中文文档 | API Reference | Architecture | Experience & Lessons

rebook is a TypeScript e-book toolkit for building fast readers and book workflows across browser, Mini Program, Node.js, and worker-like runtimes.

It parses EPUB, MOBI/AZW3, FB2, and CBZ into a normalized Book contract. Browser and WeChat Mini Program renderers, exporters, search, plugins, and AI-oriented document workflows all build on that same contract.

• Fast page turns: rebook lays sections out into Pretext-backed line ranges, then renders only the visible lines. Normal page turns update a page index and a small DOM/snapshot window instead of reflowing a full chapter DOM or iframe.
• Cross-platform rendering: browser rendering uses lightweight DOM rows; WeChat Mini Program rendering emits serializable snapshots for WXML.
• Multi-format parsing: EPUB 2/3, MOBI/AZW/AZW3, FictionBook 2, and CBZ.
• Environment-agnostic parsers: parser adapters make the same parser code run in browsers, Node.js, Mini Programs, and workers.
• Modular architecture: parsers, renderers, exporters, plugins, and adapters are independent.
• AI-ready content model: sections can expose structured blocks, styled segments, searchable text, and a mutable document tree.
• Built-in workflow pieces: search, first-section export, trial-reading limits, TTS playback hooks, professional translation pipelines, and an MCP server.

npm install rebook

import { registry, createReader } from 'rebook'
import { epub } from 'rebook/parsers/epub'
import { mobi } from 'rebook/parsers/mobi'
import { fb2 } from 'rebook/parsers/fb2'
import { cbz } from 'rebook/parsers/cbz'

registry.register('epub', epub)
registry.register('mobi', mobi)
registry.register('fb2', fb2)
registry.register('cbz', cbz)

const reader = createReader({
    container: document.getElementById('viewer')!,
    layout: 'paginated',
    maxColumnCount: 2,
    styles: {
        fontSize: '18px',
        lineHeight: 1.7,
        minColumnWidth: '320px',
        maxColumnWidth: '720px',
        margin: '32px',
    },
})

const book = await reader.open(file)
await reader.next()
await reader.goTo('chapter.xhtml#section')

Use the Mini Program reader when there is no DOM. It installs Mini Program parser adapters by default and can use wx.createOffscreenCanvas for text measurement.

import { registry } from 'rebook'
import { epub } from 'rebook/parsers/epub'
import { createWechatMiniProgramReader } from 'rebook/renderers/wechat-miniprogram'

registry.register('epub', epub)

const fs = wx.getFileSystemManager()
const arrayBuffer = fs.readFileSync(filePath) as ArrayBuffer
const unitlessStyles = new Set(['fontWeight', 'opacity', 'zIndex'])
const toStyleText = (style: Record<string, string | number> = {}) =>
    Object.entries(style)
        .map(([key, value]) => {
            const cssKey = key.replace(/[A-Z]/g, match => `-${match.toLowerCase()}`)
            const cssValue = typeof value === 'number' && !unitlessStyles.has(key)
                ? `${value}px`
                : String(value)
            return `${cssKey}:${cssValue}`
        })
        .join(';')

const reader = createWechatMiniProgramReader({
    width: 375,
    height: 667,
    wx,
    layout: 'paginated',
    styles: { fontSize: '18px', lineHeight: 1.7 },
    setData: snapshot => this.setData({
        reader: {
            ...snapshot,
            lines: snapshot.lines.map(line => ({
                ...line,
                styleText: toStyleText(line.style),
                fragments: 'fragments' in line
                    ? line.fragments.map(fragment => ({
                        ...fragment,
                        styleText: toStyleText(fragment.style),
                    }))
                    : undefined,
            })),
        },
    }),
})

await reader.open(arrayBuffer)
await reader.next()

Render reader.lines from the snapshot in WXML. Each line node includes layout style and text/image/table data. Convert style objects to CSS strings before passing them to WXML. See API Reference: WeChat Mini Program Reader for a fuller integration example.

rebook-mcp exposes a local book to AI assistants through the Model Context Protocol. It supports EPUB, MOBI/AZW3, FB2, and CBZ.

{
  "mcpServers": {
    "book": {
      "command": "npx",
      "args": ["-y", "--package", "rebook", "rebook-mcp", "/absolute/path/book.epub"]
    }
  }
}

Built-in tools include chapter listing, chapter text reading, metadata lookup, and full-book or chapter-scoped search. See MCP Tools for embedding APIs.

• Trial reading: withTrialLimit({ maxPages }), trial-aware TOC items, and guarded navigation. See Plugins.
• Export: exportFirstSections() and exportBook() support EPUB, CBZ, TXT, and HTML output. See First Sections Export.
• Search: searchBook(), searchChapters(), reader.search(), and reader.searchChapters(). See Search.
• Document Model: query and mutate section trees for AI workflows, annotation, transformation, and serialization. See Document Model.
• Pretext layout: use prepareBlocks(), layout(), and getVisibleLines() directly for custom renderers. See Pretext Layout.

• API Reference - API details for readers, parsers, renderers, plugins, exporters, adapters, search, MCP, and document APIs.
• Extension Development - Manifest v1, Host API, permissions, packaging, examples, and publishing.
• Architecture - Parser/renderer separation, adapter design, rendering pipeline, and project layout.
• Experience & Lessons - Design rationale, performance notes, AI workflow ideas, and implementation lessons.

npm install
npm run dev
npm run typecheck
npm run build
npm test

MIT

Based on the excellent foliate-js by John Factotum.