Using translated cheetah templates, allowing - in filenames.

[treecutter.git] / src / tree-cutter.py

diff --git a/src/tree-cutter.py b/src/tree-cutter.py
index 5f411496488e72178985fa8bf7c973f4bc1e0dfb..bc4ae7e0db92400f8f33019e7d0f09e33d0b1a19 100755 (executable)
--- a/src/tree-cutter.py
+++ b/src/tree-cutter.py
@@ -11,6 +11,8 @@ import argparse
 import shutil
 import pygraphviz as pgv
 import glob
+import gettext
+import shutil
 from amara import bindery
 from amara.xslt import transform
 from Cheetah.Template import Template
@@ -23,7 +25,6 @@ parser.add_argument('--output', nargs='?',
 args = parser.parse_args()
 
 style_xslt = args.style+"docbook.xsl"
-style_tmpl = args.style+"index.en.html.tmpl"
 outputdir = args.output
 
 tmptarget = tempfile.mkdtemp()+'/'
@@ -74,7 +75,8 @@ class Directory():
 
 class Page():
     """Class representing a version of a webpage"""
-    def __init__(self,page):
+    def __init__(self,link,page):
+        self._link = link
         self._file = page[1]
         self._lang = page[0]
         self._doc = None
@@ -86,6 +88,9 @@ class Page():
     def language(self):
         return self._lang
 
+    def resources(self):
+        return set(self._resources)
+
     def menu(self):
         return self._menu
 
@@ -105,7 +110,7 @@ class Page():
             for c in code:
                 (p, ext) = os.path.splitext(c.href)
                 if ext in valid_scripts:
-                    exe = os.path.join(os.path.abspath(dirname+c.href))
+                    exe = os.path.join(os.path.abspath(dirname)+'/'+c.href)
                     xml = subprocess.Popen([exe],stdout=subprocess.PIPE)
                     xstr = bindery.parse(str(xml.stdout.read()))
                     idp = c.xml_index_on_parent
@@ -146,14 +151,15 @@ class Page():
         os.chdir(cwd)
 
     def template(self,sitemap):
-        htmlmenu =  sitemap.gen_menu(self._lang,None,None)
+        htmlmenu =  sitemap.gen_menu(self._lang,None,"menu")
         levelmenu = sitemap.gen_menu(self._lang,self,"tree")
-        template = Template(file=style_tmpl,
+        langmenu = sitemap.lang_menu(self._lang,self._link)
+        template = Template(file=args.style+'index.'+self._lang+'.html.tmpl',
                             searchList=[{'title':self._title},
                                         {'menu':htmlmenu},
                                         {'article':self._rendered_article},
                                         {'levelmenu':levelmenu},
-                                        {'levelname':'Menu'}])
+                                        {'langmenu':langmenu}])
         outfile = tmptarget+'html'.join(self._file.rsplit('xml',1))
         mkdir_p(os.path.dirname(outfile))
         out = open(outfile, 'w')
@@ -172,7 +178,10 @@ class Link():
             path = path+'index'
         lang = self._scan_languages(path)
         for l in lang:
-            self._pages.append(Page(l))
+            self._pages.append(Page(self,l))
+
+    def add_page(self,l):
+        self._pages.append(Page(self,l))
 
     def _scan_languages(self,path):
         lang = []
@@ -207,6 +216,14 @@ class Link():
         for page in self._pages:
             if page.language()==lang:
                 return page
+        return None
+
+    def resources(self):
+        res  = set()
+        for page in self._pages:
+            res = res.union(page.resources())
+        return res
+
 
 class Node():
     def __init__(self,token,value):
@@ -271,11 +288,17 @@ class Trie():
         html = "<ul%s>\n" % css
         for l in trie:
             sel = ''
-            if l.value().page(lang) == page:
+            p = l.value().page(lang)
+            if p == page:
                 sel = ' class="selected"'
-            html += '<li%s><a href="%s">%s</a>\n' \
-            % (sel,l.value().link(),l.value().page(lang).menu())
-            html += self._menu(l.children(), lang, page, "")
+            if p != None:
+                html += '<li%s><a href="%s">%s</a>\n' \
+                    % (sel,l.value().link(),p.menu())
+            else:
+                html += '<li%s><a href="%s.en" hreflang="en">%s</a>*\n' \
+                    % (sel,l.value().link(), l.value().page('en').menu())
+            if l.children():
+                html += self._menu(l.children(), lang, page, "")
         html += "</ul>\n"
         return html
 
@@ -290,9 +313,12 @@ class Sitemap():
     def __init__(self):
         self._file = 'sitemap.txt'
         self._tree = Trie()
+        self._sitelang = set()
+        self._isocode = bindery.parse('/usr/share/xml/iso-codes/iso_639_3.xml')
+        self._tranlang = {}
 
     def add_link(self, link):
-        tokens = filter(None,re.split(r'(^/\w*/|\w*/)',link))
+        tokens = filter(None,re.split(r'(^/[\w-]*/|[\w-]*/)',link))
         self._tree.add(tokens,Link(link))
 
     def write_map(self):
@@ -319,9 +345,12 @@ class Sitemap():
             link.prepare()
         t2 = time.time()
         print "Prepare  [%5.2f s]" % (round(t2-t1,2))
-        sitelang = set()
         for link in self._tree:
-            sitelang = lang.union(set(link.languages()))
+            self._sitelang = self._sitelang.union(set(link.languages()))
+        for tran in self._sitelang:
+            if tran != 'en':
+                self._tranlang[tran] = gettext.translation('iso_639_3',
+                                                           languages=[tran])
         t3 = time.time()
         print "Language [%5.2f s]" % (round(t3-t2,2))
         for link in self._tree:
@@ -332,13 +361,24 @@ class Sitemap():
             link.template(self)
         t5 = time.time()
         print "Template [%5.2f s]" % (round(t5-t4,2))
-        sm = {}
-        for l in sitelang:
-            sm[l] = Page((l,'/sitemap'))
-            sm[l].set_article(self.gen_menu(l,None,"tree sitemap"))
-            sm[l].template(self)
         t6 = time.time()
-        print "Sitemap [%5.2f s]" % (round(t6-t5,2))
+        res = set()
+        cwd = os.getcwd()
+        for link in self._tree:
+            res = res.union(link.resources())
+        for f in res:
+            outfile = tmptarget+f
+            mkdir_p(os.path.dirname(outfile))
+            shutil.copyfile(f,outfile)
+        print "Resources[%5.2f s]" % (round(t6-t5,2))
+        sitmaplink = Link('/sitemap')
+        for l in self._sitelang:
+            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)
+        t7 = time.time()
+        print "Sitemap  [%5.2f s]" % (round(t7-t6,2))
 
     def graph(self):
         self._tree.graph()
@@ -346,6 +386,21 @@ class Sitemap():
     def gen_menu(self,lang,page,cssclass):
         return self._tree.menu(lang,page,cssclass)
 
+    def lang_menu(self,lang,link):
+        html = "<ul>"
+        for l in link.languages():
+            isoxml = u"//iso_639_3_entry[@*='"+l+"']"
+            ln = self._isocode.xml_select(isoxml)[0].name
+            if lang != 'en':
+                ln = self._tranlang[lang].gettext(ln)
+            p = link.link()
+            if p[-1] == '/':
+                p = p +'index'
+            p = p+'.'+l
+            html += '<li><a href="%s" hreflang="%s">%s</a></li>' % (p, l, ln)
+        html += "</ul>"
+        return html
+
     def publish(self):
         publish(tmptarget, args.output)
         publish(args.style+"css", args.output)
@@ -542,11 +597,11 @@ sitemap.read_map()
 missing = dir_.set() - sitemap.set()
 removed = sitemap.set() - dir_.set()
 for page in removed:
-    print removed+' pages missing!!'
+    print page+' pages missing!!'
 for page in missing:
     print 'adding missing page '+page
     sitemap.add_link(page)
-if len(missing & removed) != 0:
+if len(missing)+len(removed) != 0:
     print 'writing new sitemap - please adjust if needed'
     sitemap.write_map()
 sitemap.graph()