jiang13-bbs/frontend/lib/toc.ts

import type { TocItem } from "@/components/PostToc";
import type { ContentSegment } from "@/lib/directiveBlocks";

const TIMELINE_DAY_RE = /^##\s+(\d{4}-\d{2}-\d{2})\s+(.+)$/;
const HEADING_RE = /^(#{2,3})\s+(.+?)\s*#*\s*$/;
const HIDE_OPEN_RE = /^\[hide(?:\s+.+)?\]$/i;

/** 时间线日分组锚点 id（与 MdTimeline 一致） */
export function timelineDayId(date: string): string {
  if (/^\d{4}-\d{2}-\d{2}$/.test(date)) return `tl-${date}`;
  return `tl-${date.replace(/[^a-zA-Z0-9_-]+/g, "-").replace(/^-+|-+$/g, "") || "unknown"}`;
}

/**
 * 从正文提取目录：
 * - 块外 ## / ### → h-{n}（供 headingRenderer 按序挂 id）
 * - [timeline] 内按日期去重 → tl-YYYY-MM-DD（日分组锚点）
 * - 代码围栏 / [hide] / [entries] 块内标题不计入
 */
export function extractToc(content: string): TocItem[] {
  const items: TocItem[] = [];
  let inCode = false;
  let inTimeline = false;
  let inHide = false;
  let inEntries = false;
  let headingCount = 0;
  /** 当前 timeline 块内出现过的日期（保持文件顺序，收块时再按日倒序发出） */
  let timelineDates: string[] = [];
  const timelineSeen = new Set<string>();

  const flushTimelineDates = () => {
    if (timelineDates.length === 0) return;
    const sorted = [...timelineDates].sort((a, b) => (a < b ? 1 : a > b ? -1 : 0));
    for (const date of sorted) {
      items.push({ level: 2, text: date, id: timelineDayId(date) });
    }
    timelineDates = [];
    timelineSeen.clear();
  };

  for (const raw of content.replace(/\r\n/g, "\n").replace(/\r/g, "\n").split("\n")) {
    const line = raw.trim();

    if (line.startsWith("```")) {
      inCode = !inCode;
      continue;
    }
    if (inCode) continue;

    const lower = line.toLowerCase();

    if (!inHide && !inTimeline && lower === "[timeline]") {
      inTimeline = true;
      continue;
    }
    if (inTimeline && lower === "[/timeline]") {
      inTimeline = false;
      flushTimelineDates();
      continue;
    }
    if (inTimeline) {
      const tm = TIMELINE_DAY_RE.exec(line);
      if (tm && !timelineSeen.has(tm[1])) {
        timelineSeen.add(tm[1]);
        timelineDates.push(tm[1]);
      }
      continue;
    }

    if (!inHide && !inTimeline && !inEntries && lower === "[entries]") {
      inEntries = true;
      continue;
    }
    if (inEntries && lower === "[/entries]") {
      inEntries = false;
      continue;
    }
    // 条目卡块内标题不进目录（渲染时也不挂 h-{n} 锚点，避免错位）
    if (inEntries) continue;

    if (!inHide && HIDE_OPEN_RE.test(line)) {
      inHide = true;
      continue;
    }
    if (inHide && lower === "[/hide]") {
      inHide = false;
      continue;
    }
    if (inHide) continue;

    const m = HEADING_RE.exec(line);
    if (m) {
      items.push({
        level: m[1].length as 2 | 3,
        text: m[2],
        id: `h-${headingCount++}`,
      });
    }
  }

  // 未闭合的 timeline 也尽量发出已收集日期
  flushTimelineDates();
  return items;
}

/** 仅正文 Markdown 标题（排除时间线日锚点），供 headingRenderer 按序挂 id */
export function markdownHeadingToc(toc: TocItem[]): TocItem[] {
  return toc.filter((t) => /^h-\d+$/.test(t.id));
}

/** buildTocFromSegments 的产物：目录项 + 供正文挂锚点的三组 id */
export interface SegmentsToc {
  /** 按渲染顺序的目录项（md 标题 / 条目标题 / 时间线日），id 已由 nextAnchor 分配 */
  items: TocItem[];
  /** md 段标题 id 序列（供 headingComponentsFor 按序挂到 h2/h3） */
  mdHeadIds: string[];
  /** 按段索引组织的条目标题 id（传给 MdEntries 挂到卡内标题） */
  entriesTitleIds: (string[] | undefined)[];
}

/**
 * 基于「渲染分段」构建章内目录（条目卡模式下的目录来源）：
 * - md 段标题 → h2/h3 级目录项（顺序与 headingRenderer 一致）
 * - entries 段条目标题 → 二级目录项（与卡内标题共享锚点，目录可跳转）
 * - timeline 段 → 日分组锚点（与 extractToc 同为日倒序）；hide 段内容不进目录
 * nextAnchor 与阅读页章内锚点计数器共享，保证目录、正文、卡内标题同一份 id
 */
export function buildTocFromSegments(
  segs: ContentSegment[],
  nextAnchor: () => string
): SegmentsToc {
  const items: TocItem[] = [];
  const mdHeadIds: string[] = [];
  const entriesTitleIds: (string[] | undefined)[] = [];

  for (let si = 0; si < segs.length; si++) {
    const seg = segs[si];
    if (seg.type === "md") {
      for (const raw of seg.text.replace(/\r\n/g, "\n").replace(/\r/g, "\n").split("\n")) {
        const m = HEADING_RE.exec(raw.trim());
        if (!m) continue;
        const id = nextAnchor();
        items.push({ level: m[1].length as 2 | 3, text: m[2], id });
        mdHeadIds.push(id);
      }
      continue;
    }
    if (seg.type === "entries") {
      const ids = seg.items.map(() => nextAnchor());
      seg.items.forEach((it, i) => items.push({ level: 2, text: it.title, id: ids[i] }));
      entriesTitleIds[si] = ids;
      continue;
    }
    if (seg.type === "timeline") {
      const dates = [...new Set(seg.items.map((t) => t.date))].sort((a, b) => (a < b ? 1 : a > b ? -1 : 0));
      for (const date of dates) {
        items.push({ level: 2, text: date, id: timelineDayId(date) });
      }
    }
    // hide 段内容（未解锁/未登录不可见）不进目录
  }
  return { items, mdHeadIds, entriesTitleIds };
}

/** 粗略阅读时长：约 400 字/分钟，至少 1 分钟 */
export function estimateReadingMinutes(text: string, fallbackLen = 0): number {
  const len = text.trim().length > 0 ? text.trim().length : fallbackLen;
  return Math.max(1, Math.ceil(len / 400));
}