TumblrのhtmlをVuePressでもみたい

convert.rb

require 'nokogiri'
require 'erb'
require 'time'
require 'html2md'
require 'pry-byebug'
require 'fileutils'

def make_dir_if_not_exists(path)
  FileUtils.mkdir_p(path) unless Dir.exist?(path)
end

def main
  targets = Dir.chdir('../posts/html/') { |path| Dir.glob('*.html').map { |f| path + f } }
  template = File.open('template.md.erb', &:read)
  errors = []
  targets.each do |path|
    html = File.open(path, &:read)
    converter = Converter.new(template)
    result = converter.convert(html)
    if result.nil?
      errors << path
      next
    end

    output_dir = File.join('dist', result[:date].strftime('%Y/%m'))

    filename = "#{result[:title].gsub(%r{[/-]|\s}, '_')}.md"
    filepath = File.join(output_dir, filename)

    make_dir_if_not_exists(output_dir)
    File.open(filepath, 'w') do |f|
      f.write result[:markdown]
    end
  end
  puts errors
end

# Converter
class Converter
  def initialize(template)
    @template = template
  end

  def convert(html)
    doc = Nokogiri::HTML.parse(html)
    timestamp = doc.css('#timestamp').text
    date = Time.parse(timestamp)
    title = doc.at('h1')&.text

    return nil if title.nil?

    body = doc.xpath('//body')
    body.at('h1').remove
    body = body.inner_html.gsub(/\n/) { '' }.strip

    post = {
      title: title,
      date: date,
      body: body
    }
    b = binding
    markdown = ERB.new(@template).result b
    {
      date: date,
      title: title,
      markdown: markdown
    }
  end
end

main

Gemfile

# frozen_string_literal: true

source "https://rubygems.org"

git_source(:github) {|repo_name| "https://github.com/#{repo_name}" }

# gem "rails"

gem 'nokogiri'
gem 'html2md'

group :development do
  gem 'pry-byebug'
  gem 'rubocop'
end

template.md.erb

---
title: <%= post[:title] %>
date: <%= post[:date] %>
feed:
  enable: true
layout: post
---
<%= post[:body] %>