Generate Stripe Blog Archive in markdown

stripe-blog-archive

#!/usr/bin/env ruby
# frozen_string_literal: true

require "nokogiri"
require "http"

class Article
  def initialize(article)
    @article = article
  end

  def to_s
    if category.empty?
      %(- #{title_link} by #{author} #{date})
    else
      %(- #{title_link} by #{author} in #{category} #{date})
    end
  end

  private
  attr_reader :article, :title_link, :author, :date, :category

  def title_link
    post = article.css("header h1 a").first
    title_link = %([#{post.text}](https://stripe.com#{post.attr("href")}))
  end

  def author
    authors = article.css("a.author")
    if authors.empty?
      article.css("span.author").text
    else
      authors.map { |a| "[#{a.text}](#{a[:href]})" }.join(" with ")
    end
  end

  def date
    article.css("header span:last-child").text.lstrip
  end

  def category
    @_category ||= article.css("header p a:last-child").text
  end
end

class Page
  def initialize(page)
    @page = page
    @document = Nokogiri::HTML.parse(HTTP.follow.get(page).to_s)
  end

  def to_s
    [heading, posts].join("\n\n")
  end

  def heading
    "### [Page #{number}](#{page})"
  end

  def posts
    articles_by_human.map do |article|
      Article.new(article).to_s
    end.join("\n")
  end

  private
  attr_reader :page, :document

  def number
    page.end_with?("/blog") ? "1" : File.basename(page)
  end

  def articles_by_human
    document.
      css("article:not(.changelog-set)").
      css("article:not(.link)").
      css("article:not(.globalFooterNav)")
  end
end

module StripeBlog
  URL = "https://stripe.com/blog"

  def self.pages
    [*2..last_page].map do |page_number|
      File.join(URL, "page/#{page_number}")
    end.prepend(URL)
  end

  def self.last_page
    [*26..100].each do |page_number|
      response_uri = HTTP.follow.get("#{URL}/page/#{page_number}").uri.to_s
      if response_uri.end_with?("/blog")
        break(page_number-1)
      end
    end
  end
end

markdown = StripeBlog.pages.map do |page|
  puts "Parsing #{page}"
  Page.new(page).to_s
end

puts "Generated in tmp/stripe-blog.md."
IO.write("tmp/stripe-blog.md", markdown.join("\n\n"))