focus/src/core/epub.ts

export interface EpubChapterItem {
  id: string;
  title: string;
  charOffset: number;
}

export interface ParsedEpub {
  title: string;
  author: string;
  fullText: string;
  rawChapters: EpubChapterItem[];
}

/**
 * Client-side EPUB parser with dynamic JSZip import to reduce initial bundle size.
 */
export async function parseEpubFile(file: File | Blob): Promise<ParsedEpub> {
  // Dynamic import JSZip to split ~100KB off the initial load bundle
  const { default: JSZip } = await import('jszip');

  const arrayBuffer = await file.arrayBuffer();
  const zip = await JSZip.loadAsync(arrayBuffer);

  // 1. Locate container.xml
  const containerXmlFile = zip.file('META-INF/container.xml');
  if (!containerXmlFile) {
    throw new Error('无效的 EPUB 格式：缺少 META-INF/container.xml');
  }

  const containerXml = await containerXmlFile.async('text');
  const domParser = new DOMParser();
  const containerDoc = domParser.parseFromString(containerXml, 'application/xml');
  const rootfileEl = containerDoc.querySelector('rootfile');
  const opfPath = rootfileEl?.getAttribute('full-path');

  if (!opfPath) {
    throw new Error('无法在 container.xml 中找到 OPF 元数据路径');
  }

  // 2. Read OPF file
  const opfFile = zip.file(opfPath);
  if (!opfFile) {
    throw new Error(`无法找到 OPF 文件：${opfPath}`);
  }

  const opfDir = opfPath.includes('/') ? opfPath.substring(0, opfPath.lastIndexOf('/') + 1) : '';
  const opfXml = await opfFile.async('text');
  const opfDoc = domParser.parseFromString(opfXml, 'application/xml');

  // Metadata
  const title =
    opfDoc.querySelector('title, dc\\:title')?.textContent?.trim() ||
    (file instanceof File ? file.name.replace(/\.epub$/i, '') : '未命名书籍');
  const author =
    opfDoc.querySelector('creator, dc\\:creator')?.textContent?.trim() || '未知作者';

  // Manifest: id -> href
  const manifestMap = new Map<string, string>();
  opfDoc.querySelectorAll('manifest > item').forEach((item) => {
    const id = item.getAttribute('id');
    const href = item.getAttribute('href');
    if (id && href) {
      manifestMap.set(id, decodeURIComponent(href));
    }
  });

  // Spine: ordered list of itemrefs
  const spineItemIds: string[] = [];
  opfDoc.querySelectorAll('spine > itemref').forEach((itemref) => {
    const idref = itemref.getAttribute('idref');
    if (idref) {
      spineItemIds.push(idref);
    }
  });

  const rawChapters: EpubChapterItem[] = [];
  let fullText = '';
  let chapterIndex = 1;

  for (const idref of spineItemIds) {
    const relativeHref = manifestMap.get(idref);
    if (!relativeHref) continue;

    const filePath = opfDir ? resolvePath(opfDir, relativeHref) : relativeHref;
    const chapterFile = zip.file(filePath);
    if (!chapterFile) continue;

    const contentHtml = await chapterFile.async('text');
    const chapterDoc = domParser.parseFromString(contentHtml, 'text/html');

    // Extract title from headings or title tag
    const heading = chapterDoc.querySelector('h1, h2, h3, title');
    let chapterTitle = heading?.textContent?.trim() || '';
    if (!chapterTitle || chapterTitle.length > 50) {
      chapterTitle = `第 ${chapterIndex} 节`;
    }

    // Extract text content cleanly
    const body = chapterDoc.body || chapterDoc.documentElement;
    body.querySelectorAll('script, style, svg').forEach((el) => el.remove());

    const paragraphs: string[] = [];
    const elements = body.querySelectorAll('p, h1, h2, h3, h4, h5, h6, li, blockquote');

    if (elements.length > 0) {
      elements.forEach((el) => {
        const text = el.textContent?.trim();
        if (text) paragraphs.push(text);
      });
    } else {
      const text = body.textContent?.trim();
      if (text) paragraphs.push(text);
    }

    const chapterText = paragraphs.join('\n\n');
    if (chapterText.trim().length > 0) {
      const startCharOffset = fullText.length;
      rawChapters.push({
        id: `chap-${chapterIndex}`,
        title: chapterTitle,
        charOffset: startCharOffset,
      });

      fullText += (fullText ? '\n\n' : '') + chapterText;
      chapterIndex++;
    }
  }

  if (!fullText.trim()) {
    throw new Error('未能从 EPUB 提取到可读文本');
  }

  return {
    title,
    author,
    fullText,
    rawChapters,
  };
}

/**
 * Resolves a relative path against a base directory.
 */
function resolvePath(baseDir: string, relativePath: string): string {
  const parts = (baseDir + relativePath).split('/');
  const resolved: string[] = [];

  for (const part of parts) {
    if (part === '.' || part === '') continue;
    if (part === '..') {
      resolved.pop();
    } else {
      resolved.push(part);
    }
  }

  return resolved.join('/');
}