Support pending/loading message while waiting for response (#821)

* support pending message draft

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* styling + pending message for /fix

* change default pending message

* remove persona groups

* inline styling

* single timestamp

* use message id as component key

Co-authored-by: david qiu <[email protected]>

* fix conditional useEffect

* prefer MUI Typography in PendingMessageElement to match font size

* merge 2 outer div elements into 1

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: david qiu <[email protected]>

Author: 3 people

packages/jupyter-ai/jupyter_ai/chat_handlers/ask.py

@@ -71,8 +71,9 @@ async def process_message(self, message: HumanChatMessage):
         self.get_llm_chain()
 
         try:
-            result = await self.llm_chain.acall({"question": query})
-            response = result["answer"]
+            with self.pending("Searching learned documents"):
+                result = await self.llm_chain.acall({"question": query})
+                response = result["answer"]
             self.reply(response, message)
         except AssertionError as e:
             self.log.error(e)

packages/jupyter-ai/jupyter_ai/chat_handlers/base.py

@@ -1,4 +1,5 @@
 import argparse
+import contextlib
 import os
 import time
 import traceback
@@ -17,7 +18,13 @@
 
 from dask.distributed import Client as DaskClient
 from jupyter_ai.config_manager import ConfigManager, Logger
-from jupyter_ai.models import AgentChatMessage, ChatMessage, HumanChatMessage
+from jupyter_ai.models import (
+    AgentChatMessage,
+    ChatMessage,
+    ClosePendingMessage,
+    HumanChatMessage,
+    PendingMessage,
+)
 from jupyter_ai_magics import Persona
 from jupyter_ai_magics.providers import BaseProvider
 from langchain.pydantic_v1 import BaseModel
@@ -193,6 +200,57 @@ def reply(self, response: str, human_msg: Optional[HumanChatMessage] = None):
             handler.broadcast_message(agent_msg)
             break
 
+    def start_pending(self, text: str, ellipsis: bool = True) -> str:
+        """
+        Sends a pending message to the client.
+
+        Returns the pending message ID.
+        """
+        persona = self.config_manager.persona
+
+        pending_msg = PendingMessage(
+            id=uuid4().hex,
+            time=time.time(),
+            body=text,
+            persona=Persona(name=persona.name, avatar_route=persona.avatar_route),
+            ellipsis=ellipsis,
+        )
+
+        for handler in self._root_chat_handlers.values():
+            if not handler:
+                continue
+
+            handler.broadcast_message(pending_msg)
+            break
+        return pending_msg
+
+    def close_pending(self, pending_msg: PendingMessage):
+        """
+        Closes a pending message.
+        """
+        close_pending_msg = ClosePendingMessage(
+            id=pending_msg.id,
+        )
+
+        for handler in self._root_chat_handlers.values():
+            if not handler:
+                continue
+
+            handler.broadcast_message(close_pending_msg)
+            break
+
+    @contextlib.contextmanager
+    def pending(self, text: str, ellipsis: bool = True):
+        """
+        Context manager that sends a pending message to the client, and closes
+        it after the block is executed.
+        """
+        pending_msg = self.start_pending(text, ellipsis=ellipsis)
+        try:
+            yield
+        finally:
+            self.close_pending(pending_msg)
+
     def get_llm_chain(self):
         lm_provider = self.config_manager.lm_provider
         lm_provider_params = self.config_manager.lm_provider_params

packages/jupyter-ai/jupyter_ai/chat_handlers/default.py

@@ -45,5 +45,8 @@ def create_llm_chain(
 
     async def process_message(self, message: HumanChatMessage):
         self.get_llm_chain()
-        response = await self.llm_chain.apredict(input=message.body, stop=["\nHuman:"])
+        with self.pending("Generating response"):
+            response = await self.llm_chain.apredict(
+                input=message.body, stop=["\nHuman:"]
+            )
         self.reply(response, message)

packages/jupyter-ai/jupyter_ai/chat_handlers/fix.py

@@ -92,12 +92,13 @@ async def process_message(self, message: HumanChatMessage):
         extra_instructions = message.body[4:].strip() or "None."
 
         self.get_llm_chain()
-        response = await self.llm_chain.apredict(
-            extra_instructions=extra_instructions,
-            stop=["\nHuman:"],
-            cell_content=selection.source,
-            error_name=selection.error.name,
-            error_value=selection.error.value,
-            traceback="\n".join(selection.error.traceback),
-        )
+        with self.pending("Analyzing error"):
+            response = await self.llm_chain.apredict(
+                extra_instructions=extra_instructions,
+                stop=["\nHuman:"],
+                cell_content=selection.source,
+                error_name=selection.error.name,
+                error_value=selection.error.value,
+                traceback="\n".join(selection.error.traceback),
+            )
         self.reply(response, message)

packages/jupyter-ai/jupyter_ai/chat_handlers/learn.py

@@ -151,19 +151,17 @@ async def process_message(self, message: HumanChatMessage):
         # delete and relearn index if embedding model was changed
         await self.delete_and_relearn()
 
-        if args.verbose:
-            self.reply(f"Loading and splitting files for {load_path}", message)
-
-        try:
-            await self.learn_dir(
-                load_path, args.chunk_size, args.chunk_overlap, args.all_files
-            )
-        except Exception as e:
-            response = f"""Learn documents in **{load_path}** failed. {str(e)}."""
-        else:
-            self.save()
-            response = f"""🎉 I have learned documents at **{load_path}** and I am ready to answer questions about them.
-                You can ask questions about these docs by prefixing your message with **/ask**."""
+        with self.pending(f"Loading and splitting files for {load_path}"):
+            try:
+                await self.learn_dir(
+                    load_path, args.chunk_size, args.chunk_overlap, args.all_files
+                )
+            except Exception as e:
+                response = f"""Learn documents in **{load_path}** failed. {str(e)}."""
+            else:
+                self.save()
+                response = f"""🎉 I have learned documents at **{load_path}** and I am ready to answer questions about them.
+                    You can ask questions about these docs by prefixing your message with **/ask**."""
         self.reply(response, message)
 
     def _build_list_response(self):

packages/jupyter-ai/jupyter_ai/models.py

@@ -87,13 +87,34 @@ class ClearMessage(BaseModel):
     type: Literal["clear"] = "clear"
 
 
+class PendingMessage(BaseModel):
+    type: Literal["pending"] = "pending"
+    id: str
+    time: float
+    body: str
+    persona: Persona
+    ellipsis: bool = True
+
+
+class ClosePendingMessage(BaseModel):
+    type: Literal["pending"] = "close-pending"
+    id: str
+
+
 # the type of messages being broadcast to clients
 ChatMessage = Union[
     AgentChatMessage,
     HumanChatMessage,
 ]
 
-Message = Union[AgentChatMessage, HumanChatMessage, ConnectionMessage, ClearMessage]
+Message = Union[
+    AgentChatMessage,
+    HumanChatMessage,
+    ConnectionMessage,
+    ClearMessage,
+    PendingMessage,
+    ClosePendingMessage,
+]
 
 
 class ChatHistory(BaseModel):

packages/jupyter-ai/src/components/chat.tsx

@@ -9,6 +9,7 @@ import { IRenderMimeRegistry } from '@jupyterlab/rendermime';
 
 import { JlThemeProvider } from './jl-theme-provider';
 import { ChatMessages } from './chat-messages';
+import { PendingMessages } from './pending-messages';
 import { ChatInput } from './chat-input';
 import { ChatSettings } from './chat-settings';
 import { AiService } from '../handler';
@@ -38,6 +39,9 @@ function ChatBody({
   rmRegistry: renderMimeRegistry
 }: ChatBodyProps): JSX.Element {
   const [messages, setMessages] = useState<AiService.ChatMessage[]>([]);
+  const [pendingMessages, setPendingMessages] = useState<
+    AiService.PendingMessage[]
+  >([]);
   const [showWelcomeMessage, setShowWelcomeMessage] = useState<boolean>(false);
   const [includeSelection, setIncludeSelection] = useState(true);
   const [replaceSelection, setReplaceSelection] = useState(false);
@@ -73,14 +77,24 @@ function ChatBody({
    */
   useEffect(() => {
     function handleChatEvents(message: AiService.Message) {
-      if (message.type === 'connection') {
-        return;
-      } else if (message.type === 'clear') {
-        setMessages([]);
-        return;
+      switch (message.type) {
+        case 'connection':
+          return;
+        case 'clear':
+          setMessages([]);
+          return;
+        case 'pending':
+          setPendingMessages(pendingMessages => [...pendingMessages, message]);
+          return;
+        case 'close-pending':
+          setPendingMessages(pendingMessages =>
+            pendingMessages.filter(p => p.id !== message.id)
+          );
+          return;
+        default:
+          setMessages(messageGroups => [...messageGroups, message]);
+          return;
       }
-
-      setMessages(messageGroups => [...messageGroups, message]);
     }
 
     chatHandler.addListener(handleChatEvents);
@@ -157,6 +171,7 @@ function ChatBody({
     <>
       <ScrollContainer sx={{ flexGrow: 1 }}>
         <ChatMessages messages={messages} rmRegistry={renderMimeRegistry} />
+        <PendingMessages messages={pendingMessages} />
       </ScrollContainer>
       <ChatInput
         value={input}

packages/jupyter-ai/src/components/pending-messages.tsx

@@ -0,0 +1,115 @@
+import React, { useState, useEffect } from 'react';
+
+import { Box, Typography } from '@mui/material';
+import { AiService } from '../handler';
+import { ChatMessageHeader } from './chat-messages';
+
+type PendingMessagesProps = {
+  messages: AiService.PendingMessage[];
+};
+
+type PendingMessageElementProps = {
+  text: string;
+  ellipsis: boolean;
+};
+
+function PendingMessageElement(props: PendingMessageElementProps): JSX.Element {
+  const [dots, setDots] = useState('');
+
+  useEffect(() => {
+    const interval = setInterval(() => {
+      setDots(dots => (dots.length < 3 ? dots + '.' : ''));
+    }, 500);
+
+    return () => clearInterval(interval);
+  }, []);
+
+  let text = props.text;
+  if (props.ellipsis) {
+    text = props.text + dots;
+  }
+
+  return (
+    <Box>
+      {text.split('\n').map((line, index) => (
+        <Typography key={index} sx={{ lineHeight: 0.6 }}>
+          {line}
+        </Typography>
+      ))}
+    </Box>
+  );
+}
+
+export function PendingMessages(
+  props: PendingMessagesProps
+): JSX.Element | null {
+  const [timestamp, setTimestamp] = useState<string>('');
+  const [agentMessage, setAgentMessage] =
+    useState<AiService.AgentChatMessage | null>(null);
+
+  useEffect(() => {
+    if (props.messages.length === 0) {
+      setAgentMessage(null);
+      setTimestamp('');
+      return;
+    }
+    const lastMessage = props.messages[props.messages.length - 1];
+    setAgentMessage({
+      type: 'agent',
+      id: lastMessage.id,
+      time: lastMessage.time,
+      body: '',
+      reply_to: '',
+      persona: lastMessage.persona
+    });
+
+    // timestamp format copied from ChatMessage
+    const newTimestamp = new Date(lastMessage.time * 1000).toLocaleTimeString(
+      [],
+      {
+        hour: 'numeric',
+        minute: '2-digit'
+      }
+    );
+    setTimestamp(newTimestamp);
+  }, [props.messages]);
+
+  if (!agentMessage) {
+    return null;
+  }
+
+  return (
+    <Box
+      sx={{
+        padding: 4,
+        borderTop: '1px solid var(--jp-border-color2)'
+      }}
+    >
+      <ChatMessageHeader
+        message={agentMessage}
+        timestamp={timestamp}
+        sx={{
+          marginBottom: 4
+        }}
+      />
+      <Box
+        sx={{
+          marginBottom: 1,
+          paddingRight: 0,
+          color: 'var(--jp-ui-font-color2)',
+          '& > :not(:last-child)': {
+            marginBottom: '2em'
+          }
+        }}
+      >
+        {props.messages.map(message => (
+          <PendingMessageElement
+            key={message.id}
+            text={message.body}
+            ellipsis={message.ellipsis}
+          />
+        ))}
+      </Box>
+    </Box>
+  );
+}

packages/jupyter-ai/src/handler.ts

@@ -115,12 +115,28 @@ export namespace AiService {
     type: 'clear';
   };
 
+  export type PendingMessage = {
+    type: 'pending';
+    id: string;
+    time: number;
+    body: string;
+    persona: Persona;
+    ellipsis: boolean;
+  };
+
+  export type ClosePendingMessage = {
+    type: 'close-pending';
+    id: string;
+  };
+
   export type ChatMessage = AgentChatMessage | HumanChatMessage;
   export type Message =
     | AgentChatMessage
     | HumanChatMessage
     | ConnectionMessage
-    | ClearMessage;
+    | ClearMessage
+    | PendingMessage
+    | ClosePendingMessage;
 
   export type ChatHistory = {
     messages: ChatMessage[];