9 from lxml.builder import ElementMaker
11 from treecutter import constants as const
12 from treecutter.trie import Trie
13 from treecutter.link import Link
14 from treecutter.tools import ssh_cmd, publish, mkdir_p
17 """Class keeping the internal site structure"""
18 def __init__(self,args):
19 self._output = args.output
20 self._style = args.style
21 self._subdir = args.subdir
22 self._file = 'sitemap.txt'
24 self._sitelang = set()
25 self._isocode = etree.parse('/usr/share/xml/iso-codes/iso_639_3.xml')
27 self._tmptarget = tempfile.mkdtemp()+'/'
29 # The sitemap uses a trie structure to keep track of links
30 # A link represents the path to the document and the link
31 # representing the text on the site.
32 # A link can have several pages in different languages.
33 def add_link(self, link):
34 tokens = filter(None,re.split(r'(^/[\w:-]*$|^/[\w:-]*/|[\w:-]*/)',link,flags=re.UNICODE))
35 self._tree.add(tokens,Link(link))
38 f = codecs.open(self._file,'w','utf-8')
39 s = '\n'.join(link.link() for link in self._tree)
45 f = codecs.open(self._file, 'r', 'utf-8')
46 sml = f.read().split()
50 except IOError, what_error:
51 print 'INFO: Could not read sitemap.txt - one will be created'
53 # Create a set of the current tree for comparison with the
56 return set(link.link() for link in self._tree)
58 # Main driver in the application processing the documents
59 # in the collected sitemap
62 print "Prepareing the input"
63 for link in self._tree:
66 print "Prepare [%5.2f s]" % (round(t2-t1,2))
67 for link in self._tree:
68 self._sitelang = self._sitelang.union(set(link.languages()))
69 for tran in self._sitelang:
71 self._tranlang[tran] = gettext.translation('iso_639_3',
74 print "Language [%5.2f s]" % (round(t3-t2,2))
75 for link in self._tree:
76 link.render(self._style)
78 print "Render [%5.2f s]" % (round(t4-t3,2))
79 for link in self._tree:
80 link.template(self, self._style, self._tmptarget)
82 print "Template [%5.2f s]" % (round(t5-t4,2))
85 # Collect all files used by the documents
86 for link in self._tree:
87 res = res.union(link.resources())
89 outfile = self._tmptarget+f
90 mkdir_p(os.path.dirname(outfile))
91 shutil.copyfile(f,outfile)
92 print "Resources[%5.2f s]" % (round(t6-t5,2))
93 # TODO: Improve the sitemap, it is a page that is generated from
94 # the ground up and added a bit adhoc.
95 sitmaplink = Link('/sitemap')
96 for l in self._sitelang:
97 sitmaplink.add_page((l,'/sitemap.'+l+'.xml'))
98 for l in self._sitelang:
99 sitmaplink.page(l).set_article(self.gen_menu(l,None,"tree sitemap"))
100 sitmaplink.page(l).template(self,self._style,self._tmptarget)
102 print "Sitemap [%5.2f s]" % (round(t7-t6,2))
107 def gen_menu(self,lang,page,cssclass):
108 return self._tree.menu(lang,page,cssclass,self._subdir)
110 def lang_menu(self,lang,link):
111 html = ElementMaker()
113 for l in link.languages():
114 isoxml = u"//iso_639_3_entry[@*='"+l+"']"
115 ln = self._isocode.xpath(isoxml)[0].get('name')
117 ln = self._tranlang[lang].gettext(ln)
118 p = unicode(link.link())
122 li = html.li(html.a(ln.decode('utf-8'),
123 href=self._subdir+p,hreflang=l))
125 return etree.tostring(menu,encoding='UTF-8',pretty_print=False)
128 ssh_cmd(self._output,"mkdir -p")
129 publish(self._tmptarget, self._output)
130 for res in ["css","images","js","fonts","favicon.ico"]:
131 if (os.path.exists(self._style+res)):
132 publish(self._style+res, self._output)
133 ssh_cmd(self._output,"chmod a+rx")