mechanize_doi_lookup.rb

require "mechanize"

url = "http://www.crossref.org/guestquery/"
bibfile = ARGV[0]
outfile = File.open(ARGV[1], 'w') || STDOUT

puts "Searching for #{bibfile}, writing to #{outfile}"

agent = Mechanize.new
page = agent.get url
form = page.form_with name: "form2"
author_field = form.field_with name: "auth2"
title_field = form.field_with name: "atitle2"

title = author = nil
contents = File.read(bibfile).each_line.to_a

titler = /^\s*title\s*=\s*\{(.*)\},?\s*$/
authorr = /^\s*author\s*=\s*\{(.*)\},?\s*$/
namer = /((:?De )?(:?D')?\w+(?:-\w+)?)/
author1r = /^\s*#{namer},.*$/
author2r = /.*?#{namer} and.*/

myidx = 0
contents.dup.each_with_index do |line, idx|
  line = line.gsub(/\{\"([aou])\}/, "\\1")
  if line =~ titler
    title = $1
  elsif line =~ authorr
    author = $1
    if author =~ author1r
      author = $1
    elsif author =~ author2r
      author = $1
    end
  end

  if title && author
    print "#{author}: #{title}: "
    author_field.value = author
    title_field.value = title
    newpage = form.submit
    links = newpage.links_with(href: /dx.doi.org/)
    links.each do |link|
      doi = link.href.sub('http://dx.doi.org/', '')
      print "  #{doi}  "
      myidx += 1
      contents.insert(idx + myidx, "doi = {#{doi}},\n")
    end
    puts
    title = nil
    author = nil
  end
end

outfile << contents.join