censor student files more

[librarian.git] / librarian / xmlutils.py

diff --git a/librarian/xmlutils.py b/librarian/xmlutils.py
index 523ad8b..bbcc884 100644 (file)
--- a/librarian/xmlutils.py
+++ b/librarian/xmlutils.py
@@ -8,14 +8,39 @@ from collections import defaultdict
 
 
 class Xmill(object):
-    """Transforms XML to some text. 
+    """Transforms XML to some text.
     Used instead of XSLT which is difficult and cumbersome.
-    
+
     """
-    def __init__(self, options=None):
+    def __init__(self, options=None, state=None):
         self._options = []
+        self.state = state or {}
         if options:
             self._options.append(options)
+        self.text_filters = []
+        self.escaped_text_filters = []
+
+    def register_text_filter(self, fun):
+        self.text_filters.append(fun)
+
+    def register_escaped_text_filter(self, fun):
+        self.escaped_text_filters.append(fun)
+
+    def filter_text(self, text):
+        for flt in self.text_filters:
+            if text is None:
+                return None
+            else:
+                text = flt(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.escaped_text_filters:
+            output = flt(output)
+        return output
+
 
     def generate(self, document):
         """Generate text from node using handlers defined in class."""
@@ -26,9 +51,9 @@ class Xmill(object):
     def options(self):
         """Returnes merged scoped options for current node.
         """
-        # Here we can see how a decision not to return the modified map 
+        # 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):
@@ -42,8 +67,6 @@ class Xmill(object):
         tagname = None
 #        from nose.tools import set_trace
 
-        if isinstance(element, etree._Comment): return None
-
         if element.tag[0] == '{':
             for nshort, nhref in element.nsmap.items():
                 try:
@@ -54,8 +77,8 @@ class Xmill(object):
                 except ValueError:
                     pass
             if not ns:
-                raise ValueError("Strange ns for tag: %s, nsmap: %s" % 
-                                 (element.tag, element.nsmap)) 
+                raise ValueError("Strange ns for tag: %s, nsmap: %s" %
+                                 (element.tag, element.nsmap))
         else:
             tagname = element.tag
 
@@ -63,7 +86,7 @@ class Xmill(object):
             meth_name = "handle_%s__%s" % (ns, tagname)
         else:
             meth_name = "handle_%s" % (tagname,)
-        
+
         handler = getattr(self, meth_name, None)
         return handler
 
@@ -78,48 +101,62 @@ class Xmill(object):
             if element is None: return None  # end of tree
 
     def _handle_element(self, element):
+        if isinstance(element, etree._Comment): return None
+        
         handler = self._handle_for_element(element)
+        if self.state.get('mute') and not getattr(handler, 'unmuter', False): return None
         # How many scopes
         try:
             options_scopes = len(self._options)
 
             if handler is None:
-                pre = [element.text]
-                post = []
+                pre = [self.filter_text(element.text)]
+                post = [self.filter_text(element.tail)]
             else:
                 vals = handler(element)
                 # depending on number of returned values, vals can be None, a value, or a tuple.
                 # how poorly designed is that? 9 lines below are needed just to unpack this.
                 if vals is None:
-                    return []
+                    return [self.filter_text(element.tail)]
                 else:
                     if not isinstance(vals, tuple):
-                        pre = [vals]
-                        post = []
+                        return [vals, self.filter_text(element.tail)]
                     else:
-                        pre = [vals[0], element.text]
-                        post = [vals[1]]
-
-            if element.tail:
-                post.append(element.tail)
+                        pre = [vals[0], self.filter_text(element.text)]
+                        post = [vals[1], self.filter_text(element.tail)]
 
             out = pre + [self._handle_element(child) for child in element] + post
         finally:
             # clean up option scopes if necessary
             self._options = self._options[0:options_scopes]
+            
         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