Workaround for nasty literal HTML inserting issue.

[librarian.git] / librarian / xmlutils.py

diff --git a/librarian/xmlutils.py b/librarian/xmlutils.py
index 37a719b..8890f4e 100644 (file)
--- a/librarian/xmlutils.py
+++ b/librarian/xmlutils.py
@@ -22,11 +22,15 @@ class Xmill(object):
         self.text_filters.append(fun)
 
     def filter_text(self, text):
+        # TODO: just work on the tree and let lxml handle escaping.
+        e = etree.Element("x")
+        e.text = text
+        # This whole mixing text with ML is so wrong.
+        output = etree.tostring(e, encoding=unicode)[3:-4]
         for flt in self.text_filters:
-            if text is None:
-                return None
-            text = flt(text)
-        return text
+            output = flt(output)
+        return output
+
 
     def generate(self, document):
         """Generate text from node using handlers defined in class."""
@@ -39,7 +43,7 @@ class Xmill(object):
         """
         # Here we can see how a decision not to return the modified map
         # leads to a need for a hack.
-        return reduce(lambda a, b: a.update(b) or a, self._options, defaultdict(lambda: False))
+        return reduce(lambda a, b: a.update(b) or a, self._options, defaultdict(lambda: None))
 
     @options.setter
     def options(self, opts):
@@ -117,16 +121,30 @@ class Xmill(object):
         return out
 
 
-def tag(name, classes=None, **attrs):
+def tag_open_close(name_, classes_=None, **attrs):
+    u"""Creates tag beginning and end.
+    
+    >>> tag_open_close("a", "klass", x=u"ą<")
+    (u'<a x="\\u0105&lt;" class="klass">', u'</a>')
+
+    """
+    if classes_:
+        if isinstance(classes_, (tuple, list)): classes_ = ' '.join(classes_)
+        attrs['class'] = classes_
+
+    e = etree.Element(name_)
+    e.text = " "
+    for k, v in attrs.items():
+        e.attrib[k] = v
+    pre, post = etree.tostring(e, encoding=unicode).split(u"> <")
+    return pre + u">", u"<" + post
+
+def tag(name_, classes_=None, **attrs):
     """Returns a handler which wraps node contents in tag `name', with class attribute
     set to `classes' and other attributes according to keyword paramters
     """
-    if classes:
-        if isinstance(classes, (tuple, list)): classes = ' '.join(classes)
-        attrs['class'] = classes
-    a = ''.join([' %s="%s"' % (k,v) for (k,v) in attrs.items()])
     def _hnd(self, element):
-        return "<%s%s>" % (name, a), "</%s>" % name
+        return tag_open_close(name_, classes_, **attrs)
     return _hnd