Files
Agent-System/web/src/engine/renderMarkdown.ts
T
weichenandCursor 63b0f72081 feat(web): load the formula engine only when math appears
The login entry no longer ships KaTeX. Chat bubbles and document text render formulas after a dynamic import, and rich-text copy does the same.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-19 10:57:49 +00:00

267 lines
10 KiB
TypeScript

export interface RenderMarkdownOptions {
preserveNewlines?: boolean;
}
type MathRenderer = (source: string, displayMode: boolean) => string;
let mathRenderer: MathRenderer | null = null;
let mathRendererLoading: Promise<void> | null = null;
export function contentHasMath(content: string): boolean {
return /\$\$[\s\S]*?\$\$|\$[^$\n]+?\$/.test(content);
}
/** 首次遇到公式才加载 KaTeX 及其样式。没有公式的会话不请求这个包。 */
export function ensureMathRenderer(): Promise<void> {
if (mathRenderer) return Promise.resolve();
if (!mathRendererLoading) {
mathRendererLoading = (async () => {
if (typeof document !== 'undefined') {
await import('katex/dist/katex.min.css');
}
const mod = await import('./katexRender');
mathRenderer = mod.renderKatexHtml;
})();
}
return mathRendererLoading;
}
function renderMathSource(math: string, displayMode: boolean): string {
const trimmed = math.trim();
if (mathRenderer) {
try {
return mathRenderer(trimmed, displayMode);
} catch {
// 落到转义文本,避免把未渲染的源码再送进 Markdown 规则。
}
}
const escaped = escapeHtml(trimmed);
if (displayMode) {
return `<span class="math-pending">$$${escaped}$$</span>`;
}
return `<span class="math-pending">$${escaped}$</span>`;
}
/**
* 健壮的行内 Markdown、LaTeX 公式与有序/无序列表渲染引擎
* 支持:
* - 块级数学公式 ($$...$$) 与行内数学公式 ($...$)
* - 粗体 (**text** / __text__)
* - 斜体 (*text* / _text_)
* - 粗斜体 (***text***)
* - 删除线 (~~text~~)
* - 下划线 (<u>text</u>)
* - 行内代码 (`code`)
* - 超链接 ([text](url))
* - 有序列表 (1. item, 2. item ...)
* - 无序列表 (- item, * item ...)
* - 荧光高亮 (<mark> / 预置语法)
* - 安全内联样式标签 (<span style="...">)
*/
export function renderInlineMarkdown(content: string, options: RenderMarkdownOptions = {}): string {
if (!content) return '';
let text = content;
// 1. 占位保护预处理:优先解析并提取数学公式,避免公式内部的 * 或 _ 被 Markdown 解析器误伤
// 注意:占位符严禁使用包含 _ 或 * 等 Markdown 特殊字符,采用纯大写字母标识符
const mathPlaceholders: string[] = [];
// 每次渲染换盐,避免用户文本伪造 XMATHTOKENX0XENDX 在 escapeHtml 之后被还原成 HTML
const tokenSalt = Math.random().toString(36).slice(2, 10).toUpperCase();
const placeholderPrefix = `XMATHTOKENX${tokenSalt}X`;
// 1.1 块级独立公式 $$ ... $$
text = text.replace(/\$\$([\s\S]*?)\$\$/g, (_, math) => {
const idx = mathPlaceholders.length;
mathPlaceholders.push(renderMathSource(math, true));
return `${placeholderPrefix}${idx}XENDX`;
});
// 1.2 行内公式 $ ... $
text = text.replace(/\$([^\$\n]+?)\$/g, (_, math) => {
const idx = mathPlaceholders.length;
mathPlaceholders.push(renderMathSource(math, false));
return `${placeholderPrefix}${idx}XENDX`;
});
// 2. 行内代码保护 `code`
const codePlaceholders: string[] = [];
const codePlaceholderPrefix = `XCODETOKENX${tokenSalt}X`;
text = text.replace(/`([^`\n]+)`/g, (_, code) => {
const rendered = `<code style="font-family:var(--font-mono);font-size:12px;background:var(--bg-subtle);color:var(--brand-500);padding:2px 5px;border-radius:4px;border:1px solid var(--border)">${escapeHtml(code)}</code>`;
const idx = codePlaceholders.length;
codePlaceholders.push(rendered);
return `${codePlaceholderPrefix}${idx}XENDX`;
});
// 2.5 转义剩余文本中的 HTML,防止用户/LLM 输入经 dangerouslySetInnerHTML 执行脚本
text = escapeHtml(text);
// 3. 粗斜体 ***text***
text = text.replace(/\*\*\*([^\*\n]+?)\*\*\*/g, '<strong><em>$1</em></strong>');
// 4. 粗体 **text** 或 __text__
text = text.replace(/\*\*([^\*\n]+?)\*\*/g, '<strong>$1</strong>');
text = text.replace(/__([^_\n]+?)__/g, '<strong>$1</strong>');
// 5. 斜体 *text* 或 _text_
text = text.replace(/\*([^\*\n]+?)\*/g, '<em>$1</em>');
text = text.replace(/_([^_\n]+?)_/g, '<em>$1</em>');
// 6. 删除线 ~~text~~
text = text.replace(/~~([^~\n]+?)~~/g, '<del style="color:var(--text-muted)">$1</del>');
// 7. 链接 [text](url)
text = text.replace(
/\[([^\]\n]+)\]\(([^)\s]+)\)/g,
(_, label, url) =>
`<a href="${sanitizeHref(url)}" target="_blank" rel="noopener noreferrer" style="color:var(--brand-500);text-decoration:underline">${label}</a>`
);
// 8. 还原行内代码占位符
codePlaceholders.forEach((codeHtml, idx) => {
text = text.replace(`${codePlaceholderPrefix}${idx}XENDX`, () => codeHtml);
});
// 9. 还原数学公式占位符
mathPlaceholders.forEach((mathHtml, idx) => {
text = text.replace(`${placeholderPrefix}${idx}XENDX`, () => mathHtml);
});
// 10. 处理标题 / 分隔线 / 表格 / 引用 / 有序无序列表与换行
if (options.preserveNewlines !== false) {
const rawLines = text.split('\n');
const resultParts: string[] = [];
let i = 0;
const isBlockHtml = (s: string) =>
/<(ol|ul|h[1-6]|table|hr|blockquote)\b/i.test(s) || s.startsWith('<ol') || s.startsWith('<ul');
while (i < rawLines.length) {
const line = rawLines[i];
const headingMatch = line.match(/^(#{1,6})\s+(.*)$/);
const hrMatch = line.match(/^\s*(?:-{3,}|\*{3,}|_{3,})\s*$/);
const quoteMatch = line.match(/^>\s?(.*)$/);
const tableMatch = line.match(/^\s*\|.+\|\s*$/);
const orderedMatch = line.match(/^(\s*)(\d+)\.\s+(.*)$/);
const unorderedMatch = line.match(/^(\s*)[-*+]\s+(.*)$/);
if (headingMatch) {
const level = Math.min(headingMatch[1].length, 6);
const sizes = ['1.35em', '1.2em', '1.1em', '1.05em', '1em', '0.95em'];
resultParts.push(
`<h${level} class="chat-md-h" style="margin:10px 0 6px;font-size:${sizes[level - 1]};font-weight:700;line-height:1.35;color:inherit;">${headingMatch[2]}</h${level}>`
);
i++;
} else if (hrMatch) {
resultParts.push(
`<hr class="chat-md-hr" style="border:none;border-top:1px solid var(--border);margin:10px 0;opacity:0.85;" />`
);
i++;
} else if (tableMatch) {
const tableLines: string[] = [];
while (i < rawLines.length && /^\s*\|.+\|\s*$/.test(rawLines[i])) {
tableLines.push(rawLines[i].trim());
i++;
}
const parsedRows: string[][] = [];
for (const tl of tableLines) {
if (/^\|?\s*[:\-| ]+\s*\|?$/.test(tl)) continue;
const cells = tl
.replace(/^\|/, '')
.replace(/\|$/, '')
.split('|')
.map((c) => c.trim());
if (cells.length > 0) parsedRows.push(cells);
}
if (parsedRows.length > 0) {
const head = parsedRows[0];
const body = parsedRows.slice(1);
const thead = `<thead><tr>${head.map((c) => `<th style="border:1px solid var(--border);padding:6px 8px;background:var(--bg-subtle);text-align:left;font-weight:600;">${c}</th>`).join('')}</tr></thead>`;
const tbody =
body.length > 0
? `<tbody>${body.map((r) => `<tr>${r.map((c) => `<td style="border:1px solid var(--border);padding:6px 8px;vertical-align:top;">${c}</td>`).join('')}</tr>`).join('')}</tbody>`
: '';
resultParts.push(
`<div style="overflow-x:auto;margin:8px 0;"><table class="chat-md-table" style="border-collapse:collapse;width:100%;font-size:12.5px;line-height:1.45;">${thead}${tbody}</table></div>`
);
}
} else if (quoteMatch) {
const quotes: string[] = [];
while (i < rawLines.length) {
const qm = rawLines[i].match(/^>\s?(.*)$/);
if (!qm) break;
quotes.push(qm[1] || '');
i++;
}
resultParts.push(
`<blockquote class="chat-md-quote" style="margin:8px 0;padding:6px 10px;border-left:3px solid var(--brand-400);background:var(--bg-subtle);border-radius:0 6px 6px 0;">${quotes.join('<br/>')}</blockquote>`
);
} else if (orderedMatch) {
const listItems: string[] = [];
const startNum = parseInt(orderedMatch[2], 10) || 1;
while (i < rawLines.length) {
const curMatch = rawLines[i].match(/^\s*\d+\.\s+(.*)$/);
if (!curMatch) break;
listItems.push(curMatch[1]);
i++;
}
resultParts.push(
`<ol class="doc-ol" start="${startNum}" style="margin: 6px 0; padding-left: 24px; line-height: 1.65; list-style-type: decimal;">` +
listItems.map((li) => `<li style="margin-bottom: 4px;">${li}</li>`).join('') +
`</ol>`
);
} else if (unorderedMatch) {
const listItems: string[] = [];
while (i < rawLines.length) {
const curMatch = rawLines[i].match(/^\s*[-*+]\s+(.*)$/);
if (!curMatch) break;
listItems.push(curMatch[1]);
i++;
}
resultParts.push(
`<ul class="doc-ul" style="margin: 6px 0; padding-left: 20px; line-height: 1.65; list-style-type: disc;">` +
listItems.map((li) => `<li style="margin-bottom: 4px;">${li}</li>`).join('') +
`</ul>`
);
} else {
resultParts.push(line);
i++;
}
}
let outHtml = '';
for (let p = 0; p < resultParts.length; p++) {
const cur = resultParts[p];
if (p > 0) {
const prev = resultParts[p - 1];
if (!isBlockHtml(prev) && !isBlockHtml(cur)) {
outHtml += '<br/>';
}
}
outHtml += cur;
}
text = outHtml;
}
return text;
}
function escapeHtml(str: string): string {
return str
.replace(/&/g, '&amp;')
.replace(/</g, '&lt;')
.replace(/>/g, '&gt;')
.replace(/"/g, '&quot;')
.replace(/'/g, '&#39;');
}
/** Block executable URL schemes in markdown link hrefs. */
function sanitizeHref(url: string): string {
const trimmed = url.trim();
if (/^(javascript|data|vbscript):/i.test(trimmed)) {
return '#';
}
return escapeHtml(trimmed);
}