type: architecture title: Application state and event handling description: How the App struct in src/app.rs holds UI state, routes AppEvent values, handles keys, ingests messages and runs background tasks. tags: [app, state, events, keybindings, tokio] verified:
9fec9d1af4
resource: repo://src/app.rs
generated: { by: "claude-code", at: "2026-10-19T19:53:56.480Z" }
---src/app.rs owns the whole application state. The event loop described in
Runtime and event loop feeds every input into
App::handle_event, and the terminal UI renders the public
fields of App after each event. App is the only component that mutates the
history store in response to user and WhatsApp
activity.
AppEvent (src/app.rs:56-79) is the single message type of the main channel:
| Variant | Producer | Meaning |
|---|---|---|
Key(KeyEvent) |
input thread | A key press (release/repeat events are filtered out upstream). |
Redraw |
input thread | Terminal resized; nothing to change, just redraw. |
Wa(WaEvent) |
WhatsApp forwarder | An event from the WhatsApp client. |
AiReply { chat_jid, result } |
AI background task | Generated reply text, or an error string. |
CooldownElapsed { chat_jid } |
cooldown timer task | The cooldown of a chat with a deferred reply has ended. |
Sent { chat_jid, text, is_ai, result } |
send background task | Outcome of a text send. |
handle_event (src/app.rs:206-220) dispatches each variant. Only fatal
history (SQLite) errors propagate as Err; every other failure is turned into
a status text shown in the footer.
App (src/app.rs:128-158) exposes UI fields read by the renderer: screen,
qr_code, focus, list_mode, chats, contacts, selected, open_chat,
open_chat_name, open_chat_autoreply, messages, input, status and
should_quit. Private fields hold the History, the Config, an optional
AiClient, the autoreply ReplyGate, the own_jids_known and
catch_up_done flags, the WaHandle and a clone of the event sender used by
background tasks.
Screen (src/app.rs:81-87): Connecting → Pairing (QR shown) → Main.Focus (src/app.rs:89-94): List or Input.ListMode (src/app.rs:96-101): Chats or Contacts.App::new (src/app.rs:165-199) builds the AI client with build_ai, loads
chats and contacts from history, starts on Connecting with list focus in
chat mode, and creates the ReplyGate with the current time and the
Cooldowns { private, group } taken from cooldown_seconds and
group_cooldown_seconds, so that backlog messages are never auto-answered
live.
build_ai (src/app.rs:875-883) returns no client when system_prompt is
missing (status STATUS_AI_NOT_CONFIGURED) and reports a client build error
with the AI non disponibile prefix.
handle_wa_event (src/app.rs:226-251):
Qr → switch to Pairing and store the QR payload.Connected → switch to Main, status Connesso, refresh own JIDs.Disconnected / Error → update status only.LoggedOut → back to Connecting, status Sessione scollegata.OfflineSyncCompleted → on_offline_sync_completed, the startup catch-up
(see below).Contact → on_contact: upsert the contact, reload contacts and chats and
refresh the open chat name (src/app.rs:257-265).Message { message, image, addressed_to } → on_message
(src/app.rs:273-282).on_message saves the message and stops immediately when
History::save_message reports a duplicate, so history sync replays never
trigger UI reloads or autoreplies twice. New messages reload the chat list, the
open conversation and then go through maybe_autoreply together with the
addressed_to JIDs (mentions/quotes) used by the group cooldown (see
AI autoreply).
Own JIDs are fetched by refresh_own_jids (src/app.rs:456-463); the list is
empty while the device is not paired, so the refresh is retried later
(own_jids_known stays false until it succeeds).
Three paths start an AI reply, all through start_reply
(src/app.rs:403-416), which records the reply in the gate (cancelling any
pending deferral), loads the context with
History::messages_since(chat_jid, context_start(now), CONTEXT_MESSAGES) and
calls spawn_reply (src/app.rs:426-449) to build the turns and spawn the AI
task. ai_setup (src/app.rs:393-395) returns nothing, and so stops every
path, when the AI client or the system prompt is missing.
maybe_autoreply (src/app.rs:293-315) replies when
ReplyGate::should_reply allows it. Otherwise, if
ReplyGate::deferral_due returns a due time (private chats blocked only by
the cooldown), the chat is marked pending and spawn_cooldown_timer
(src/app.rs:749-755) sleeps until then and emits CooldownElapsed.handle_cooldown_elapsed (src/app.rs:324-332)
consumes the pending mark with take_pending; if it was still pending and
autoreply is still enabled it sends one text-only reply (no image).on_offline_sync_completed (src/app.rs:338-344)
runs catch_up once per process (catch_up_done). catch_up
(src/app.rs:353-369) asks catch_up_chats (src/app.rs:376-387) for the
autoreply-enabled chats whose latest stored message passes
ReplyGate::catch_up_due within catch_up_seconds, starts a text-only
reply for each and sets the status Risposte di recupero: N when N > 0.handle_key (src/app.rs:528-550):
| Key | Where | Action |
|---|---|---|
| Ctrl+C | any screen | Quit (should_quit = true). |
| Ctrl+A | main screen | Toggle autoreply for the open chat. |
| Tab | main screen | Switch focus between list and input. |
| F2 | main screen | Switch chat list / contact list; selection resets to 0. |
| Up / Down | list focus | Move selection, clamped to the list. |
| Enter | list focus | Open the selected chat or contact. |
| characters / Backspace | input focus | Edit the input line. |
| Esc | input focus | Return focus to the list. |
| Enter | input focus | Send the trimmed input to the open chat. |
Non-main screens ignore everything except Ctrl+C. is_command
(src/app.rs:905-907) treats Ctrl only as a command when Alt is not held,
because Windows reports AltGr as Ctrl+Alt (needed for @, #, € on the
Italian layout). Switching list mode with F2 also makes the event loop clear
the terminal (see Runtime and event loop).
open_selected (src/app.rs:630-644) loads display name, autoreply flag and
the last MESSAGES_SHOWN (200) messages, then moves focus to the input. The
reload_* helpers (src/app.rs:661-686) keep selected inside the active list.
toggle_autoreply (src/app.rs:610-624) always allows disabling. Enabling
goes through can_enable_autoreply (src/app.rs:890-898), which first
requires an AI client and then a chat accepted by supports_autoreply
(private chats and groups; broadcasts, status and newsletters are refused with
Autoreply disponibile solo nelle chat private e nei gruppi). The flag is
persisted with History::set_autoreply.
All network work runs in tokio tasks that report back through the event
channel; send_event ignores a closed channel because that only happens on
shutdown (src/app.rs:739-742).
submit_input (src/app.rs:586-594) ignores blank input or no open chat,
clears the input and calls spawn_send with is_ai: false.spawn_send (src/app.rs:859-868) calls WaHandle::send_text and emits
Sent.handle_sent (src/app.rs:498-522) stores a successful send as a
from_me message (with is_ai preserved) and reloads the views; a failure
sets Invio fallito: ….spawn_ai_request (src/app.rs:762-772) runs a text completion and emits
AiReply.spawn_vision_request (src/app.rs:781-791) runs one task that downloads
the image and calls the vision model, falling back to the text courtesy
turns on failure (complete_with_fallback, src/app.rs:799-812). When both
fail, combine_errors joins the two messages. Image bytes stay in memory.send_ai_reply_if_enabled (src/app.rs:483-492) re-checks the persisted
autoreply flag before sending an AI reply, discarding it with
Risposta AI scartata: autoreply disattivato if the user turned it off
meanwhile.Unit tests at src/app.rs:909 onward cover can_enable_autoreply (private,
self-chat LID, groups allowed; broadcasts/newsletters and missing AI refused),
vision_applies, combine_errors, selection clamping/stepping, the toggle
helpers and is_command. See Unit tests.