X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/fe8e5b5e224d32baebbdaa2fecf4a847ed4e5354..3b0b98465bc1862306b05bb8305a1abbf40ca310:/librarian/xmlutils.py

diff --git a/librarian/xmlutils.py b/librarian/xmlutils.py
index d762320..ae3512a 100644
--- a/librarian/xmlutils.py
+++ b/librarian/xmlutils.py
@@ -12,21 +12,34 @@ class Xmill(object):
     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
-            text = flt(text)
-        return text
+            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."""
@@ -39,7 +52,7 @@ 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):
@@ -47,20 +60,17 @@ class Xmill(object):
         """
         self._options.append(opts)
 
-
     def _handle_for_element(self, element):
         ns = None
         tagname = None
-#        from nose.tools import set_trace
-
-        if isinstance(element, etree._Comment): return None
+        # from nose.tools import set_trace
 
         if element.tag[0] == '{':
             for nshort, nhref in element.nsmap.items():
                 try:
                     if element.tag.index('{%s}' % nhref) == 0:
                         ns = nshort
-                        tagname  = element.tag[len('{%s}' % nhref):]
+                        tagname = element.tag[len('{%s}' % nhref):]
                         break
                 except ValueError:
                     pass
@@ -84,25 +94,31 @@ class Xmill(object):
 
         while True:
             sibling = element.getnext()
-            if sibling is not None: return sibling  # found a new branch to dig into
+            if sibling is not None:
+                return sibling  # found a new branch to dig into
             element = element.getparent()
-            if element is None: return None  # end of tree
+            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
+        options_scopes = len(self._options)
         try:
-            options_scopes = len(self._options)
-
             if handler is None:
                 pre = [self.filter_text(element.text)]
-                post = []
+                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):
                         return [vals, self.filter_text(element.tail)]
@@ -114,19 +130,36 @@ class Xmill(object):
         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
 
 
@@ -135,13 +168,16 @@ def tagged(name, classes=None, **attrs):
     set to `classes' and other attributes according to keyword paramters
     """
     if classes:
-        if isinstance(classes, (tuple,list)): classes = ' '.join(classes)
+        if isinstance(classes, (tuple, list)):
+            classes = ' '.join(classes)
         attrs['class'] = classes
-    a = ''.join([' %s="%s"' % (k,v) for (k,v) in attrs.items()])
+    a = ''.join([' %s="%s"' % (k, v) for (k, v) in attrs.items()])
+
     def _decor(f):
         def _wrap(self, element):
             r = f(self, element)
-            if r is None: return
+            if r is None:
+                return
 
             prepend = "<%s%s>" % (name, a)
             append = "</%s>" % name
@@ -166,6 +202,7 @@ def ifoption(**options):
         return _handler
     return _decor
 
+
 def flatten(l, ltypes=(list, tuple)):
     """flatten function from BasicPropery/BasicTypes package
     """