@armadra/agent 0.0.0-stage → 0.3.0

package/dist/tui/components/loader.js

@@ -0,0 +1,71 @@
+/**
+ * 旋转指示器（设计 §12.1、§12.6：运行中显示 Loader 与已用时）。[B4]
+ *
+ * 帧由定时器推进并调用 `requestRender`；`stop()` 后不再触发渲染。时钟可注入便于测试。
+ */
+import { truncateToWidth } from "../ansi.js";
+export const LOADER_FRAMES = ["⠋", "⠙", "⠹", "⠸", "⠼", "⠴", "⠦", "⠧", "⠇", "⠏"];
+export function formatElapsed(ms) {
+    const total = Math.max(0, Math.floor(ms / 1000));
+    if (total < 60)
+        return `${total}s`;
+    const m = Math.floor(total / 60);
+    const s = total % 60;
+    return `${m}m${String(s).padStart(2, "0")}s`;
+}
+export class Loader {
+    requestRender;
+    options;
+    frame = 0;
+    timer = null;
+    startedAt;
+    message;
+    now;
+    constructor(requestRender, options = {}) {
+        this.requestRender = requestRender;
+        this.options = options;
+        this.now = options.now ?? Date.now;
+        this.message = options.message ?? "";
+        this.startedAt = this.now();
+    }
+    get running() {
+        return this.timer !== null;
+    }
+    setMessage(message) {
+        this.message = message;
+        this.requestRender();
+    }
+    start() {
+        if (this.timer !== null)
+            return;
+        this.startedAt = this.now();
+        this.frame = 0;
+        this.timer = setInterval(() => this.tick(), this.options.intervalMs ?? 80);
+        this.timer.unref?.();
+        this.requestRender();
+    }
+    stop() {
+        if (this.timer === null)
+            return;
+        clearInterval(this.timer);
+        this.timer = null;
+    }
+    /** 推进一帧（定时器调用；测试可直接调用）。 */
+    tick() {
+        this.frame = (this.frame + 1) % LOADER_FRAMES.length;
+        this.requestRender();
+    }
+    render(width) {
+        const theme = this.options.theme;
+        const spinner = LOADER_FRAMES[this.frame];
+        const parts = [theme ? theme.fg("accent", spinner) : spinner];
+        if (this.message !== "")
+            parts.push(this.message);
+        if (this.options.showElapsed !== false) {
+            const elapsed = `(${formatElapsed(this.now() - this.startedAt)})`;
+            parts.push(theme ? theme.fg("dim", elapsed) : elapsed);
+        }
+        return [truncateToWidth(parts.join(" "), width)];
+    }
+    invalidate() { }
+}

package/dist/tui/components/markdown.d.ts

@@ -0,0 +1,90 @@
+/**
+ * Markdown → 行（设计 §12.5）。[B4]
+ *
+ * 支持：ATX 标题、段落（软换行合并为空格，行尾两个空格或 `\` 为硬换行）、无序 / 有序列表（缩进嵌套）、
+ * 围栏代码块（边框 + 语言标签；流式时未闭合的围栏按代码块渲染）、引用、分隔线、
+ * 行内 `code` / **粗体** / *斜体* / ~~删除线~~ / [链接](url) / <自动链接> / 反斜杠转义。
+ * 表格降级为等宽对齐文本（不换行、超宽截断）；不做语法高亮（§12.9）。
+ *
+ * 缓存：按块缓存渲染结果（键 = 宽度 + 块源文本），流式追加时只有末块重新渲染；解析也是增量的
+ * （`Markdown.append` 只从尾部可能变化的块重新解析）。
+ */
+import type { Component, Theme } from "../component.js";
+export type MarkdownBlock = {
+    kind: "heading";
+    level: number;
+    text: string;
+    source: string;
+} | {
+    kind: "paragraph";
+    text: string;
+    source: string;
+} | {
+    kind: "code";
+    lang: string;
+    lines: string[];
+    source: string;
+} | {
+    kind: "list";
+    items: ListItem[];
+    source: string;
+} | {
+    kind: "quote";
+    text: string;
+    source: string;
+} | {
+    kind: "table";
+    rows: string[][];
+    source: string;
+} | {
+    kind: "hr";
+    source: string;
+};
+export interface ListItem {
+    depth: number;
+    marker: string;
+    text: string;
+}
+/** 块占的行区间 [start, end)（行号，相对整篇）。 */
+export interface BlockSpan {
+    start: number;
+    end: number;
+}
+/** 把源文本切成块。 */
+export declare function parseMarkdown(source: string): MarkdownBlock[];
+/** 行内标记 → 带样式文本。 */
+export declare function renderInline(text: string, theme: Theme): string;
+/** 单个块 → 行（不含块间空行）。 */
+export declare function renderBlock(block: MarkdownBlock, width: number, theme: Theme): string[];
+export interface MarkdownOptions {
+    theme?: Theme;
+    /** 左右内边距。 */
+    paddingX?: number;
+}
+export declare class Markdown implements Component {
+    private text;
+    private readonly options;
+    private blocks;
+    /** 与 blocks 一一对应的行区间（增量解析用）。 */
+    private spans;
+    /** 规范化（\r\n → \n）后的行。 */
+    private lines;
+    private cacheWidth;
+    private readonly blockCache;
+    private lastLines;
+    private readonly theme;
+    /** 累计解析过的行数（诊断与测试：流式追加应只重解析尾部）。 */
+    parsedLines: number;
+    constructor(text?: string, options?: MarkdownOptions);
+    getText(): string;
+    setText(text: string): void;
+    /**
+     * 流式追加：只有最后一行会变长，此前的行不变。一个块的结束取决于它之后的一行（列表再多看一行），
+     * 所以「结束行 + 1」早于旧的最后一行的块保持不变，从第一个不满足的块的起始行重新解析
+     * （未闭合的围栏就是从围栏起点）。含 `\r` 时回落全量（`\r\n` 可能被拆在两次追加之间）。
+     */
+    append(delta: string): void;
+    private reparseFrom;
+    render(width: number): string[];
+    invalidate(): void;
+}

package/dist/tui/components/markdown.js

@@ -0,0 +1,462 @@
+/**
+ * Markdown → 行（设计 §12.5）。[B4]
+ *
+ * 支持：ATX 标题、段落（软换行合并为空格，行尾两个空格或 `\` 为硬换行）、无序 / 有序列表（缩进嵌套）、
+ * 围栏代码块（边框 + 语言标签；流式时未闭合的围栏按代码块渲染）、引用、分隔线、
+ * 行内 `code` / **粗体** / *斜体* / ~~删除线~~ / [链接](url) / <自动链接> / 反斜杠转义。
+ * 表格降级为等宽对齐文本（不换行、超宽截断）；不做语法高亮（§12.9）。
+ *
+ * 缓存：按块缓存渲染结果（键 = 宽度 + 块源文本），流式追加时只有末块重新渲染；解析也是增量的
+ * （`Markdown.append` 只从尾部可能变化的块重新解析）。
+ */
+import { padToWidth, truncateToWidth, visibleWidth, wrapTextWithAnsi } from "../ansi.js";
+import { plainTheme } from "../theme.js";
+const HEADING_RE = /^ {0,3}(#{1,6})(?:\s+(.*?))?\s*#*\s*$/;
+const FENCE_RE = /^ {0,3}(`{3,}|~{3,})\s*([^\s`]*)/;
+const HR_RE = /^ {0,3}([-*_])(?:\s*\1){2,}\s*$/;
+const LIST_RE = /^(\s*)([-*+]|\d{1,9}[.)])\s+(.*)$/;
+const QUOTE_RE = /^ {0,3}>\s?(.*)$/;
+const TABLE_RE = /^\s*\|.*\|\s*$/;
+const TABLE_SEP_RE = /^\s*\|?\s*:?-{2,}:?\s*(\|\s*:?-{2,}:?\s*)*\|?\s*$/;
+function isBlockStart(line) {
+    return (HEADING_RE.test(line) ||
+        FENCE_RE.test(line) ||
+        HR_RE.test(line) ||
+        LIST_RE.test(line) ||
+        QUOTE_RE.test(line) ||
+        TABLE_RE.test(line));
+}
+function splitLines(source) {
+    return source.replace(/\r\n?/g, "\n").split("\n");
+}
+/** 把源文本切成块。 */
+export function parseMarkdown(source) {
+    return parseLines(splitLines(source), 0, [], []);
+}
+/**
+ * 从第 `from` 行起解析，块与行区间分别追加到 `blocks` / `spans`。块与块之间不带状态（空行跳过），
+ * 所以从任一块的起始行重新解析，结果与全量解析的对应部分相同（增量解析的前提）。
+ */
+function parseLines(lines, from, blocks, spans) {
+    let i = from;
+    while (i < lines.length) {
+        const line = lines[i];
+        if (line.trim() === "") {
+            i++;
+            continue;
+        }
+        const start = i;
+        const before = blocks.length;
+        i = parseBlock(lines, i, blocks);
+        for (let k = before; k < blocks.length; k++)
+            spans.push({ start, end: i });
+    }
+    return blocks;
+}
+/** 解析从 `i` 起（非空行）的一个块，返回下一行号。 */
+function parseBlock(lines, i, blocks) {
+    const line = lines[i];
+    const start = i;
+    const fence = FENCE_RE.exec(line);
+    if (fence) {
+        const marker = fence[1];
+        const body = [];
+        i++;
+        while (i < lines.length && !lines[i].trimStart().startsWith(marker))
+            body.push(lines[i++]);
+        if (i < lines.length)
+            i++;
+        const src = lines.slice(start, i).join("\n");
+        blocks.push({ kind: "code", lang: fence[2] ?? "", lines: body, source: src });
+        return i;
+    }
+    const heading = HEADING_RE.exec(line);
+    if (heading) {
+        blocks.push({
+            kind: "heading",
+            level: heading[1].length,
+            text: heading[2] ?? "",
+            source: line,
+        });
+        i++;
+        return i;
+    }
+    if (HR_RE.test(line)) {
+        blocks.push({ kind: "hr", source: line });
+        i++;
+        return i;
+    }
+    if (TABLE_RE.test(line)) {
+        const rows = [];
+        while (i < lines.length && TABLE_RE.test(lines[i])) {
+            const row = lines[i++];
+            if (TABLE_SEP_RE.test(row))
+                continue;
+            rows.push(splitTableRow(row));
+        }
+        blocks.push({ kind: "table", rows, source: lines.slice(start, i).join("\n") });
+        return i;
+    }
+    if (QUOTE_RE.test(line)) {
+        const body = [];
+        while (i < lines.length && QUOTE_RE.test(lines[i]))
+            body.push(QUOTE_RE.exec(lines[i++])[1]);
+        blocks.push({
+            kind: "quote",
+            text: body.join("\n"),
+            source: lines.slice(start, i).join("\n"),
+        });
+        return i;
+    }
+    if (LIST_RE.test(line)) {
+        return parseList(lines, i, blocks);
+    }
+    const body = [line];
+    i++;
+    while (i < lines.length && lines[i].trim() !== "" && !isBlockStart(lines[i])) {
+        body.push(lines[i++]);
+    }
+    blocks.push({ kind: "paragraph", text: joinSoftBreaks(body), source: body.join("\n") });
+    return i;
+}
+function parseList(lines, start, blocks) {
+    const items = [];
+    const indents = [];
+    let i = start;
+    while (i < lines.length) {
+        const line = lines[i];
+        const m = LIST_RE.exec(line);
+        if (m) {
+            const indent = m[1].replace(/\t/g, "    ").length;
+            while (indents.length > 0 && indent < indents[indents.length - 1])
+                indents.pop();
+            if (indents.length === 0 || indent > indents[indents.length - 1])
+                indents.push(indent);
+            items.push({ depth: indents.length - 1, marker: m[2], text: m[3] });
+            i++;
+            continue;
+        }
+        // 续行：缩进的非空行并入上一项；空行后接列表项则继续，否则结束
+        if (line.trim() !== "" && /^\s+/.test(line) && items.length > 0) {
+            const last = items[items.length - 1];
+            last.text = joinSoftBreaks([last.text, line.trim()]);
+            i++;
+            continue;
+        }
+        if (line.trim() === "" && i + 1 < lines.length && LIST_RE.test(lines[i + 1])) {
+            i++;
+            continue;
+        }
+        break;
+    }
+    blocks.push({ kind: "list", items, source: lines.slice(start, i).join("\n") });
+    return i;
+}
+function joinSoftBreaks(lines) {
+    let out = "";
+    lines.forEach((line, index) => {
+        const text = index === lines.length - 1 ? line.trim() : line.trimStart();
+        if (index === 0) {
+            out = text;
+            return;
+        }
+        const hard = / {2,}$/.test(out) || out.endsWith("\\");
+        out = out.replace(/(?: {2,}|\\)$/, "").trimEnd() + (hard ? "\n" : " ") + text;
+    });
+    return out.trimEnd();
+}
+function splitTableRow(row) {
+    const trimmed = row.trim().replace(/^\|/, "").replace(/\|$/, "");
+    return trimmed.split(/(?<!\\)\|/).map((cell) => cell.trim().replace(/\\\|/g, "|"));
+}
+/** 行内标记 → 带样式文本。 */
+export function renderInline(text, theme) {
+    let out = "";
+    let i = 0;
+    while (i < text.length) {
+        const ch = text[i];
+        if (ch === "\\" && i + 1 < text.length && /[\\`*_{}[\]()#+\-.!~|<>]/.test(text[i + 1])) {
+            out += text[i + 1];
+            i += 2;
+            continue;
+        }
+        if (ch === "`") {
+            const ticks = /^`+/.exec(text.slice(i))[0];
+            const end = text.indexOf(ticks, i + ticks.length);
+            if (end !== -1) {
+                out += theme.fg("code", text.slice(i + ticks.length, end).trim() || " ");
+                i = end + ticks.length;
+                continue;
+            }
+        }
+        const emphasis = matchDelimited(text, i, ["**", "__", "~~", "*", "_"]);
+        if (emphasis) {
+            const inner = renderInline(emphasis.inner, theme);
+            if (emphasis.delim === "**" || emphasis.delim === "__")
+                out += theme.bold(inner);
+            else if (emphasis.delim === "~~")
+                out += `\x1b[9m${inner}\x1b[29m`;
+            else
+                out += theme.italic(inner);
+            i = emphasis.end;
+            continue;
+        }
+        if (ch === "[") {
+            const link = /^\[([^\]]*)\]\(([^)\s]*)(?:\s+"[^"]*")?\)/.exec(text.slice(i));
+            if (link) {
+                const label = renderInline(link[1], theme);
+                const url = link[2];
+                out +=
+                    link[1] === url || url === ""
+                        ? theme.underline(label || url)
+                        : theme.underline(label) + theme.fg("dim", ` (${url})`);
+                i += link[0].length;
+                continue;
+            }
+        }
+        if (ch === "<") {
+            const auto = /^<((?:https?|mailto):[^>\s]+)>/.exec(text.slice(i));
+            if (auto) {
+                out += theme.underline(auto[1]);
+                i += auto[0].length;
+                continue;
+            }
+        }
+        out += ch;
+        i++;
+    }
+    return out;
+}
+function matchDelimited(text, i, delims) {
+    for (const delim of delims) {
+        if (!text.startsWith(delim, i))
+            continue;
+        const after = text[i + delim.length];
+        if (after === undefined || /\s/.test(after))
+            continue;
+        // 下划线只在词边界起作用（snake_case 不当斜体）
+        if (delim.startsWith("_") && i > 0 && /[\p{L}\p{N}]/u.test(text[i - 1]))
+            continue;
+        let search = i + delim.length + 1;
+        while (search <= text.length) {
+            const close = text.indexOf(delim, search);
+            if (close === -1)
+                break;
+            const before = text[close - 1];
+            const next = text[close + delim.length];
+            const wordAfter = next !== undefined && /[\p{L}\p{N}]/u.test(next);
+            if (!/\s/.test(before) && !(delim.startsWith("_") && wordAfter)) {
+                // `*` 不吃 `**` 的一半
+                if (delim.length === 1 && text[close + 1] === delim) {
+                    search = close + 2;
+                    continue;
+                }
+                return { delim, inner: text.slice(i + delim.length, close), end: close + delim.length };
+            }
+            search = close + 1;
+        }
+    }
+    return null;
+}
+/** 单个块 → 行（不含块间空行）。 */
+export function renderBlock(block, width, theme) {
+    switch (block.kind) {
+        case "heading": {
+            const text = renderInline(block.text, theme);
+            const styled = block.level <= 2 ? theme.fg("accent", theme.bold(text)) : theme.bold(text);
+            return wrapTextWithAnsi(styled, width);
+        }
+        case "paragraph":
+            return wrapTextWithAnsi(renderInline(block.text, theme), width);
+        case "hr":
+            return [theme.fg("border", "─".repeat(Math.max(1, width)))];
+        case "quote": {
+            const bar = theme.fg("border", "│ ");
+            const inner = wrapTextWithAnsi(theme.italic(renderInline(block.text, theme)), width - 2);
+            return inner.map((line) => bar + line);
+        }
+        case "list":
+            return renderList(block.items, width, theme);
+        case "code":
+            return renderCode(block.lang, block.lines, width, theme);
+        case "table":
+            return renderTable(block.rows, width, theme);
+    }
+}
+const BULLETS = ["•", "◦", "▪"];
+function renderList(items, width, theme) {
+    const lines = [];
+    for (const item of items) {
+        const indent = "  ".repeat(item.depth);
+        const ordered = /\d/.test(item.marker);
+        const bullet = ordered ? item.marker : BULLETS[item.depth % BULLETS.length];
+        const head = indent + theme.fg("accent", bullet) + " ";
+        const headWidth = visibleWidth(head);
+        const body = wrapTextWithAnsi(renderInline(item.text, theme), Math.max(1, width - headWidth));
+        body.forEach((line, index) => {
+            lines.push((index === 0 ? head : " ".repeat(headWidth)) + line);
+        });
+    }
+    return lines;
+}
+function renderCode(lang, body, width, theme) {
+    const code = body.map((line) => line.replace(/\t/g, "    "));
+    if (width < 8)
+        return code.map((line) => truncateToWidth(line, width));
+    const inner = width - 4;
+    const border = (s) => theme.fg("border", s);
+    const label = lang === "" ? "" : truncateToWidth(` ${lang} `, inner - 1);
+    const top = border("╭─") +
+        theme.fg("dim", label) +
+        border("─".repeat(width - 3 - visibleWidth(label)) + "╮");
+    const lines = [top];
+    for (const line of code) {
+        for (const piece of wrapCode(line, inner)) {
+            lines.push(border("│ ") + padToWidth(theme.fg("code", piece), inner) + border(" │"));
+        }
+    }
+    lines.push(border("╰" + "─".repeat(width - 2) + "╯"));
+    return lines;
+}
+/** 代码按列硬断行（不按词），空行保留。 */
+function wrapCode(line, width) {
+    if (visibleWidth(line) <= width)
+        return [line];
+    const out = [];
+    let current = "";
+    let w = 0;
+    for (const ch of line) {
+        const cw = visibleWidth(ch);
+        if (w + cw > width) {
+            out.push(current);
+            current = "";
+            w = 0;
+        }
+        current += ch;
+        w += cw;
+    }
+    out.push(current);
+    return out;
+}
+function renderTable(rows, width, theme) {
+    const cols = Math.max(0, ...rows.map((r) => r.length));
+    const widths = new Array(cols).fill(0);
+    const rendered = rows.map((row) => row.map((cell) => renderInline(cell, theme)));
+    for (const row of rendered) {
+        row.forEach((cell, c) => (widths[c] = Math.max(widths[c], visibleWidth(cell))));
+    }
+    const sep = theme.fg("border", " │ ");
+    return rendered.map((row, r) => {
+        const cells = widths.map((w, c) => padToWidth(row[c] ?? "", w));
+        const line = cells.join(sep).replace(/\s+$/, "");
+        return truncateToWidth(r === 0 && rows.length > 1 ? theme.bold(line) : line, width);
+    });
+}
+export class Markdown {
+    text;
+    options;
+    blocks = [];
+    /** 与 blocks 一一对应的行区间（增量解析用）。 */
+    spans = [];
+    /** 规范化（\r\n → \n）后的行。 */
+    lines = [];
+    cacheWidth = -1;
+    blockCache = new Map();
+    lastLines = null;
+    theme;
+    /** 累计解析过的行数（诊断与测试：流式追加应只重解析尾部）。 */
+    parsedLines = 0;
+    constructor(text = "", options = {}) {
+        this.text = text;
+        this.options = options;
+        this.theme = options.theme ?? plainTheme();
+        this.reparseFrom(0, splitLines(text));
+    }
+    getText() {
+        return this.text;
+    }
+    setText(text) {
+        if (text === this.text)
+            return;
+        if (text.startsWith(this.text) && this.text !== "") {
+            this.append(text.slice(this.text.length));
+            return;
+        }
+        this.text = text;
+        this.reparseFrom(0, splitLines(text));
+    }
+    /**
+     * 流式追加：只有最后一行会变长，此前的行不变。一个块的结束取决于它之后的一行（列表再多看一行），
+     * 所以「结束行 + 1」早于旧的最后一行的块保持不变，从第一个不满足的块的起始行重新解析
+     * （未闭合的围栏就是从围栏起点）。含 `\r` 时回落全量（`\r\n` 可能被拆在两次追加之间）。
+     */
+    append(delta) {
+        if (delta === "")
+            return;
+        const previous = this.text;
+        this.text = previous + delta;
+        if (delta.includes("\r") || previous.endsWith("\r")) {
+            this.reparseFrom(0, splitLines(this.text));
+            return;
+        }
+        const lines = this.lines;
+        const oldLast = lines.length - 1;
+        const pieces = delta.split("\n");
+        lines[oldLast] = (lines[oldLast] ?? "") + pieces[0];
+        for (let k = 1; k < pieces.length; k++)
+            lines.push(pieces[k]);
+        let keep = this.spans.findIndex((span) => span.end + 1 >= oldLast);
+        if (keep === -1)
+            keep = this.spans.length;
+        const from = keep < this.spans.length ? this.spans[keep].start : (this.spans.at(-1)?.end ?? 0);
+        this.blocks.length = keep;
+        this.spans.length = keep;
+        this.reparseFrom(from, lines);
+    }
+    reparseFrom(from, lines) {
+        if (from === 0) {
+            this.blocks = [];
+            this.spans = [];
+        }
+        this.lines = lines;
+        parseLines(lines, from, this.blocks, this.spans);
+        this.parsedLines += lines.length - from;
+        this.lastLines = null;
+    }
+    render(width) {
+        if (width !== this.cacheWidth) {
+            this.blockCache.clear();
+            this.cacheWidth = width;
+            this.lastLines = null;
+        }
+        if (this.lastLines)
+            return this.lastLines;
+        const padX = this.options.paddingX ?? 0;
+        const inner = Math.max(1, width - padX * 2);
+        const pad = " ".repeat(padX);
+        const lines = [];
+        const used = new Set();
+        this.blocks.forEach((block, index) => {
+            const key = `${block.kind}\u0000${block.source}`;
+            used.add(key);
+            let rendered = this.blockCache.get(key);
+            if (!rendered) {
+                rendered = renderBlock(block, inner, this.theme).map((line) => pad + line);
+                this.blockCache.set(key, rendered);
+            }
+            if (index > 0)
+                lines.push("");
+            for (const line of rendered)
+                lines.push(line);
+        });
+        for (const key of [...this.blockCache.keys()])
+            if (!used.has(key))
+                this.blockCache.delete(key);
+        this.lastLines = lines;
+        return lines;
+    }
+    invalidate() {
+        this.blockCache.clear();
+        this.lastLines = null;
+    }
+}

package/dist/tui/components/meter.d.ts

@@ -0,0 +1,29 @@
+/**
+ * 单行余量表（第三波 A8）：`ctx ▮▮▮▮▮▮▮▯▯▯ 72%`。[W3-B9a-2]
+ *
+ * - `value` 是 0–1 的比例（越界夹紧）；undefined 显示 `—`，格子全空；
+ * - 阈值着色（需要 theme）：≥ `dangerAt`（缺省 0.9）error，≥ `warnAt`（缺省 0.7）warning，否则 success；
+ *   空格子 dim；
+ * - 宽度不够时先去掉格子，只剩「标签 百分比」，再不够就截断。
+ */
+import type { Component, SemanticColor, Theme } from "../component.js";
+export interface MeterOptions {
+    label?: string;
+    /** 格子数，缺省 10。 */
+    cells?: number;
+    warnAt?: number;
+    dangerAt?: number;
+    theme?: Theme;
+}
+export declare const METER_FULL = "\u25AE";
+export declare const METER_EMPTY = "\u25AF";
+export declare class Meter implements Component {
+    private value;
+    private readonly options;
+    constructor(value?: number | undefined, options?: MeterOptions);
+    setValue(value: number | undefined): void;
+    /** 当前阈值档位对应的语义色。 */
+    level(): SemanticColor;
+    render(width: number): string[];
+    invalidate(): void;
+}