| 1 | import { lazy, memo, startTransition, Suspense, useCallback, useEffect, useLayoutEffect, useRef, useState } from "react"; |
| 2 | import { t } from "../lib/i18n"; |
| 3 | |
| 4 | async function loadMarkdownView<T>(component: Promise<T>): Promise<T> { |
| 5 | await Promise.all([import("./MarkdownImage.css"), import("./CodeBlock.css")]); |
| 6 | return component; |
| 7 | } |
| 8 | |
| 9 | let historyView: typeof import("./MarkdownHistory").default | undefined; |
| 10 | let historyViewPromise: Promise<typeof import("./MarkdownHistory")> | undefined; |
| 11 | export function preloadMarkdownHistory(): Promise<typeof import("./MarkdownHistory")> { |
| 12 | return historyViewPromise ??= loadMarkdownView(import("./MarkdownHistory")).then(module => { |
| 13 | historyView = module.default; |
| 14 | return module; |
| 15 | }); |
| 16 | } |
| 17 | const LazyMarkdownHistory = lazy(preloadMarkdownHistory); |
| 18 | const LiveMarkdownCode = lazy(() => loadMarkdownView(import("./markdownComponents")).then(module => ({ default: module.MarkdownCode }))); |
| 19 | const STREAMING_TAIL_THRESHOLD = 8_000; |
| 20 | const FINALIZE_SETTLE_MS = 50; |
| 21 | const FINALIZE_IDLE_TIMEOUT_MS = 1_000; |
| 22 | const MARKDOWN_SECTION_TARGET_CHARS = 12_000; |
| 23 | const CROSS_SECTION_REFERENCE_RE = /(?:^ {0,3}\[[^\]\n]+\]:|\[\^[^\]\n]+\])/m; |
| 24 | const CROSS_SECTION_CONTAINER_RE = /^ {0,3}(?:>|(?:[-+*]|\d{1,9}[.)])(?:[ \t]+|$)|<)/; |
| 25 | |
| 26 | function scanMarkdownSections(text: string): { boundaries: number[]; hasCrossSectionContainer: boolean } { |
| 27 | const boundaries = [0]; |
| 28 | let lineStart = 0; |
| 29 | let fence: { marker: string; length: number } | null = null; |
| 30 | let displayMath = false; |
| 31 | let boundaryAfterFence = false; |
| 32 | let hasCrossSectionContainer = false; |
| 33 | |
| 34 | const addBoundary = (offset: number) => { |
| 35 | if (offset > 0 && boundaries[boundaries.length - 1] !== offset) boundaries.push(offset); |
| 36 | }; |
| 37 | |
| 38 | while (lineStart < text.length) { |
| 39 | const newline = text.indexOf("\n", lineStart); |
| 40 | const lineEnd = newline === -1 ? text.length : newline + 1; |
| 41 | const line = text.slice(lineStart, newline === -1 ? text.length : newline).replace(/\r$/, ""); |
| 42 | const trimmed = line.trim(); |
| 43 | |
| 44 | if (fence) { |
| 45 | const close = new RegExp(`^ {0,3}${fence.marker}{${fence.length},}[ \\t]*$`); |
| 46 | if (close.test(line)) { |
| 47 | fence = null; |
| 48 | boundaryAfterFence = true; |
| 49 | } |
| 50 | lineStart = lineEnd; |
| 51 | continue; |
| 52 | } |
| 53 | |
| 54 | if (displayMath) { |
| 55 | if (trimmed === "$$") { |
| 56 | displayMath = false; |
| 57 | boundaryAfterFence = true; |
| 58 | } |
| 59 | lineStart = lineEnd; |
| 60 | continue; |
| 61 | } |
| 62 | |
| 63 | if (boundaryAfterFence && trimmed !== "") { |
| 64 | addBoundary(lineStart); |
| 65 | boundaryAfterFence = false; |
| 66 | } |
| 67 | |
| 68 | if (CROSS_SECTION_CONTAINER_RE.test(line)) { |
| 69 | hasCrossSectionContainer = true; |
| 70 | } |
| 71 | const fenceMatch = /^ {0,3}(`{3,}|~{3,})/.exec(line); |
| 72 | if (fenceMatch) { |
| 73 | addBoundary(lineStart); |
| 74 | fence = { marker: fenceMatch[1][0], length: fenceMatch[1].length }; |
| 75 | } else if (trimmed === "$$") { |
| 76 | addBoundary(lineStart); |
| 77 | displayMath = true; |
| 78 | } else if (/^ {0,3}#{1,6}(?:[ \t]+|$)/.test(line)) { |
| 79 | addBoundary(lineStart); |
| 80 | } |
| 81 | lineStart = lineEnd; |
| 82 | } |
| 83 | |
| 84 | return { boundaries, hasCrossSectionContainer }; |
| 85 | } |
| 86 | |
| 87 | // Stable top-level sections let React.memo retain completed Markdown while the |
| 88 | // streaming tail changes. Cross-section references intentionally stay in one |
| 89 | // renderer because their definitions can affect nodes anywhere in the document. |
| 90 | export function splitStableMarkdownSections(text: string): string[] { |
| 91 | if (text.length < MARKDOWN_SECTION_TARGET_CHARS || CROSS_SECTION_REFERENCE_RE.test(text)) return [text]; |
| 92 | const { boundaries, hasCrossSectionContainer } = scanMarkdownSections(text); |
| 93 | if (hasCrossSectionContainer) return [text]; |
| 94 | if (boundaries.length === 1) return [text]; |
| 95 | |
| 96 | const sections = boundaries.map((start, index) => text.slice(start, boundaries[index + 1] ?? text.length)); |
| 97 | const chunks: string[] = []; |
| 98 | let current = ""; |
| 99 | for (const section of sections) { |
| 100 | if (current && current.length + section.length > MARKDOWN_SECTION_TARGET_CHARS) { |
| 101 | chunks.push(current); |
| 102 | current = section; |
| 103 | } else { |
| 104 | current += section; |
| 105 | } |
| 106 | } |
| 107 | if (current) chunks.push(current); |
| 108 | return chunks.length > 1 ? chunks : [text]; |
| 109 | } |
| 110 | |
| 111 | type IdleWindow = Window & { |
| 112 | requestIdleCallback?: (callback: () => void, options?: { timeout: number }) => number; |
| 113 | cancelIdleCallback?: (handle: number) => void; |
| 114 | }; |
| 115 | |
| 116 | function scheduleMarkdownFinalization(callback: () => void): () => void { |
| 117 | const idleWindow = window as IdleWindow; |
| 118 | let cancelled = false; |
| 119 | let idleHandle: number | null = null; |
| 120 | let frameHandle: number | null = null; |
| 121 | const timeoutHandle = window.setTimeout(() => { |
| 122 | if (cancelled) return; |
| 123 | const run = () => { |
| 124 | if (cancelled) return; |
| 125 | startTransition(callback); |
| 126 | }; |
| 127 | if (idleWindow.requestIdleCallback) { |
| 128 | idleHandle = idleWindow.requestIdleCallback(run, { timeout: FINALIZE_IDLE_TIMEOUT_MS }); |
| 129 | } else { |
| 130 | frameHandle = requestAnimationFrame(run); |
| 131 | } |
| 132 | }, FINALIZE_SETTLE_MS); |
| 133 | |
| 134 | return () => { |
| 135 | cancelled = true; |
| 136 | window.clearTimeout(timeoutHandle); |
| 137 | if (idleHandle !== null) idleWindow.cancelIdleCallback?.(idleHandle); |
| 138 | if (frameHandle !== null) cancelAnimationFrame(frameHandle); |
| 139 | }; |
| 140 | } |
| 141 | |
| 142 | export function streamingMarkdownCommitInterval(textLength: number): number { |
| 143 | if (textLength >= 32_000) return 300; |
| 144 | if (textLength >= 8_000) return 150; |
| 145 | return 50; |
| 146 | } |
| 147 | |
| 148 | const STREAMING_LIST_ITEM_RE = /^ {0,3}(?:[*+-]|\d{1,9}[.)])(?:[ \t]+|$)/; |
| 149 | const STREAMING_THEMATIC_BREAK_RE = /^ {0,3}(?:(?:-[ \t]*){3,}|(?:\*[ \t]*){3,}|(?:_[ \t]*){3,})[ \t]*$/; |
| 150 | |
| 151 | function isStreamingListItemLine(line: string): boolean { |
| 152 | return STREAMING_LIST_ITEM_RE.test(line) && !STREAMING_THEMATIC_BREAK_RE.test(line); |
| 153 | } |
| 154 | |
| 155 | // Live parse prefix: last completed block. A later list marker commits prior |
| 156 | // items only — the new item stays in the tail so indented continuations can join. |
| 157 | // An open code fence commits only up to the fence line: the tail renders the |
| 158 | // growing code with code styling (splitStreamingTailFence), so streaming a |
| 159 | // large block no longer re-parses the whole document on every commit. |
| 160 | export function streamingCommitTarget(text: string): string { |
| 161 | let lineStart = 0; |
| 162 | let fence: { marker: string; length: number } | null = null; |
| 163 | let fenceStart = 0; |
| 164 | let displayMath = false; |
| 165 | let boundary = 0; |
| 166 | while (lineStart < text.length) { |
| 167 | const newline = text.indexOf("\n", lineStart); |
| 168 | const lineEnd = newline === -1 ? text.length : newline + 1; |
| 169 | const line = text.slice(lineStart, newline === -1 ? text.length : newline).replace(/\r$/, ""); |
| 170 | const terminated = newline !== -1; |
| 171 | if (fence) { |
| 172 | if (new RegExp(`^ {0,3}${fence.marker}{${fence.length},}[ \\t]*$`).test(line)) { |
| 173 | fence = null; |
| 174 | if (terminated) boundary = lineEnd; |
| 175 | } |
| 176 | } else if (displayMath) { |
| 177 | if (line.trim() === "$$") { |
| 178 | displayMath = false; |
| 179 | if (terminated) boundary = lineEnd; |
| 180 | } |
| 181 | } else { |
| 182 | const fenceMatch = /^ {0,3}(`{3,}|~{3,})/.exec(line); |
| 183 | if (fenceMatch) { |
| 184 | fence = { marker: fenceMatch[1][0], length: fenceMatch[1].length }; |
| 185 | fenceStart = lineStart; |
| 186 | } else if (line.trim() === "$$") displayMath = true; |
| 187 | else if (terminated && line.trim() === "") boundary = lineEnd; |
| 188 | // A heading interrupts a paragraph, so a partial heading line already |
| 189 | // completes everything before it; a terminated one is itself complete. |
| 190 | else if (/^ {0,3}#{1,6}[ \t]+/.test(line)) boundary = terminated ? lineEnd : lineStart; |
| 191 | else if (isStreamingListItemLine(line)) boundary = lineStart; |
| 192 | } |
| 193 | lineStart = lineEnd; |
| 194 | } |
| 195 | return fence ? text.slice(0, fenceStart) : displayMath ? text : text.slice(0, boundary); |
| 196 | } |
| 197 | |
| 198 | type StreamingTailFence = { head: string; lang: string; code: string }; |
| 199 | |
| 200 | // Split a streaming tail around an unclosed code fence so the fence body can |
| 201 | // render with code styling before the closing fence arrives. Bail out cheaply |
| 202 | // when no fence marker exists; otherwise mirror the fence state machine from |
| 203 | // streamingCommitTarget in one forward pass over the tail. |
| 204 | export function splitStreamingTailFence(text: string, includeJustClosed = false): StreamingTailFence | null { |
| 205 | if (!text.includes("```") && !text.includes("~~~")) return null; |
| 206 | let lineStart = 0; |
| 207 | let fence: { marker: string; length: number } | null = null; |
| 208 | let fenceStart = 0; |
| 209 | let fenceBodyStart = 0; |
| 210 | let lang = ""; |
| 211 | while (lineStart < text.length) { |
| 212 | const newline = text.indexOf("\n", lineStart); |
| 213 | const lineEnd = newline === -1 ? text.length : newline + 1; |
| 214 | const line = text.slice(lineStart, newline === -1 ? text.length : newline).replace(/\r$/, ""); |
| 215 | if (fence) { |
| 216 | if (new RegExp(`^ {0,3}${fence.marker}{${fence.length},}[ \\t]*$`).test(line)) { |
| 217 | // Keep the styled fence until the committed parser takes it over. |
| 218 | if (includeJustClosed && !text.slice(lineEnd).trim()) { |
| 219 | return { head: text.slice(0, fenceStart), lang, code: text.slice(fenceBodyStart, lineStart) }; |
| 220 | } |
| 221 | fence = null; |
| 222 | } |
| 223 | } else { |
| 224 | const fenceMatch = /^ {0,3}(`{3,}|~{3,})([^\n]*)$/.exec(line); |
| 225 | if (fenceMatch && newline !== -1) { |
| 226 | fence = { marker: fenceMatch[1][0], length: fenceMatch[1].length }; |
| 227 | fenceStart = lineStart; |
| 228 | fenceBodyStart = lineEnd; |
| 229 | lang = fenceMatch[2].trim().split(/\s+/)[0]; |
| 230 | } |
| 231 | } |
| 232 | lineStart = lineEnd; |
| 233 | } |
| 234 | if (!fence) return null; |
| 235 | return { head: text.slice(0, fenceStart), lang, code: text.slice(fenceBodyStart) }; |
| 236 | } |
| 237 | |
| 238 | export function useRenderedMarkdownText(text: string, streaming: boolean, holdIdleFinalization = false): string { |
| 239 | const [renderedText, setRenderedText] = useState(() => streaming && splitStreamingTailFence(text) ? streamingCommitTarget(text) : text); |
| 240 | const latestTextRef = useRef(text); |
| 241 | const frameRef = useRef<number | null>(null); |
| 242 | const timeoutRef = useRef<number | null>(null); |
| 243 | const lastCommitAtRef = useRef(0); |
| 244 | const wasStreamingRef = useRef(streaming); |
| 245 | const finalizingTextRef = useRef<string | null>(null); |
| 246 | const cancelFinalizationRef = useRef<(() => void) | null>(null); |
| 247 | const finalizationStartedAtRef = useRef(0); |
| 248 | const finalizationLengthRef = useRef(0); |
| 249 | |
| 250 | latestTextRef.current = text; |
| 251 | |
| 252 | useLayoutEffect(() => { |
| 253 | const endedStreaming = wasStreamingRef.current && !streaming; |
| 254 | wasStreamingRef.current = streaming; |
| 255 | if (streaming) { |
| 256 | cancelFinalizationRef.current?.(); |
| 257 | cancelFinalizationRef.current = null; |
| 258 | finalizingTextRef.current = null; |
| 259 | // A bounded live preview occasionally advances its window and drops an |
| 260 | // old prefix. Discard the stale parsed tree before paint; the complete |
| 261 | // replacement stays visible through StreamingMarkdownTail and is parsed |
| 262 | // later under the normal adaptive budget. |
| 263 | if (renderedText !== "" && !text.startsWith(renderedText)) { |
| 264 | setRenderedText(""); |
| 265 | } |
| 266 | return; |
| 267 | } |
| 268 | lastCommitAtRef.current = 0; |
| 269 | if (frameRef.current !== null) { |
| 270 | cancelAnimationFrame(frameRef.current); |
| 271 | frameRef.current = null; |
| 272 | } |
| 273 | if (timeoutRef.current !== null) { |
| 274 | window.clearTimeout(timeoutRef.current); |
| 275 | timeoutRef.current = null; |
| 276 | } |
| 277 | if (renderedText === text) { |
| 278 | cancelFinalizationRef.current?.(); |
| 279 | cancelFinalizationRef.current = null; |
| 280 | finalizingTextRef.current = null; |
| 281 | if (finalizationStartedAtRef.current > 0) { |
| 282 | performance.measure("reasonix:markdown-finalize", { |
| 283 | start: finalizationStartedAtRef.current, |
| 284 | end: performance.now(), |
| 285 | detail: { textLength: finalizationLengthRef.current }, |
| 286 | }); |
| 287 | finalizationStartedAtRef.current = 0; |
| 288 | finalizationLengthRef.current = 0; |
| 289 | } |
| 290 | return; |
| 291 | } |
| 292 | |
| 293 | const canFinalizeWhenIdle = |
| 294 | (endedStreaming || finalizingTextRef.current !== null) && |
| 295 | text.length >= STREAMING_TAIL_THRESHOLD && |
| 296 | text.startsWith(renderedText); |
| 297 | if (canFinalizeWhenIdle) { |
| 298 | // The worker path owns the final parse of a completed stream: holding |
| 299 | // here keeps the committed prefix frozen instead of re-parsing the full |
| 300 | // document on the main thread at idle time. |
| 301 | if (holdIdleFinalization) return; |
| 302 | if (finalizingTextRef.current === text) return; |
| 303 | cancelFinalizationRef.current?.(); |
| 304 | finalizingTextRef.current = text; |
| 305 | cancelFinalizationRef.current = scheduleMarkdownFinalization(() => { |
| 306 | cancelFinalizationRef.current = null; |
| 307 | finalizationStartedAtRef.current = performance.now(); |
| 308 | finalizationLengthRef.current = latestTextRef.current.length; |
| 309 | setRenderedText(latestTextRef.current); |
| 310 | }); |
| 311 | return; |
| 312 | } |
| 313 | |
| 314 | cancelFinalizationRef.current?.(); |
| 315 | cancelFinalizationRef.current = null; |
| 316 | finalizingTextRef.current = null; |
| 317 | setRenderedText(text); |
| 318 | }, [renderedText, streaming, text, holdIdleFinalization]); |
| 319 | |
| 320 | useLayoutEffect(() => { |
| 321 | if (streaming) lastCommitAtRef.current = performance.now(); |
| 322 | }, [renderedText, streaming]); |
| 323 | |
| 324 | useEffect(() => { |
| 325 | if (!streaming || frameRef.current !== null || timeoutRef.current !== null) return; |
| 326 | if (streamingCommitTarget(text).length <= renderedText.length) return; |
| 327 | const commit = () => { |
| 328 | timeoutRef.current = null; |
| 329 | frameRef.current = requestAnimationFrame(() => { |
| 330 | frameRef.current = null; |
| 331 | // Recompute at commit time: only ever advance to a newer boundary. |
| 332 | const target = streamingCommitTarget(latestTextRef.current); |
| 333 | setRenderedText((prev) => (target.length > prev.length ? target : prev)); |
| 334 | }); |
| 335 | }; |
| 336 | const now = performance.now(); |
| 337 | const elapsed = lastCommitAtRef.current === 0 ? Number.POSITIVE_INFINITY : now - lastCommitAtRef.current; |
| 338 | const delay = streamingMarkdownCommitInterval(text.length) - elapsed; |
| 339 | if (delay <= 0) commit(); |
| 340 | else timeoutRef.current = window.setTimeout(commit, delay); |
| 341 | }, [renderedText, streaming, text]); |
| 342 | |
| 343 | useEffect(() => () => { |
| 344 | if (frameRef.current !== null) cancelAnimationFrame(frameRef.current); |
| 345 | if (timeoutRef.current !== null) window.clearTimeout(timeoutRef.current); |
| 346 | cancelFinalizationRef.current?.(); |
| 347 | }, []); |
| 348 | |
| 349 | return renderedText; |
| 350 | } |
| 351 | |
| 352 | function StreamingMarkdownTail({ text }: { text: string }) { |
| 353 | const fence = splitStreamingTailFence(text, true); |
| 354 | if (!fence) return <span className="md" style={{ whiteSpace: "pre-wrap" }}>{text}</span>; |
| 355 | const value = fence.code.replace(/\n$/, ""); |
| 356 | return <div className="md"> |
| 357 | {fence.head && <span style={{ whiteSpace: "pre-wrap" }}>{fence.head}</span>} |
| 358 | <Suspense fallback={<pre className="code">{value}</pre>}> |
| 359 | <LiveMarkdownCode value={value} language={fence.lang} /> |
| 360 | </Suspense> |
| 361 | </div>; |
| 362 | } |
| 363 | |
| 364 | export const Markdown = memo(function Markdown({ |
| 365 | text, plainStatusBlocks = false, streaming = false, cacheKey, |
| 366 | }: { text: string; plainStatusBlocks?: boolean; streaming?: boolean; cacheKey?: string; wasStreamed?: boolean }) { |
| 367 | const renderedText = useRenderedMarkdownText(text, streaming, false); |
| 368 | const [failed, setFailed] = useState<string>(); |
| 369 | const onError = useCallback(() => setFailed(text), [text]); |
| 370 | const History = historyView ?? LazyMarkdownHistory; |
| 371 | const fallback = <div className="md" style={{ whiteSpace: "pre-wrap" }}>{text}</div>; |
| 372 | if (failed === text) return <><span className="chat-notice" role="status">{t("chat.parseFailed")}</span>{fallback}</>; |
| 373 | return <Suspense fallback={fallback}> |
| 374 | <History text={streaming ? renderedText : text} streaming={streaming} |
| 375 | plainStatusBlocks={plainStatusBlocks} cacheKey={cacheKey} |
| 376 | fallback={<span style={{ whiteSpace: "pre-wrap" }}>{streaming ? renderedText : text}</span>} onError={onError} /> |
| 377 | {streaming && text.startsWith(renderedText) && text.length > renderedText.length && |
| 378 | <StreamingMarkdownTail text={text.slice(renderedText.length)} />} |
| 379 | </Suspense>; |
| 380 | }); |
| 381 |