feat(keyboard): ship AI mode surface with streaming search answers

Add the AI keyboard tab, Agent settings, and user-owned LLM key path for 1.7.0, including streaming answers and web-search transports without the built-in DeepSeek fallback.
2026-08-11 01:06:27 +08:00
parent f6212dcd2c
commit 3de665d254
81 changed files with 4467 additions and 398 deletions
@@ -0,0 +1,162 @@
+// AIKeyboardCoordinator.swift
+// OSGKeyboard · Keyboard Extension
+//
+// Owns the temporary AI-mode UI state. Audio, ASR, and LLM work remain in the
+// shared Flow transport and host app; this coordinator never performs network
+// work and never inserts an answer before explicit user confirmation.
+
+import Foundation
+import OSGKeyboardShared
+
+@MainActor
+final class AIKeyboardCoordinator {
+    private let state: KeyboardState
+    private let flow: KeyboardFlowCoordinator
+    private let insertAnswer: (AIAnswer) -> Bool
+    private let performReturn: () -> Void
+
+    init(
+        state: KeyboardState,
+        flow: KeyboardFlowCoordinator,
+        insertAnswer: @escaping (AIAnswer) -> Bool,
+        performReturn: @escaping () -> Void
+    ) {
+        self.state = state
+        self.flow = flow
+        self.insertAnswer = insertAnswer
+        self.performReturn = performReturn
+    }
+
+    func beginNewPresentation() {
+        endConversationIfNeeded()
+        state.aiSession.enter()
+    }
+
+    func enterIfNeeded() {
+        guard !state.aiSession.isActive else { return }
+        state.aiSession.enter()
+    }
+
+    func leave() {
+        if state.aiSession.isBusy {
+            flow.cancelAIRecording()
+        }
+        endConversationIfNeeded()
+        state.aiSession.leave()
+    }
+
+    func toggleMicrophone() {
+        switch state.aiSession.phase {
+        case .listening:
+            guard let utteranceID = state.aiSession.activeUtteranceID else { return }
+            state.aiSession.beginRecognizing(utteranceID: utteranceID)
+            flow.stopAIRecording()
+        case .idle, .ready, .awaitingSend, .inserted, .sent, .failed:
+            enterIfNeeded()
+            guard let conversationID = state.aiSession.conversationID else { return }
+            let disposition = flow.beginAIRecording(conversationID: conversationID)
+            if case .rejected(let rejection) = disposition {
+                state.aiSession.fail(message(for: rejection), utteranceID: nil)
+            }
+        case .inactive:
+            enterIfNeeded()
+            toggleMicrophone()
+        case .preparing, .recognizing, .generating:
+            break
+        }
+    }
+
+    func cancel() {
+        guard state.aiSession.isBusy else { return }
+        flow.cancelAIRecording()
+        state.aiSession.cancelCurrentWork()
+    }
+
+    func sendLatestAnswer() {
+        if state.aiSession.canInsert, let answer = state.aiSession.answer {
+            guard insertAnswer(answer) else { return }
+            state.aiSession.markAnswerInserted(
+                offersSend: state.returnKeyRole == .send
+            )
+        } else if state.aiSession.canSend {
+            state.aiSession.markAnswerSent()
+            // Let the host consume the inserted answer before issuing Return.
+            Task { @MainActor [weak self] in
+                await Task.yield()
+                self?.performReturn()
+            }
+        }
+    }
+
+    func utterancePrepared(_ utteranceID: UUID) {
+        state.aiSession.beginPreparing(utteranceID: utteranceID)
+    }
+
+    func recordingStarted(_ utteranceID: UUID) {
+        state.aiSession.beginListening(utteranceID: utteranceID)
+    }
+
+    func recognitionStarted(_ utteranceID: UUID) {
+        state.aiSession.beginRecognizing(utteranceID: utteranceID)
+    }
+
+    func receiveTranscript(
+        _ transcript: String,
+        utteranceID: UUID,
+        status: FlowResult.Status
+    ) {
+        state.aiSession.updateTranscript(transcript, utteranceID: utteranceID)
+        if status == .rawReady {
+            state.aiSession.beginGenerating(
+                question: transcript,
+                utteranceID: utteranceID
+            )
+        }
+    }
+
+    func receivePartialAnswer(_ draft: String, utteranceID: UUID) {
+        state.aiSession.receivePartialAnswer(draft, utteranceID: utteranceID)
+    }
+
+    func receive(result: FlowResult) {
+        guard result.resolvedUtteranceMode == .aiQuestion else {
+            return
+        }
+        guard result.aiConversationID == state.aiSession.conversationID,
+              let answer = result.text,
+              !answer.isEmpty else {
+            state.aiSession.fail(
+                ExtL10n.string("keyboard.ai.error.requestFailed"),
+                utteranceID: result.utteranceId
+            )
+            return
+        }
+        state.aiSession.receiveAnswer(answer, utteranceID: result.utteranceId)
+    }
+
+    func fail(_ message: String, utteranceID: UUID?) {
+        state.aiSession.fail(message, utteranceID: utteranceID)
+    }
+
+    private func endConversationIfNeeded() {
+        guard let conversationID = state.aiSession.conversationID else { return }
+        flow.endAIConversation(conversationID)
+    }
+
+    private func message(for rejection: FlowUtteranceStartRejection) -> String {
+        switch rejection {
+        case .onboardingIncomplete:
+            return ExtL10n.string("keyboard.hint.finishSetupInApp")
+        case .missingAPIKey:
+            return ExtL10n.string("keyboard.ai.error.missingAPIKey")
+        case .noFullAccess:
+            return ExtL10n.string("keyboard.error.fullAccessRequired")
+        case .appGroupUnavailable:
+            return ExtL10n.string("keyboard.error.appGroupCommunication")
+        case .hostUnavailable:
+            return ExtL10n.string("keyboard.flow.hostDisconnected")
+        case .pipelineBusy:
+            return ExtL10n.string("keyboard.ai.error.pipelineBusy")
+        }
+    }
+}