treecutter/sitemap.py

   1 #!/usr/bin/python
   2 import os
   3 import re
   4 import shutil
   5 import gettext
   6 import tempfile
   7 from amara import bindery
   8 from time import time
   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()
  47         print "Prepareing the input"
  48         for link in self._tree:
  49             link.prepare()
  50         t2 = time()
  51         print "Prepare  [%5.2f s]" % (round(t2-t1,2))
  52         for link in self._tree:
  53             self._sitelang = self._sitelang.union(set(link.languages()))
  54         for tran in self._sitelang:
  55             if tran != 'en':
  56                 self._tranlang[tran] = gettext.translation('iso_639_3',
  57                                                            languages=[tran])
  58         t3 = time()
  59         print "Language [%5.2f s]" % (round(t3-t2,2))
  60         for link in self._tree:
  61             link.render(style)
  62         t4 = time()
  63         print "Render   [%5.2f s]" % (round(t4-t3,2))
  64         for link in self._tree:
  65             link.template(self, style, self._tmptarget)
  66         t5 = time()
  67         print "Template [%5.2f s]" % (round(t5-t4,2))
  68         t6 = time()
  69         res = set()
  70         cwd = os.getcwd()
  71         for link in self._tree:
  72             res = res.union(link.resources())
  73         for f in res:
  74             outfile = self._tmptarget+f
  75             mkdir_p(os.path.dirname(outfile))
  76             shutil.copyfile(f,outfile)
  77         print "Resources[%5.2f s]" % (round(t6-t5,2))
  78         sitmaplink = Link('/sitemap')
  79         for l in self._sitelang:
  80             sitmaplink.add_page((l,'/sitemap.'+l+'.xml'))
  81         for l in self._sitelang:
  82             sitmaplink.page(l).set_article(self.gen_menu(l,None,"tree sitemap"))
  83             sitmaplink.page(l).template(self,style,self._tmptarget)
  84         t7 = time()
  85         print "Sitemap  [%5.2f s]" % (round(t7-t6,2))
  86
  87     def graph(self):
  88         self._tree.graph()
  89
  90     def gen_menu(self,lang,page,cssclass):
  91         return self._tree.menu(lang,page,cssclass)
  92
  93     def lang_menu(self,lang,link):
  94         html = "<ul>"
  95         for l in link.languages():
  96             isoxml = u"//iso_639_3_entry[@*='"+l+"']"
  97             ln = self._isocode.xml_select(isoxml)[0].name
  98             if lang != 'en':
  99                 ln = self._tranlang[lang].gettext(ln)
 100             p = link.link()
 101             if p[-1] == '/':
 102                 p = p +'index'
 103             p = p+'.'+l
 104             html += '<li><a href="%s" hreflang="%s">%s</a></li>' % (p, l, ln)
 105         html += "</ul>"
 106         return html
 107
 108     def publish(self,output,style):
 109         ssh_cmd(output,"mkdir -p")
 110         publish(self._tmptarget, output)
 111         for res in ["css","images","js","fonts","favicon.ico"]:
 112             if (os.path.exists(style+res)):
 113                 publish(style+res, output)
 114         ssh_cmd(output,"chmod a+rx")