#!/usr/local/bin/ruby

# mailfeed.rb
# reads a pop3 account and generates an RSS 2.0 feed from its messages.
#
# any feedback appreciated.
#
# by Martin Dittus, martin@dekstop.de, 2005-11-24
# last change: 2005-12-09

# =========
# = ABOUT =
# =========
#
# - the script currently downloads and displays _all_ mails in your mail account every time
#   you request the generated feed. there is no caching of content. you need to manually 
#   delete old mails to speed up the script (it gets especially slow when you receive large 
#   attachments, images etc.)
#
# - multipart messages (e.g., HTML mails and mails with attachments) are handled as follows:
#   - all parts with content-type=text/plain get chained together sequentially as one message
#   - all other content types (text/html, image/jpeg, ...) are ignored
#
# - this doesn't yet support ssl access (ssl support will only be introduced to 
#   net/pop with ruby 1.9), which means GMail is not supported as mail storage.
#

# ===========
# = INSTALL =
# =========== 
#
# - make sure your webserver has the RMail gem
# - edit settings in the "prefs" section below. at minimum you need to edit the @mailprefs
#   hash to reflect your email account (server, username, password). the @feedprefs hash
#   contains strings that are used to create the actual RSS feed; adjust to your liking.
# - copy the file to your webserver, make it executable (some webservers require a .cgi
#   filename suffix)
# - take proper security precautions if you don't want the world to see this feed -- e.g.,
#   instruct your webserver to require HTTP authentication to access this feed.
# - done.
#

# ========
# = TODO =
# ========
#
# - include support for "smart" history inclusion (but make sure to show all new mail 
#   since the last access -- i.e., include support for conditional HTTP GET/HEAD requests)
# - allow HTML mails as an option (but take proper precautions)
# - don't download binary attachments of multipart mails
# - properly decode mail header strings (this is missing in RMail) -- there currently 
#   is a naive workaround for the "Subject:" header using rfc2047.rb, but that's not 
#   very elegant (and hence commented out for now)
# - use a mailto:<sender> - link as item URL
# - handle empty subjects
# - catch exceptions and exit gracefully
#

require 'iconv'
require 'time'
require 'net/pop'
require "rexml/document"

#require 'rfc2047'

require 'rubygems'
require_gem 'rmail'


# =========
# = prefs =
# =========

@prefs = {
  :default_mail_charset => 'iso-8859-15',
  :output_charset => 'utf-8'
}

# edit these to match your mail account
@mailprefs = {
  :server => 'mail.yourdomain.com',
  :port => 110,
  :uname => 'your_username',
  :pwd => 'your_password'
}

# edit these to change the feed properties
@feedprefs = {
  :title => 'mailfeed.rb',
  :link => 'http://dekstop.de/mailfeed/',
  :description => 'RSS feed generated from an email account.',
  :language => 'en-us',
  :contact => 'http://dekstop.de/contact/',

  :guid_prefix => 'http://dekstop.de/mailfeed/ ',

  :max_entries => 20
}

# RSS 2.0 template
rssheader = "<?xml version=\"1.0\"?>
<rss version=\"2.0\">
	<channel>
		<title>#{ @feedprefs[:title] }</title>
		<link>#{ @feedprefs[:link] }</link>
		<description>#{ @feedprefs[:description] }</description>
		<language>#{ @feedprefs[:language] }</language>
		<docs>http://blogs.law.harvard.edu/tech/rss</docs>
		<generator>MicroLink 5.6 (NewsletterFeed)</generator>
		<managingEditor>#{ @feedprefs[:contact] }</managingEditor>
		<webMaster>#{ @feedprefs[:contact] }</webMaster>
	</channel>
</rss>"


# ========
# = subs =
# ========

# attempts to determine the charset for a given "Content-type"-header value
# returns default_charset if no charset encoding is found
def get_charset(content_type_value, default_charset)
  if (content_type_value != nil)
    match = content_type_value.match(/charset="?([^"\;]*)"?\;?/)
    return match[1] if (match != nil)
  end
  return default_charset
end

# basic HTML markup -- substitutes newlines for HTML-linebreaks and paragraphs
def html_markup(text)
  text.gsub!(/</, '&lt;')
  text.gsub!(/>/, '&gt;')
  '<p>' + text.split(/\n\n/).map { |paragraph|
    paragraph.gsub(/\n/, "<br>\n")
  }.join("</p>\n<p>") + '</p>'
end

# creates an xml node element
def create_node(name, value)
  node = REXML::Element.new(name)
  node.text = value
  node
end


# ========
# = main =
# ========

puts "Content-type: text/xml; charset=#{ @prefs[:output_charset] }\n\n"


rssdoc = REXML::Document.new(rssheader)
rsschannel = rssdoc.elements['rss/channel']

# insert pubDate and lastBuildDate elements
rsschannel << create_node('pubDate', Time.now.rfc2822)
rsschannel << create_node('lastBuildDate', Time.now.rfc2822)


# login to pop server
pop = Net::POP3.new(@mailprefs[:server], @mailprefs[:port])
pop.start(@mailprefs[:uname], @mailprefs[:pwd]) { |pop|

  if pop.mails.empty?
    #puts 'No mail.'
  else
    #puts "#{pop.mails.size} mails"
    
    popmails = pop.mails.reverse
    popmails = popmails[0, @feedprefs[:max_entries]] if popmails.length > @feedprefs[:max_entries]
    
    popmails.each do |mailitem|

      uid = @feedprefs[:guid_prefix] + mailitem.unique_id

      # extract header, body
      mail = nil
      if (mailitem.pop.match(/\r/))
        # RMail chokes when confronted with \r characters
        mail = RMail::Parser.read(mailitem.pop.gsub(/\r?\n/, "\n"))
      else
        mail = RMail::Parser.read(mailitem.pop)
      end
      
      # prepare charset conversions
      charset = get_charset(mail.header['Content-Type'], @prefs[:default_mail_charset])
      conv = Iconv.new(@prefs[:output_charset], charset)


      # parse header
      # TODO: determine charset, e.g. "Content-Type: text/plain; charset=ISO-8859-1"
      from = conv.iconv(mail.header.from.first.format)
      subject = conv.iconv(mail.header.subject)
      #subject = Rfc2047.decode_to(@prefs[:output_charset], mail.header.subject)
      date = mail.header.date.rfc2822

      # decode body
      message_str = nil
      
      if (mail.multipart?)
        message_str = "(This is a multipart message -- some parts may not be displayed.)\n\n"
        mail.body.each { |part|
          if (part.header.content_type == 'text/plain')
            message_str << conv.iconv(part.decode) << "\n\n"
          end
        }
      else
        message_str = conv.iconv(mail.decode)
      end
      
      message = REXML::CData.new(html_markup(message_str))
      

      # create rss item
      rssitem = REXML::Document.new('<item/>').root
      rssitem << create_node('title', subject)
      rssitem << create_node('link', @feedprefs[:link])
      rssitem << create_node('description', message)
      rssitem << create_node('author', from)
      rssitem << create_node('pubDate', date)
      rssitem << create_node('guid', uid)
      
      # insert item to feed
      rsschannel << rssitem
    end
  end
}

xml = ""
rssdoc.write(xml, 0)    # auto-indent
puts xml