gen-sitemap.py

   1 #! /usr/bin/env python
   2
   3 import argparse
   4 from fnmatch import fnmatch
   5 import os
   6 from news import write_if_changed
   7
   8 parser = argparse.ArgumentParser(description='Generate sitemap')
   9 parser.add_argument('-x', '--exclude', action='append',
  10                     help='exclude directories (pattern)')
  11 parser.add_argument('root_dir', help='Root dicrectory')
  12 args = parser.parse_args()
  13
  14 exclude = []
  15 for pat in args.exclude:
  16     exclude.append(pat)
  17     if not pat.endswith('/*'):
  18         exclude.append(pat + '/*')
  19
  20 os.chdir(args.root_dir)
  21 fullpath = os.getcwd()
  22 fp_len = len(fullpath)
  23
  24 tree = ('', [])
  25 current_subtree = tree[1]
  26 save_level = 0
  27
  28 for dirpath, dirs, files in sorted(os.walk(fullpath)):
  29     try:
  30         dirpath = dirpath[fp_len:]
  31         if not dirpath:
  32             continue
  33         for pat in exclude:
  34             if fnmatch(dirpath, pat):
  35                 raise StopIteration
  36         parts = dirpath.split('/')
  37         level = len(parts) - 2
  38         if level < save_level:
  39             current_subtree = tree[1]
  40             for i in range(level):
  41                 current_subtree = current_subtree[-1][1]
  42         elif level > save_level:
  43             assert level - save_level == 1
  44             current_subtree = current_subtree[-1][1]
  45         #else: # level == save_level:
  46         current_subtree.append((parts[-1], []))
  47         save_level = level
  48     except StopIteration:
  49         pass
  50
  51 def _tree2html(tree, level=0):
  52     indent = "  " * level
  53     strings = []
  54     strings.append(indent + "<ul>")
  55
  56     level += 1
  57     indent2 = "  " * level
  58
  59     for title, subtree in tree:
  60         if subtree:
  61             strings.append(indent2 + "<li>%s" % title)
  62             strings.append(_tree2html(subtree, level+1))
  63             strings.append(indent2 + "</li>")
  64         else:
  65             strings.append(indent2 + "<li>%s</li>" % title)
  66
  67     strings.append(indent + "</ul>")
  68     return '\n'.join(strings)
  69
  70 sitemap_tmpl = ["""\
  71 #extends phd_site
  72 #implements respond
  73 #attr $Title = 'Sitemap'
  74 #attr $Copyright = 2015
  75 ##
  76 #def body_html
  77 """]
  78
  79 sitemap_tmpl.append(_tree2html(tree[1]))
  80
  81 sitemap_tmpl.append("""
  82 #end def
  83 $phd_site.respond(self)
  84 """)
  85
  86 write_if_changed("sitemap.tmpl", ''.join(sitemap_tmpl))