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 if args.exclude:
  16     for pat in args.exclude:
  17         exclude.append(pat)
  18         if not pat.endswith('/*'):
  19             exclude.append(pat + '/*')
  20
  21 os.chdir(args.root_dir)
  22 fullpath = os.getcwd()
  23 fp_len = len(fullpath)
  24
  25 tree = ('', [])
  26 current_subtree = tree[1]
  27 save_level = 0
  28
  29 for dirpath, dirs, files in sorted(os.walk(fullpath)):
  30     try:
  31         dirpath = dirpath[fp_len:]
  32         if not dirpath:
  33             continue
  34         for pat in exclude:
  35             if fnmatch(dirpath, pat):
  36                 raise StopIteration
  37         parts = dirpath.split('/')
  38         level = len(parts) - 2
  39         if level < save_level:
  40             current_subtree = tree[1]
  41             for i in range(level):
  42                 current_subtree = current_subtree[-1][1]
  43         elif level > save_level:
  44             assert level - save_level == 1
  45             current_subtree = current_subtree[-1][1]
  46         #else: # level == save_level:
  47         current_subtree.append((parts[-1], []))
  48         save_level = level
  49     except StopIteration:
  50         pass
  51
  52 def _tree2html(tree, path='', level=0):
  53     subparts = []
  54     indent = "  " * (level + 1)
  55     for title, subtree in tree:
  56         subpath = "%s/%s" % (path, title)
  57         if subpath.startswith('/'):
  58             subpath = subpath[1:]
  59         href = '<a href="%s/">%s</a>' % (subpath, title)
  60         if subtree:
  61             subparts.append(indent + "<li>%s" % href)
  62             subparts.append(_tree2html(subtree, subpath, level+2))
  63             subparts.append(indent + "</li>")
  64         else:
  65             subparts.append(indent + "<li>%s</li>" % href)
  66     s = "\n".join(subparts)
  67
  68     parts = []
  69     indent = "  " * level
  70     parts.append(indent + "<ul>")
  71     parts.append(s)
  72     parts.append(indent + "</ul>")
  73
  74     return "\n".join(parts)
  75
  76 sitemap_tmpl = ["""\
  77 #extends phd_site
  78 #implements respond
  79 #attr $Title = 'Sitemap'
  80 #attr $Copyright = 2015
  81 ##
  82 #def body_html
  83 """]
  84
  85 sitemap_tmpl.append(_tree2html(tree[1]))
  86
  87 sitemap_tmpl.append("""
  88 #end def
  89 $phd_site.respond(self)
  90 """)
  91
  92 write_if_changed("sitemap.tmpl", ''.join(sitemap_tmpl))