#!/usr/bin/python
# -*- coding: UTF-8 -*-

import os, re, string, sys

class Tree:
  def __init__(self):
    self.link = None
    self.branches = {}
  def html(self):
    if not self.branches:
      return ''
    res = "<ul>\n"
    #for (k, v) in self.branches.iteritems():
    for k in sorted(self.branches.keys()):
      v = self.branches[k]
      res += "<li>"
      if v.link:
        res += '<a href="%s">%s</a>' % (v.link, k)
      else:
        res += k
      res += v.html()
      res += "</li>\n"
    res += "</ul>\n"
    return res
  def __repr__(self):
    res = ''
    #for (k, v) in self.branches.iteritems():
    for k in sorted(self.branches.keys()):
      v = self.branches[k]
      res += k
      if v.link:
        res += ' => ' + v.link
      res += "\n"
      res += "\n".join(['  ' + l for l in repr(v).splitlines()]) + "\n"
    return res
  def setlink(self, p):
    if self.link is None:
      self.link = p
    else:
      raise RuntimeError(p)
  def branch(self, p):
    return self.branches.setdefault(p, Tree())

res = Tree()

def add(path, keywords, tree=res):
  if len(keywords)==0:
    tree.setlink(path)
  else:
    p = keywords[0]
    add(path, keywords[1:], tree.branch(p))

def process(filename):
  match = re.compile('^\s*<!--\s*SITEMAP:\s*(.*?\w)\s*-->', re.S|re.M).search(open(filename).read())
  if match:
    for line in match.group(1).splitlines():
      # filename hat die Form './…' ⇒ also erstes Zeichen abschneiden
      add(filename[1:], re.split('\s\|\s', line.strip()))

for root, dirs, files in os.walk('.'):
  for f in files:
    process(os.path.join(root, f))

try:
  of = file('sitemap/sitemap.inc', 'w')
except RuntimeError, e:
  print 'RuntimeError', e
  sys.exit()
print >> of, res.html()
print res