treecutter/sitemap.py

   1 #!/usr/bin/python
   2 import os
   3 import re
   4 import time
   5 import shutil
   6 import gettext
   7 import tempfile
   8 from amara import bindery
   9 from treecutter.trie import Trie
  10 from treecutter.link import Link
  11 from treecutter.tools import ssh_cmd, publish, mkdir_p
  12
  13 class Sitemap():
  14     """Class keeping the internal site structure"""
  15     def __init__(self):
  16         self._file = 'sitemap.txt'
  17         self._tree = Trie()
  18         self._sitelang = set()
  19         self._isocode = bindery.parse('/usr/share/xml/iso-codes/iso_639_3.xml')
  20         self._tranlang = {}
  21         self._tmptarget = tempfile.mkdtemp()+'/'
  22
  23     def add_link(self, link):
  24         tokens = filter(None,re.split(r'(^/[\w-]*/|[\w-]*/)',link))
  25         self._tree.add(tokens,Link(link))
  26
  27     def write_map(self):
  28         f = open(self._file,'w')
  29         f.write('\n'.join(link.link() for link in self._tree))
  30         f.close()
  31
  32     def read_map(self):
  33         try:
  34             f = open(self._file)
  35             sml = f.read().split()
  36             f.close()
  37             for line in sml:
  38                 self.add_link(line)
  39         except IOError, what_error:
  40             print 'INFO: Could not read sitemap.txt - one will be created'
  41
  42     def set(self):
  43         return set(link.link() for link in self._tree)
  44
  45     def process(self, style):
  46         t1 = time.time()
  47         for link in self._tree:
  48             link.prepare()
  49         t2 = time.time()
  50         print "Prepare  [%5.2f s]" % (round(t2-t1,2))
  51         for link in self._tree:
  52             self._sitelang = self._sitelang.union(set(link.languages()))
  53         for tran in self._sitelang:
  54             if tran != 'en':
  55                 self._tranlang[tran] = gettext.translation('iso_639_3',
  56                                                            languages=[tran])
  57         t3 = time.time()
  58         print "Language [%5.2f s]" % (round(t3-t2,2))
  59         for link in self._tree:
  60             link.render(style)
  61         t4 = time.time()
  62         print "Render   [%5.2f s]" % (round(t4-t3,2))
  63         for link in self._tree:
  64             link.template(self, style, self._tmptarget)
  65         t5 = time.time()
  66         print "Template [%5.2f s]" % (round(t5-t4,2))
  67         t6 = time.time()
  68         res = set()
  69         cwd = os.getcwd()
  70         for link in self._tree:
  71             res = res.union(link.resources())
  72         for f in res:
  73             outfile = self._tmptarget+f
  74             mkdir_p(os.path.dirname(outfile))
  75             shutil.copyfile(f,outfile)
  76         print "Resources[%5.2f s]" % (round(t6-t5,2))
  77         sitmaplink = Link('/sitemap')
  78         for l in self._sitelang:
  79             sitmaplink.add_page((l,'/sitemap.'+l+'.xml'))
  80         for l in self._sitelang:
  81             sitmaplink.page(l).set_article(self.gen_menu(l,None,"tree sitemap"))
  82             sitmaplink.page(l).template(self,style,self._tmptarget)
  83         t7 = time.time()
  84         print "Sitemap  [%5.2f s]" % (round(t7-t6,2))
  85
  86     def graph(self):
  87         self._tree.graph()
  88
  89     def gen_menu(self,lang,page,cssclass):
  90         return self._tree.menu(lang,page,cssclass)
  91
  92     def lang_menu(self,lang,link):
  93         html = "<ul>"
  94         for l in link.languages():
  95             isoxml = u"//iso_639_3_entry[@*='"+l+"']"
  96             ln = self._isocode.xml_select(isoxml)[0].name
  97             if lang != 'en':
  98                 ln = self._tranlang[lang].gettext(ln)
  99             p = link.link()
 100             if p[-1] == '/':
 101                 p = p +'index'
 102             p = p+'.'+l
 103             html += '<li><a href="%s" hreflang="%s">%s</a></li>' % (p, l, ln)
 104         html += "</ul>"
 105         return html
 106
 107     def publish(self,output,style):
 108         ssh_cmd(output,"mkdir -p")
 109         publish(self._tmptarget, output)
 110         for res in ["css","images","js","favicon.ico"]:
 111             if (os.path.exists(style+res)):
 112                 publish(style+res, output)
 113         ssh_cmd(output,"chmod a+rx")