#! /usr/bin/ruby

require 'digest/md5'
require 'net/http'
require 'uri'

$unknown_gravatar = 'd5fe5cbcc31cff5f8ac010db72eb000c'
$email_addr_regex = /[^\[{(< @"']+@[^\]})> @"']+\.[^\]})> @"']+/

def md5(str)
  Digest::MD5.hexdigest str
end

def get(url)
  Net::HTTP.get URI.parse url
end

def gravatar(email)
  email_hash = md5(email)
  "https://secure.gravatar.com/avatar/#{email_hash}?s=512" unless \
    $unknown_gravatar == md5(get('http://gravatar.com/avatar/' + email_hash))
end

$seen = {}
def find_gravatars(line)
  found = nil
  line.scan($email_addr_regex).each do |email|
    next if ($seen[email])
    $seen[email] = 1
    next unless (url = gravatar email)
    puts "#{url} #{email}"
    found = true
  end
  found
end

# if you give a bunch of non-file email addresses on the command line, try those
if ARGV.select do |maybe_not_filename|
  true if File.file? maybe_not_filename
  find_gravatars maybe_not_filename
end.length == 0 then

  # otherwise grep given files (or stdin, if none are given) for email addresses
  ARGF.each do |authors_line|
    $stderr.puts "Processing file #{ARGF.filename}" if (ARGF.file.lineno == 1)
    find_gravatars authors_line
  end
end