Merge c3b50a9c93 into 48469bd8ca

app/components/chat.tsx

@@ -18,6 +18,7 @@ import ReturnIcon from "../icons/return.svg";
 import CopyIcon from "../icons/copy.svg";
 import SpeakIcon from "../icons/speak.svg";
 import SpeakStopIcon from "../icons/speak-stop.svg";
+import LoadingIcon from "../icons/three-dots.svg";
 import LoadingButtonIcon from "../icons/loading.svg";
 import PromptIcon from "../icons/prompt.svg";
 import MaskIcon from "../icons/mask.svg";
@@ -78,6 +79,8 @@ import {
 
 import { uploadImage as uploadImageRemote } from "@/app/utils/chat";
 
+import dynamic from "next/dynamic";
+
 import { ChatControllerPool } from "../client/controller";
 import { DalleQuality, DalleStyle, ModelSize } from "../typing";
 import { Prompt, usePromptStore } from "../store/prompt";
@@ -122,15 +125,14 @@ import { getModelProvider } from "../utils/model";
 import { RealtimeChat } from "@/app/components/realtime-chat";
 import clsx from "clsx";
 import { getAvailableClientsCount, isMcpEnabled } from "../mcp/actions";
-import { Markdown } from "./markdown";
 
 const localStorage = safeLocalStorage();
 
 const ttsPlayer = createTTSPlayer();
 
-// const Markdown = dynamic(async () => (await import("./markdown")).Markdown, {
-//   loading: () => <LoadingIcon />,
-// });
+const Markdown = dynamic(async () => (await import("./markdown")).Markdown, {
+  loading: () => <LoadingIcon />,
+});
 
 const MCPAction = () => {
   const navigate = useNavigate();
@@ -1982,8 +1984,6 @@ function _Chat() {
                               fontFamily={fontFamily}
                               parentRef={scrollRef}
                               defaultShow={i >= messages.length - 6}
-                              immediatelyRender={i >= messages.length - 3}
-                              streaming={message.streaming}
                             />
                             {getMessageImages(message).length == 1 && (
                               <img

app/components/markdown.tsx

@@ -267,136 +267,6 @@ function tryWrapHtmlCode(text: string) {
     );
 }
 
-// Split content into paragraphs while preserving code blocks
-function splitContentIntoParagraphs(content: string) {
-  // Check for unclosed code blocks
-  const codeBlockStartCount = (content.match(/```/g) || []).length;
-  let processedContent = content;
-
-  // Add closing tag if there's an odd number of code block markers
-  if (codeBlockStartCount % 2 !== 0) {
-    processedContent = content + "\n```";
-  }
-
-  // Extract code blocks
-  const codeBlockRegex = /```[\s\S]*?```/g;
-  const codeBlocks: string[] = [];
-  let codeBlockCounter = 0;
-
-  // Replace code blocks with placeholders
-  const contentWithPlaceholders = processedContent.replace(
-    codeBlockRegex,
-    (match) => {
-      codeBlocks.push(match);
-      const placeholder = `__CODE_BLOCK_${codeBlockCounter++}__`;
-      return placeholder;
-    },
-  );
-
-  // Split by double newlines
-  const paragraphs = contentWithPlaceholders
-    .split(/\n\n+/)
-    .filter((p) => p.trim());
-
-  // Restore code blocks
-  return paragraphs.map((p) => {
-    if (p.match(/__CODE_BLOCK_\d+__/)) {
-      return p.replace(/__CODE_BLOCK_\d+__/g, (match) => {
-        const index = parseInt(match.match(/\d+/)?.[0] || "0");
-        return codeBlocks[index] || match;
-      });
-    }
-    return p;
-  });
-}
-
-// Lazy-loaded paragraph component
-function MarkdownParagraph({
-  content,
-  onLoad,
-}: {
-  content: string;
-  onLoad?: () => void;
-}) {
-  const [isLoaded, setIsLoaded] = useState(false);
-  const placeholderRef = useRef<HTMLDivElement>(null);
-  const [isVisible, setIsVisible] = useState(false);
-
-  useEffect(() => {
-    let observer: IntersectionObserver;
-    if (placeholderRef.current) {
-      observer = new IntersectionObserver(
-        (entries) => {
-          if (entries[0].isIntersecting) {
-            setIsVisible(true);
-          }
-        },
-        { threshold: 0.1, rootMargin: "200px 0px" },
-      );
-      observer.observe(placeholderRef.current);
-    }
-    return () => observer?.disconnect();
-  }, []);
-
-  useEffect(() => {
-    if (isVisible && !isLoaded) {
-      setIsLoaded(true);
-      onLoad?.();
-    }
-  }, [isVisible, isLoaded, onLoad]);
-
-  // Generate preview content
-  const previewContent = useMemo(() => {
-    if (content.startsWith("```")) {
-      return "```" + (content.split("\n")[0] || "").slice(3) + "...```";
-    }
-    return content.length > 60 ? content.slice(0, 60) + "..." : content;
-  }, [content]);
-
-  return (
-    <div className="markdown-paragraph" ref={placeholderRef}>
-      {!isLoaded ? (
-        <div className="markdown-paragraph-placeholder">{previewContent}</div>
-      ) : (
-        <_MarkDownContent content={content} />
-      )}
-    </div>
-  );
-}
-
-// Memoized paragraph component to prevent unnecessary re-renders
-const MemoizedMarkdownParagraph = React.memo(
-  ({ content }: { content: string }) => {
-    return <_MarkDownContent content={content} />;
-  },
-  (prevProps, nextProps) => prevProps.content === nextProps.content,
-);
-
-MemoizedMarkdownParagraph.displayName = "MemoizedMarkdownParagraph";
-
-// Specialized component for streaming content
-function StreamingMarkdownContent({ content }: { content: string }) {
-  const paragraphs = useMemo(
-    () => splitContentIntoParagraphs(content),
-    [content],
-  );
-  const lastParagraphRef = useRef<HTMLDivElement>(null);
-
-  return (
-    <div className="markdown-streaming-content">
-      {paragraphs.map((paragraph, index) => (
-        <div
-          key={`p-${index}-${paragraph.substring(0, 20)}`}
-          className="markdown-paragraph markdown-streaming-paragraph"
-          ref={index === paragraphs.length - 1 ? lastParagraphRef : null}
-        >
-          <MemoizedMarkdownParagraph content={paragraph} />
-        </div>
-      ))}
-    </div>
-  );
-}
-
 function _MarkDownContent(props: { content: string }) {
   const escapedContent = useMemo(() => {
     return tryWrapHtmlCode(escapeBrackets(props.content));
@@ -456,27 +326,9 @@ export function Markdown(
     fontFamily?: string;
     parentRef?: RefObject<HTMLDivElement>;
     defaultShow?: boolean;
-    immediatelyRender?: boolean;
-    streaming?: boolean; // Whether this is a streaming response
   } & React.DOMAttributes<HTMLDivElement>,
 ) {
   const mdRef = useRef<HTMLDivElement>(null);
-  const paragraphs = useMemo(
-    () => splitContentIntoParagraphs(props.content),
-    [props.content],
-  );
-  const [loadedCount, setLoadedCount] = useState(0);
-
-  // Determine rendering strategy based on props
-  const shouldAsyncRender =
-    !props.immediatelyRender && !props.streaming && paragraphs.length > 1;
-
-  useEffect(() => {
-    // Immediately render all paragraphs if specified
-    if (props.immediatelyRender) {
-      setLoadedCount(paragraphs.length);
-    }
-  }, [props.immediatelyRender, paragraphs.length]);
 
   return (
     <div
@@ -492,24 +344,6 @@ export function Markdown(
     >
       {props.loading ? (
         <LoadingIcon />
-      ) : props.streaming ? (
-        // Use specialized component for streaming content
-        <StreamingMarkdownContent content={props.content} />
-      ) : shouldAsyncRender ? (
-        <div className="markdown-content">
-          {paragraphs.map((paragraph, index) => (
-            <MarkdownParagraph
-              key={index}
-              content={paragraph}
-              onLoad={() => setLoadedCount((prev) => prev + 1)}
-            />
-          ))}
-          {loadedCount < paragraphs.length && loadedCount > 0 && (
-            <div className="markdown-paragraph-loading">
-              <LoadingIcon />
-            </div>
-          )}
-        </div>
       ) : (
         <MarkdownContent content={props.content} />
       )}

app/components/tts-config.tsx

@@ -1,19 +1,76 @@
 import { TTSConfig, TTSConfigValidator } from "../store";
+import React, { useState } from "react";
 
 import Locale from "../locales";
 import { ListItem, Select } from "./ui-lib";
 import {
+  ModelProvider,
   DEFAULT_TTS_ENGINE,
   DEFAULT_TTS_ENGINES,
   DEFAULT_TTS_MODELS,
   DEFAULT_TTS_VOICES,
 } from "../constant";
 import { InputRange } from "./input-range";
+import { IconButton } from "./button";
+import SpeakIcon from "../icons/speak.svg";
+import SpeakStopIcon from "../icons/speak-stop.svg";
+import { createTTSPlayer } from "../utils/audio";
+import { useAppConfig } from "../store";
+import { ClientApi } from "../client/api";
+import { showToast } from "../components/ui-lib";
 
+const ttsPlayer = createTTSPlayer();
 export function TTSConfigList(props: {
   ttsConfig: TTSConfig;
   updateConfig: (updater: (config: TTSConfig) => void) => void;
 }) {
+  const [speechLoading, setSpeechLoading] = useState(false);
+  const [speechStatus, setSpeechStatus] = useState(false);
+
+  const config = useAppConfig.getState();
+
+  function stopSpeech() {
+    ttsPlayer.stop();
+    setSpeechStatus(false);
+  }
+
+  async function playSpeech(text: string, ttsConfig: TTSConfig) {
+    try {
+      const api = new ClientApi(ModelProvider.GPT);
+      setSpeechLoading(true);
+      ttsPlayer.init();
+
+      const audioBuffer = await api.llm.speech({
+        model: ttsConfig.model,
+        input: text,
+        voice: ttsConfig.voice,
+        speed: ttsConfig.speed,
+      });
+
+      setSpeechStatus(true);
+      await ttsPlayer.play(audioBuffer, () => {
+        setSpeechStatus(false);
+      });
+    } catch (error) {
+      console.error("[OpenAI Speech]", error);
+      setSpeechStatus(false);
+      // Implement user-facing error notification here
+      if (typeof (error as Error).message === "string") {
+        showToast((error as Error).message);
+      }
+    } finally {
+      setSpeechLoading(false);
+    }
+  }
+
+  async function openaiSpeech(text: string) {
+    if (speechStatus) {
+      stopSpeech();
+    } else {
+      await playSpeech(text, config.ttsConfig);
+    }
+  }
+
   return (
     <>
       <ListItem
@@ -88,23 +145,41 @@ export function TTSConfigList(props: {
             title={Locale.Settings.TTS.Voice.Title}
             subTitle={Locale.Settings.TTS.Voice.SubTitle}
           >
-            <Select
-              value={props.ttsConfig.voice}
-              onChange={(e) => {
-                props.updateConfig(
-                  (config) =>
-                    (config.voice = TTSConfigValidator.voice(
+            <div style={{ display: "flex", gap: "10px" }}>
+              <IconButton
+                aria={Locale.Chat.Actions.Speech}
+                icon={speechStatus ? <SpeakStopIcon /> : <SpeakIcon />}
+                text={
+                  speechLoading
+                    ? "Loading..."
+                    : speechStatus
+                    ? Locale.Chat.Actions.Stop
+                    : Locale.Chat.Actions.Speech
+                }
+                onClick={() => {
+                  openaiSpeech(
+                    "NextChat,Unleash your imagination, experience the future of AI conversation.",
+                  );
+                }}
+              />
+
+              <Select
+                value={props.ttsConfig.voice}
+                onChange={(e) => {
+                  props.updateConfig((config) => {
+                    config.voice = TTSConfigValidator.voice(
                       e.currentTarget.value,
-                    )),
-                );
-              }}
-            >
-              {DEFAULT_TTS_VOICES.map((v, i) => (
-                <option value={v} key={i}>
-                  {v}
-                </option>
-              ))}
-            </Select>
+                    );
+                  });
+                }}
+              >
+                {DEFAULT_TTS_VOICES.map((v, i) => (
+                  <option value={v} key={i}>
+                    {v}
+                  </option>
+                ))}
+              </Select>
+            </div>
           </ListItem>
           <ListItem
             title={Locale.Settings.TTS.Speed.Title}

app/styles/markdown.scss

@@ -99,7 +99,6 @@
   font-size: 14px;
   line-height: 1.5;
   word-wrap: break-word;
-  margin-bottom: 0;
 }
 
 .light {
@@ -359,14 +358,8 @@
 .markdown-body kbd {
   display: inline-block;
   padding: 3px 5px;
-  font:
-    11px ui-monospace,
-    SFMono-Regular,
-    SF Mono,
-    Menlo,
-    Consolas,
-    Liberation Mono,
-    monospace;
+  font: 11px ui-monospace, SFMono-Regular, SF Mono, Menlo, Consolas,
+    Liberation Mono, monospace;
   line-height: 10px;
   color: var(--color-fg-default);
   vertical-align: middle;
@@ -455,28 +448,16 @@
 .markdown-body tt,
 .markdown-body code,
 .markdown-body samp {
-  font-family:
-    ui-monospace,
-    SFMono-Regular,
-    SF Mono,
-    Menlo,
-    Consolas,
-    Liberation Mono,
-    monospace;
+  font-family: ui-monospace, SFMono-Regular, SF Mono, Menlo, Consolas,
+    Liberation Mono, monospace;
   font-size: 12px;
 }
 
 .markdown-body pre {
   margin-top: 0;
   margin-bottom: 0;
-  font-family:
-    ui-monospace,
-    SFMono-Regular,
-    SF Mono,
-    Menlo,
-    Consolas,
-    Liberation Mono,
-    monospace;
+  font-family: ui-monospace, SFMono-Regular, SF Mono, Menlo, Consolas,
+    Liberation Mono, monospace;
   font-size: 12px;
   word-wrap: normal;
 }
@@ -1149,87 +1130,3 @@
 #dmermaid {
   display: none;
 }
-
-.markdown-content {
-  width: 100%;
-}
-
-.markdown-paragraph {
-  transition: opacity 0.3s ease;
-  margin-bottom: 0.5em;
-
-  &.markdown-paragraph-visible {
-    opacity: 1;
-  }
-
-  &.markdown-paragraph-hidden {
-    opacity: 0.7;
-  }
-}
-
-.markdown-paragraph-placeholder {
-  padding: 8px;
-  color: var(--color-fg-subtle);
-  background-color: var(--color-canvas-subtle);
-  border-radius: 6px;
-  border-left: 3px solid var(--color-border-muted);
-  white-space: nowrap;
-  overflow: hidden;
-  text-overflow: ellipsis;
-  font-family: var(--font-family-sans);
-  font-size: 14px;
-  min-height: 1.2em;
-}
-
-.markdown-paragraph-loading {
-  height: 20px;
-  background-color: var(--color-canvas-subtle);
-  border-radius: 6px;
-  margin-bottom: 8px;
-  position: relative;
-  overflow: hidden;
-
-  &::after {
-    content: "";
-    position: absolute;
-    top: 0;
-    left: 0;
-    width: 30%;
-    height: 100%;
-    background: linear-gradient(
-      90deg,
-      transparent,
-      rgba(255, 255, 255, 0.1),
-      transparent
-    );
-    animation: shimmer 1.5s infinite;
-  }
-}
-
-@keyframes shimmer {
-  0% {
-    transform: translateX(-100%);
-  }
-  100% {
-    transform: translateX(200%);
-  }
-}
-
-.markdown-streaming-content {
-  width: 100%;
-}
-
-.markdown-streaming-paragraph {
-  opacity: 1;
-  animation: fadeIn 0.3s ease-in-out;
-  margin-bottom: 0.5em;
-}
-
-@keyframes fadeIn {
-  from {
-    opacity: 0.5;
-  }
-  to {
-    opacity: 1;
-  }
-}