Roo-Code/webview-ui/src/components/chat/ChatView.tsx

1444 lines
47 KiB
TypeScript

import { VSCodeButton, VSCodeLink } from "@vscode/webview-ui-toolkit/react"
import debounce from "debounce"
import { forwardRef, useCallback, useEffect, useImperativeHandle, useMemo, useRef, useState } from "react"
import { useDeepCompareEffect, useEvent, useMount } from "react-use"
import { Virtuoso, type VirtuosoHandle } from "react-virtuoso"
import removeMd from "remove-markdown"
import { Trans } from "react-i18next"
import {
ClineAsk,
ClineMessage,
ClineSayBrowserAction,
ClineSayTool,
ExtensionMessage,
} from "@roo/shared/ExtensionMessage"
import { McpServer, McpTool } from "@roo/shared/mcp"
import { findLast } from "@roo/shared/array"
import { combineApiRequests } from "@roo/shared/combineApiRequests"
import { combineCommandSequences } from "@roo/shared/combineCommandSequences"
import { getApiMetrics } from "@roo/shared/getApiMetrics"
import { AudioType } from "@roo/shared/WebviewMessage"
import { getAllModes } from "@roo/shared/modes"
import { useExtensionState } from "@src/context/ExtensionStateContext"
import { vscode } from "@src/utils/vscode"
import { useSelectedModel } from "@/components/ui/hooks/useSelectedModel"
import { validateCommand } from "@src/utils/command-validation"
import { useAppTranslation } from "@src/i18n/TranslationContext"
import TelemetryBanner from "../common/TelemetryBanner"
import HistoryPreview from "../history/HistoryPreview"
import RooHero from "@src/components/welcome/RooHero"
import RooTips from "@src/components/welcome/RooTips"
import Announcement from "./Announcement"
import BrowserSessionRow from "./BrowserSessionRow"
import ChatRow from "./ChatRow"
import ChatTextArea from "./ChatTextArea"
import TaskHeader from "./TaskHeader"
import AutoApproveMenu from "./AutoApproveMenu"
import SystemPromptWarning from "./SystemPromptWarning"
import { useTaskSearch } from "../history/useTaskSearch"
export interface ChatViewProps {
isHidden: boolean
showAnnouncement: boolean
hideAnnouncement: () => void
}
export interface ChatViewRef {
acceptInput: () => void
}
export const MAX_IMAGES_PER_MESSAGE = 20 // Anthropic limits to 20 images
const isMac = navigator.platform.toUpperCase().indexOf("MAC") >= 0
const ChatViewComponent: React.ForwardRefRenderFunction<ChatViewRef, ChatViewProps> = (
{ isHidden, showAnnouncement, hideAnnouncement },
ref,
) => {
const { t } = useAppTranslation()
const modeShortcutText = `${isMac ? "⌘" : "Ctrl"} + . ${t("chat:forNextMode")}`
const {
clineMessages: messages,
taskHistory,
apiConfiguration,
mcpServers,
alwaysAllowBrowser,
alwaysAllowReadOnly,
alwaysAllowReadOnlyOutsideWorkspace,
alwaysAllowWrite,
alwaysAllowWriteOutsideWorkspace,
alwaysAllowExecute,
alwaysAllowMcp,
allowedCommands,
writeDelayMs,
mode,
setMode,
autoApprovalEnabled,
alwaysAllowModeSwitch,
alwaysAllowSubtasks,
customModes,
telemetrySetting,
hasSystemPromptOverride,
historyPreviewCollapsed, // Added historyPreviewCollapsed
} = useExtensionState()
const { tasks } = useTaskSearch()
// Initialize expanded state based on the persisted setting (default to expanded if undefined)
const [isExpanded, setIsExpanded] = useState(
historyPreviewCollapsed === undefined ? true : !historyPreviewCollapsed,
)
const toggleExpanded = useCallback(() => {
const newState = !isExpanded
setIsExpanded(newState)
// Send message to extension to persist the new collapsed state
vscode.postMessage({ type: "setHistoryPreviewCollapsed", bool: !newState })
}, [isExpanded])
// Leaving this less safe version here since if the first message is not a
// task, then the extension is in a bad state and needs to be debugged (see
// Cline.abort).
const task = useMemo(() => messages.at(0), [messages])
const modifiedMessages = useMemo(() => combineApiRequests(combineCommandSequences(messages.slice(1))), [messages])
// Has to be after api_req_finished are all reduced into api_req_started messages.
const apiMetrics = useMemo(() => getApiMetrics(modifiedMessages), [modifiedMessages])
const [inputValue, setInputValue] = useState("")
const textAreaRef = useRef<HTMLTextAreaElement>(null)
const [textAreaDisabled, setTextAreaDisabled] = useState(false)
const [selectedImages, setSelectedImages] = useState<string[]>([])
// we need to hold on to the ask because useEffect > lastMessage will always let us know when an ask comes in and handle it, but by the time handleMessage is called, the last message might not be the ask anymore (it could be a say that followed)
const [clineAsk, setClineAsk] = useState<ClineAsk | undefined>(undefined)
const [enableButtons, setEnableButtons] = useState<boolean>(false)
const [primaryButtonText, setPrimaryButtonText] = useState<string | undefined>(undefined)
const [secondaryButtonText, setSecondaryButtonText] = useState<string | undefined>(undefined)
const [didClickCancel, setDidClickCancel] = useState(false)
const virtuosoRef = useRef<VirtuosoHandle>(null)
const [expandedRows, setExpandedRows] = useState<Record<number, boolean>>({})
const scrollContainerRef = useRef<HTMLDivElement>(null)
const disableAutoScrollRef = useRef(false)
const [showScrollToBottom, setShowScrollToBottom] = useState(false)
const [isAtBottom, setIsAtBottom] = useState(false)
const lastTtsRef = useRef<string>("")
const [wasStreaming, setWasStreaming] = useState<boolean>(false)
const [showCheckpointWarning, setShowCheckpointWarning] = useState<boolean>(false)
// UI layout depends on the last 2 messages
// (since it relies on the content of these messages, we are deep comparing. i.e. the button state after hitting button sets enableButtons to false, and this effect otherwise would have to true again even if messages didn't change
const lastMessage = useMemo(() => messages.at(-1), [messages])
const secondLastMessage = useMemo(() => messages.at(-2), [messages])
function playSound(audioType: AudioType) {
vscode.postMessage({ type: "playSound", audioType })
}
function playTts(text: string) {
vscode.postMessage({ type: "playTts", text })
}
useDeepCompareEffect(() => {
// if last message is an ask, show user ask UI
// if user finished a task, then start a new task with a new conversation history since in this moment that the extension is waiting for user response, the user could close the extension and the conversation history would be lost.
// basically as long as a task is active, the conversation history will be persisted
if (lastMessage) {
switch (lastMessage.type) {
case "ask":
const isPartial = lastMessage.partial === true
switch (lastMessage.ask) {
case "api_req_failed":
playSound("progress_loop")
setTextAreaDisabled(true)
setClineAsk("api_req_failed")
setEnableButtons(true)
setPrimaryButtonText(t("chat:retry.title"))
setSecondaryButtonText(t("chat:startNewTask.title"))
break
case "mistake_limit_reached":
playSound("progress_loop")
setTextAreaDisabled(false)
setClineAsk("mistake_limit_reached")
setEnableButtons(true)
setPrimaryButtonText(t("chat:proceedAnyways.title"))
setSecondaryButtonText(t("chat:startNewTask.title"))
break
case "followup":
setTextAreaDisabled(isPartial)
setClineAsk("followup")
// setting enable buttons to `false` would trigger a focus grab when
// the text area is enabled which is undesirable.
// We have no buttons for this tool, so no problem having them "enabled"
// to workaround this issue. See #1358.
setEnableButtons(true)
setPrimaryButtonText(undefined)
setSecondaryButtonText(undefined)
break
case "tool":
if (!isAutoApproved(lastMessage)) {
playSound("notification")
}
setTextAreaDisabled(isPartial)
setClineAsk("tool")
setEnableButtons(!isPartial)
const tool = JSON.parse(lastMessage.text || "{}") as ClineSayTool
switch (tool.tool) {
case "editedExistingFile":
case "appliedDiff":
case "newFileCreated":
case "insertContent":
setPrimaryButtonText(t("chat:save.title"))
setSecondaryButtonText(t("chat:reject.title"))
break
case "finishTask":
setPrimaryButtonText(t("chat:completeSubtaskAndReturn"))
setSecondaryButtonText(undefined)
break
default:
setPrimaryButtonText(t("chat:approve.title"))
setSecondaryButtonText(t("chat:reject.title"))
break
}
break
case "browser_action_launch":
if (!isAutoApproved(lastMessage)) {
playSound("notification")
}
setTextAreaDisabled(isPartial)
setClineAsk("browser_action_launch")
setEnableButtons(!isPartial)
setPrimaryButtonText(t("chat:approve.title"))
setSecondaryButtonText(t("chat:reject.title"))
break
case "command":
if (!isAutoApproved(lastMessage)) {
playSound("notification")
}
setTextAreaDisabled(isPartial)
setClineAsk("command")
setEnableButtons(!isPartial)
setPrimaryButtonText(t("chat:runCommand.title"))
setSecondaryButtonText(t("chat:reject.title"))
break
case "command_output":
setTextAreaDisabled(false)
setClineAsk("command_output")
setEnableButtons(true)
setPrimaryButtonText(t("chat:proceedWhileRunning.title"))
setSecondaryButtonText(t("chat:killCommand.title"))
break
case "use_mcp_server":
setTextAreaDisabled(isPartial)
setClineAsk("use_mcp_server")
setEnableButtons(!isPartial)
setPrimaryButtonText(t("chat:approve.title"))
setSecondaryButtonText(t("chat:reject.title"))
break
case "completion_result":
// extension waiting for feedback. but we can just present a new task button
playSound("celebration")
setTextAreaDisabled(isPartial)
setClineAsk("completion_result")
setEnableButtons(!isPartial)
setPrimaryButtonText(t("chat:startNewTask.title"))
setSecondaryButtonText(undefined)
break
case "resume_task":
setTextAreaDisabled(false)
setClineAsk("resume_task")
setEnableButtons(true)
setPrimaryButtonText(t("chat:resumeTask.title"))
setSecondaryButtonText(t("chat:terminate.title"))
setDidClickCancel(false) // special case where we reset the cancel button state
break
case "resume_completed_task":
setTextAreaDisabled(false)
setClineAsk("resume_completed_task")
setEnableButtons(true)
setPrimaryButtonText(t("chat:startNewTask.title"))
setSecondaryButtonText(undefined)
setDidClickCancel(false)
break
}
break
case "say":
// don't want to reset since there could be a "say" after an "ask" while ask is waiting for response
switch (lastMessage.say) {
case "api_req_retry_delayed":
setTextAreaDisabled(true)
break
case "api_req_started":
if (secondLastMessage?.ask === "command_output") {
// If the last ask is a command_output, and we
// receive an api_req_started, then that means
// the command has finished and we don't need
// input from the user anymore (in every other
// case, the user has to interact with input
// field or buttons to continue, which does the
// following automatically).
setInputValue("")
setTextAreaDisabled(true)
setSelectedImages([])
setClineAsk(undefined)
setEnableButtons(false)
}
break
case "api_req_finished":
case "task":
case "error":
case "text":
case "browser_action":
case "browser_action_result":
case "command_output":
case "mcp_server_request_started":
case "mcp_server_response":
case "completion_result":
case "tool":
break
}
break
}
} else {
// this would get called after sending the first message, so we have to watch messages.length instead
// No messages, so user has to submit a task
// setTextAreaDisabled(false)
// setClineAsk(undefined)
// setPrimaryButtonText(undefined)
// setSecondaryButtonText(undefined)
}
}, [lastMessage, secondLastMessage])
useEffect(() => {
if (messages.length === 0) {
setTextAreaDisabled(false)
setClineAsk(undefined)
setEnableButtons(false)
setPrimaryButtonText(undefined)
setSecondaryButtonText(undefined)
}
}, [messages.length])
useEffect(() => {
setExpandedRows({})
}, [task?.ts])
const isStreaming = useMemo(() => {
const isLastAsk = !!modifiedMessages.at(-1)?.ask // checking clineAsk isn't enough since messages effect may be called again for a tool for example, set clineAsk to its value, and if the next message is not an ask then it doesn't reset. This is likely due to how much more often we're updating messages as compared to before, and should be resolved with optimizations as it's likely a rendering bug. but as a final guard for now, the cancel button will show if the last message is not an ask
const isToolCurrentlyAsking =
isLastAsk && clineAsk !== undefined && enableButtons && primaryButtonText !== undefined
if (isToolCurrentlyAsking) {
return false
}
const isLastMessagePartial = modifiedMessages.at(-1)?.partial === true
if (isLastMessagePartial) {
return true
} else {
const lastApiReqStarted = findLast(modifiedMessages, (message) => message.say === "api_req_started")
if (
lastApiReqStarted &&
lastApiReqStarted.text !== null &&
lastApiReqStarted.text !== undefined &&
lastApiReqStarted.say === "api_req_started"
) {
const cost = JSON.parse(lastApiReqStarted.text).cost
if (cost === undefined) {
// api request has not finished yet
return true
}
}
}
return false
}, [modifiedMessages, clineAsk, enableButtons, primaryButtonText])
const handleChatReset = useCallback(() => {
// Only reset message-specific state, preserving mode.
setInputValue("")
setTextAreaDisabled(true)
setSelectedImages([])
setClineAsk(undefined)
setEnableButtons(false)
// Do not reset mode here as it should persist.
// setPrimaryButtonText(undefined)
// setSecondaryButtonText(undefined)
disableAutoScrollRef.current = false
}, [])
const handleSendMessage = useCallback(
(text: string, images: string[]) => {
text = text.trim()
if (text || images.length > 0) {
if (messages.length === 0) {
vscode.postMessage({ type: "newTask", text, images })
} else if (clineAsk) {
switch (clineAsk) {
case "followup":
case "tool":
case "browser_action_launch":
case "command": // User can provide feedback to a tool or command use.
case "command_output": // User can send input to command stdin.
case "use_mcp_server":
case "completion_result": // If this happens then the user has feedback for the completion result.
case "resume_task":
case "resume_completed_task":
case "mistake_limit_reached":
vscode.postMessage({ type: "askResponse", askResponse: "messageResponse", text, images })
break
// There is no other case that a textfield should be enabled.
}
}
handleChatReset()
}
},
[messages.length, clineAsk, handleChatReset],
)
const handleSetChatBoxMessage = useCallback(
(text: string, images: string[]) => {
// Avoid nested template literals by breaking down the logic
let newValue = text
if (inputValue !== "") {
newValue = inputValue + " " + text
}
setInputValue(newValue)
setSelectedImages([...selectedImages, ...images])
},
[inputValue, selectedImages],
)
const startNewTask = useCallback(() => {
vscode.postMessage({ type: "clearTask" })
}, [])
/*
This logic depends on the useEffect[messages] above to set clineAsk, after which buttons are shown and we then send an askResponse to the extension.
*/
const handlePrimaryButtonClick = useCallback(
(text?: string, images?: string[]) => {
const trimmedInput = text?.trim()
switch (clineAsk) {
case "api_req_failed":
case "command":
case "tool":
case "browser_action_launch":
case "use_mcp_server":
case "resume_task":
case "mistake_limit_reached":
// Only send text/images if they exist
if (trimmedInput || (images && images.length > 0)) {
vscode.postMessage({
type: "askResponse",
askResponse: "yesButtonClicked",
text: trimmedInput,
images: images,
})
} else {
vscode.postMessage({ type: "askResponse", askResponse: "yesButtonClicked" })
}
// Clear input state after sending
setInputValue("")
setSelectedImages([])
break
case "completion_result":
case "resume_completed_task":
// Waiting for feedback, but we can just present a new task button
startNewTask()
break
case "command_output":
vscode.postMessage({ type: "terminalOperation", terminalOperation: "continue" })
break
}
setTextAreaDisabled(true)
setClineAsk(undefined)
setEnableButtons(false)
},
[clineAsk, startNewTask],
)
const handleSecondaryButtonClick = useCallback(
(text?: string, images?: string[]) => {
const trimmedInput = text?.trim()
if (isStreaming) {
vscode.postMessage({ type: "cancelTask" })
setDidClickCancel(true)
return
}
switch (clineAsk) {
case "api_req_failed":
case "mistake_limit_reached":
case "resume_task":
startNewTask()
break
case "command":
case "tool":
case "browser_action_launch":
case "use_mcp_server":
// Only send text/images if they exist
if (trimmedInput || (images && images.length > 0)) {
vscode.postMessage({
type: "askResponse",
askResponse: "noButtonClicked",
text: trimmedInput,
images: images,
})
} else {
// Responds to the API with a "This operation failed" and lets it try again
vscode.postMessage({ type: "askResponse", askResponse: "noButtonClicked" })
}
// Clear input state after sending
setInputValue("")
setSelectedImages([])
break
case "command_output":
vscode.postMessage({ type: "terminalOperation", terminalOperation: "abort" })
break
}
setTextAreaDisabled(true)
setClineAsk(undefined)
setEnableButtons(false)
},
[clineAsk, startNewTask, isStreaming],
)
const handleTaskCloseButtonClick = useCallback(() => {
startNewTask()
}, [startNewTask])
const { info: model } = useSelectedModel(apiConfiguration)
const selectImages = useCallback(() => {
vscode.postMessage({ type: "selectImages" })
}, [])
const shouldDisableImages =
!model?.supportsImages || textAreaDisabled || selectedImages.length >= MAX_IMAGES_PER_MESSAGE
const handleMessage = useCallback(
(e: MessageEvent) => {
const message: ExtensionMessage = e.data
switch (message.type) {
case "action":
switch (message.action!) {
case "didBecomeVisible":
if (!isHidden && !textAreaDisabled && !enableButtons) {
textAreaRef.current?.focus()
}
break
case "focusInput":
textAreaRef.current?.focus()
break
}
break
case "selectedImages":
const newImages = message.images ?? []
if (newImages.length > 0) {
setSelectedImages((prevImages) =>
[...prevImages, ...newImages].slice(0, MAX_IMAGES_PER_MESSAGE),
)
}
break
case "invoke":
switch (message.invoke!) {
case "newChat":
handleChatReset()
break
case "sendMessage":
handleSendMessage(message.text ?? "", message.images ?? [])
break
case "setChatBoxMessage":
handleSetChatBoxMessage(message.text ?? "", message.images ?? [])
break
case "primaryButtonClick":
handlePrimaryButtonClick(message.text ?? "", message.images ?? [])
break
case "secondaryButtonClick":
handleSecondaryButtonClick(message.text ?? "", message.images ?? [])
break
}
}
// textAreaRef.current is not explicitly required here since react gaurantees that ref will be stable across re-renders, and we're not using its value but its reference.
},
[
isHidden,
textAreaDisabled,
enableButtons,
handleChatReset,
handleSendMessage,
handleSetChatBoxMessage,
handlePrimaryButtonClick,
handleSecondaryButtonClick,
],
)
useEvent("message", handleMessage)
useMount(() => {
// NOTE: the vscode window needs to be focused for this to work
textAreaRef.current?.focus()
})
useEffect(() => {
const timer = setTimeout(() => {
if (!isHidden && !textAreaDisabled && !enableButtons) {
textAreaRef.current?.focus()
}
}, 50)
return () => {
clearTimeout(timer)
}
}, [isHidden, textAreaDisabled, enableButtons])
const visibleMessages = useMemo(() => {
return modifiedMessages.filter((message) => {
switch (message.ask) {
case "completion_result":
// don't show a chat row for a completion_result ask without text. This specific type of message only occurs if cline wants to execute a command as part of its completion result, in which case we interject the completion_result tool with the execute_command tool.
if (message.text === "") {
return false
}
break
case "api_req_failed": // this message is used to update the latest api_req_started that the request failed
case "resume_task":
case "resume_completed_task":
return false
}
switch (message.say) {
case "api_req_finished": // combineApiRequests removes this from modifiedMessages anyways
case "api_req_retried": // this message is used to update the latest api_req_started that the request was retried
case "api_req_deleted": // aggregated api_req metrics from deleted messages
return false
case "api_req_retry_delayed":
// Only show the retry message if it's the last message or the last messages is api_req_retry_delayed+resume_task
const last1 = modifiedMessages.at(-1)
const last2 = modifiedMessages.at(-2)
if (last1?.ask === "resume_task" && last2 === message) {
return true
}
return message === last1
case "text":
// Sometimes cline returns an empty text message, we don't want to render these. (We also use a say text for user messages, so in case they just sent images we still render that)
if ((message.text ?? "") === "" && (message.images?.length ?? 0) === 0) {
return false
}
break
case "mcp_server_request_started":
return false
}
return true
})
}, [modifiedMessages])
const isReadOnlyToolAction = useCallback((message: ClineMessage | undefined) => {
if (message?.type === "ask") {
if (!message.text) {
return true
}
const tool = JSON.parse(message.text)
return [
"readFile",
"listFiles",
"listFilesTopLevel",
"listFilesRecursive",
"listCodeDefinitionNames",
"searchFiles",
].includes(tool.tool)
}
return false
}, [])
const isWriteToolAction = useCallback((message: ClineMessage | undefined) => {
if (message?.type === "ask") {
if (!message.text) {
return true
}
const tool = JSON.parse(message.text)
return [
"editedExistingFile",
"appliedDiff",
"newFileCreated",
"searchAndReplace",
"insertContent",
].includes(tool.tool)
}
return false
}, [])
const isMcpToolAlwaysAllowed = useCallback(
(message: ClineMessage | undefined) => {
if (message?.type === "ask" && message.ask === "use_mcp_server") {
if (!message.text) {
return true
}
const mcpServerUse = JSON.parse(message.text) as { type: string; serverName: string; toolName: string }
if (mcpServerUse.type === "use_mcp_tool") {
const server = mcpServers?.find((s: McpServer) => s.name === mcpServerUse.serverName)
const tool = server?.tools?.find((t: McpTool) => t.name === mcpServerUse.toolName)
return tool?.alwaysAllow || false
}
}
return false
},
[mcpServers],
)
// Check if a command message is allowed
const isAllowedCommand = useCallback(
(message: ClineMessage | undefined): boolean => {
if (message?.type !== "ask") return false
return validateCommand(message.text || "", allowedCommands || [])
},
[allowedCommands],
)
const isAutoApproved = useCallback(
(message: ClineMessage | undefined) => {
if (!autoApprovalEnabled || !message || message.type !== "ask") return false
if (message.ask === "browser_action_launch") {
return alwaysAllowBrowser
}
if (message.ask === "use_mcp_server") {
return alwaysAllowMcp && isMcpToolAlwaysAllowed(message)
}
if (message.ask === "command") {
return alwaysAllowExecute && isAllowedCommand(message)
}
// For read/write operations, check if it's outside workspace and if we have permission for that
if (message.ask === "tool") {
let tool: any = {}
try {
tool = JSON.parse(message.text || "{}")
} catch (error) {
console.error("Failed to parse tool:", error)
}
if (!tool) {
return false
}
if (tool?.tool === "fetchInstructions") {
if (tool.content === "create_mode") {
return alwaysAllowModeSwitch
}
if (tool.content === "create_mcp_server") {
return alwaysAllowMcp
}
}
if (tool?.tool === "switchMode") {
return alwaysAllowModeSwitch
}
if (["newTask", "finishTask"].includes(tool?.tool)) {
return alwaysAllowSubtasks
}
const isOutsideWorkspace = !!tool.isOutsideWorkspace
if (isReadOnlyToolAction(message)) {
return alwaysAllowReadOnly && (!isOutsideWorkspace || alwaysAllowReadOnlyOutsideWorkspace)
}
if (isWriteToolAction(message)) {
return alwaysAllowWrite && (!isOutsideWorkspace || alwaysAllowWriteOutsideWorkspace)
}
}
return false
},
[
autoApprovalEnabled,
alwaysAllowBrowser,
alwaysAllowReadOnly,
alwaysAllowReadOnlyOutsideWorkspace,
isReadOnlyToolAction,
alwaysAllowWrite,
alwaysAllowWriteOutsideWorkspace,
isWriteToolAction,
alwaysAllowExecute,
isAllowedCommand,
alwaysAllowMcp,
isMcpToolAlwaysAllowed,
alwaysAllowModeSwitch,
alwaysAllowSubtasks,
],
)
useEffect(() => {
// this ensures the first message is not read, future user messages are labelled as user_feedback
if (lastMessage && messages.length > 1) {
//console.log(JSON.stringify(lastMessage))
if (
lastMessage.text && // has text
(lastMessage.say === "text" || lastMessage.say === "completion_result") && // is a text message
!lastMessage.partial && // not a partial message
!lastMessage.text.startsWith("{") // not a json object
) {
let text = lastMessage?.text || ""
const mermaidRegex = /```mermaid[\s\S]*?```/g
// remove mermaid diagrams from text
text = text.replace(mermaidRegex, "")
// remove markdown from text
text = removeMd(text)
// ensure message is not a duplicate of last read message
if (text !== lastTtsRef.current) {
try {
playTts(text)
lastTtsRef.current = text
} catch (error) {
console.error("Failed to execute text-to-speech:", error)
}
}
}
}
// Only execute when isStreaming changes from true to false
if (wasStreaming && !isStreaming && lastMessage) {
// Play appropriate sound based on lastMessage content
if (lastMessage.type === "ask") {
// Don't play sounds for auto-approved actions
if (!isAutoApproved(lastMessage)) {
switch (lastMessage.ask) {
case "api_req_failed":
case "mistake_limit_reached":
playSound("progress_loop")
break
case "followup":
if (!lastMessage.partial) {
playSound("notification")
}
break
case "tool":
case "browser_action_launch":
case "resume_task":
case "use_mcp_server":
playSound("notification")
break
case "completion_result":
case "resume_completed_task":
playSound("celebration")
break
}
}
}
}
// Update previous value
setWasStreaming(isStreaming)
}, [isStreaming, lastMessage, wasStreaming, isAutoApproved, messages.length])
const isBrowserSessionMessage = (message: ClineMessage): boolean => {
// which of visible messages are browser session messages, see above
if (message.type === "ask") {
return ["browser_action_launch"].includes(message.ask!)
}
if (message.type === "say") {
return ["api_req_started", "text", "browser_action", "browser_action_result"].includes(message.say!)
}
return false
}
const groupedMessages = useMemo(() => {
const result: (ClineMessage | ClineMessage[])[] = []
let currentGroup: ClineMessage[] = []
let isInBrowserSession = false
const endBrowserSession = () => {
if (currentGroup.length > 0) {
result.push([...currentGroup])
currentGroup = []
isInBrowserSession = false
}
}
visibleMessages.forEach((message) => {
if (message.ask === "browser_action_launch") {
// complete existing browser session if any
endBrowserSession()
// start new
isInBrowserSession = true
currentGroup.push(message)
} else if (isInBrowserSession) {
// end session if api_req_started is cancelled
if (message.say === "api_req_started") {
// get last api_req_started in currentGroup to check if it's cancelled. If it is then this api req is not part of the current browser session
const lastApiReqStarted = [...currentGroup].reverse().find((m) => m.say === "api_req_started")
if (lastApiReqStarted?.text !== null && lastApiReqStarted?.text !== undefined) {
const info = JSON.parse(lastApiReqStarted.text)
const isCancelled = info.cancelReason !== null && info.cancelReason !== undefined
if (isCancelled) {
endBrowserSession()
result.push(message)
return
}
}
}
if (isBrowserSessionMessage(message)) {
currentGroup.push(message)
// Check if this is a close action
if (message.say === "browser_action") {
const browserAction = JSON.parse(message.text || "{}") as ClineSayBrowserAction
if (browserAction.action === "close") {
endBrowserSession()
}
}
} else {
// complete existing browser session if any
endBrowserSession()
result.push(message)
}
} else {
result.push(message)
}
})
// Handle case where browser session is the last group
if (currentGroup.length > 0) {
result.push([...currentGroup])
}
return result
}, [visibleMessages])
// scrolling
const scrollToBottomSmooth = useMemo(
() =>
debounce(
() => {
virtuosoRef.current?.scrollTo({
top: Number.MAX_SAFE_INTEGER,
behavior: "smooth",
})
},
10,
{ immediate: true },
),
[],
)
const scrollToBottomAuto = useCallback(() => {
virtuosoRef.current?.scrollTo({
top: Number.MAX_SAFE_INTEGER,
behavior: "auto", // instant causes crash
})
}, [])
// scroll when user toggles certain rows
const toggleRowExpansion = useCallback(
(ts: number) => {
const isCollapsing = expandedRows[ts] ?? false
const lastGroup = groupedMessages.at(-1)
const isLast = Array.isArray(lastGroup) ? lastGroup[0].ts === ts : lastGroup?.ts === ts
const secondToLastGroup = groupedMessages.at(-2)
const isSecondToLast = Array.isArray(secondToLastGroup)
? secondToLastGroup[0].ts === ts
: secondToLastGroup?.ts === ts
const isLastCollapsedApiReq =
isLast &&
!Array.isArray(lastGroup) && // Make sure it's not a browser session group
lastGroup?.say === "api_req_started" &&
!expandedRows[lastGroup.ts]
setExpandedRows((prev) => ({
...prev,
[ts]: !prev[ts],
}))
// disable auto scroll when user expands row
if (!isCollapsing) {
disableAutoScrollRef.current = true
}
if (isCollapsing && isAtBottom) {
const timer = setTimeout(() => {
scrollToBottomAuto()
}, 0)
return () => clearTimeout(timer)
} else if (isLast || isSecondToLast) {
if (isCollapsing) {
if (isSecondToLast && !isLastCollapsedApiReq) {
return
}
const timer = setTimeout(() => {
scrollToBottomAuto()
}, 0)
return () => clearTimeout(timer)
} else {
const timer = setTimeout(() => {
virtuosoRef.current?.scrollToIndex({
index: groupedMessages.length - (isLast ? 1 : 2),
align: "start",
})
}, 0)
return () => clearTimeout(timer)
}
}
},
[groupedMessages, expandedRows, scrollToBottomAuto, isAtBottom],
)
const handleRowHeightChange = useCallback(
(isTaller: boolean) => {
if (!disableAutoScrollRef.current) {
if (isTaller) {
scrollToBottomSmooth()
} else {
setTimeout(() => {
scrollToBottomAuto()
}, 0)
}
}
},
[scrollToBottomSmooth, scrollToBottomAuto],
)
useEffect(() => {
if (!disableAutoScrollRef.current) {
setTimeout(() => {
scrollToBottomSmooth()
}, 50)
// return () => clearTimeout(timer) // dont cleanup since if visibleMessages.length changes it cancels.
}
}, [groupedMessages.length, scrollToBottomSmooth])
const handleWheel = useCallback((event: Event) => {
const wheelEvent = event as WheelEvent
if (wheelEvent.deltaY && wheelEvent.deltaY < 0) {
if (scrollContainerRef.current?.contains(wheelEvent.target as Node)) {
// user scrolled up
disableAutoScrollRef.current = true
}
}
}, [])
useEvent("wheel", handleWheel, window, { passive: true }) // passive improves scrolling performance
// Effect to handle showing the checkpoint warning after a delay
useEffect(() => {
// Only show the warning when there's a task but no visible messages yet
if (task && modifiedMessages.length === 0 && !isStreaming) {
const timer = setTimeout(() => {
setShowCheckpointWarning(true)
}, 5000) // 5 seconds
return () => clearTimeout(timer)
}
}, [task, modifiedMessages.length, isStreaming])
// Effect to hide the checkpoint warning when messages appear
useEffect(() => {
if (modifiedMessages.length > 0 || isStreaming) {
setShowCheckpointWarning(false)
}
}, [modifiedMessages.length, isStreaming])
// Checkpoint warning component
const CheckpointWarningMessage = useCallback(
() => (
<div className="flex items-center p-3 my-3 bg-vscode-inputValidation-warningBackground border border-vscode-inputValidation-warningBorder rounded">
<span className="codicon codicon-loading codicon-modifier-spin mr-2" />
<span className="text-vscode-foreground">
<Trans
i18nKey="chat:checkpoint.initializingWarning"
components={{
settingsLink: (
<VSCodeLink
href="#"
onClick={(e) => {
e.preventDefault()
window.postMessage(
{
type: "action",
action: "settingsButtonClicked",
values: { section: "checkpoints" },
},
"*",
)
}}
className="inline px-0.5"
/>
),
}}
/>
</span>
</div>
),
[],
)
const placeholderText = task ? t("chat:typeMessage") : t("chat:typeTask")
const itemContent = useCallback(
(index: number, messageOrGroup: ClineMessage | ClineMessage[]) => {
// browser session group
if (Array.isArray(messageOrGroup)) {
return (
<BrowserSessionRow
messages={messageOrGroup}
isLast={index === groupedMessages.length - 1}
lastModifiedMessage={modifiedMessages.at(-1)}
onHeightChange={handleRowHeightChange}
isStreaming={isStreaming}
// Pass handlers for each message in the group
isExpanded={(messageTs: number) => expandedRows[messageTs] ?? false}
onToggleExpand={(messageTs: number) => {
setExpandedRows((prev) => ({
...prev,
[messageTs]: !prev[messageTs],
}))
}}
/>
)
}
// regular message
return (
<ChatRow
key={messageOrGroup.ts}
message={messageOrGroup}
isExpanded={expandedRows[messageOrGroup.ts] || false}
onToggleExpand={() => toggleRowExpansion(messageOrGroup.ts)}
lastModifiedMessage={modifiedMessages.at(-1)}
isLast={index === groupedMessages.length - 1}
onHeightChange={handleRowHeightChange}
isStreaming={isStreaming}
onSuggestionClick={(answer: string, event?: React.MouseEvent) => {
if (event?.shiftKey) {
// Always append to existing text, don't overwrite
setInputValue((currentValue) => {
return currentValue !== "" ? `${currentValue} \n${answer}` : answer
})
} else {
handleSendMessage(answer, [])
}
}}
/>
)
},
[
expandedRows,
modifiedMessages,
groupedMessages.length,
handleRowHeightChange,
isStreaming,
toggleRowExpansion,
handleSendMessage,
],
)
useEffect(() => {
// Only proceed if we have an ask and buttons are enabled
if (!clineAsk || !enableButtons) return
const autoApprove = async () => {
if (isAutoApproved(lastMessage)) {
// Add delay for write operations
if (lastMessage?.ask === "tool" && isWriteToolAction(lastMessage)) {
await new Promise((resolve) => setTimeout(resolve, writeDelayMs))
}
handlePrimaryButtonClick()
}
}
autoApprove()
}, [
clineAsk,
enableButtons,
handlePrimaryButtonClick,
alwaysAllowBrowser,
alwaysAllowReadOnly,
alwaysAllowReadOnlyOutsideWorkspace,
alwaysAllowWrite,
alwaysAllowWriteOutsideWorkspace,
alwaysAllowExecute,
alwaysAllowMcp,
messages,
allowedCommands,
mcpServers,
isAutoApproved,
lastMessage,
writeDelayMs,
isWriteToolAction,
])
// Function to handle mode switching
const switchToNextMode = useCallback(() => {
const allModes = getAllModes(customModes)
const currentModeIndex = allModes.findIndex((m) => m.slug === mode)
const nextModeIndex = (currentModeIndex + 1) % allModes.length
// Update local state and notify extension to sync mode change
setMode(allModes[nextModeIndex].slug)
vscode.postMessage({
type: "mode",
text: allModes[nextModeIndex].slug,
})
}, [mode, setMode, customModes])
// Add keyboard event handler
const handleKeyDown = useCallback(
(event: KeyboardEvent) => {
// Check for Command + . (period)
if ((event.metaKey || event.ctrlKey) && event.key === ".") {
event.preventDefault() // Prevent default browser behavior
switchToNextMode()
}
},
[switchToNextMode],
)
// Add event listener
useEffect(() => {
window.addEventListener("keydown", handleKeyDown)
return () => {
window.removeEventListener("keydown", handleKeyDown)
}
}, [handleKeyDown])
useImperativeHandle(ref, () => ({
acceptInput: () => {
if (enableButtons && primaryButtonText) {
handlePrimaryButtonClick(inputValue, selectedImages)
} else if (!textAreaDisabled && (inputValue.trim() || selectedImages.length > 0)) {
handleSendMessage(inputValue, selectedImages)
}
},
}))
return (
<div className={isHidden ? "hidden" : "fixed top-0 left-0 right-0 bottom-0 flex flex-col overflow-hidden"}>
{showAnnouncement && <Announcement hideAnnouncement={hideAnnouncement} />}
{task ? (
<>
<TaskHeader
task={task}
tokensIn={apiMetrics.totalTokensIn}
tokensOut={apiMetrics.totalTokensOut}
doesModelSupportPromptCache={model?.supportsPromptCache ?? false}
cacheWrites={apiMetrics.totalCacheWrites}
cacheReads={apiMetrics.totalCacheReads}
totalCost={apiMetrics.totalCost}
contextTokens={apiMetrics.contextTokens}
onClose={handleTaskCloseButtonClick}
/>
{/* System prompt override warning */}
{hasSystemPromptOverride && (
<div className="px-3">
<SystemPromptWarning />
</div>
)}
{/* Checkpoint warning message */}
{showCheckpointWarning && (
<div className="px-3">
<CheckpointWarningMessage />
</div>
)}
</>
) : (
<div className="flex-1 min-h-0 overflow-y-auto flex flex-col gap-4">
{/* Moved Task Bar Header Here */}
{tasks.length !== 0 && (
<div className="flex text-vscode-descriptionForeground w-full mx-auto px-5 pt-3">
<div className="flex items-center gap-1 cursor-pointer" onClick={toggleExpanded}>
{tasks.length < 10 && (
<span className={`font-medium text-xs `}>{t("history:recentTasks")}</span>
)}
<span
className={`codicon ${isExpanded ? "codicon-eye" : "codicon-eye-closed"} scale-90`}
/>
</div>
</div>
)}
<div
className={` w-full flex flex-col gap-4 m-auto ${isExpanded && tasks.length > 0 ? "mt-0" : ""} p-10 pt-5`}>
<RooHero />
{telemetrySetting === "unset" && <TelemetryBanner />}
{/* Show the task history preview if expanded and tasks exist */}
{taskHistory.length > 0 && isExpanded && <HistoryPreview />}
<p className="ext-vscode-editor-foreground leading-tight font-vscode text-center">
<Trans
i18nKey="chat:about"
components={{
DocsLink: (
<a href="https://docs.roocode.com/" target="_blank" rel="noopener noreferrer">
the docs
</a>
),
}}
/>
</p>
<RooTips cycle={false} />
</div>
</div>
)}
{/*
// Flex layout explanation:
// 1. Content div above uses flex: "1 1 0" to:
// - Grow to fill available space (flex-grow: 1)
// - Shrink when AutoApproveMenu needs space (flex-shrink: 1)
// - Start from zero size (flex-basis: 0) to ensure proper distribution
// minHeight: 0 allows it to shrink below its content height
//
// 2. AutoApproveMenu uses flex: "0 1 auto" to:
// - Not grow beyond its content (flex-grow: 0)
// - Shrink when viewport is small (flex-shrink: 1)
// - Use its content size as basis (flex-basis: auto)
// This ensures it takes its natural height when there's space
// but becomes scrollable when the viewport is too small
*/}
{!task && (
<div className="mb-[-2px] flex-initial min-h-0">
<AutoApproveMenu />
</div>
)}
{task && (
<>
<div className="grow flex" ref={scrollContainerRef}>
<Virtuoso
ref={virtuosoRef}
key={task.ts} // trick to make sure virtuoso re-renders when task changes, and we use initialTopMostItemIndex to start at the bottom
className="scrollable grow overflow-y-scroll"
components={{
Footer: () => <div className="h-[5px]" />, // Add empty padding at the bottom
}}
// increasing top by 3_000 to prevent jumping around when user collapses a row
increaseViewportBy={{ top: 3_000, bottom: Number.MAX_SAFE_INTEGER }} // hack to make sure the last message is always rendered to get truly perfect scroll to bottom animation when new messages are added (Number.MAX_SAFE_INTEGER is safe for arithmetic operations, which is all virtuoso uses this value for in src/sizeRangeSystem.ts)
data={groupedMessages} // messages is the raw format returned by extension, modifiedMessages is the manipulated structure that combines certain messages of related type, and visibleMessages is the filtered structure that removes messages that should not be rendered
itemContent={itemContent}
atBottomStateChange={(isAtBottom) => {
setIsAtBottom(isAtBottom)
if (isAtBottom) {
disableAutoScrollRef.current = false
}
setShowScrollToBottom(disableAutoScrollRef.current && !isAtBottom)
}}
atBottomThreshold={10} // anything lower causes issues with followOutput
initialTopMostItemIndex={groupedMessages.length - 1}
/>
</div>
<AutoApproveMenu />
{showScrollToBottom ? (
<div className="flex px-[15px] pt-[10px]">
<div
className="bg-[color-mix(in_srgb,_var(--vscode-toolbar-hoverBackground)_55%,_transparent)] rounded-[3px] overflow-hidden cursor-pointer flex justify-center items-center flex-1 h-[25px] hover:bg-[color-mix(in_srgb,_var(--vscode-toolbar-hoverBackground)_90%,_transparent)] active:bg-[color-mix(in_srgb,_var(--vscode-toolbar-hoverBackground)_70%,_transparent)]"
onClick={() => {
scrollToBottomSmooth()
disableAutoScrollRef.current = false
}}
title={t("chat:scrollToBottom")}>
<span className="codicon codicon-chevron-down text-[18px]"></span>
</div>
</div>
) : (
<div
className={`flex ${
primaryButtonText || secondaryButtonText || isStreaming ? "px-[15px] pt-[10px]" : "p-0"
} ${
primaryButtonText || secondaryButtonText || isStreaming
? enableButtons || (isStreaming && !didClickCancel)
? "opacity-100"
: "opacity-50"
: "opacity-0"
}`}>
{primaryButtonText && !isStreaming && (
<VSCodeButton
appearance="primary"
disabled={!enableButtons}
className={secondaryButtonText ? "flex-1 mr-[6px]" : "flex-[2] mr-0"}
title={
primaryButtonText === t("chat:retry.title")
? t("chat:retry.tooltip")
: primaryButtonText === t("chat:save.title")
? t("chat:save.tooltip")
: primaryButtonText === t("chat:approve.title")
? t("chat:approve.tooltip")
: primaryButtonText === t("chat:runCommand.title")
? t("chat:runCommand.tooltip")
: primaryButtonText === t("chat:startNewTask.title")
? t("chat:startNewTask.tooltip")
: primaryButtonText === t("chat:resumeTask.title")
? t("chat:resumeTask.tooltip")
: primaryButtonText === t("chat:proceedAnyways.title")
? t("chat:proceedAnyways.tooltip")
: primaryButtonText ===
t("chat:proceedWhileRunning.title")
? t("chat:proceedWhileRunning.tooltip")
: undefined
}
onClick={() => handlePrimaryButtonClick(inputValue, selectedImages)}>
{primaryButtonText}
</VSCodeButton>
)}
{(secondaryButtonText || isStreaming) && (
<VSCodeButton
appearance="secondary"
disabled={!enableButtons && !(isStreaming && !didClickCancel)}
className={isStreaming ? "flex-[2] ml-0" : "flex-1 ml-[6px]"}
title={
isStreaming
? t("chat:cancel.tooltip")
: secondaryButtonText === t("chat:startNewTask.title")
? t("chat:startNewTask.tooltip")
: secondaryButtonText === t("chat:reject.title")
? t("chat:reject.tooltip")
: secondaryButtonText === t("chat:terminate.title")
? t("chat:terminate.tooltip")
: undefined
}
onClick={() => handleSecondaryButtonClick(inputValue, selectedImages)}>
{isStreaming ? t("chat:cancel.title") : secondaryButtonText}
</VSCodeButton>
)}
</div>
)}
</>
)}
<ChatTextArea
ref={textAreaRef}
inputValue={inputValue}
setInputValue={setInputValue}
textAreaDisabled={textAreaDisabled}
selectApiConfigDisabled={textAreaDisabled && clineAsk !== "api_req_failed"}
placeholderText={placeholderText}
selectedImages={selectedImages}
setSelectedImages={setSelectedImages}
onSend={() => handleSendMessage(inputValue, selectedImages)}
onSelectImages={selectImages}
shouldDisableImages={shouldDisableImages}
onHeightChange={() => {
if (isAtBottom) {
scrollToBottomAuto()
}
}}
mode={mode}
setMode={setMode}
modeShortcutText={modeShortcutText}
/>
<div id="roo-portal" />
</div>
)
}
const ChatView = forwardRef(ChatViewComponent)
export default ChatView