Files
codex/codex-rs/tui_app_server/src/app_event.rs
T
Felipe Coury 334164a6f7 feat(tui): restore composer history in app-server tui (#14945)
## Problem

The app-server TUI (`tui_app_server`) lacked composer history support.
Pressing Up/Down to recall previous prompts hit a stub that logged a
warning and displayed "Not available in app-server TUI yet." New
submissions were silently dropped from the shared history file, so
nothing persisted for future sessions.

## Mental model

Codex maintains a single, append-only history file
(`$CODEX_HOME/history.jsonl`) shared across all TUI processes on the
same machine. The legacy (in-process) TUI already reads/writes this file
through `codex_core::message_history`. The app-server TUI delegates most
operations to a separate process over RPC, but history is intentionally
*not* an RPC concern — it's a client-local file.

This PR makes the app-server TUI access the same history file directly,
bypassing the app-server process entirely. The composer's Up/Down
navigation and submit-time persistence now follow the same code paths as
the legacy TUI, with the only difference being *where* the call is
dispatched (locally in `App`, rather than inside `CodexThread`).

The branch is rebuilt directly on top of `upstream/main`, so it keeps
the
existing app-server restore architecture intact.
`AppServerStartedThread`
still restores transcript history from the server `Thread` snapshot via
`thread_snapshot_events`; this PR only adds composer-history support.

## Non-goals

- Adding history support to the app-server protocol. History remains
client-local.
- Changing the on-disk format or location of `history.jsonl`.
- Surfacing history I/O errors to the user (failures are logged and
silently swallowed, matching the legacy TUI).

## Tradeoffs

| Decision | Why | Risk |
|----------|-----|------|
| Widen `message_history` from `pub(crate)` to `pub` | Avoids
duplicating file I/O logic; the module already has a clean, minimal API
surface. | Other workspace crates can now call these functions — the
contract is no longer crate-private. However, this is consistent with
recent precedent: `590cfa617` exposed `mention_syntax` for TUI
consumption, `752402c4f` exposed plugin APIs (`PluginsManager`), and
`14fcb6645`/`edacbf7b6` widened internal core APIs for other crates.
These were all narrow, intentional exposures of specific APIs — not
broad "make internals public" moves. `1af2a37ad` even went the other
direction, reducing broad re-exports to tighten boundaries. This change
follows the same pattern: a small, deliberate API surface (3 functions)
rather than a wholesale visibility change. |
| Intercept `AddToHistory` / `GetHistoryEntryRequest` in `App` before
RPC fallback | Keeps history ops out of the "unsupported op" error path
without changing app-server protocol. | This now routes through a single
`submit_thread_op` entry point, which is safer than the original
duplicated dispatch. The remaining risk is organizational: future
thread-op submission paths need to keep using that shared entry point. |
| `session_configured_from_thread_response` is now `async` | Needs
`await` on `history_metadata()` to populate real `history_log_id` /
`history_entry_count`. | Adds an async file-stat + full-file newline
scan to the session bootstrap path. The scan is bounded by
`history.max_bytes` and matches the legacy TUI's cost profile, but
startup latency still scales with file size. |

## Architecture

```
User presses Up                     User submits a prompt
       │                                    │
       ▼                                    ▼
ChatComposerHistory                 ChatWidget::do_submit_turn
  navigate_up()                       encode_history_mentions()
       │                                    │
       ▼                                    ▼
  AppEvent::CodexOp                  Op::AddToHistory { text }
  (GetHistoryEntryRequest)                  │
       │                                    ▼
       ▼                            App::try_handle_local_history_op
  App::try_handle_local_history_op    message_history::append_entry()
    spawn_blocking {                        │
      message_history::lookup()             ▼
    }                                $CODEX_HOME/history.jsonl
       │
       ▼
  AppEvent::ThreadEvent
  (GetHistoryEntryResponse)
       │
       ▼
  ChatComposerHistory::on_entry_response()
```

## Observability

- `tracing::warn` on `append_entry` failure (includes thread ID).
- `tracing::warn` on `spawn_blocking` lookup join error.
- `tracing::warn` from `message_history` internals on file-open, lock,
or parse failures.

## Tests

- `chat_composer_history::tests::navigation_with_async_fetch` — verifies
that Up emits `Op::GetHistoryEntryRequest` (was: checked for stub error
cell).
- `app::tests::history_lookup_response_is_routed_to_requesting_thread` —
verifies multi-thread composer recall routes the lookup result back to
the originating thread.
-
`app_server_session::tests::resume_response_relies_on_snapshot_replay_not_initial_messages`
— verifies app-server session restore still uses the upstream
thread-snapshot path.
-
`app_server_session::tests::session_configured_populates_history_metadata`
— verifies bootstrap sets nonzero `history_log_id` /
`history_entry_count` from the shared local history file.
2026-03-18 11:54:11 -06:00

496 lines
16 KiB
Rust

//! Application-level events used to coordinate UI actions.
//!
//! `AppEvent` is the internal message bus between UI components and the top-level `App` loop.
//! Widgets emit events to request actions that must be handled at the app layer (like opening
//! pickers, persisting configuration, or shutting down the agent), without needing direct access to
//! `App` internals.
//!
//! Exit is modelled explicitly via `AppEvent::Exit(ExitMode)` so callers can request shutdown-first
//! quits without reaching into the app loop or coupling to shutdown/exit sequencing.
use std::path::PathBuf;
use codex_app_server_protocol::McpServerStatus;
use codex_chatgpt::connectors::AppInfo;
use codex_file_search::FileMatch;
use codex_protocol::ThreadId;
use codex_protocol::openai_models::ModelPreset;
use codex_protocol::protocol::GetHistoryEntryResponseEvent;
use codex_protocol::protocol::Op;
use codex_protocol::protocol::RateLimitSnapshot;
use codex_utils_approval_presets::ApprovalPreset;
use crate::bottom_pane::ApprovalRequest;
use crate::bottom_pane::StatusLineItem;
use crate::history_cell::HistoryCell;
use codex_core::config::types::ApprovalsReviewer;
use codex_core::features::Feature;
use codex_protocol::config_types::CollaborationModeMask;
use codex_protocol::config_types::Personality;
use codex_protocol::config_types::ServiceTier;
use codex_protocol::openai_models::ReasoningEffort;
use codex_protocol::protocol::AskForApproval;
use codex_protocol::protocol::SandboxPolicy;
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub(crate) enum RealtimeAudioDeviceKind {
Microphone,
Speaker,
}
impl RealtimeAudioDeviceKind {
pub(crate) fn title(self) -> &'static str {
match self {
Self::Microphone => "Microphone",
Self::Speaker => "Speaker",
}
}
pub(crate) fn noun(self) -> &'static str {
match self {
Self::Microphone => "microphone",
Self::Speaker => "speaker",
}
}
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
pub(crate) enum WindowsSandboxEnableMode {
Elevated,
Legacy,
}
#[derive(Debug, Clone)]
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
pub(crate) struct ConnectorsSnapshot {
pub(crate) connectors: Vec<AppInfo>,
}
#[allow(clippy::large_enum_variant)]
#[derive(Debug)]
pub(crate) enum AppEvent {
/// Open the agent picker for switching active threads.
OpenAgentPicker,
/// Switch the active thread to the selected agent.
SelectAgentThread(ThreadId),
/// Submit an op to the specified thread, regardless of current focus.
SubmitThreadOp {
thread_id: ThreadId,
op: Op,
},
/// Deliver a synthetic history lookup response to a specific thread channel.
ThreadHistoryEntryResponse {
thread_id: ThreadId,
event: GetHistoryEntryResponseEvent,
},
/// Start a new session.
NewSession,
/// Clear the terminal UI (screen + scrollback), start a fresh session, and keep the
/// previous chat resumable.
ClearUi,
/// Open the resume picker inside the running TUI session.
OpenResumePicker,
/// Fork the current session into a new thread.
ForkCurrentSession,
/// Request to exit the application.
///
/// Use `ShutdownFirst` for user-initiated quits so core cleanup runs and the
/// UI exits only after `ShutdownComplete`. `Immediate` is a last-resort
/// escape hatch that skips shutdown and may drop in-flight work (e.g.,
/// background tasks, rollout flush, or child process cleanup).
Exit(ExitMode),
/// Request to exit the application due to a fatal error.
#[allow(dead_code)]
FatalExitRequest(String),
/// Forward an `Op` to the Agent. Using an `AppEvent` for this avoids
/// bubbling channels through layers of widgets.
CodexOp(Op),
/// Kick off an asynchronous file search for the given query (text after
/// the `@`). Previous searches may be cancelled by the app layer so there
/// is at most one in-flight search.
StartFileSearch(String),
/// Result of a completed asynchronous file search. The `query` echoes the
/// original search term so the UI can decide whether the results are
/// still relevant.
FileSearchResult {
query: String,
matches: Vec<FileMatch>,
},
/// Result of refreshing rate limits
#[allow(dead_code)]
RateLimitSnapshotFetched(RateLimitSnapshot),
/// Result of prefetching connectors.
ConnectorsLoaded {
result: Result<ConnectorsSnapshot, String>,
is_final: bool,
},
/// Result of computing a `/diff` command.
DiffResult(String),
/// Open the app link view in the bottom pane.
OpenAppLink {
app_id: String,
title: String,
description: Option<String>,
instructions: String,
url: String,
is_installed: bool,
is_enabled: bool,
},
/// Open the provided URL in the user's browser.
OpenUrlInBrowser {
url: String,
},
/// Refresh app connector state and mention bindings.
RefreshConnectors {
force_refetch: bool,
},
/// Fetch MCP inventory via app-server RPCs and render it into history.
FetchMcpInventory,
/// Result of fetching MCP inventory via app-server RPCs.
McpInventoryLoaded {
result: Result<Vec<McpServerStatus>, String>,
},
InsertHistoryCell(Box<dyn HistoryCell>),
/// Apply rollback semantics to local transcript cells.
///
/// This is emitted when rollback was not initiated by the current
/// backtrack flow so trimming occurs in AppEvent queue order relative to
/// inserted history cells.
ApplyThreadRollback {
num_turns: u32,
},
StartCommitAnimation,
StopCommitAnimation,
CommitTick,
/// Update the current reasoning effort in the running app and widget.
UpdateReasoningEffort(Option<ReasoningEffort>),
/// Update the current model slug in the running app and widget.
UpdateModel(String),
/// Update the active collaboration mask in the running app and widget.
UpdateCollaborationMode(CollaborationModeMask),
/// Update the current personality in the running app and widget.
UpdatePersonality(Personality),
/// Persist the selected model and reasoning effort to the appropriate config.
PersistModelSelection {
model: String,
effort: Option<ReasoningEffort>,
},
/// Persist the selected personality to the appropriate config.
PersistPersonalitySelection {
personality: Personality,
},
/// Persist the selected service tier to the appropriate config.
PersistServiceTierSelection {
service_tier: Option<ServiceTier>,
},
/// Open the device picker for a realtime microphone or speaker.
OpenRealtimeAudioDeviceSelection {
kind: RealtimeAudioDeviceKind,
},
/// Persist the selected realtime microphone or speaker to top-level config.
#[cfg_attr(
any(target_os = "linux", not(feature = "voice-input")),
allow(dead_code)
)]
PersistRealtimeAudioDeviceSelection {
kind: RealtimeAudioDeviceKind,
name: Option<String>,
},
/// Restart the selected realtime microphone or speaker locally.
RestartRealtimeAudioDevice {
kind: RealtimeAudioDeviceKind,
},
/// Open the reasoning selection popup after picking a model.
OpenReasoningPopup {
model: ModelPreset,
},
/// Open the Plan-mode reasoning scope prompt for the selected model/effort.
OpenPlanReasoningScopePrompt {
model: String,
effort: Option<ReasoningEffort>,
},
/// Open the full model picker (non-auto models).
OpenAllModelsPopup {
models: Vec<ModelPreset>,
},
/// Open the confirmation prompt before enabling full access mode.
OpenFullAccessConfirmation {
preset: ApprovalPreset,
return_to_permissions: bool,
},
/// Open the Windows world-writable directories warning.
/// If `preset` is `Some`, the confirmation will apply the provided
/// approval/sandbox configuration on Continue; if `None`, it performs no
/// policy change and only acknowledges/dismisses the warning.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
OpenWorldWritableWarningConfirmation {
preset: Option<ApprovalPreset>,
/// Up to 3 sample world-writable directories to display in the warning.
sample_paths: Vec<String>,
/// If there are more than `sample_paths`, this carries the remaining count.
extra_count: usize,
/// True when the scan failed (e.g. ACL query error) and protections could not be verified.
failed_scan: bool,
},
/// Prompt to enable the Windows sandbox feature before using Agent mode.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
OpenWindowsSandboxEnablePrompt {
preset: ApprovalPreset,
},
/// Open the Windows sandbox fallback prompt after declining or failing elevation.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
OpenWindowsSandboxFallbackPrompt {
preset: ApprovalPreset,
},
/// Begin the elevated Windows sandbox setup flow.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
BeginWindowsSandboxElevatedSetup {
preset: ApprovalPreset,
},
/// Begin the non-elevated Windows sandbox setup flow.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
BeginWindowsSandboxLegacySetup {
preset: ApprovalPreset,
},
/// Begin a non-elevated grant of read access for an additional directory.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
BeginWindowsSandboxGrantReadRoot {
path: String,
},
/// Result of attempting to grant read access for an additional directory.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
WindowsSandboxGrantReadRootCompleted {
path: PathBuf,
error: Option<String>,
},
/// Enable the Windows sandbox feature and switch to Agent mode.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
EnableWindowsSandboxForAgentMode {
preset: ApprovalPreset,
mode: WindowsSandboxEnableMode,
},
/// Update the Windows sandbox feature mode without changing approval presets.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
/// Update the current approval policy in the running app and widget.
UpdateAskForApprovalPolicy(AskForApproval),
/// Update the current sandbox policy in the running app and widget.
UpdateSandboxPolicy(SandboxPolicy),
/// Update the current approvals reviewer in the running app and widget.
UpdateApprovalsReviewer(ApprovalsReviewer),
/// Update feature flags and persist them to the top-level config.
UpdateFeatureFlags {
updates: Vec<(Feature, bool)>,
},
/// Update whether the full access warning prompt has been acknowledged.
UpdateFullAccessWarningAcknowledged(bool),
/// Update whether the world-writable directories warning has been acknowledged.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
UpdateWorldWritableWarningAcknowledged(bool),
/// Update whether the rate limit switch prompt has been acknowledged for the session.
UpdateRateLimitSwitchPromptHidden(bool),
/// Update the Plan-mode-specific reasoning effort in memory.
UpdatePlanModeReasoningEffort(Option<ReasoningEffort>),
/// Persist the acknowledgement flag for the full access warning prompt.
PersistFullAccessWarningAcknowledged,
/// Persist the acknowledgement flag for the world-writable directories warning.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
PersistWorldWritableWarningAcknowledged,
/// Persist the acknowledgement flag for the rate limit switch prompt.
PersistRateLimitSwitchPromptHidden,
/// Persist the Plan-mode-specific reasoning effort.
PersistPlanModeReasoningEffort(Option<ReasoningEffort>),
/// Persist the acknowledgement flag for the model migration prompt.
PersistModelMigrationPromptAcknowledged {
from_model: String,
to_model: String,
},
/// Skip the next world-writable scan (one-shot) after a user-confirmed continue.
#[cfg_attr(not(target_os = "windows"), allow(dead_code))]
SkipNextWorldWritableScan,
/// Re-open the approval presets popup.
OpenApprovalsPopup,
/// Open the skills list popup.
OpenSkillsList,
/// Open the skills enable/disable picker.
OpenManageSkillsPopup,
/// Enable or disable a skill by path.
SetSkillEnabled {
path: PathBuf,
enabled: bool,
},
/// Enable or disable an app by connector ID.
SetAppEnabled {
id: String,
enabled: bool,
},
/// Notify that the manage skills popup was closed.
ManageSkillsClosed,
/// Re-open the permissions presets popup.
OpenPermissionsPopup,
/// Live update for the in-progress voice recording placeholder. Carries
/// the placeholder `id` and the text to display (e.g., an ASCII meter).
#[cfg(not(target_os = "linux"))]
UpdateRecordingMeter {
id: String,
text: String,
},
/// Voice transcription finished for the given placeholder id.
#[cfg(not(target_os = "linux"))]
TranscriptionComplete {
id: String,
text: String,
},
/// Voice transcription failed; remove the placeholder identified by `id`.
#[cfg(not(target_os = "linux"))]
TranscriptionFailed {
id: String,
#[allow(dead_code)]
error: String,
},
/// Open the branch picker option from the review popup.
OpenReviewBranchPicker(PathBuf),
/// Open the commit picker option from the review popup.
OpenReviewCommitPicker(PathBuf),
/// Open the custom prompt option from the review popup.
OpenReviewCustomPrompt,
/// Submit a user message with an explicit collaboration mask.
SubmitUserMessageWithMode {
text: String,
collaboration_mode: CollaborationModeMask,
},
/// Open the approval popup.
FullScreenApprovalRequest(ApprovalRequest),
/// Open the feedback note entry overlay after the user selects a category.
OpenFeedbackNote {
category: FeedbackCategory,
include_logs: bool,
},
/// Open the upload consent popup for feedback after selecting a category.
OpenFeedbackConsent {
category: FeedbackCategory,
},
/// Launch the external editor after a normal draw has completed.
LaunchExternalEditor,
/// Async update of the current git branch for status line rendering.
StatusLineBranchUpdated {
cwd: PathBuf,
branch: Option<String>,
},
/// Apply a user-confirmed status-line item ordering/selection.
StatusLineSetup {
items: Vec<StatusLineItem>,
},
/// Dismiss the status-line setup UI without changing config.
StatusLineSetupCancelled,
/// Apply a user-confirmed syntax theme selection.
SyntaxThemeSelected {
name: String,
},
}
/// The exit strategy requested by the UI layer.
///
/// Most user-initiated exits should use `ShutdownFirst` so core cleanup runs and the UI exits only
/// after core acknowledges completion. `Immediate` is an escape hatch for cases where shutdown has
/// already completed (or is being bypassed) and the UI loop should terminate right away.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub(crate) enum ExitMode {
/// Shutdown core and exit after completion.
ShutdownFirst,
/// Exit the UI loop immediately without waiting for shutdown.
///
/// This skips `Op::Shutdown`, so any in-flight work may be dropped and
/// cleanup that normally runs before `ShutdownComplete` can be missed.
Immediate,
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub(crate) enum FeedbackCategory {
BadResult,
GoodResult,
Bug,
SafetyCheck,
Other,
}