implement mwt.feats

fix #132

Author: martinpopel

udapi/block/read/conllu.py

@@ -223,6 +223,6 @@ def read_tree_from_lines(self, lines):
                 logging.warning(f"Wrong MWT range in\n{fields[0]}\n\n{lines}")
                 raise
             words = nodes[int(range_start):int(range_end) + 1]
-            root.create_multiword_token(words, form=fields[1], misc=fields[-1])
+            root.create_multiword_token(words, form=fields[1], feats=fields[5], misc=fields[9])
 
         return root

udapi/block/write/conllu.py

@@ -117,7 +117,9 @@ def process_tree(self, tree):  # pylint: disable=too-many-branches
             if mwt and node._ord > last_mwt_id:
                 print('\t'.join((mwt.ord_range,
                                  '_' if mwt.form is None else mwt.form,
-                                 '_\t_\t_\t_\t_\t_\t_',
+                                 '_\t_\t_',
+                                 '_' if mwt._feats is None else str(mwt.feats),
+                                 '_\t_\t_',
                                  '_' if mwt._misc is None else str(mwt.misc))))
                 last_mwt_id = mwt.words[-1]._ord
 

udapi/core/mwt.py

@@ -1,19 +1,38 @@
 """MWT class represents a multi-word token."""
 from udapi.core.dualdict import DualDict
-
+from udapi.core.feats import Feats
 
 class MWT(object):
     """Class for representing multi-word tokens in UD trees."""
-    __slots__ = ['words', 'form', '_misc', 'root']
+    __slots__ = ['words', 'form', '_feats', '_misc', 'root']
 
-    def __init__(self, words=None, form=None, misc=None, root=None):
+    def __init__(self, words=None, form=None, feats=None, misc=None, root=None):
         self.words = words if words is not None else []
         self.form = form
+        self._feats = Feats(feats) if feats and feats != '_' else None
         self._misc = DualDict(misc) if misc and misc != '_' else None
         self.root = root
         for word in self.words:
             word._mwt = self  # pylint: disable=W0212
 
+    @property
+    def feats(self):
+        """Property `feats` in MWT should be used only for `Typo=Yes`.
+
+        See https://universaldependencies.org/changes.html#typos-in-multiword-tokens
+        However, Udapi does not enforce this restriction and mwt.feats works exactly the same as node.feats.
+        """
+        if self._feats is None:
+            self._feats = Feats()
+        return self._feats
+
+    @feats.setter
+    def feats(self, value):
+        if self._feats is None:
+            self._feats = Feats(value)
+        else:
+            self._feats.set_mapping(value)
+
     @property
     def misc(self):
         """Property for MISC attributes stored as a `DualDict` object.

udapi/core/root.py

@@ -167,22 +167,23 @@ def create_empty_child(self, **kwargs):
         return new_node
 
     # TODO document whether misc is a string or dict or it can be both
-    def create_multiword_token(self, words=None, form=None, misc=None):
+    def create_multiword_token(self, words=None, form=None, feats=None, misc=None):
         """Create and return a new multi-word token (MWT) in this tree.
 
         The new MWT can be optionally initialized using the following args.
         Args:
         words: a list of nodes which are part of the new MWT
         form: string representing the surface form of the new MWT
-        misc: misc attribute of the new MWT
+        misc: FEATS attribute of the new MWT (only `Typo=Yes` allowed there in UD guidelines)
+        misc: MISC attribute of the new MWT
         """
         # Nested or overlapping MWTs are not allowed in CoNLL-U,
         # so first remove all previous MWTs containing any of words.
         for w in words:
             if w.multiword_token:
                 w.multiword_token.remove()
         # Now, create the new MWT.
-        mwt = MWT(words, form, misc, root=self)
+        mwt = MWT(words, form, feats, misc, root=self)
         self._mwts.append(mwt)
         if words[-1].misc["SpaceAfter"] == "No":
             mwt.misc["SpaceAfter"] = "No"