require 'date'

# Variables
wp_prefix = 'wp_' # WP Table Prefix
pid = 10          # Post ID to start at
cid = 10          # Comment ID to start at

class Post
  attr_accessor :title
  attr_accessor :body
  attr_accessor :date
  attr_accessor :author
  attr_accessor :comments

  def initialize
    @comments = Array.new
  end
end

# Recursive function to grab all entries
def process_dir posts, dirname, root = "."
  entries = Dir.entries(root + "/" + dirname)
  entries.each do |fn|
    if (fn != "." && fn != "..") then 
      fullname = root + "/" + dirname + "/" + fn
      if (File.directory?(fullname)) then
        process_dir(posts, fn, root + "/" + dirname) if !(fn.match("entry"))
        
      else
        post = parse_post(fullname, fn)

        # Get comments
        #puts fullname[0..-5]
        commentdir = fullname[0..-5] + "/comments/"
        if File.directory?(commentdir) then
          Dir.entries(commentdir).each do |cn|
            if cn != "." && cn != ".." && cn.match("comment") then  
              comment = (parse_post(commentdir + "/" + cn, cn, true))
              post.comments.push(comment)
            end
          end
        end
        posts.push(post)
      end
    end
  end
end

# Parses an entry*.txt file or a comment*.txt file
def parse_post(fullname, fn, iscomment = false)
  buffer = ""
  File.open(fullname).each do |line|
    buffer = buffer + line + "\n"
  end
  #puts fullname + "\n"
  tokens = buffer.split("|"); #TODO cope with escaped |

  post = Post.new
  if (tokens[0] == "VERSION") then
    # New version post
    i = 0
    while (i < tokens.size) do
      case tokens[i]
        when 'SUBJECT'
          post.title = tokens[i+1]
        when 'CONTENT'
          post.body = tokens[i+1]
        when 'NAME'
          post.author = tokens[i+1]
      end
      i = i + 2
    end
  else
    # Old post
    post.title = tokens[0] if !iscomment;
    post.author = tokens[0] if iscomment;
    post.body = tokens[2];
  end

  # Not sure the best way to parse dates...
  offset = fn.match("entry") ? 5 : 7
  year =   fn[offset+0,2]
  month =  fn[offset+2,2]
  day =    fn[offset+4,2]
  hour =   fn[offset+7,2]
  minute = fn[offset+9,2]
  second = fn[offset+11,2]

  post.date = Date.strptime(year + " " + month + " " + day + " " + hour + " " + minute + " " + second, "%y %m %d %H %M %S");
  return post;
end

# Create object tree
posts = Array.new
process_dir posts, "content", "."

# Process BBCode
def bb2html str
  str = str.gsub(/\[([a-zA-Z]+)\](.*?)\[\/\1\]/m, '<\1>\2</\1>')
  str = str.gsub(/\[([a-zA-Z]+)\](.*?)\[\/\1\]/m, '<\1>\2</\1>')
  str = str.gsub(/\[url=([^\]]*)\](.*?)\[\/url\]/m, '<a href="\1">\2</a>')
  str = str.gsub(/\[img=([^\]]*)\]/m, '<img src="\1" />')
end

# Create a url friendly title
def titleize str
  str = str.gsub(/[^A-Za-z\s]/, '')
  str = str.downcase
  str = str.gsub(/\s/, '-')
end

# Output SQL
posts.each do |p|
sql = "INSERT INTO #{wp_prefix}posts (id, post_author, post_date, post_date_gmt, "
  sql = sql + "post_content, post_title, post_name, comment_count) "
  sql = sql + " VALUES (#{pid}, 1, '#{p.date}', '#{p.date}', "
  sql = sql + "'#{bb2html(p.body).gsub(/'/,'\'')}', "
  sql = sql + "'#{bb2html(p.title).gsub(/'/,'\'')}', "
  sql = sql + "'#{titleize(p.title).gsub(/'/,'\'')}', "
  sql = sql + "#{p.comments.size});\n"
  puts sql

  p.comments.each do |c|
    sql = "INSERT INTO #{wp_prefix}comments (comment_id, comment_post_id, comment_content, "
    sql = sql + "comment_author, comment_date, comment_date_gmt) VALUES ("
    sql = sql + "#{cid}, #{pid}, '#{bb2html(c.body).gsub(/'/,'\'')}', "
    sql = sql + "'#{c.author.gsub(/'/,'\'')}', '#{c.date}', "
    sql = sql + "'#{p.date}');\n"
    puts sql
    cid = cid + 1
  end
  pid = pid + 1
end