pkgsrc_license_audit.rb

#!/usr/bin/ruby
#
# Quick and dirty process license information from pkg_src summary files

# Package categories to skip
SKIP_CATEGORIES = [
  "games",
  "emulators",
  "japanese",
  "chinese",
  "chat",
  "cad",
  "biology"
]

def audited_category(categories)
  categories.split(" ").each do |c|
    if SKIP_CATEGORIES.include?(c)
      return false
    end
  end
  return true
end

class PackageInfo
  attr_accessor :data

  # Type of value if not present we assume
  # it;s a single value type ie it does not appear
  # multiple times
  KEY_TYPES = {
    "REQUIRES" => :MULTI_VALUE,
    "DESCRIPTION" => :MULTI_VALUE,
  }

  def initialize()
    @data = {}
  end

  def self.package_from_summary(data)
    pkg = PackageInfo.new

    data.each_line do |l|
      if l =~ /=/
        key, value = l.split('=')
        pkg.set_property(key, value)
      end
    end

    return pkg
  end

  def set_property(key, value)
    case KEY_TYPES[key]
    when :MULTI_VALUE
      set_multivalue_property(key, value)
    else
      set_singlevalue_property(key, value)
    end
  end

  def set_multivalue_property(key, value)
    if @data.key? key
      @data[key] << value.strip!
    else
      @data[key] = [value.strip!]
    end
  end

  def set_singlevalue_property(key, value)
    @data[key] = value.strip!
  end
end

in_pkg = false
pkg_data_buffer = ""
packages = []
File.open("pkg_summary").each_line do |l|
  if l.length > 1
    in_pkg = true
    pkg_data_buffer << l
  else
    in_pkg = false
    packages << PackageInfo.package_from_summary(pkg_data_buffer)
    pkg_data_buffer = ""
  end
end

licenses = {}
packages.each do |p|
  license_data = p.data["LICENSE"].split(" ")
  license_data.each do |l|
    next if l == "AND" || l == "OR"
    licenses[l] = 1
  end
end

puts "There are #{licenses.keys.length} unique licenses"
licenses.keys.each do |l|
  puts l
end

unlicensed = []
packages.each do |p|
  if audited_category(p.data["CATEGORIES"])
    if p.data["LICENSE"].length > 1
      puts "#{p.data["PKGNAME"]} #{p.data["LICENSE"]}"
    else
      unlicensed << p
    end
  end
end

puts "Unlincensed packages or which there are #{unlicensed.length}"
unlicensed.each do |p|
  puts "#{p.data["PKGNAME"]} #{p.data["HOMEPAGE"]}"
end