feedBlogger.rb

#!/usr/bin/env ruby

#-------------------------------------------------------------------------------
# A script to load up a blogger blog with entries from a feed on disk.
# Basically allows you to move a feed to Blogger, if you have the complete
# feed as an existing atom feed.  Doesn't handle comments or anything, just
# the feed data.
#-------------------------------------------------------------------------------

require 'cgi'
require 'net/http'
require 'net/https'
require 'rexml/document'

#-------------------------------------------------------------------------------
# POST something to a URL
#-------------------------------------------------------------------------------
def httpPostForm(url, formData) 
  elementArray = []
  formData.each_pair { |k,v| elementArray << "#{CGI.escape(k)}=#{CGI.escape(v)}"}
  formData = elementArray.join "&"
  
  url  = URI.parse(url)
  http = Net::HTTP.new(url.host, url.port)
  
  if (url.scheme == "https")
    http.use_ssl = true
    http.verify_mode = OpenSSL::SSL::VERIFY_NONE
  end
  
  return http.post(url.path, formData)
end

#-------------------------------------------------------------------------------
# Get Google ClientAuth key
#-------------------------------------------------------------------------------
def getGoogleClientAuth(email, password)
  formData = {
    "Email" => email,
    "Passwd" => password,
    "source" => "MuellerWare-feedBlogger-0",
    "service" => "blogger"
  }
  
  response = httpPostForm("https://www.google.com/accounts/ClientLogin", formData)
  unless ("200" == response.code) 
    puts "Error getting profile: #{response.code} #{response.message}"
    exit
  end

  re = /.\nAuth=(.*)(\n|$)/m
  match = re.match(response.body)

  return nil unless match

  return match[1]
end

#-------------------------------------------------------------------------------
# Get the URL to post blog entries to 
#-------------------------------------------------------------------------------
def getPostingURL(blogURL)
  html = Net::HTTP.get(URI.parse(blogURL))
  pattern = /<link.*?rel="service.post".*?href="(.*?)"/m
  match = pattern.match(html)
  return nil unless match
  return match[1]
end

#-------------------------------------------------------------------------------
# Post a blog entry
#-------------------------------------------------------------------------------
def postBlogEntry(postingURL, googleAuth, entry)
  headers = {
    "Authorization" => "GoogleLogin auth=#{googleAuth}",
    "Content-Type"  => "application/atom+xml"
  }
  
  url  = URI.parse(postingURL)
  http = Net::HTTP.new(url.host, url.port)
  
  if (url.scheme == "https")
    http.use_ssl = true
    http.verify_mode = OpenSSL::SSL::VERIFY_NONE
  end
  
  response = http.post(url.path, entry, headers)
  
  return ["200","201"].include?(response.code)
end

##-------------------------------------------------------------------------------
# main program
#-------------------------------------------------------------------------------

# check parameters
if (ARGV.length < 4)
  puts "Expecting [blogURL] [email] [password] [feedFile]"
  puts
  puts "Adds the entries in [feedFile] to the blog at [blogURL] with the specified"
  puts "credentials.  For blogger.com"
  exit
end

# set parameters
blogURL  = ARGV[0]
email    = ARGV[1]
password = ARGV[2]
feedFile = ARGV[3]

# sanity check parameters
unless File.exist?(feedFile)
  puts "File not found: #{feedFile}"
  exit
end

# get the posting url of the blog
postingURL = getPostingURL(blogURL)

# read the feed file
xml = REXML::Document.new(File.open(feedFile))

# authenticate
auth = getGoogleClientAuth(email, password)

# process the entries
xml.elements.each("//entry") do | entry |
  title = entry.get_text("title")
  puts "Processing: #{title}"
  
  entry.add_namespace("http://www.w3.org/2005/Atom")
  result = postBlogEntry(postingURL, auth, entry.to_s)
  unless result 
    puts "Error posting blog entry"
  end
end