Hatena Blog から Jekyll へのコンバーター

# frozen_string_literal: true

require 'erb'
require 'reverse_markdown'
require 'time'
require 'uri'

contents_all = File.readlines('xxx.hatenablog.com.export.txt')
puts contents_all.length

chunked = contents_all.chunk_while do |i, _|
  !(i =~ /^--------\Z/)
end

puts chunked.to_a.size

class Article
  # @return [Array]
  attr_accessor :raw
  # @return [Hash]
  attr_accessor :meta
  # @return [Array]
  attr_accessor :body
  # @return [Array]
  attr_accessor :comment
  # @return [String]
  attr_accessor :body_markdown

  def initialize(arr)
    @raw = arr
    @meta = {}
    @body = []
    @comment = []

    @raw.pop # rm term

    parse!
    convert_body!
  end

  def parse!
    state = :meta
    original = @raw.dup
    while true
      line = original.shift
      break unless line

      case state
      when :meta
        if line =~ /-----/
          state = :body
          next
        end
        line =~ /([A-Z ]+): (.+)?/
        @meta[$1] ||= []
        @meta[$1] << $2
      when :body
        if line =~ /-----/
          state = :comment
          next
        end

        @body << line
      when :comment
        @comment << line
      end
    end
  end

  def convert_body!
    b = @body.dup
    b.shift # rm BODY: line
    @body_markdown = ReverseMarkdown.convert(b.join, github_flavored: true)
  end
end

articles = chunked.map do |chunk|
  Article.new(chunk)
end

class Publisher
  attr_accessor :article
  attr_accessor :title
  attr_accessor :publish_datetime
  attr_accessor :categories
  attr_accessor :body
  attr_accessor :content
  attr_accessor :filename

  # @param [Article] article
  def initialize(article)
    @article = article

    @title = article.meta['TITLE']
    @publish_datetime = article.meta['DATE']
    @categories = article.meta['CATEGORY']
    @body = article.body_markdown

    normalize!
  end

  def publish
    template = <<EOS
---
layout: post
title:  "<%= @title %>"
date:   <%= @publish_datetime.to_s %>
categories: [<%= @categories.join(', ') %>]
---

<%= @body %>
EOS

    b = binding
    @content = ERB.new(template).result(b)
    @filename = generate_filename
  end

  def normalize!
    @title = @title[0] || '無題'
    @publish_datetime = Time.strptime(@publish_datetime[0], '%m/%d/%Y %H:%M:%S')
    @categories = @categories || ['post']
  end

  def generate_filename
    ymd = @publish_datetime.strftime("%Y-%m-%d")
    title = @title.lstrip[0, 16].gsub(/[\/.:]/, '_')
    "#{ymd}-#{URI.escape(title)}.markdown"
  end
end

publishes = articles.map do |article|
  p = Publisher.new(article)
  p.publish
  p
end

publishes.each do |pub|
  puts pub.filename

  File.open('_posts/' + pub.filename, 'w') do |f|
    f.write(pub.content)
  end
end