web-ansol.org/import.rb

#!/usr/bin/env ruby

require 'yaml'
require 'fileutils'
require 'set'

Post = Struct.new(:filename, :frontmatter, :body) do
  def to_s
    "#{YAML.dump(frontmatter)}---\n#{body}"
  end

  def self.load(filename)
    parts = File.read(filename).split("---\n")
    frontmatter = YAML.load(parts[1])

    self.new(filename, frontmatter, parts[2])
  end

  def dump_hugo
    File.write("content/post/#{File.basename(filename)}", to_s)
  end
end

posts = Dir["imported-content/_posts/*.md"].map do |filename|
  Post.load(filename)
end

# Idempotency check
# just checking if the ruby serialization isn't changing files unnecessarily
posts.each do |post|
  if File.read(post.filename) != post.to_s
    raise "not idempotent!"
  end
end

# Handle attachments

FileUtils.mkdir_p("static/attachments")
files_to_copy = Set.new
posts.select { |p| p.frontmatter.dig("metadata", "anexos")&.any? }.each do |post|
  anexos = post.frontmatter.dig("metadata", "anexos")

  anexos.filter! { |anexo| anexo["anexos_display"] == 1 }

  if anexos.empty?
    post.frontmatter["metadata"].delete("anexos")
  end

  anexos.each do |anexo|
    files_to_copy.add(anexo["anexos_uri"].gsub("public://", ""))

    anexo["anexos_uri"].gsub!("public://", "/attachments/")
  end

  post.dump_hugo
end


# Handle links
posts.each do |post|
  post.body.gsub!(/href="(.*?)"/) do |match|
    url = $1.sub(/^http:\/\/ansol\.org/, "https://ansol.org")

    url =
      if url.start_with?("https://ansol.org/sites/ansol.org/files/")
        filename = url.gsub("https://ansol.org/sites/ansol.org/files/", "")
        files_to_copy.add(filename.gsub("%20", " "))
        "/attachments/#{filename}"
      elsif url.start_with?("https://ansol.org")
        pp url

        url
      else
        url
      end

    "href=\"#{url}\""
  end

  post.dump_hugo
end

files_to_copy.each do |file|
  FileUtils.mkdir_p("static/attachments/#{File.dirname(file)}")
  FileUtils.copy_file("imported-files/#{file}", "static/attachments/#{file}")
end