generate URI Schemes RFC list

urischemes.rb

#! /usr/bin/env ruby

require 'rubygems'
require 'open-uri'
require 'nokogiri'
require 'pp'

styles = {
  "PROPOSED STANDARD" => "background-color:#6666FF;color:white;",
  "EXPERIMENTAL"      => "background-color:#EEEE00;color:white",
  "INFORMATIONAL"     => "background-color:#FFAA00;color:white",
  "DRAFT STANDARD"    => "background-color:#44DDDD;color:white",
  "HISTORIC"          => "background-color:#666666;color:white",
}

source = open("http://www.ietf.org/download/rfc-index.txt") {|io| io.read }

status = {}
source.scan(/^(\d{4}) (.+?)\n\n/m).each do |data|
  num = data[0]
  txt = data[1]
  txt.gsub!(/\s+/m, " ")

  txt =~ /\(Status: (.+?)\)/
  stat = $1
  status[num] = [txt, stat]
end


doc = Nokogiri::HTML(open("http://www.iana.org/assignments/uri-schemes.html"))
trs = doc.xpath("//tbody//tbody/tr")

schemes = []
trs.each do |tr|
  tds = tr.xpath("td")
  next unless tds[2].text =~ /\A\[RFC(\d+)\]\z/
  num    = $1
  scheme = tds[0].text
  stat   = status[num][1].strip
  style  = styles[stat]
  txt    = status[num][0]
  
  print %{- <a href="http://tools.ietf.org/html/rfc#{num}">RFC#{num}</a> <strong>#{scheme}</strong> }
  puts %{[http://tools.ietf.org/html/rfc#{num}:bookmark]}
  puts %{-- <span style="#{style}">#{stat}</span>}
  puts %{-- #{txt}}
end