fixed bug with could_be formats, and cleaned up value parsing

microtron/__init__.py

@@ -71,7 +71,7 @@ def _parse_node(self, node, format):
                                 if '__type__' in value:
                                     value['__type__'] += ' ' + format_results[0].pop('__type__')
 
-                                value.update(results[0])
+                                value.update(format_results[0])
 
                         except:
                             pass
@@ -157,18 +157,17 @@ def _parse_node(self, node, format):
         return result
 
     def _parse_value(self, node):
-        text_expr = 'normalize-space(string(.))'
         value_expr = 'descendant::*[contains(concat(" ", normalize-space(@class), " "), " value ")]'
         value_nodes = node.xpath(value_expr)
 
         if value_nodes:
-            return self._parse_text(" ".join(value_node.xpath(text_expr) for value_node in value_nodes))
+            return " ".join(self._parse_text(value_node) for value_node in value_nodes)
 
         elif node.tag == 'abbr' and 'title' in node.attrib:
             return node.attrib['title']
 
         else:
-            return node.xpath(text_expr)
+            return self._parse_text(node)
 
     def _parse_text(self, node):
         text_expr = 'normalize-space(string(.))'

parse.py

@@ -17,6 +17,7 @@ def parse(argv = None):
     options, arguments = parser.parse_args(argv[1:])
     if len(arguments) != 3:
         parser.error('Incorrect number of arguments')
+
     source_filename = os.path.abspath(arguments[0])
     formats_filename = os.path.abspath(arguments[1])
     format = arguments[2]