基于NowInAndroid架构：如何从AppState向ModalBottomSheet传语音文本？

1. 调整ViewModel，管理朗读文本状态 #

在UniversalisViewModel中添加状态流保存当前朗读文本，并提供更新方法，符合UI层状态管理的职责：

@HiltViewModel
class UniversalisViewModel @Inject constructor(
    private val savedStateHandle: SavedStateHandle,
    val userDataRepository: UserDataRepository,
    getTopicWithDate: GetUniversalisUseCase,
) : ViewModel() {

    // 保存当前需朗读的文本，对外暴露只读流
    private val _currentVoiceText = MutableStateFlow("")
    val currentVoiceText: StateFlow<String> = _currentVoiceText.asStateFlow()

    // 更新朗读文本的方法
    fun updateCurrentVoiceText(text: String) {
        _currentVoiceText.value = text
    }

    // 原有uiState等代码保持不变
}

2. 修改Universalis可组合项，同步文本到ViewModel #

在Universalis可组合项中，利用LaunchedEffect监听data变化，将getAllForRead()的结果同步到ViewModel：

@Composable
fun Universalis(
    data: Universalis,
    topicId: String?,
    userData: UserDataDynamic,
    viewModel: UniversalisViewModel // 传入ViewModel
) {
    val itemId = topicId!!.toInt()
    val onTap = { point: Offset -> }
    
    if (data.liturgia!!.liturgiaTypus is Sortable) {
        data.liturgia!!.liturgiaTypus?.sort()
    }

    // 当data实例变化时，自动更新ViewModel中的朗读文本
    LaunchedEffect(data) {
        viewModel.updateCurrentVoiceText(data.getAllForRead())
    }

    // 原有UI渲染代码保持不变
    when (itemId) {
        1 -> MixtusScreen(
            data = data.liturgia!!.liturgiaTypus as LHMixtus,
            calendarTime = data.timeFK,
            userData = userData,
            onTap = onTap
        )
        2 -> OfficiumScreen(
            data = data.liturgia!!.liturgiaTypus as LHOfficium,
            calendarTime = data.timeFK,
            userData = userData,
            onTap = onTap
        )
    }
}

3. 连接AppState与ViewModel状态，控制BottomSheet #

在顶层组合中，订阅ViewModel的朗读文本状态，同步到AppState，并处理顶部栏点击逻辑：

@Composable
fun UniversalisScreen(
    viewModel: UniversalisViewModel = hiltViewModel(),
    appState: AppState = rememberAppState()
) {
    // 订阅ViewModel中的朗读文本
    val currentVoiceText by viewModel.currentVoiceText.collectAsState()

    // 同步文本到AppState
    LaunchedEffect(currentVoiceText) {
        appState.textToRead = currentVoiceText
    }

    // 顶部栏按钮逻辑
    NiaTopAppBar(
        // ... 其他配置
        readerIcon = NiaIcons.Reader,
        onReaderClick = { 
            appState.showBottomSheet = true
            // 此时appState.textToRead已同步为当前页面的朗读文本
        },
    )

    // 展示ModalBottomSheet
    if (appState.showBottomSheet) {
        ModalBottomSheet(onDismissRequest = { appState.showBottomSheet = false }) {
            TextToSpeechScreen(appState.textToRead)
        }
    }

    // 渲染Universalis内容
    val uiState by viewModel.uiState.collectAsState()
    when (uiState) {
        is UniversalisUiState.UniversalisData -> {
            val uiData = uiState as UniversalisUiState.UniversalisData
            // 根据selectedTopicId筛选当前显示的Universalis实例
            val currentUniversalis = uiData.topics.firstOrNull { 
                it.id == uiData.selectedTopicId?.toInt() 
            }
            currentUniversalis?.let {
                Universalis(
                    data = it,
                    topicId = uiData.selectedTopicId,
                    userData = // 从userDataRepository获取用户数据
                    viewModel = viewModel
                )
            }
        }
        UniversalisUiState.Loading -> LoadingScreen()
        UniversalisUiState.Error -> ErrorScreen()
        UniversalisUiState.Empty -> EmptyScreen()
    }
}

规范说明 #

• Clean Architecture：ViewModel作为UI层状态中枢，负责接收UI层的文本更新请求并维护状态，业务方法getAllForRead()属于领域层逻辑，仅在可组合项中调用后传递给ViewModel，符合单向数据流原则。
• Compose规范：使用LaunchedEffect处理副作用（同步文本到ViewModel），通过StateFlow和collectAsState实现状态订阅，保持可组合项的无状态性，仅依赖参数传递数据。

内容的提问来源于stack exchange，提问作者A. Cedano