writebook-to-pdf

install.sh

#!/bin/bash
set -e

curl -s https://gist.githubusercontent.com/danielwestendorf/cda938b1ffb0a203ec3a415079d56efa/raw/a99d03110521fb3c1a21e0412614831a96b024ea/writebook-to-pdf > /tmp/writebook-to-pdf
docker run -d -p 5001:5001 --name breezy-pdf-lite -e "DEBUG=breezy-pdf-lite:*" -e "PORT=5001" -e "PRIVATE_TOKEN=YOURSUPERSECRETTOKEN" danielwestendorf/breezy-pdf-lite:latest

echo "Try: ruby /tmp/writebook-to-pdf export https://books.37signals.com/2/the-writebook-manual"

writebook-to-pdf

#!/usr/bin/env ruby
#
# Convert a Writebook book to PDF
#
# Depends upon breezy-pdf-lite running via docker
# docker run -d -p 5001:5001 --name breezy-pdf-lite -e "DEBUG=breezy-pdf-lite:*" -e "PORT=5001" -e "PRIVATE_TOKEN=YOURSUPERSECRETTOKEN" danielwestendorf/breezy-pdf-lite:latest
#
# Example
# $ ruby path_to_this_file export http://yourdomain.com/the_book --debug --output /Users/yourusername/Desktop/mybook.pdf
#
# Install:
# curl -s > 
require "bundler/inline"
require "net/http"
require "erb"
require "forwardable"
require "fileutils"

gemfile do
  source "https://rubygems.org"

  gem "thor"
  gem "nokogiri"
  gem "breezy_pdf_lite"
end

BreezyPDFLite.setup do |config|
  config.secret_api_key = ENV.fetch("BREEZYPDF_LITE_SECRET_API_KEY", "YOURSUPERSECRETTOKEN")
  config.base_url = ENV.fetch("BREEZYPDF_BASE_URL", "http://localhost:5001")
end

class Book
  attr_reader :url

  def initialize(url)
    @url = url
  end

  def uri
    @uri ||= URI(url)
  end

  def html
    @html ||= Net::HTTP.get(uri)
  end

  def doc
    @doc ||= Nokogiri(html)
  end

  def style_link_tags
    @style_link_tags ||= doc.xpath("//head/link[@rel='stylesheet']").collect do |link|
      <<~HTML
        <link rel="stylesheet" href="#{URI("#{uri.origin}#{link['href']}")}">
      HTML
    end.join("\n")
  end

  def cover_img_url
    cover_img_path = doc.xpath("//main//aside//div/a").first["href"]
    URI("#{uri.origin}#{cover_img_path}").to_s
  end

  def page_uris
    @page_uris ||= doc.xpath("//main//div/menu//li//a[@class='toc__link']").collect { |a| URI("#{uri.origin}#{a['href']}") }
  end

  def toc_html
    @toc_html ||= doc.css("main menu li.toc__leaf").each.with_index do |node, i|
      node.css("a.toc__title").first["href"] = "#page-#{i + 1}"
    end.collect(&:to_s).join("\n")
  end
end

class Render
  extend Forwardable

  attr_reader :book
  def_delegators :book, :cover_img_url, :style_link_tags, :page_uris, :toc_html

  def initialize(book)
    @book = book
  end

  def to_pdf
    BreezyPDFLite::RenderRequest.new(html).to_file
  end

  def to_html
    Tempfile.new(["book", ".html"]).tap { |f| f.write(html); f.flush }
  end

  def html
    @html ||= tmpl.result(binding)
  end

  def page_htmls
    return @page_htmls if @page_htmls

    @page_htmls = {}

    page_uris.collect do |page_uri|
      Thread.new do
        page_html = Net::HTTP.get(page_uri)
        page_doc = Nokogiri(page_html)
        @page_htmls[page_uri] = page_doc.xpath("//main/*").collect(&:to_s).join
      end
    end.each(&:join)

    @page_htmls
  end

  def tmpl
    @tmpl ||= ERB.new(DATA.read)
  end
end

class Writebook < Thor
  package_name "Writebook"

  desc "export URL", "exports Writebook at given URL to a PDF"
  method_options output: :string, debug: :boolean
  def export(url)
    begin
      Net::HTTP.get(URI(BreezyPDFLite.base_url)) == "OK"
      say "BreezyPDFLite server is accessible...", :green
    rescue StandardError => e
      error "Unable to access breezy pdf lite server: #{e.inspect}\n\rMake sure it's running? See: https://github.com/danielwestendorf/breezy-pdf-lite"
      exit(1)
    end

    say "Attempting to render book..."
    render = Render.new(Book.new(url))
    render.to_html
    say "Book HTML collected..."

    if options[:debug]
      file = render.to_html
      say "Debug HTML: #{file.path}", :yellow
      yes? "Press enter to delete debug HTML and continue..."
    end

    say "Rendering as pdf..."
    pdf = render.to_pdf
    say "PDF Render complete...", :green
    if options[:output]
      FileUtils.cp pdf.path, options[:output]
      say "Saved to #{options[:output]}", :green
    else
      system "open #{pdf.path}"
      sleep 1 # wait to open before tmpfile unlink
    end

    say "Done."
  end
end

Writebook.start

__END__
<!DOCTYPE html>
<html>
<head>
  <meta name="breezy-pdf-displayBackground" content="true">
  <%= style_link_tags %>

  <style type="text/css">
    html, body { margin: 0; padding: 0; }

    #cover {
      background-image: url("<%= cover_img_url %>");
      background-size: cover;
      background-size: 100% auto;
      background-repeat: no-repeat;

      width: 100vw;
      height: 100vh;
      page-break-after: always;
    }

    #toc {
      page-break-after: always;
    }

    .page--page, .page--section {
      page-break-before: left;
    }

    p, pre { page-break-inside: avoid; }

    .heading__link { display: none; }

    .toc__container { max-inline-size: unset; }

    .toc {
      display: flex;
      flex-direction: column;
      row-gap: calc(var(--block-space)* 0.3);
    }

    .toc__thumbnail { display: none !important; }

    .toc__container .toc__leaf {
      align-items: center;
      column-gap: 1ch;
      flex-direction: row;
      justify-content: center;
      row-gap: calc(var(--block-space)* 0.3);
      text-align: start;
    }

    .toc__title {
      flex-direction: row;
      flex-grow: 1;
      font-size: inherit;
      white-space: nowrap;
    }

    .toc__title:after {
      border-block-end: 1px dotted var(--title-border-color);
      content: "";
      flex-grow: 1;
      margin-block-end: 0.25em;
    }

    @page {
      margin: 0.15in;
    }
  </style>
</head>
<body>
  <div id="cover"></div>
  <div id="toc" class="toc__container">
    <menu class="toc">
      <%= toc_html %>
    </menu>
  </div>
  <main id="main">
    <% page_uris.each.with_index do |uri, i| %>
      <div id="page-<%= i + 1 %>">
        <%= page_htmls[uri] %>
      </div>
    <% end %>
  </main>
</body>
</html>

Thanks for sharing this 🙂

thanks!

Great work! Much appreciated!