sitemap/resource: adding minus in path, adding video
[treecutter.git] / treecutter / directory.py
index 19c4d8b4182cf4d4764d9ec77442a0473fd76839..1b7a3e9f3ee6562ff9bfedd51ea0bdac409b1d7c 100644 (file)
@@ -3,12 +3,14 @@ import os
 import fnmatch
 from lxml import etree
 import treecutter.constants as const
+import re
 
 class Directory():
     """Class containing the state of the directory with articles"""
     def __init__(self):
-        self._cwd = '.'
+        self._cwd = u'.'
         self._tree = []
+        self._basepath = re.compile('[/\w\._-]*/[\w-]+',re.UNICODE)
 
     def scan(self):
         for dirname, dirnames, filenames in os.walk(self._cwd):
@@ -19,8 +21,8 @@ class Directory():
                     title = doc.xpath(u'/db:article/db:info/db:title',namespaces=const.XPATH)
                     menu  = doc.xpath(u'/db:article/db:info/db:titleabbrev',namespaces=const.XPATH)
                     if title and menu:
-                        base = file_.split('.')[1]
-                        link = base.replace('index','')
+                        base = self._basepath.match(file_).group()
+                        link = base.replace('index','')[1:]
                         self._tree.append(link)
 
     def set(self):