Copies assessments from an old working set to a new working set

reptilescan.rb

require 'rubygems'
require 'find'
require 'rainbow'
require 'nokogiri'
require 'fileutils'

def lsl (dir)
  Find.find dir do |path|
    if File.directory?(path)
      next
    else
      yield path
    end
  end
end

# Not needed any more?
def node_scan (path, mapping)
  
  # Act only on nodes
  if path =~ /nodes.*\.xml/
    
    id = File.basename(path).chomp(File.extname(path))

    # read the XML
    doc = Nokogiri::XML(File.open(path))
    species = doc.css("species")
    if !species.empty? then
      specie = species.first
      name = specie[:name]
      parent = specie[:parent]
      level = specie[:level]
      footprint = specie.css("footprint").inner_text()
      mapping["#{parent} #{name}"] = id;
    end
    
  end

end

def draft_scan (path, mapping)
  
  if path =~ /drafts.*\.xml/
    
    id = File.basename(path).chomp(File.extname(path))

    doc = Nokogiri::XML(File.open(path))
    
    mapping[doc.css("speciesName").inner_text] = path;
    
  end

end

#FIXME load this from command line please
left = "madrep_july10"
right = "madrep_fixed"
#TODO extract zips

left_drafts = Hash.new
right_drafts = Hash.new

puts "Loading left (bad taxa) data set"
lsl(left) {|path| draft_scan path, left_drafts}

puts "Loading right (good taxa) data set"
lsl(right) {|path| draft_scan path, right_drafts}

count = 0

puts "Remapping drafts from left to right:"
left_drafts.each do |taxon,from_draft|
  if right_drafts.has_key?(taxon) then
    doc = Nokogiri::XML(File.open(from_draft))

    to_draft = right_drafts[taxon]
    to_draft_doc = Nokogiri::XML(File.open(to_draft))
    remap_asm = to_draft_doc.css("assessmentID").first.content
    remap_species = to_draft_doc.css("speciesID").first.content

    # copy new species and assessment IDs
    doc.css("speciesID").first.content = remap_species
    doc.css("assessmentID").first.content = remap_asm
    doc.css("assessment").first["id"] = remap_asm

    # copy region information from original
    doc.css('field[id="RegionInformation"]').first.replace(
      to_draft_doc.css('field[id="RegionInformation"]').first.to_xml
    )
    
    File.open(to_draft, 'w') {|f| f.write(doc.to_xml) }
    count = count + 1
  else
    puts "No taxon in right for #{taxon}, will not remap it".color(:red)
  end
end

puts "Remapped #{count} drafts"

useful shell commands

This shouldn't be necessary any more, because I think I taught the software how to deal with a normal zip, but here it is anyway:

find . -type f -print | zip ../madrep-remapped3 -@