INSDC structured comment parser

parse_st.rb

#!/usr/bin/env ruby

require 'rubygems'
require 'bio'

def parse_st_comment(comment)
  comment.scan(/##(.+)-START##\n(.*)\n##\1-END##/m).each_with_object({}) {|(tagset, block), memo|
    i = 0

    memo[tagset] = block.lines.chunk {|line|
      line.include?('::') ? i += 1 : i
    }.map {|_, lines|
      lines.map(&:strip).join(' ').split(/\s*::\s*/)
    }.to_h
  }
end

io = ARGF
Bio::FlatFile.auto(io).each do |entry|
  #puts entry.comment
  st = parse_st_comment(entry.comment)
  st.each do |k, v|
    v.each do |vk, vv|
      puts ["#{entry.entry_id}.#{entry.version}", k, vk, vv].join("\t")
    end
  end
  #@entry = entry
  #@features = entry.features
  #@source = @features.shift
  #parse_sequence
  #parse_source
  #parse_genes
  #parse_features
end