MessageType type,
    String message,
    List<String> actionLabels,
    CancellationToken cancellationToken,
  ) async {
    assert(supportsShowMessageRequest);
    var requestId = (nextServerRequestId++).toString();
    var actions = actionLabels.map((label) => MessageAction(label)).toList();
    var request = ServerShowMessageRequestParams(
      type.forLegacy,
      message,
      actions,
    ).toRequest(requestId, clientUriConverter: uriConverter);
    var responseFuture = sendRequest(request);

    // Wait for either the result, or cancellation.
    await Future.any([responseFuture, cancellationToken.whenCancelled]);

    if (cancellationToken.isCancellationRequested) {
      return null;
    } else {
      // If we didn't enter the branch above, we know this future completed.
      var response = await responseFuture;
      return response.result?['action'] as String?;
    }
  }

  @override
  Future<void> shutdown() async {
    await super.shutdown();

    pubApi.close();

    detachableFileSystemManager?.dispose();

    // Defer closing the channel and shutting down the instrumentation server so
    // that the shutdown response can be sent and logged.
    unawaited(
      Future(() {
        instrumentationService.shutdown();
        channel.close();
      }),
    );
  }

  /// Implementation for `analysis.updateContent`.
  void updateContent(String id, Map<String, Object?> changes) {
    _onAnalysisSetChangedController.add(null);
    changes.forEach((file, change) {
      // Prepare the old overlay contents.
      String? oldContents;
      try {
        if (resourceProvider.hasOverlay(file)) {
          oldContents = resourceProvider.getFile(file).readAsStringSync();
        }
      } catch (_) {}

      // Prepare the new contents.
      String? newContents;
      int? newVersion;
      if (change is AddContentOverlay) {
        newContents = change.content;
        newVersion = change.version;
      } else if (change is ChangeContentOverlay) {
        if (oldContents == null) {
          // The client may only send a ChangeContentOverlay if there is
          // already an existing overlay for the source.
          throw RequestFailure(
            Response(
              id,
              error: RequestError(
                RequestErrorCode.INVALID_OVERLAY_CHANGE,
                'Invalid overlay change',
              ),
            ),
          );
        }
        try {
          newContents = SourceEdit.applySequence(oldContents, change.edits);
          newVersion = change.version;
        } on RangeError {
          throw RequestFailure(
            Response(
              id,
              error: RequestError(
                RequestErrorCode.INVALID_OVERLAY_CHANGE,
                'Invalid overlay change',
              ),
            ),
          );
        }
      } else if (change is RemoveContentOverlay) {
        newContents = null;
        newVersion = null;
      } else {
        // Protocol parsing should have ensured that we never get here.
        throw AnalysisException('Illegal change type');
      }

      if (newContents != null) {
        resourceProvider.setOverlay(
          file,
          content: newContents,
          modificationStamp: overlayModificationStamp++,
        );
      } else {
        resourceProvider.removeOverlay(file);
      }
      if (newVersion != null) {
        documentVersions[file] = newVersion;
      } else {
        documentVersions.remove(file);
      }

      for (var driver in driverMap.values) {
        driver.changeFile(file);
      }

      // If the file did not exist, and is "overlay only", it still should be
      // analyzed. Add it to driver to which it should have been added.
      contextManager.getDriverFor(file)?.addFile(file);

      notifyFlutterWidgetDescriptions(file);

      // TODO(scheglov): implement other cases
    });
  }

  /// Use the given updaters to configure the analysis options builders for
  /// existing analysis contexts.
  void updateOptions(List<AnalysisOptionsBuilderUpdater> builderUpdaters) {
    // TODO(scheglov): implement for the new analysis driver
    //    //
    //    // Update existing contexts.
    //    //
    //    for (AnalysisContext context in analysisContexts) {
    //      var builder = AnalysisOptionsBuilder.from(context.analysisOptions);
    //      builderUpdaters.forEach((
    //        AnalysisOptionsBuilderUpdater builderUpdater,
    //      ) {
    //        builderUpdater(builder);
    //      });
    //      context.analysisOptions = builder.build();
    //      // `TODO`(brianwilkerson) As far as I can tell, this doesn't cause analysis
    //      // to be scheduled for this context.
    //    }
    //    //
    //    // Update the defaults used to create new contexts.
    //    //
    //    builderUpdaters.forEach((
    //      AnalysisOptionsBuilderUpdater builderUpdater,
    //    ) {
    //      builderUpdater(defaultContextOptions);
    //    });
  }

  /// Returns `true` if there is a subscription for the given [service] and
  /// [file].
  bool _hasAnalysisServiceSubscription(AnalysisService service, String file) {
    return analysisServices[service]?.contains(file) ?? false;
  }

  bool _hasFlutterServiceSubscription(FlutterService service, String file) {
    return flutterServices[service]?.contains(file) ?? false;
  }

  /// Initializes [_refactoringManager] with a new instance.
  void _newRefactoringManager() {
    _refactoringManager = RefactoringManager(this, refactoringWorkspace);
  }

  /// Performs "on idle" actions, given either a new status for whether the
  /// server is analyzing, or a new status for whether the plugin isolate is
  /// analyzing.
  void _performOnIdleActions({required bool isPluginAnalyzing}) {
    // Perform on-idle actions.
    var isAnalyzing =
        serverStatusAnalyzing ||
        isPluginAnalyzing ||
        !pluginManager.initializedCompleter.isCompleted;
    if (!serverStatusAnalyzing) {
      if (generalAnalysisServices.contains(
        GeneralAnalysisService.ANALYZED_FILES,
      )) {
        sendAnalysisNotificationAnalyzedFiles(this);
      }
      _scheduleAnalysisImplementedNotification();
      filesResolvedSinceLastIdle.clear();
    }
    // Only send status when subscribed.
    if (!serverServices.contains(ServerService.STATUS)) {
      return;
    }

    // Only send status when it changes.
    if (statusAnalyzing == isAnalyzing) {
      return;
    }
    statusAnalyzing = isAnalyzing;
    if (!serverStatusAnalyzing) {
      // Only send analysis analytics after analysis is complete.
      reportAnalysisAnalytics();
    }
    var analysis = AnalysisStatus(isAnalyzing);
    channel.sendNotification(
      ServerStatusParams(analysis: analysis)
          .toNotification(clientUriConverter: uriConverter),
    );
  }

  void _scheduleAnalysisImplementedNotification() {
    var subscribed = analysisServices[AnalysisService.IMPLEMENTED];
    if (subscribed == null) {
      return;
    }

    var toSend = subscribed.intersection(filesResolvedSinceLastIdle);
    if (toSend.isEmpty) {
      return;
    }

    scheduleImplementedNotification(this, toSend);
  }

  void _sendSubscriptions({bool analysis = false, bool flutter = false}) {
    var files = <String>{};

    if (analysis) {
      for (var serviceFiles in analysisServices.values) {
        files.addAll(serviceFiles);
      }
    }

    if (flutter) {
      for (var serviceFiles in flutterServices.values) {
        files.addAll(serviceFiles);
      }
    }

    for (var file in files) {
      // The result will be produced by the "results" stream with
      // the fully resolved unit, and processed with sending analysis
      // notifications as it happens after content changes.
      if (file_paths.isDart(resourceProvider.pathContext, file)) {
        getResolvedUnit(file, sendCachedToStream: true, interactive: false);
      }
    }
  }
}

class ServerContextManagerCallbacks
    extends CommonServerContextManagerCallbacks {
  @override
  final LegacyAnalysisServer analysisServer;

  new(this.analysisServer, super.resourceProvider);

  AbstractNotificationManager get _notificationManager =>
      analysisServer.notificationManager;

  @override
  void afterContextsCreated() {
    super.afterContextsCreated();
    analysisServer._sendSubscriptions(analysis: true, flutter: true);
  }

  @override
  void afterWatchEvent(WatchEvent event) {
    analysisServer._onAnalysisSetChangedController.add(null);
  }

  @override
  void flushResults(List<String> files) {
    sendAnalysisNotificationFlushResults(analysisServer, files);
  }

  @override
  void handleResolvedUnitResult(ResolvedUnitResult result) {
    var path = result.path;

    var unit = result.unit;
    if (analysisServer._hasAnalysisServiceSubscription(
      AnalysisService.HIGHLIGHTS,
      path,
    )) {
      _runDelayed(() {
        _notificationManager.recordHighlightRegions(
          NotificationManager.serverId,
          path,
          _computeHighlightRegions(unit),
        );
      });
    }
    if (analysisServer._hasAnalysisServiceSubscription(
      AnalysisService.NAVIGATION,
      path,
    )) {
      _runDelayed(() {
        _notificationManager.recordNavigationParams(
          NotificationManager.serverId,
          path,
          _computeNavigationParams(path, result),
        );
      });
    }
    if (analysisServer._hasAnalysisServiceSubscription(
      AnalysisService.OCCURRENCES,
      path,
    )) {
      _runDelayed(() {
        _notificationManager.recordOccurrences(
          NotificationManager.serverId,
          path,
          _computeOccurrences(unit),
        );
      });
    }
    // if (analysisServer._hasAnalysisServiceSubscription(
    //     AnalysisService.OUTLINE, path)) {
    //   _runDelayed(() {
    //     // `TODO`(brianwilkerson) Change NotificationManager to store params
    //     // so that fileKind and libraryName can be recorded / passed along.
    //     notificationManager.recordOutlines(NotificationManager.serverId, path,
    //         _computeOutlineParams(path, unit, result.lineInfo));
    //   });
    // }
    if (analysisServer._hasAnalysisServiceSubscription(
      AnalysisService.CLOSING_LABELS,
      path,
    )) {
      // Closing labels are controlled by the subscription for the legacy server
      // even if we are configured to send the notifications over LSP. For
      // the real LSP server, closing labels are are automatically sent for all
      // open (priority files).
      _runDelayed(() {
        sendAnalysisNotificationClosingLabels(
          analysisServer,
          path,
          result.lineInfo,
          unit,
        );
      });
    }
    if (analysisServer._hasAnalysisServiceSubscription(
      AnalysisService.FOLDING,
      path,
    )) {
      _runDelayed(() {
        sendAnalysisNotificationFolding(
          analysisServer,
          path,
          result.lineInfo,
          unit,
        );
      });
    }
    if (analysisServer._hasAnalysisServiceSubscription(
      AnalysisService.OUTLINE,
      path,
    )) {
      _runDelayed(() {
        sendAnalysisNotificationOutline(analysisServer, result);
      });
    }
    if (analysisServer._hasAnalysisServiceSubscription(
      AnalysisService.OVERRIDES,
      path,
    )) {
      _runDelayed(() {
        sendAnalysisNotificationOverrides(analysisServer, path, unit);
      });
    }
    if (analysisServer._hasFlutterServiceSubscription(
      FlutterService.OUTLINE,
      path,
    )) {
      _runDelayed(() {
        sendFlutterNotificationOutline(analysisServer, result);
      });
    }
  }

  List<HighlightRegion> _computeHighlightRegions(CompilationUnit unit) {
    return DartUnitHighlightsComputer(unit).compute();
  }

  server.AnalysisNavigationParams _computeNavigationParams(
    String path,
    ParsedUnitResult result,
  ) {
    var collector = NavigationCollectorImpl();
    computeDartNavigation(resourceProvider, collector, result, null, null);
    collector.createRegions();
    return server.AnalysisNavigationParams(
      path,
      collector.regions,
      collector.targets,
      collector.files,
    );
  }

  List<Occurrences> _computeOccurrences(CompilationUnit unit) {
    var collector = OccurrencesCollectorImpl();
    addDartOccurrences(collector, unit);
    return collector.allOccurrences;
  }

  /// Run [f] in a new [Future].
  ///
  /// This method is used to delay sending notifications. If there is a more
  /// important consumer of an analysis results, specifically a code completion
  /// computer, we want it to run before spending time of sending notifications.
  ///
  // TODO(scheglov): Consider replacing this with full priority based scheduler.
  //
  // TODO(scheglov): Alternatively, if code completion work in a way that does
  // not produce (at first) fully resolved unit, but only part of it - a single
  // method, or a top-level declaration, we would not have this problem - the
  // completion computer would be the only consumer of the partial analysis
  // result.
  void _runDelayed(void Function() f) {
    Future(f);
  }
}

/// Used to record server exceptions.
class ServerException {
  final String message;
  final dynamic exception;
  final StackTrace stackTrace;
  final bool fatal;

  new(this.message, this.exception, this.stackTrace, this.fatal);

  @override
  String toString() => message;
}

/// A class used by [LegacyAnalysisServer] to record performance information
/// such as request latency.
class ServerPerformance {
  /// The creation time and the time when performance information
  /// started to be recorded here.
  final int startTime = DateTime.now().millisecondsSinceEpoch;

  /// The number of requests.
  int requestCount = 0;

  /// The number of requests that recorded latency information.
  int latencyCount = 0;

  /// The total latency (milliseconds) for all recorded requests.
  int requestLatency = 0;

  /// The maximum latency (milliseconds) for all recorded requests.
  int maxLatency = 0;

  /// The number of requests with latency > 150 milliseconds.
  int slowRequestCount = 0;

  /// Log timing information for a request.
  void logRequestTiming(int? clientRequestTime) {
    ++requestCount;
    if (clientRequestTime != null) {
      var latency = DateTime.now().millisecondsSinceEpoch - clientRequestTime;
      ++latencyCount;
      requestLatency += latency;
      maxLatency = max(maxLatency, latency);
      if (latency > 150) {
        ++slowRequestCount;
      }
    }
  }
}