ncbi_biosample_attributes_extend2ttl.rb

#!/usr/bin/env ruby

require 'nokogiri'
#require 'erb'
require 'pp'
require 'json'
#require 'thor'

# 1. wget https://www.ncbi.nlm.nih.gov/biosample/docs/attributes/?format=xml -O ncbi_biosample_attributes.xml
# 2. %ruby ncbi_biosample_attributes_extend2ttl.rb ncbi_biosample_attributes.xml  > biosample_attributes_extend.ttl 

class BioSampleAttributes 
  include Enumerable

  def initialize(xml)
    @xml =xml
  end

  def each
    @doc = []
    IO.foreach(@xml) do |line|
      next if line =~/\<\?xml|BioSampleAttributes/
      @doc.push('<?xml version="1.0" encoding="UTF-8"?>') if line =~/^\s*<Attribute/
      @doc.push(line.chomp)
      if line =~/\<\/Attribute\>/
        docs = @doc.join("\n").to_s
        yield Attribute.new(docs)
        @doc = []
      end
    end
  end

  def to_ttl
    puts "@base <http://ddbj.nig.ac.jp/ontologies/biosample> .
@prefix : <http://ddbj.nig.ac.jp/ontologies/biosample/> .
@prefix skos: <http://www.w3.org/2004/02/skos/core#> .

"    
    self.each_with_index do |attr,i|
      attr.to_ttl
    end
  end

  def format
    self.each_with_index do |attr,i|
      puts attr.harmonizedName + "\t" +  attr.format
    end
  end

  def to_s
    self.each_with_index do |attr,i|
      puts attr.to_s
    end
  end

  def to_json
    self.each_with_index do |attr,i|
      puts attr.to_json
    end
  end
end

class Attribute
  def initialize(xml)
    @attr = Nokogiri::XML(xml).css("Attribute")
    raise NameError, "attribute element not found" unless @attr
    doc = Nokogiri::XML(xml)
    package = doc.xpath("/Attribute")
  end

  def name
      @attr.css('Name').inner_text
  end

  def harmonizedName
      @attr.css('HarmonizedName').inner_text
  end

  def format
      @attr.css('Format').inner_text
  end

  def preferred_format 
      case self.harmonizedName
      when "air_temp_regm", "annual_season_temp", "host_body_temp", "ph",
           "samp_store_temp", "temp", "typ_occupant_dens", "water_temp_regm"
          "numeric"
      when "birth_date","collection_date","death_date"
          "date"
      else
          "text"
      end
  end

  def synonym
      #@attr.css('Synonym').inner_text
      @attr.css('Synonym').to_a.join("; ")
  end

  def description
      @attr.css('Description').inner_text
  end

  def to_ttl
      class_name = self.harmonizedName.capitalize + "_Attribute"
      print ":" + class_name + "\tskos:altLabel\t\"" + self.name + "\"" 
      if @attr.css('Synonym').size > 0 
        puts ";"
        puts @attr.css('Synonym').map{ |s|
          "    skos:hiddenLabel \"#{s.inner_text}\"" 
        }.join("; \n") +"."
      else
        puts "."
      end
      puts
      puts ":" + class_name + "\t:preferred_format\t\"" + preferred_format + "\"." 
      puts
  end

  def to_s
      #[self.name,  self.harmonizedName,  self.synonym,  self.format,  self.description].join("\t")
      [self.name,  @attr.css('Synonym').to_a.join("; ")].join("\t")
  end

  def to_json
      {
          'name': self.name,
          'harmonizedName': self.harmonizedName,
          'synonym': self.synonym,
          'format': self.format,
          'description': self.description
      }.to_json
  end
end

xml =  ARGV[0] || 'ncbi_biosample_attributes.xml'
attrs = BioSampleAttributes.new(xml)
attrs.to_ttl