Add command interface and status indicator.

This commit is contained in:
Drew 2026-02-24 14:23:33 -08:00
parent 5cb6647513
commit 7b9525ef95
7 changed files with 358 additions and 22 deletions

View file

@ -61,7 +61,7 @@ Events use parent IDs forming a tree (not a flat list). This enables future bran
- Unit tests go in the same file as the code (`#[cfg(test)] mod tests`) - Unit tests go in the same file as the code (`#[cfg(test)] mod tests`)
- Integration tests go in `tests/` - Integration tests go in `tests/`
- TUI widget tests use `ratatui::backend::TestBackend` + `insta` snapshots - TUI widget tests use `ratatui::backend::TestBackend`
- Provider tests replay recorded SSE fixtures from `tests/fixtures/` - Provider tests replay recorded SSE fixtures from `tests/fixtures/`
- Sandbox tests use `tempdir` and skip Landlock-specific assertions if kernel < 5.13 - Sandbox tests use `tempdir` and skip Landlock-specific assertions if kernel < 5.13
- Run `cargo test` before every commit - Run `cargo test` before every commit

View file

@ -67,7 +67,7 @@
- **`sandbox`:** Landlock policy construction, path validation logic (without applying kernel rules) - **`sandbox`:** Landlock policy construction, path validation logic (without applying kernel rules)
- **`core`:** Conversation tree operations (insert, query by parent, turn computation, token totals), orchestrator state machine transitions against mock `StreamEvent` sequences - **`core`:** Conversation tree operations (insert, query by parent, turn computation, token totals), orchestrator state machine transitions against mock `StreamEvent` sequences
- **`session`:** JSONL serialization roundtrips, parent ID chain reconstruction - **`session`:** JSONL serialization roundtrips, parent ID chain reconstruction
- **`tui`:** Widget rendering via Ratatui `TestBackend`, snapshot tests with `insta` crate for layout/mode indicator/token display - **`tui`:** Widget rendering via Ratatui `TestBackend`
### Integration Tests — Component Boundaries ### Integration Tests — Component Boundaries
- **Core ↔ Provider:** Mock `ModelProvider` replaying recorded API sessions (full SSE streams with tool use). Tests the complete orchestration loop deterministically without network. - **Core ↔ Provider:** Mock `ModelProvider` replaying recorded API sessions (full SSE streams with tool use). Tests the complete orchestration loop deterministically without network.
@ -78,11 +78,6 @@
- **Recorded session replay:** Capture real Claude API HTTP request/response pairs, replay deterministically. Exercises full stack (core + channel + mock TUI) without cost or network dependency. Primary E2E test strategy. - **Recorded session replay:** Capture real Claude API HTTP request/response pairs, replay deterministically. Exercises full stack (core + channel + mock TUI) without cost or network dependency. Primary E2E test strategy.
- **Live API tests:** Small suite behind feature flag / env var. Verifies real API integration. Run manually before releases, not in CI. - **Live API tests:** Small suite behind feature flag / env var. Verifies real API integration. Run manually before releases, not in CI.
### Snapshot Testing
- `insta` crate for TUI visual regression testing from Phase 2 onward
- Capture rendered `TestBackend` buffers as string snapshots
- Catches layout, mode indicator, and token display regressions
### Benchmarking — SWE-bench ### Benchmarking — SWE-bench
- **Target:** SWE-bench Verified (500 curated problems) as primary benchmark - **Target:** SWE-bench Verified (500 curated problems) as primary benchmark
- **Secondary:** SWE-bench Pro for testing planning mode on longer-horizon tasks - **Secondary:** SWE-bench Pro for testing planning mode on longer-horizon tasks
@ -93,7 +88,6 @@
### Test Sequencing ### Test Sequencing
- Phase 1: Unit tests for SSE parser, event types, message serialization - Phase 1: Unit tests for SSE parser, event types, message serialization
- Phase 2: Snapshot tests for TUI with `insta`
- Phase 4: Recorded session replay infrastructure (core loop complex enough to warrant it) - Phase 4: Recorded session replay infrastructure (core loop complex enough to warrant it)
- Phase 6-7: Headless mode + first SWE-bench Verified run - Phase 6-7: Headless mode + first SWE-bench Verified run

View file

@ -28,7 +28,6 @@
- Key dispatch: correct handler called per mode - Key dispatch: correct handler called per mode
- Command parser: `:quit`, `:clear`, `:q`, unknown command - Command parser: `:quit`, `:clear`, `:q`, unknown command
- Scroll clamping: j/k at boundaries, G/gg - Scroll clamping: j/k at boundaries, G/gg
- `insta` snapshot tests: mode indicator rendering for each mode, layout with status bar
- Existing Phase 1 input tests still pass (insert mode behavior unchanged) - Existing Phase 1 input tests still pass (insert mode behavior unchanged)
## Phase 3: Tool Execution ## Phase 3: Tool Execution

View file

@ -34,6 +34,11 @@ impl ConversationHistory {
pub fn messages(&self) -> &[ConversationMessage] { pub fn messages(&self) -> &[ConversationMessage] {
&self.messages &self.messages
} }
/// Remove all messages from the history.
pub fn clear(&mut self) {
self.messages.clear();
}
} }
impl Default for ConversationHistory { impl Default for ConversationHistory {
@ -73,6 +78,17 @@ mod tests {
assert_eq!(msgs[1].content, "hi there"); assert_eq!(msgs[1].content, "hi there");
} }
#[test]
fn clear_empties_history() {
let mut history = ConversationHistory::new();
history.push(ConversationMessage {
role: Role::User,
content: "hello".to_string(),
});
history.clear();
assert!(history.messages().is_empty());
}
#[test] #[test]
fn messages_preserves_insertion_order() { fn messages_preserves_insertion_order() {
let mut history = ConversationHistory::new(); let mut history = ConversationHistory::new();

View file

@ -65,6 +65,9 @@ impl<P: ModelProvider> Orchestrator<P> {
while let Some(action) = self.action_rx.recv().await { while let Some(action) = self.action_rx.recv().await {
match action { match action {
UserAction::Quit => break, UserAction::Quit => break,
UserAction::ClearHistory => {
self.history.clear();
}
UserAction::SendMessage(text) => { UserAction::SendMessage(text) => {
// Push the user message before snapshotting, so providers // Push the user message before snapshotting, so providers

View file

@ -20,6 +20,8 @@ pub enum UserAction {
SendMessage(String), SendMessage(String),
/// The user has requested to quit. /// The user has requested to quit.
Quit, Quit,
/// The user has requested to clear conversation history.
ClearHistory,
} }
/// An event sent from the core orchestrator to the TUI. /// An event sent from the core orchestrator to the TUI.

View file

@ -20,7 +20,7 @@ use crossterm::terminal::{
use futures::StreamExt; use futures::StreamExt;
use ratatui::backend::CrosstermBackend; use ratatui::backend::CrosstermBackend;
use ratatui::layout::{Constraint, Layout, Rect}; use ratatui::layout::{Constraint, Layout, Rect};
use ratatui::style::{Color, Style}; use ratatui::style::{Color, Modifier, Style};
use ratatui::text::{Line, Span}; use ratatui::text::{Line, Span};
use ratatui::widgets::{Block, Paragraph, Wrap}; use ratatui::widgets::{Block, Paragraph, Wrap};
use ratatui::{Frame, Terminal}; use ratatui::{Frame, Terminal};
@ -70,6 +70,8 @@ pub struct AppState {
pub pending_keys: Vec<char>, pub pending_keys: Vec<char>,
/// Last-known viewport height (output pane lines). Updated each frame by `update_scroll`. /// Last-known viewport height (output pane lines). Updated each frame by `update_scroll`.
pub viewport_height: u16, pub viewport_height: u16,
/// Transient error message shown in the status bar, cleared on next keypress.
pub status_error: Option<String>,
} }
impl AppState { impl AppState {
@ -81,6 +83,7 @@ impl AppState {
mode: Mode::Insert, mode: Mode::Insert,
pending_keys: Vec::new(), pending_keys: Vec::new(),
viewport_height: 0, viewport_height: 0,
status_error: None,
} }
} }
} }
@ -125,6 +128,8 @@ enum LoopControl {
SendMessage(String), SendMessage(String),
/// The user pressed Ctrl+C or Ctrl+D; exit the event loop. /// The user pressed Ctrl+C or Ctrl+D; exit the event loop.
Quit, Quit,
/// The user ran `:clear`; wipe the conversation.
ClearHistory,
} }
/// Map a key event to a [`LoopControl`] signal, mutating `state` as a side-effect. /// Map a key event to a [`LoopControl`] signal, mutating `state` as a side-effect.
@ -133,6 +138,8 @@ enum LoopControl {
/// handler for the current [`Mode`]. /// handler for the current [`Mode`].
fn handle_key(key: Option<KeyEvent>, state: &mut AppState) -> Option<LoopControl> { fn handle_key(key: Option<KeyEvent>, state: &mut AppState) -> Option<LoopControl> {
let key = key?; let key = key?;
// Clear any transient status error on the next keypress.
state.status_error = None;
// Ctrl+C quits from any mode. // Ctrl+C quits from any mode.
if key.modifiers.contains(KeyModifiers::CONTROL) && key.code == KeyCode::Char('c') { if key.modifiers.contains(KeyModifiers::CONTROL) && key.code == KeyCode::Char('c') {
return Some(LoopControl::Quit); return Some(LoopControl::Quit);
@ -269,6 +276,28 @@ fn handle_insert_key(key: KeyEvent, state: &mut AppState) -> Option<LoopControl>
} }
} }
/// Execute a parsed command string and return the appropriate control signal.
///
/// | Command | Effect |
/// |---------------|-------------------------------------------------|
/// | `quit` / `q` | Return `LoopControl::Quit` |
/// | `clear` | Clear messages and scroll, return `ClearHistory` |
/// | anything else | Set `status_error` on state |
fn execute_command(buf: &str, state: &mut AppState) -> Option<LoopControl> {
match buf.trim() {
"quit" | "q" => Some(LoopControl::Quit),
"clear" => {
state.messages.clear();
state.scroll = 0;
Some(LoopControl::ClearHistory)
}
other => {
state.status_error = Some(format!("Unknown command: {other}"));
None
}
}
}
/// Handle a keypress in Command mode (ex-style `:` command entry). /// Handle a keypress in Command mode (ex-style `:` command entry).
/// ///
/// | Key | Effect | /// | Key | Effect |
@ -289,14 +318,11 @@ fn handle_command_key(key: KeyEvent, state: &mut AppState) -> Option<LoopControl
// Already set to Normal above. // Already set to Normal above.
} }
KeyCode::Enter => { KeyCode::Enter => {
// Command execution deferred to phase 2.3; just return to Normal. return execute_command(&buf, state);
} }
KeyCode::Backspace => { KeyCode::Backspace => {
buf.pop(); buf.pop();
if !buf.is_empty() { state.mode = Mode::Command(buf);
state.mode = Mode::Command(buf);
}
// else stay in Normal (already set)
} }
KeyCode::Char(c) => { KeyCode::Char(c) => {
buf.push(c); buf.push(c);
@ -351,8 +377,8 @@ fn drain_ui_events(event_rx: &mut mpsc::Receiver<UIEvent>, state: &mut AppState)
/// - `ceil(chars / width).max(1)` lines per newline-separated content line /// - `ceil(chars / width).max(1)` lines per newline-separated content line
/// - 1 blank separator line /// - 1 blank separator line
fn update_scroll(state: &mut AppState, area: Rect) { fn update_scroll(state: &mut AppState, area: Rect) {
// 3 = height of the input pane (border top + content + border bottom) // 4 = input pane (3: border top + content + border bottom) + status bar (1)
let viewport_height = area.height.saturating_sub(3); let viewport_height = area.height.saturating_sub(4);
state.viewport_height = viewport_height; state.viewport_height = viewport_height;
let width = area.width.max(1) as usize; let width = area.width.max(1) as usize;
@ -392,11 +418,19 @@ fn update_scroll(state: &mut AppState, area: Rect) {
/// | Input | Length(3) /// | Input | Length(3)
/// | > _ | /// | > _ |
/// +--------------------------------+ /// +--------------------------------+
/// | NORMAL tokens: -- | Length(1)
/// +--------------------------------+
/// ``` /// ```
/// ///
/// Role headers are coloured: `"You:"` in cyan, `"Assistant:"` in green. /// Role headers are coloured: `"You:"` in cyan, `"Assistant:"` in green.
/// In Command mode, a one-line overlay appears at row 1 of the output pane.
fn render(frame: &mut Frame, state: &AppState) { fn render(frame: &mut Frame, state: &AppState) {
let chunks = Layout::vertical([Constraint::Fill(1), Constraint::Length(3)]).split(frame.area()); let chunks = Layout::vertical([
Constraint::Fill(1),
Constraint::Length(3),
Constraint::Length(1),
])
.split(frame.area());
// --- Output pane --- // --- Output pane ---
let mut lines: Vec<Line> = Vec::new(); let mut lines: Vec<Line> = Vec::new();
@ -417,10 +451,108 @@ fn render(frame: &mut Frame, state: &AppState) {
.scroll((state.scroll, 0)); .scroll((state.scroll, 0));
frame.render_widget(output, chunks[0]); frame.render_widget(output, chunks[0]);
// --- Command overlay (floating box centered on output pane) ---
if let Mode::Command(ref buf) = state.mode {
let out = chunks[0];
let overlay_w = (out.width / 2).max(80).min(out.width);
let overlay_h = 3; // border + content + border
let overlay_x = out.x + (out.width.saturating_sub(overlay_w)) / 2;
let overlay_y = out.y + (out.height.saturating_sub(overlay_h)) / 2;
let overlay_area = Rect {
x: overlay_x,
y: overlay_y,
width: overlay_w,
height: overlay_h.min(out.height),
};
// Clear the area behind the overlay so it appears floating.
let clear = Paragraph::new("");
frame.render_widget(clear, overlay_area);
let overlay = Paragraph::new(format!(":{buf}")).block(
Block::bordered()
.border_style(Style::default().fg(Color::Yellow))
.title("Command"),
);
frame.render_widget(overlay, overlay_area);
}
// --- Input pane --- // --- Input pane ---
let (input_title, input_style) = match state.mode {
Mode::Normal | Mode::Command(_) => (
"Input (normal)",
Style::default()
.fg(Color::DarkGray)
.add_modifier(Modifier::ITALIC),
),
Mode::Insert => ("Input", Style::default()),
};
let input_text = format!("> {}", state.input); let input_text = format!("> {}", state.input);
let input_widget = Paragraph::new(input_text).block(Block::bordered().title("Input")); let input_widget = Paragraph::new(input_text)
.style(input_style)
.block(Block::bordered().title(input_title));
frame.render_widget(input_widget, chunks[1]); frame.render_widget(input_widget, chunks[1]);
// --- Cursor positioning ---
match state.mode {
Mode::Insert => {
// Cursor at end of input text: border(1) + "> " (2) + input len
let cursor_x = chunks[1].x + 1 + 2 + state.input.len() as u16;
let cursor_y = chunks[1].y + 1; // inside the border
frame.set_cursor_position((cursor_x, cursor_y));
}
Mode::Command(ref buf) => {
// Cursor in the floating overlay: recalculate overlay position
let out = chunks[0];
let overlay_w = (out.width / 2).max(80).min(out.width);
let overlay_x = out.x + (out.width.saturating_sub(overlay_w)) / 2;
let overlay_y = out.y + (out.height.saturating_sub(3)) / 2;
// border(1) + ":" (1) + buf len
let cursor_x = overlay_x + 1 + 1 + buf.len() as u16;
let cursor_y = overlay_y + 1; // inside the border
frame.set_cursor_position((cursor_x, cursor_y));
}
Mode::Normal => {} // no cursor
}
// --- Status bar ---
let (mode_label, mode_style) = match state.mode {
Mode::Normal => (
" NORMAL ",
Style::default().bg(Color::Blue).fg(Color::White),
),
Mode::Insert => (
" INSERT ",
Style::default().bg(Color::Green).fg(Color::White),
),
Mode::Command(_) => (
" COMMAND ",
Style::default().bg(Color::Yellow).fg(Color::Black),
),
};
let right_text = if let Some(ref err) = state.status_error {
err.clone()
} else {
"tokens: --".to_string()
};
let right_style = if state.status_error.is_some() {
Style::default().fg(Color::Red)
} else {
Style::default()
};
let bar_width = chunks[2].width as usize;
let left_len = mode_label.len();
let right_len = right_text.len();
let pad = bar_width.saturating_sub(left_len + right_len);
let status_line = Line::from(vec![
Span::styled(mode_label, mode_style),
Span::raw(" ".repeat(pad)),
Span::styled(right_text, right_style),
]);
let status_widget = Paragraph::new(status_line);
frame.render_widget(status_widget, chunks[2]);
} }
/// Run the TUI event loop. /// Run the TUI event loop.
@ -482,6 +614,9 @@ pub async fn run(
let _ = action_tx.send(UserAction::Quit).await; let _ = action_tx.send(UserAction::Quit).await;
break; break;
} }
Some(LoopControl::ClearHistory) => {
let _ = action_tx.send(UserAction::ClearHistory).await;
}
None => {} None => {}
} }
} }
@ -694,11 +829,17 @@ mod tests {
} }
#[test] #[test]
fn command_backspace_empty_exits() { fn command_backspace_empty_stays_in_command() {
let mut state = AppState::new(); let mut state = AppState::new();
state.mode = Mode::Command(String::new()); state.mode = Mode::Command("ab".to_string());
handle_key(make_key(KeyCode::Backspace), &mut state); handle_key(make_key(KeyCode::Backspace), &mut state);
assert_eq!(state.mode, Mode::Normal); assert_eq!(state.mode, Mode::Command("a".to_string()));
handle_key(make_key(KeyCode::Backspace), &mut state);
assert_eq!(state.mode, Mode::Command(String::new()));
handle_key(make_key(KeyCode::Backspace), &mut state);
assert_eq!(state.mode, Mode::Command(String::new()));
handle_key(make_key(KeyCode::Backspace), &mut state);
assert_eq!(state.mode, Mode::Command(String::new()));
} }
#[test] #[test]
@ -804,4 +945,185 @@ mod tests {
handle_key(ctrl_key('u'), &mut state); handle_key(ctrl_key('u'), &mut state);
assert_eq!(state.scroll, 0); assert_eq!(state.scroll, 0);
} }
// --- execute_command tests ---
#[test]
fn command_quit_returns_quit() {
let mut state = AppState::new();
let result = execute_command("quit", &mut state);
assert!(matches!(result, Some(LoopControl::Quit)));
}
#[test]
fn command_q_returns_quit() {
let mut state = AppState::new();
let result = execute_command("q", &mut state);
assert!(matches!(result, Some(LoopControl::Quit)));
}
#[test]
fn command_clear_empties_messages() {
let mut state = AppState::new();
state.messages.push((Role::User, "hi".to_string()));
state.messages.push((Role::Assistant, "hello".to_string()));
state.scroll = 10;
let result = execute_command("clear", &mut state);
assert!(matches!(result, Some(LoopControl::ClearHistory)));
assert!(state.messages.is_empty());
assert_eq!(state.scroll, 0);
}
#[test]
fn command_unknown_sets_status_error() {
let mut state = AppState::new();
let result = execute_command("foo", &mut state);
assert!(result.is_none());
assert_eq!(state.status_error.as_deref(), Some("Unknown command: foo"));
}
#[test]
fn status_error_cleared_on_next_keypress() {
let mut state = AppState::new();
state.status_error = Some("some error".to_string());
handle_key(make_key(KeyCode::Char('h')), &mut state);
assert!(state.status_error.is_none());
}
#[test]
fn command_enter_executes_quit() {
let mut state = AppState::new();
state.mode = Mode::Command("q".to_string());
let result = handle_key(make_key(KeyCode::Enter), &mut state);
assert!(matches!(result, Some(LoopControl::Quit)));
assert_eq!(state.mode, Mode::Normal);
}
// --- render snapshot tests ---
#[test]
fn render_status_bar_normal_mode() {
let backend = TestBackend::new(80, 24);
let mut terminal = Terminal::new(backend).unwrap();
let mut state = AppState::new();
state.mode = Mode::Normal;
terminal.draw(|frame| render(frame, &state)).unwrap();
let buf = terminal.backend().buffer().clone();
let all_text: String = buf
.content()
.iter()
.map(|c| c.symbol().to_string())
.collect();
assert!(all_text.contains("NORMAL"), "expected 'NORMAL' in buffer");
assert!(
all_text.contains("tokens: --"),
"expected 'tokens: --' in buffer"
);
}
#[test]
fn render_status_bar_insert_mode() {
let backend = TestBackend::new(80, 24);
let mut terminal = Terminal::new(backend).unwrap();
let state = AppState::new(); // starts in Insert
terminal.draw(|frame| render(frame, &state)).unwrap();
let buf = terminal.backend().buffer().clone();
let all_text: String = buf
.content()
.iter()
.map(|c| c.symbol().to_string())
.collect();
assert!(all_text.contains("INSERT"), "expected 'INSERT' in buffer");
}
#[test]
fn render_status_bar_command_mode() {
let backend = TestBackend::new(80, 24);
let mut terminal = Terminal::new(backend).unwrap();
let mut state = AppState::new();
state.mode = Mode::Command("quit".to_string());
terminal.draw(|frame| render(frame, &state)).unwrap();
let buf = terminal.backend().buffer().clone();
let all_text: String = buf
.content()
.iter()
.map(|c| c.symbol().to_string())
.collect();
assert!(all_text.contains("COMMAND"), "expected 'COMMAND' in buffer");
}
#[test]
fn render_command_overlay_visible_in_command_mode() {
let backend = TestBackend::new(80, 24);
let mut terminal = Terminal::new(backend).unwrap();
let mut state = AppState::new();
state.mode = Mode::Command("quit".to_string());
terminal.draw(|frame| render(frame, &state)).unwrap();
let buf = terminal.backend().buffer().clone();
let all_text: String = buf
.content()
.iter()
.map(|c| c.symbol().to_string())
.collect();
assert!(
all_text.contains(":quit"),
"expected ':quit' overlay in buffer"
);
}
#[test]
fn render_no_overlay_in_normal_mode() {
let backend = TestBackend::new(80, 24);
let mut terminal = Terminal::new(backend).unwrap();
let mut state = AppState::new();
state.mode = Mode::Normal;
terminal.draw(|frame| render(frame, &state)).unwrap();
let buf = terminal.backend().buffer().clone();
// Row 1 should not have a ":" prefix from the overlay
let row1: String = (0..80)
.map(|x| buf.cell((x, 1)).unwrap().symbol().to_string())
.collect();
assert!(
!row1.starts_with(':'),
"overlay should not appear in Normal mode"
);
}
#[test]
fn render_input_pane_dimmed_in_normal() {
let backend = TestBackend::new(80, 24);
let mut terminal = Terminal::new(backend).unwrap();
let mut state = AppState::new();
state.mode = Mode::Normal;
terminal.draw(|frame| render(frame, &state)).unwrap();
let buf = terminal.backend().buffer().clone();
let all_text: String = buf
.content()
.iter()
.map(|c| c.symbol().to_string())
.collect();
assert!(
all_text.contains("Input (normal)"),
"expected 'Input (normal)' title in Normal mode"
);
}
#[test]
fn render_status_bar_shows_error() {
let backend = TestBackend::new(80, 24);
let mut terminal = Terminal::new(backend).unwrap();
let mut state = AppState::new();
state.status_error = Some("Unknown command: foo".to_string());
terminal.draw(|frame| render(frame, &state)).unwrap();
let buf = terminal.backend().buffer().clone();
let all_text: String = buf
.content()
.iter()
.map(|c| c.symbol().to_string())
.collect();
assert!(
all_text.contains("Unknown command: foo"),
"expected error in status bar"
);
}
} }