w3m/Bonus/makeref

#!/usr/local/bin/ruby

#  HTML reference generator
#  by A.Ito 1999/3/30

require 'kconv'

###########################################################################
class URL
  attr 'scheme'
  attr 'host'
  attr 'port'
  attr 'file'
  attr 'label'
  def initialize(str)
    if /([a-zA-Z+\-]+):(.*)/ =~ str then
      @scheme = $1
      str = $2
    else
      @scheme = 'unknown'
    end
    hostpart = ''
    if %r'//([^/]*)(/.*)' =~ str then
      hostpart = $1
      str = $2
    elsif %r'//([^/]*)$' =~ str then
      hostpart = str
      str = ''
    end
    if hostpart != '' then
      if /(.*):(\d+)/ =~ hostpart then
        @host = $1
        @port = $2
      else
        @host = hostpart
        @port = ''
      end
    else
      @host = @port = ''
    end
    if /(.*)#(.*)/ =~ str then
      @file = $1
      @label = $2
    else
      @file = str
      @label = ''
    end
  end
  def to_s
    s = "#{@scheme}:"
    if s == 'news' or s == 'mailto' then
      return s+@file
    end
    s += "//"+@host
    s += ":"+@port if @port.size > 0
    s += @file
    s += "#"+@label if @label.size > 0
    s
  end
  def complete(current)
    @scheme = current.scheme  if @scheme == 'unknown'
    @port = current.port if @host == '' and @port == ''
    @host = current.host if @host == ''
    unless @file =~ %r'^/' then
      @file = File.expand_path(File.dirname(current.file)+'/'+@file)
    end
    self
  end
end

class Tag
  def initialize(str)
    if str =~ /<(.+)>/ then
      str = $1
    end
    tags = str.split
    @tagname = tags.shift.downcase
    @vals = {}
    tags.each do |t|
      if t =~ /=/ then
	tn,tv = t.split(/\s*=\s*/,2)
	tv.sub!(/^"/,"")
	tv.sub!(/"$/,"")
	@vals[tn.downcase] = tv
      else
	@vals[t.downcase] = TRUE
      end
    end
  end
  def tagname
    return @tagname
  end
  def each
    @vals.each do |k,v|
      yield k,v
    end
  end
  def switch(k)
    return @vals[k]
  end
  def to_s
    if tagname =~ /!--/ then
      return ''
    end
    t = "<"+tagname
    if @vals.size == 0 then
      return t+">"
    end
    each do |a,v|
      if v == true then
        t += " #{a}"
      else
        t += " #{a}=\"#{v}\""
      end
    end
    t+">"
  end
end

class TokenStream
  TAG_START = ?<
  TAG_END = ?>
  AMP_START = ?&
  AMP_END = ?;
  
  def initialize(file)
    if file.kind_of?(IO) then
      @f = file
    else
      @f = File.new(file)
    end
    @buf = nil
    @bpos = 0
  end
  
  def read_until(endsym)
    complete = FALSE
    tag = []
    begin
      while @bpos < @buf.size
	c = @buf[@bpos]
	if c == endsym then
	  tag.push(c.chr)
	  complete = TRUE
	  @bpos += 1
	  break
	end
	if c == 10 || c == 13 then
	  tag.push(' ')
	else
	  tag.push(c.chr)
	end
	@bpos += 1
      end
      unless complete
	@buf = @f.gets
	@bpos = 0
	break if @f.eof?
      end
    end until complete
    return tag.join('')
  end
    
  def get
    while TRUE
      if @buf.nil? then
	@buf = @f.gets
	if @f.eof? then
	  return nil
	end
        @buf = Kconv.toeuc(@buf)
	@bpos = 0
      end
      if @buf[@bpos] == TAG_START then
	return Tag.new(read_until(TAG_END))
      elsif @buf[@bpos] == AMP_START then
	return read_until(AMP_END)
      else
	i = @bpos
	while i < @buf.size && @buf[i] != TAG_START && @buf[i] != AMP_START
	  i += 1
	end
	r = @buf[@bpos,i-@bpos]
	if i == @buf.size then
	  @buf = nil
	else
	  @bpos = i
	end
	redo if r =~ /^\s+$/
	return r
      end
    end
  end
  public :eof?
  def eof?
    @f.eof?
  end
end

################################ MAIN ####################################

refs = []
refnum = 0
body_finished = false
html_finished = false
currentURL = nil
immediate_ref = false

while ARGV[0] =~ /^-/
  case ARGV.shift
  when '-url'
    currentURL = URL.new(ARGV.shift)
  when '-u'
    immediate_ref = true
  end
end

if ARGV.size > 0 then
  f = TokenStream.new(ARGV[0])
else
  f = TokenStream.new(STDIN)
end

until f.eof?
  tok = f.get
  if tok.kind_of?(Tag) then
    if tok.tagname == 'a' and !tok.switch('href').nil? then
      refs[refnum] = tok.switch('href')
      refnum += 1
    elsif tok.tagname == '/a' then
      if immediate_ref then
        r = refs[refnum-1]
        if !currentURL.nil? then
          r = URL.new(r).complete(currentURL).to_s
        end
        print "[#{r}]"
      else
        print "[#{refnum}]"
      end
    elsif tok.tagname == '/body' then
      body_finished = true
      break
    elsif tok.tagname == '/html' then
      html_finished = true
      break
    end
    print tok.to_s
  elsif !tok.nil? then
    print tok
  end
end
if !immediate_ref and refs.size > 0 then
  print "<hr><h2>References</h2>\n"
  for i in 0..refs.size-1
    if currentURL.nil? then
      r = refs[i]
    else
      r = URL.new(refs[i])
      r.complete(currentURL)
      r = r.to_s
    end
    print "[#{i+1}] #{r}<br>\n"
  end
end
print "</body>\n" unless body_finished
print "</html>\n" unless html_finished