require 'nokogiri'
require 'rbbt-util'

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

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


    interactions = {}
    doc.xpath("//Interaction").each do |interaction|
      id = interaction.attribute('id').value
      type = interaction.attribute('interaction_type').value
      molecule_ids = interaction.xpath('InteractionComponentList/InteractionComponent').collect{|c| c.attribute('molecule_idref').value}.flatten.compact
      pathway_ids = interaction.xpath('Abstraction').collect{|c| c.attribute('pathway_idref').value}.flatten.compact

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

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

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

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

      pathway_uniprot_ids = pathway_molecule_ids.collect do |i| 
        next unless molecules.include? i
        molecules[i][:proteins]
      end
      if get_short_name
        pathways[id] = [[name], [pathway_uniprot_ids.flatten.compact.uniq], [short_name]]
      else
        pathways[id] = [[name], [pathway_uniprot_ids.flatten.compact.uniq]]
      end
    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("PathwayComponentList/PathwayComponent").collect{|component| component.attribute("interaction_idref").value}

      pathway_interactions = interaction_ids.collect{|i| interactions[i]}
      pathway_subnet_ids = pathway_interactions.collect{|info| info[:pathway_ids]}.compact.flatten

      pathway_subnet_ids.collect do |nid| 
        next unless pathways.include? nid
        new_genes = pathways[id].last
        pathways[nid][1] = (pathways[nid][1] + new_genes).uniq
      end
    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", true)

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

file 'biocarta_pathways_fixed_ids' => 'biocarta_pathways' do |t|
  orig = TSV.open(Open.open(t.prerequisites.first))
  tsv = TSV.setup({}, :type => :double, :key_field => "BioCarta Pathway ID", :fields => ["Pathway Name", "Entrez Gene ID"])

  orig.through do |key, values|
    name, genes, short = values
    code = "h_" + short.first
    tsv[code] = [name, genes]
  end

  Open.write(t.name, tsv.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