#! /usr/local/bin/ruby

##################################################################
#### RSS2NNTP module, to convert RSS articles to NNTP articles
####
#### Written by F. Senault (fred@lacave.net), free to use,
####   redistribute, modify.
#### No warranty at all, implied or whatever.  Use at your own
####   risk.
##################################################################

begin
  require 'rubygems'
rescue Exception
end
require 'feed_tools'
require 'uri'
require 'open3'
require 'tempfile'
require 'iconv'
require 'html/htmltokenizer'
require 'htmlentities'

$PATH = 'innxmit-lacave!rss-nntp-lacave'
$ROOTGROUP = 'lacave.rss.'

class RSS2NNTPConf
  def initialize(f = nil)
    @conf = {
        :path       => 'rss2nntp',
        :hierarchy  => 'nonexistent.test.',
        :converter  => '/usr/local/bin/lynx -dump -force_html ' +
            '-assume_charset iso-8859-15 -nolist ',
        :host       => Socket.gethostname,
        :server     => 'localhost',
        :abuse      => 'abuse@' + Socket.gethostname.gsub(/^[^.]*\./, ''),
        :program    => 'Lacave.net RSS2NNTP aggregator',
        :version    => '0.5.3',
        :usesig     => true,
        :debug      => false,
        :verbose    => false,
        :post       => true,
      }
    @feeds = {}
    if(f.nil?) then
      f = $0
      f.gsub!(/\/[^\/]*$/,'/')
      f += "rss2nntp.conf"
    end
    File.open(f) do |h|
      a = ''
      while true
        l = h.gets
        if !l.nil?
          l.chomp!
          l.gsub!(/#.*$/, '')
          if l =~ /^\s/
            a += l.gsub(/^\s+/, ' ')
            next
          end
        end
        if a != ''
          a.strip!
          if a =~ /^([A-Z0-9]+)\s*=\s*(.+)$/
            key = $1.downcase.to_sym
            val = $2
            @conf[key] = val
          elsif a =~ /^([^\s]+)$/
            key = $1
            val = ($2.nil? ? '' : $2)
            #@feeds[key] = val
            setgroup(key, val)
          elsif a =~ /^(.+?)\s+(.+)$/
          #elsif a =~ /^([^\s]+?)(?:\s+([^\s]+))?(?:\s+([^\s]+))?$/
            key = $1
            val = ($2.nil? ? '' : $2)
            #@feeds[key] = val
            setgroup(key, val)
          end
        end
        a = l
        break if l.nil?
      end
      [:usesig, :debug, :verbose, :post].each do |k|
        if @conf[k].is_a? String
          @conf[k] = (@conf[k].downcase == 'yes' ||
                @conf[k].downcase == 'true')
        end
      end
    end
  end
  def []=(key, val)
    @conf[key] = val
  end
  def [](key)
    @conf[key]
  end
  def setgroup(url, group)
    if(group != '') then
      if(group[0..0] == '/') then
        @feeds[url] = group[1..-1]
      else
        @feeds[url] = @conf[:hierarchy] + group
      end
    else
      @feeds[url] = ''
    end
  end
  def feed(url)
    @feeds[url]
  end
  def feeds
    @feeds
  end
end

class RSSArticle < Net::NNTP::Article
  @@conf = nil
  @@conv = nil
  def initialize(item, feed, url)
    super()
    @@conf = RSS2NNTPConf.new() if(@@conf.nil?)
    @@conv = Iconv.new('ISO-8859-15//TRANSLIT', 'UTF-8') if(@@conv.nil?)
    @tags = []
    @body = parse_body(item, feed)
    g = @@conf.feed(url)
    if(g == '') then
      g = RSSArticle.make_group(url)
      @@conf.setgroup(url, g)
    end
    self['Path'] = @@conf[:path]
    self['Subject'] = if(item.title.nil? || item.title == '') then
      'Pas de sujet'
    else
      @@conv.iconv(HTMLEntities.decode_entities(item.title))
    end
    self['From'] = @@conv.iconv(HTMLEntities.decode_entities(make_from(item, feed)))
    self['Newsgroups'] = g #@@conf[:hierarchy] + g
    self['Message-ID'] = RSSArticle.make_mid(item, g)
    self['Date'] = make_rss_date(item)
    self['NNTP-Posting-Host'] = @@conf[:host]
    self['NNTP-Posting-Date'] = RSSArticle.make_date(Time.now)
    self['Lines'] = @body.scan(/\n/).length
    self['Mime-Version'] = '1.0'
    self['Content-Type'] = 'text/plain; charset="iso-8859-15"'
    self['Content-Transfer-Encoding'] = '8bit'
    self['X-Trace'] = "#{@@conf[:host]} #{Time.now.to_i} "+
          "#{Process.pid} 127.0.0.1 (#{Time.now.to_s})"
    self['X-Complaints-To'] = @@conf[:abuse]
    self['User-Agent'] = @@conf[:program] + ' ' + @@conf[:version]
    self['Keywords'] = make_keywords(item.categories)
    self
  end
  def self.loadconfig(conf)
    @@conf = conf
  end
  def self.make_group(url)
    u = URI.parse(url)
    g = u.host.split(/\./).reverse
    g.shift
    if(g[0] == 'co' || g[0] == 'uc') then
      g.shift
    end
    if(g[-1] == 'www') then
      g.pop
    end
    g.join('.')
  end
  def self.make_mid(item, group)
    if(item.guid.nil?) then
      guid = item.link
    else
      guid = item.guid
    end
    if(guid =~ /^http:\/\//) then
      u = URI.parse(guid)
      l = u.path.gsub(/\//, '.')
      l.gsub!(/^./, '')
      l.gsub!(/.$/, '')
      l.gsub!(/index.[a-z]{2,6}\./, '')
      if(!u.query.nil? && u.query != '') then
        l += (l == '' ? '' : '.') + "#{u.query}"
      end
      if(!u.fragment.nil? && u.fragment != '') then
        l += (l == '' ? '' : '.') + "#{u.fragment}"
      end
      mid = "#{l}"
    else
      mid = guid
      if(mid =~ /^tag:([^,])+,/)
        mid.gsub!(/^tag:([^,])+,/,'')
      end
    end
    if(mid == '' && !item.published.nil?) then
      mid = item.published.to_s
    end
    if(mid == '' && !item.time.nil?) then
      mid = item.time.to_s
    end
    if(mid == '') then
      mid = item.title
    end
    mid.gsub!(/[:;,=+]/, '.')
    mid.gsub!(/[^a-zA-Z0-9._-]/, '')
    mid = "<#{mid}.1@#{group}>"
  end
  def parse_body(item, feed)
    b = ''
    tf = Tempfile.new('rss2news')
    tf.puts pre_parse_links(@@conv.iconv(item.content))
    tf.close
    Open3.popen3(@@conf[:converter] + " #{tf.path}") do |i, o, e|
      b.each do |l|
        i.puts(b)
      end
      i.close_write
      b = ''
      while(l = o.gets)
        l.gsub!(/^   /, '')
        b += l
      end
    end
    tf.close(true)
    b = post_parse_links(b)
    b << "\n" unless(b[-1,1] == "\n")
    if(@@conf[:usesig]) then
      b << "-- \n"
    end
    b << "Original : #{item.link}\nSite     : #{feed.link}\n"
    b
  end
  def pre_parse_links(content)
    r = ''
    lok = -1
    ll = false
    tk = HTMLTokenizer.new(content)
    acc = ''
    while(t = tk.getNextToken)
      if(t.is_a?(HTMLTag) && t.tag_name == 'a') then
        if(t.attr_hash['href'] && \
           t.attr_hash['href'] =~ /^(http|mailto|ftp|news|nntp)/) then
          @tags.push(t.attr_hash['href'])
          lok = @tags.length
        else
          lok = -1
        end
      elsif(t.is_a?(HTMLTag) && (t.tag_name == '/a' || t.tag_name == 'img')) then
        if(acc != '') then
          r << acc if(lok == -1 || acc.strip != @tags[lok - 1].strip)
          acc = ''
        end
        if(t.attr_hash['alt'] && t.attr_hash['alt'] != '') then
          r << ' ' unless(r[-1,1] == ' ')
          r << '[' + t.attr_hash['alt'] + ']'
        end
        if(t.tag_name == 'img') then
          @tags.push(t.attr_hash['src'])
          r << '&-' + @tags.length.to_s
        else
          if(lok != -1) then
            r << '&-' + lok.to_s
          else
            r << '&-lien'
          end
          lok = -1
        end
        r << '-&'
      else
        if(lok == -1) then
          r << t.to_s
        else
          acc << t.to_s
        end
      end
    end
    r
  end
  def post_parse_links(content)
    return content if(@tags.length == 0)
    r = content.gsub(/( +)?&-((?:lien|img|\d+)(?::.*?)?)-&([ ()'`´"])?/) do |m|
      e = ''
      e << ($1.nil? ? ' ' : $1.squeeze) unless($`[-1,1] == "\n" || $`.length == 0)
      if($2[0..2] == 'img' && $2.length > 3) then
        e << "[#{$2[3..-1]}]"
      elsif($2[0..2] == 'img') then
        e << "[image]"
      else
        e << "[#{$2}]"
      end
      e << ($3.nil? ? ' ' : $3)
    end
    r << "\n" unless(r =~ /\n\n$/)
    f = '%' + (Math.log10(@tags.length).to_i + 1).to_s + 'd'
    @tags.each_index do |i|
      @tags[i].gsub!(/&amp;/, '&')
      @tags[i].gsub!(/&apos;/, '\'')
      r << "[#{f % [ i + 1 ] }] : #{@tags[i]}\n"
    end
    r
  end
  def make_from(item, feed)
    if(item.author.nil? || item.author.name.nil? || item.author.name == '') then
      if(feed.title.nil?) then
        "Unknown"
      else
        feed.title
      end
    else
      item.author.name
    end
  end
  def make_keywords(c)
    if(c.nil?) then
      ''
    else
      c.collect { |c| ( c.label.nil? ? c.term : c.label ) }.join(',')
    end
  end
  def make_rss_date(item)
    d = nil
    if(!item.published.nil?) then
      d = item.published
    end
    if(d.nil? && !item.time.nil?) then
      d = item.time
    end
    if(d.nil?) then
      d = Time.now
    end
    Net::NNTP::Article.make_date(d)
  end
end