require 'nokogiri'

module NCI
  def self.get_pathways(xml, format = "UP")
    doc = Nokogiri::XML(xml)
    pathways = {}

    molecules = {}
    doc.xpath("//Molecule").each do |molecule|
      id = molecule.attribute('id').value
      type = molecule.attribute('molecule_type').value
      next unless type == "protein"
      names = molecule.xpath("Name[@name_type='#{format}']").collect{|name| name.attribute("value").value}
      next if names.empty?
      molecules[id] = {:xml => molecule, :uniprot => names.first}
    end

    interactions = {}
    doc.xpath("//Interaction").each do |interaction|
      id = interaction.attribute('id').value
      molecule_ids = interaction.xpath('*/InteractionComponent').collect{|c| c.attribute('molecule_idref').value}

      interactions[id] = {:xml => interaction, :molecule_ids => molecule_ids}
    end

    doc.xpath("//Pathway").each do |pathway|
      id = pathway.attribute('id').value
      subnet = pathway.attribute('subnet').value
      name = pathway.xpath('LongName').first.content

      interaction_ids = pathway.xpath("*/PathwayComponent").collect{|component| component.attribute("interaction_idref").value}

      pathway_interactions = interaction_ids.collect{|id| interactions[id]}
      pathway_molecule_ids = pathway_interactions.collect{|info| info[:molecule_ids]}.flatten

      pathway_uniprot_ids = pathway_molecule_ids.collect do |id| 
        next unless molecules.include? id
        molecules[id][:uniprot]
      end
      pathways[id] = [[name], [pathway_uniprot_ids.compact.uniq]]
    end
    pathways
  end
end

file 'nature_pathways' do |t|

  url = "ftp://ftp1.nci.nih.gov/pub/PID/XML/NCI-Nature_Curated.xml.gz"

  xml = Open.read(url)

  pathways = NCI.get_pathways(xml)

  Open.write(t.name, TSV.setup(pathways, :type => :double, :key_field => "NCI Nature Pathway ID", :fields => ["Pathway Name","UniProt/SwissProt Accession"]).to_s)
end

file 'biocarta_pathways' do |t|

  url = "ftp://ftp1.nci.nih.gov/pub/PID/XML/BioCarta.xml.gz"

  xml = Open.read(url)

  pathways = NCI.get_pathways(xml, "LL")

  Open.write(t.name, TSV.setup(pathways, :type => :double, :key_field => "NCI BioCarta Pathway ID", :fields => ["Pathway Name","Entrez Gene ID"]).to_s)
end

file 'reactome_pathways' do |t|

  url = "ftp://ftp1.nci.nih.gov/pub/PID/XML/Reactome.xml.gz"

  xml = Open.read(url)

  pathways = NCI.get_pathways(xml, "UP")

  Open.write(t.name, TSV.setup(pathways, :type => :double, :key_field => "NCI Reactome Pathway ID", :fields => ["Pathway Name","UniProt/SwissProt Accession"]).to_s)
end