Timeline Perf Improvement (#1521)

* emojify msg txt find&replace instead of recursion

* move findAndReplace func in its own file

* improve find and replace

* move markdown file to plugins

* make find and replace work without g flag regex

* fix pagination stop on msg arrive

* render blurhash in small size

src/app/components/editor/output.ts

@@ -3,7 +3,8 @@ import { Descendant, Text } from 'slate';
 import { sanitizeText } from '../../utils/sanitize';
 import { BlockType } from './types';
 import { CustomElement } from './slate';
-import { parseBlockMD, parseInlineMD, replaceMatch } from '../../utils/markdown';
+import { parseBlockMD, parseInlineMD } from '../../plugins/markdown';
+import { findAndReplace } from '../../utils/findAndReplace';
 
 export type OutputOptions = {
   allowTextFormatting?: boolean;
@@ -69,14 +70,14 @@ const elementToCustomHtml = (node: CustomElement, children: string): string => {
   }
 };
 
-const HTML_TAG_REG = /<([\w-]+)(?: [^>]*)?(?:(?:\/>)|(?:>.*?<\/\1>))/;
+const HTML_TAG_REG_G = /<([\w-]+)(?: [^>]*)?(?:(?:\/>)|(?:>.*?<\/\1>))/g;
-const ignoreHTMLParseInlineMD = (text: string): string => {
+const ignoreHTMLParseInlineMD = (text: string): string =>
-  if (text === '') return text;
+  findAndReplace(
-  const match = text.match(HTML_TAG_REG);
+    text,
-  if (!match) return parseInlineMD(text);
+    HTML_TAG_REG_G,
-  const [matchedTxt] = match;
+    (match) => match[0],
-  return replaceMatch((txt) => [ignoreHTMLParseInlineMD(txt)], text, match, matchedTxt).join('');
+    (txt) => parseInlineMD(txt)
-};
+  ).join('');
 
 export const toMatrixCustomHTML = (
   node: Descendant | Descendant[],

src/app/organisms/room/RoomTimeline.tsx

@@ -345,7 +345,6 @@ const useTimelinePagination = (
 
     return async (backwards: boolean) => {
       if (fetching) return;
-      const targetTimeline = timelineRef.current;
       const { linkedTimelines: lTimelines } = timelineRef.current;
       const timelinesEventsCount = lTimelines.map(timelineToEventsCount);
 
@@ -385,7 +384,6 @@ const useTimelinePagination = (
       }
 
       fetching = false;
-      if (targetTimeline !== timelineRef.current) return;
       if (alive()) {
         recalibratePagination(lTimelines, timelinesEventsCount, backwards);
       }

src/app/organisms/room/message/ImageContent.tsx

@@ -98,7 +98,13 @@ export const ImageContent = as<'div', ImageContentProps>(
           </Overlay>
         )}
         {typeof blurHash === 'string' && !load && (
-          <BlurhashCanvas style={{ width: '100%', height: '100%' }} hash={blurHash} punch={1} />
+          <BlurhashCanvas
+            style={{ width: '100%', height: '100%' }}
+            width={32}
+            height={32}
+            hash={blurHash}
+            punch={1}
+          />
         )}
         {!autoPlay && srcState.status === AsyncStatus.Idle && (
           <Box className={css.AbsoluteContainer} alignItems="Center" justifyContent="Center">

src/app/organisms/room/message/VideoContent.tsx

@@ -88,7 +88,13 @@ export const VideoContent = as<'div', VideoContentProps>(
     return (
       <Box className={classNames(css.RelativeBase, className)} {...props} ref={ref}>
         {typeof blurHash === 'string' && !load && (
-          <BlurhashCanvas style={{ width: '100%', height: '100%' }} hash={blurHash} punch={1} />
+          <BlurhashCanvas
+            style={{ width: '100%', height: '100%' }}
+            width={32}
+            height={32}
+            hash={blurHash}
+            punch={1}
+          />
         )}
         {thumbSrcState.status === AsyncStatus.Success && !load && (
           <Box className={css.AbsoluteContainer} alignItems="Center" justifyContent="Center">

src/app/plugins/markdown.ts

@@ -0,0 +1,368 @@
+export type MatchResult = RegExpMatchArray | RegExpExecArray;
+export type RuleMatch = (text: string) => MatchResult | null;
+
+export const beforeMatch = (text: string, match: RegExpMatchArray | RegExpExecArray): string =>
+  text.slice(0, match.index);
+export const afterMatch = (text: string, match: RegExpMatchArray | RegExpExecArray): string =>
+  text.slice((match.index ?? 0) + match[0].length);
+
+export const replaceMatch = <C>(
+  convertPart: (txt: string) => Array<string | C>,
+  text: string,
+  match: MatchResult,
+  content: C
+): Array<string | C> => [
+  ...convertPart(beforeMatch(text, match)),
+  content,
+  ...convertPart(afterMatch(text, match)),
+];
+
+/*
+ *****************
+ * INLINE PARSER *
+ *****************
+ */
+
+export type InlineMDParser = (text: string) => string;
+
+export type InlineMatchConverter = (parse: InlineMDParser, match: MatchResult) => string;
+
+export type InlineMDRule = {
+  match: RuleMatch;
+  html: InlineMatchConverter;
+};
+
+export type InlineRuleRunner = (
+  parse: InlineMDParser,
+  text: string,
+  rule: InlineMDRule
+) => string | undefined;
+export type InlineRulesRunner = (
+  parse: InlineMDParser,
+  text: string,
+  rules: InlineMDRule[]
+) => string | undefined;
+
+const MIN_ANY = '(.+?)';
+const URL_NEG_LB = '(?<!(https?|ftp|mailto|magnet):\\/\\/\\S*)';
+
+const BOLD_MD_1 = '**';
+const BOLD_PREFIX_1 = '\\*{2}';
+const BOLD_NEG_LA_1 = '(?!\\*)';
+const BOLD_REG_1 = new RegExp(
+  `${URL_NEG_LB}${BOLD_PREFIX_1}${MIN_ANY}${BOLD_PREFIX_1}${BOLD_NEG_LA_1}`
+);
+const BoldRule: InlineMDRule = {
+  match: (text) => text.match(BOLD_REG_1),
+  html: (parse, match) => {
+    const [, , g2] = match;
+    return `<strong data-md="${BOLD_MD_1}">${parse(g2)}</strong>`;
+  },
+};
+
+const ITALIC_MD_1 = '*';
+const ITALIC_PREFIX_1 = '\\*';
+const ITALIC_NEG_LA_1 = '(?!\\*)';
+const ITALIC_REG_1 = new RegExp(
+  `${URL_NEG_LB}${ITALIC_PREFIX_1}${MIN_ANY}${ITALIC_PREFIX_1}${ITALIC_NEG_LA_1}`
+);
+const ItalicRule1: InlineMDRule = {
+  match: (text) => text.match(ITALIC_REG_1),
+  html: (parse, match) => {
+    const [, , g2] = match;
+    return `<i data-md="${ITALIC_MD_1}">${parse(g2)}</i>`;
+  },
+};
+
+const ITALIC_MD_2 = '_';
+const ITALIC_PREFIX_2 = '_';
+const ITALIC_NEG_LA_2 = '(?!_)';
+const ITALIC_REG_2 = new RegExp(
+  `${URL_NEG_LB}${ITALIC_PREFIX_2}${MIN_ANY}${ITALIC_PREFIX_2}${ITALIC_NEG_LA_2}`
+);
+const ItalicRule2: InlineMDRule = {
+  match: (text) => text.match(ITALIC_REG_2),
+  html: (parse, match) => {
+    const [, , g2] = match;
+    return `<i data-md="${ITALIC_MD_2}">${parse(g2)}</i>`;
+  },
+};
+
+const UNDERLINE_MD_1 = '__';
+const UNDERLINE_PREFIX_1 = '_{2}';
+const UNDERLINE_NEG_LA_1 = '(?!_)';
+const UNDERLINE_REG_1 = new RegExp(
+  `${URL_NEG_LB}${UNDERLINE_PREFIX_1}${MIN_ANY}${UNDERLINE_PREFIX_1}${UNDERLINE_NEG_LA_1}`
+);
+const UnderlineRule: InlineMDRule = {
+  match: (text) => text.match(UNDERLINE_REG_1),
+  html: (parse, match) => {
+    const [, , g2] = match;
+    return `<u data-md="${UNDERLINE_MD_1}">${parse(g2)}</u>`;
+  },
+};
+
+const STRIKE_MD_1 = '~~';
+const STRIKE_PREFIX_1 = '~{2}';
+const STRIKE_NEG_LA_1 = '(?!~)';
+const STRIKE_REG_1 = new RegExp(
+  `${URL_NEG_LB}${STRIKE_PREFIX_1}${MIN_ANY}${STRIKE_PREFIX_1}${STRIKE_NEG_LA_1}`
+);
+const StrikeRule: InlineMDRule = {
+  match: (text) => text.match(STRIKE_REG_1),
+  html: (parse, match) => {
+    const [, , g2] = match;
+    return `<del data-md="${STRIKE_MD_1}">${parse(g2)}</del>`;
+  },
+};
+
+const CODE_MD_1 = '`';
+const CODE_PREFIX_1 = '`';
+const CODE_NEG_LA_1 = '(?!`)';
+const CODE_REG_1 = new RegExp(`${URL_NEG_LB}${CODE_PREFIX_1}(.+?)${CODE_PREFIX_1}${CODE_NEG_LA_1}`);
+const CodeRule: InlineMDRule = {
+  match: (text) => text.match(CODE_REG_1),
+  html: (parse, match) => {
+    const [, , g2] = match;
+    return `<code data-md="${CODE_MD_1}">${g2}</code>`;
+  },
+};
+
+const SPOILER_MD_1 = '||';
+const SPOILER_PREFIX_1 = '\\|{2}';
+const SPOILER_NEG_LA_1 = '(?!\\|)';
+const SPOILER_REG_1 = new RegExp(
+  `${URL_NEG_LB}${SPOILER_PREFIX_1}${MIN_ANY}${SPOILER_PREFIX_1}${SPOILER_NEG_LA_1}`
+);
+const SpoilerRule: InlineMDRule = {
+  match: (text) => text.match(SPOILER_REG_1),
+  html: (parse, match) => {
+    const [, , g2] = match;
+    return `<span data-md="${SPOILER_MD_1}" data-mx-spoiler>${parse(g2)}</span>`;
+  },
+};
+
+const LINK_ALT = `\\[${MIN_ANY}\\]`;
+const LINK_URL = `\\((https?:\\/\\/.+?)\\)`;
+const LINK_REG_1 = new RegExp(`${LINK_ALT}${LINK_URL}`);
+const LinkRule: InlineMDRule = {
+  match: (text) => text.match(LINK_REG_1),
+  html: (parse, match) => {
+    const [, g1, g2] = match;
+    return `<a data-md href="${g2}">${parse(g1)}</a>`;
+  },
+};
+
+const runInlineRule: InlineRuleRunner = (parse, text, rule) => {
+  const matchResult = rule.match(text);
+  if (matchResult) {
+    const content = rule.html(parse, matchResult);
+    return replaceMatch((txt) => [parse(txt)], text, matchResult, content).join('');
+  }
+  return undefined;
+};
+
+/**
+ * Runs multiple rules at the same time to better handle nested rules.
+ * Rules will be run in the order they appear.
+ */
+const runInlineRules: InlineRulesRunner = (parse, text, rules) => {
+  const matchResults = rules.map((rule) => rule.match(text));
+
+  let targetRule: InlineMDRule | undefined;
+  let targetResult: MatchResult | undefined;
+
+  for (let i = 0; i < matchResults.length; i += 1) {
+    const currentResult = matchResults[i];
+    if (currentResult && typeof currentResult.index === 'number') {
+      if (
+        !targetResult ||
+        (typeof targetResult?.index === 'number' && currentResult.index < targetResult.index)
+      ) {
+        targetResult = currentResult;
+        targetRule = rules[i];
+      }
+    }
+  }
+
+  if (targetRule && targetResult) {
+    const content = targetRule.html(parse, targetResult);
+    return replaceMatch((txt) => [parse(txt)], text, targetResult, content).join('');
+  }
+  return undefined;
+};
+
+const LeveledRules = [
+  BoldRule,
+  ItalicRule1,
+  UnderlineRule,
+  ItalicRule2,
+  StrikeRule,
+  SpoilerRule,
+  LinkRule,
+];
+
+export const parseInlineMD: InlineMDParser = (text) => {
+  if (text === '') return text;
+  let result: string | undefined;
+  if (!result) result = runInlineRule(parseInlineMD, text, CodeRule);
+
+  if (!result) result = runInlineRules(parseInlineMD, text, LeveledRules);
+
+  return result ?? text;
+};
+
+/*
+ ****************
+ * BLOCK PARSER *
+ ****************
+ */
+
+export type BlockMDParser = (test: string, parseInline?: (txt: string) => string) => string;
+
+export type BlockMatchConverter = (
+  match: MatchResult,
+  parseInline?: (txt: string) => string
+) => string;
+
+export type BlockMDRule = {
+  match: RuleMatch;
+  html: BlockMatchConverter;
+};
+
+export type BlockRuleRunner = (
+  parse: BlockMDParser,
+  text: string,
+  rule: BlockMDRule,
+  parseInline?: (txt: string) => string
+) => string | undefined;
+
+const HEADING_REG_1 = /^(#{1,6}) +(.+)\n?/m;
+const HeadingRule: BlockMDRule = {
+  match: (text) => text.match(HEADING_REG_1),
+  html: (match, parseInline) => {
+    const [, g1, g2] = match;
+    const level = g1.length;
+    return `<h${level} data-md="${g1}">${parseInline ? parseInline(g2) : g2}</h${level}>`;
+  },
+};
+
+const CODEBLOCK_MD_1 = '```';
+const CODEBLOCK_REG_1 = /^`{3}(\S*)\n((.+\n)+)`{3} *(?!.)\n?/m;
+const CodeBlockRule: BlockMDRule = {
+  match: (text) => text.match(CODEBLOCK_REG_1),
+  html: (match) => {
+    const [, g1, g2] = match;
+    const classNameAtt = g1 ? ` class="language-${g1}"` : '';
+    return `<pre data-md="${CODEBLOCK_MD_1}"><code${classNameAtt}>${g2}</code$></pre>`;
+  },
+};
+
+const BLOCKQUOTE_MD_1 = '>';
+const QUOTE_LINE_PREFIX = /^> */;
+const BLOCKQUOTE_TRAILING_NEWLINE = /\n$/;
+const BLOCKQUOTE_REG_1 = /(^>.*\n?)+/m;
+const BlockQuoteRule: BlockMDRule = {
+  match: (text) => text.match(BLOCKQUOTE_REG_1),
+  html: (match, parseInline) => {
+    const [blockquoteText] = match;
+
+    const lines = blockquoteText
+      .replace(BLOCKQUOTE_TRAILING_NEWLINE, '')
+      .split('\n')
+      .map((lineText) => {
+        const line = lineText.replace(QUOTE_LINE_PREFIX, '');
+        if (parseInline) return `${parseInline(line)}<br/>`;
+        return `${line}<br/>`;
+      })
+      .join('');
+    return `<blockquote data-md="${BLOCKQUOTE_MD_1}">${lines}</blockquote>`;
+  },
+};
+
+const ORDERED_LIST_MD_1 = '-';
+const O_LIST_ITEM_PREFIX = /^(-|[\da-zA-Z]\.) */;
+const O_LIST_START = /^([\d])\./;
+const O_LIST_TYPE = /^([aAiI])\./;
+const O_LIST_TRAILING_NEWLINE = /\n$/;
+const ORDERED_LIST_REG_1 = /(^(-|[\da-zA-Z]\.) +.+\n?)+/m;
+const OrderedListRule: BlockMDRule = {
+  match: (text) => text.match(ORDERED_LIST_REG_1),
+  html: (match, parseInline) => {
+    const [listText] = match;
+    const [, listStart] = listText.match(O_LIST_START) ?? [];
+    const [, listType] = listText.match(O_LIST_TYPE) ?? [];
+
+    const lines = listText
+      .replace(O_LIST_TRAILING_NEWLINE, '')
+      .split('\n')
+      .map((lineText) => {
+        const line = lineText.replace(O_LIST_ITEM_PREFIX, '');
+        const txt = parseInline ? parseInline(line) : line;
+        return `<li><p>${txt}</p></li>`;
+      })
+      .join('');
+
+    const dataMdAtt = `data-md="${listType || listStart || ORDERED_LIST_MD_1}"`;
+    const startAtt = listStart ? ` start="${listStart}"` : '';
+    const typeAtt = listType ? ` type="${listType}"` : '';
+    return `<ol ${dataMdAtt}${startAtt}${typeAtt}>${lines}</ol>`;
+  },
+};
+
+const UNORDERED_LIST_MD_1 = '*';
+const U_LIST_ITEM_PREFIX = /^\* */;
+const U_LIST_TRAILING_NEWLINE = /\n$/;
+const UNORDERED_LIST_REG_1 = /(^\* +.+\n?)+/m;
+const UnorderedListRule: BlockMDRule = {
+  match: (text) => text.match(UNORDERED_LIST_REG_1),
+  html: (match, parseInline) => {
+    const [listText] = match;
+
+    const lines = listText
+      .replace(U_LIST_TRAILING_NEWLINE, '')
+      .split('\n')
+      .map((lineText) => {
+        const line = lineText.replace(U_LIST_ITEM_PREFIX, '');
+        const txt = parseInline ? parseInline(line) : line;
+        return `<li><p>${txt}</p></li>`;
+      })
+      .join('');
+
+    return `<ul data-md="${UNORDERED_LIST_MD_1}">${lines}</ul>`;
+  },
+};
+
+const runBlockRule: BlockRuleRunner = (parse, text, rule, parseInline) => {
+  const matchResult = rule.match(text);
+  if (matchResult) {
+    const content = rule.html(matchResult, parseInline);
+    return replaceMatch((txt) => [parse(txt, parseInline)], text, matchResult, content).join('');
+  }
+  return undefined;
+};
+
+export const parseBlockMD: BlockMDParser = (text, parseInline) => {
+  if (text === '') return text;
+  let result: string | undefined;
+
+  if (!result) result = runBlockRule(parseBlockMD, text, CodeBlockRule, parseInline);
+  if (!result) result = runBlockRule(parseBlockMD, text, BlockQuoteRule, parseInline);
+  if (!result) result = runBlockRule(parseBlockMD, text, OrderedListRule, parseInline);
+  if (!result) result = runBlockRule(parseBlockMD, text, UnorderedListRule, parseInline);
+  if (!result) result = runBlockRule(parseBlockMD, text, HeadingRule, parseInline);
+
+  // replace \n with <br/> because want to preserve empty lines
+  if (!result) {
+    if (parseInline) {
+      result = text
+        .split('\n')
+        .map((lineText) => parseInline(lineText))
+        .join('<br/>');
+    } else {
+      result = text.replace(/\n/g, '<br/>');
+    }
+  }
+
+  return result ?? text;
+};

src/app/plugins/react-custom-html-parser.tsx

@@ -18,11 +18,11 @@ import { getMxIdLocalPart, getRoomWithCanonicalAlias } from '../utils/matrix';
 import { getMemberDisplayName } from '../utils/room';
 import { EMOJI_PATTERN, URL_NEG_LB } from '../utils/regex';
 import { getHexcodeForEmoji, getShortcodeFor } from './emoji';
-import { replaceMatch } from '../utils/markdown';
+import { findAndReplace } from '../utils/findAndReplace';
 
 const ReactPrism = lazy(() => import('./react-prism/ReactPrism'));
 
-const EMOJI_REG = new RegExp(`${URL_NEG_LB}(${EMOJI_PATTERN})`);
+const EMOJI_REG_G = new RegExp(`${URL_NEG_LB}(${EMOJI_PATTERN})`, 'g');
 
 export const LINKIFY_OPTS: LinkifyOpts = {
   attributes: {
@@ -35,26 +35,22 @@ export const LINKIFY_OPTS: LinkifyOpts = {
   ignoreTags: ['span'],
 };
 
-const stringToEmojifyJSX = (text: string): (string | JSX.Element)[] => {
+const textToEmojifyJSX = (text: string): (string | JSX.Element)[] =>
-  const match = text.match(EMOJI_REG);
+  findAndReplace(
-  if (!match) return [text];
-
-  const [emoji] = match;
-
-  return replaceMatch(
-    stringToEmojifyJSX,
     text,
-    match,
+    EMOJI_REG_G,
-    <span className={css.EmoticonBase}>
+    (match, pushIndex) => (
-      <span className={css.Emoticon()} title={getShortcodeFor(getHexcodeForEmoji(emoji))}>
+      <span key={pushIndex} className={css.EmoticonBase}>
-        {emoji}
+        <span className={css.Emoticon()} title={getShortcodeFor(getHexcodeForEmoji(match[0]))}>
+          {match[0]}
         </span>
       </span>
+    ),
+    (txt) => txt
   );
-};
 
 export const emojifyAndLinkify = (text: string, linkify?: boolean) => {
-  const emojifyJSX = stringToEmojifyJSX(text);
+  const emojifyJSX = textToEmojifyJSX(text);
 
   if (linkify) {
     return <Linkify options={LINKIFY_OPTS}>{emojifyJSX}</Linkify>;

src/app/utils/findAndReplace.ts

@@ -0,0 +1,28 @@
+export type ReplaceCallback<R> = (
+  match: RegExpExecArray | RegExpMatchArray,
+  pushIndex: number
+) => R;
+export type ConvertPartCallback<R> = (text: string, pushIndex: number) => R;
+
+export const findAndReplace = <ReplaceReturnType, ConvertReturnType>(
+  text: string,
+  regex: RegExp,
+  replace: ReplaceCallback<ReplaceReturnType>,
+  convertPart: ConvertPartCallback<ConvertReturnType>
+): Array<ReplaceReturnType | ConvertReturnType> => {
+  const result: Array<ReplaceReturnType | ConvertReturnType> = [];
+  let lastEnd = 0;
+
+  let match: RegExpExecArray | RegExpMatchArray | null = regex.exec(text);
+  while (match !== null && typeof match.index === 'number') {
+    result.push(convertPart(text.slice(lastEnd, match.index), result.length));
+    result.push(replace(match, result.length));
+
+    lastEnd = match.index + match[0].length;
+    if (regex.global) match = regex.exec(text);
+  }
+
+  result.push(convertPart(text.slice(lastEnd), result.length));
+
+  return result;
+};