Script to migrate articles from PluXML CMS to Jekyll's markdown

pluxml2md.rb

require 'nokogiri'
require 'reverse_markdown'
require 'date'

ARTICLE_PATH='/Users/yadomi/Desktop/data/articles/*.xml'

def title doc
  doc.at('title').text
end

def content doc
  chapo = ReverseMarkdown.convert doc.at('chapo').text
  content = ReverseMarkdown.convert doc.at('content').text
  "#{chapo} #{content}"
end

def tags doc
  tag = doc.at('tags').text.split(',')[0]
  if tag then tag.downcase end
end

def date doc
  DateTime.parse(File.basename(doc).split('.')[3])
end

def createFile filename, header, content
  puts filename
  buffer = File.new(filename, "w")
  buffer.puts(header)
  buffer.puts(content)
  buffer.close
end

def createHeader title, date, tags
  date = date.strftime("%Y-%m-%d %H:%M:%S")
  %{---
layout: post
title: "#{title}"
date: #{date}
categories: #{tags}
---}
end

Dir.glob(ARTICLE_PATH) do |file|
  doc = Nokogiri::XML(File.open(file))

  date = Date.parse(File.basename(file).split('.')[3][0...8])
  title = title(doc).downcase.gsub(' ', '-').gsub(/[^a-zA-Z0-9\-]/,'').squeeze('-').chomp('-')
  filename = "#{date}-#{title}.markdown"

  puts filename

  header = createHeader(title(doc), date(file), tags(doc))
  createFile(filename, header, content(doc))

end