sitemap: adding sizecalculation

[treecutter.git] / treecutter / sitemap.py

diff --git a/treecutter/sitemap.py b/treecutter/sitemap.py
index 24b53fe5b4ab3bd84d65caca16cd1e1e07db54a6..707067f1f8cf28a284759248ce111ef520bf780a 100644 (file)
--- a/treecutter/sitemap.py
+++ b/treecutter/sitemap.py
@@ -1,7 +1,9 @@
 #!/usr/bin/python
 import os
+import codecs
 import re
 import shutil
+import sys
 import gettext
 import tempfile
 from lxml import etree
@@ -10,7 +12,8 @@ from time import time
 from treecutter import constants as const
 from treecutter.trie import Trie
 from treecutter.link import Link
-from treecutter.tools import ssh_cmd, publish, mkdir_p
+from treecutter.tools import ssh_cmd, publish, mkdir_p,get_folder_size,sizeof_fmt
+
 
 class Sitemap():
     """Class keeping the internal site structure"""
@@ -30,17 +33,18 @@ class Sitemap():
     # representing the text on the site.
     # A link can have several pages in different languages.
     def add_link(self, link):
-        tokens = filter(None,re.split(r'(^/[\w-]*/|[\w-]*/)',link))
+        tokens = filter(None,re.split(r'(^/[\w\.:-]*$|^/[\w\.:-]*/|[\w\.:-]*/)',link,flags=re.UNICODE))
         self._tree.add(tokens,Link(link))
 
     def write_map(self):
-        f = open(self._file,'w')
-        f.write('\n'.join(link.link() for link in self._tree))
+        f = codecs.open(self._file,'w','utf-8')
+        s = '\n'.join(link.link() for link in self._tree)
+        f.write(s)
         f.close()
 
     def read_map(self):
         try:
-            f = open(self._file)
+            f = codecs.open(self._file, 'r', 'utf-8')
             sml = f.read().split()
             f.close()
             for line in sml:
@@ -70,12 +74,14 @@ class Sitemap():
                                                            languages=[tran])
         t3 = time()
         print "Language [%5.2f s]" % (round(t3-t2,2))
+        transform = {}
+        transform['xhtml5'] = etree.XSLT(etree.parse(self._style+"docbook.xhtml5.xsl"))
         for link in self._tree:
-            link.render(self._style)
+            link.render(transform)
         t4 = time()
         print "Render   [%5.2f s]" % (round(t4-t3,2))
         for link in self._tree:
-            link.template(self, self._style, self._tmptarget)
+            link.template(self, self._style, self._tmptarget,self._subdir)
         t5 = time()
         print "Template [%5.2f s]" % (round(t5-t4,2))
         t6 = time()
@@ -95,7 +101,7 @@ class Sitemap():
             sitmaplink.add_page((l,'/sitemap.'+l+'.xml'))
         for l in self._sitelang:
             sitmaplink.page(l).set_article(self.gen_menu(l,None,"tree sitemap"))
-            sitmaplink.page(l).template(self,self._style,self._tmptarget)
+            sitmaplink.page(l).template(self,self._style,self._tmptarget,self._subdir)
         t7 = time()
         print "Sitemap  [%5.2f s]" % (round(t7-t6,2))
 
@@ -113,19 +119,20 @@ class Sitemap():
             ln = self._isocode.xpath(isoxml)[0].get('name')
             if lang != 'en':
                 ln = self._tranlang[lang].gettext(ln)
-            p = link.link()
-            if p[-1] == '/':
-                p = p +'index'
-            p = p+'.'+l
+            p = unicode(link.link())
+            if p[-1] == u'/':
+                p = p +u'index'
+            p = p+u'.'+l
             li = html.li(html.a(ln.decode('utf-8'),
                                 href=self._subdir+p,hreflang=l))
             menu.append(li)
         return etree.tostring(menu,encoding='UTF-8',pretty_print=False)
 
     def publish(self):
+        print "Size [ %7s ]" % (sizeof_fmt(get_folder_size(self._tmptarget)))
         ssh_cmd(self._output,"mkdir -p")
         publish(self._tmptarget, self._output)
-        for res in ["css","images","js","fonts","favicon.ico"]:
+        for res in ["stylesheets","images","js","fonts","favicon.ico"]:
             if (os.path.exists(self._style+res)):
                 publish(self._style+res, self._output)
         ssh_cmd(self._output,"chmod a+rx")