Updated processing of Old Czech data.

Author: dan-zeman

udapi/block/ud/cs/addmwt.py

@@ -126,7 +126,7 @@ def multiword_analysis(self, node):
             # could be masculine or neuter. We pick Gender=Masc and Animacy=Anim
             # by default, unless the original token was annotated as Animacy=Inan
             # or Gender=Neut.
-            m = re.match(r"^(na|o|pro|přěde|ski?rz[eě]|za)[nň](ž?)$", node.form.lower())
+            m = re.match(r"^(na|nade|o|pro|přěde|ski?rz[eě]|za)[nň](ž?)$", node.form.lower())
             if m:
                 node.misc['AddMwt'] = ''
                 # Remove vocalization from 'přěde' (přěd něj) but keep it in 'skrze'

udapi/block/ud/cs/markfeatsbugs.py

@@ -37,7 +37,8 @@ def process_node(self, node):
                     'Gender': ['Neut'],
                     'Number': ['Sing', 'Dual', 'Plur'],
                     'Case': ['Nom', 'Gen', 'Dat', 'Acc', 'Voc', 'Loc', 'Ins'],
-                    'Foreign': ['Yes']
+                    'Foreign': ['Yes'],
+                    'Abbr': ['Yes']
                 })
             elif node.feats['Gender'] == 'Masc':
                 self.check_required_features(node, ['Animacy'])
@@ -46,13 +47,15 @@ def process_node(self, node):
                     'Animacy': ['Anim', 'Inan'],
                     'Number': ['Sing', 'Dual', 'Plur'],
                     'Case': ['Nom', 'Gen', 'Dat', 'Acc', 'Voc', 'Loc', 'Ins'],
-                    'Foreign': ['Yes']})
+                    'Foreign': ['Yes'],
+                    'Abbr': ['Yes']})
             else:
                 self.check_allowed_features(node, {
                     'Gender': ['Masc', 'Fem', 'Neut'],
                     'Number': ['Sing', 'Dual', 'Plur'],
                     'Case': ['Nom', 'Gen', 'Dat', 'Acc', 'Voc', 'Loc', 'Ins'],
-                    'Foreign': ['Yes']})
+                    'Foreign': ['Yes'],
+                    'Abbr': ['Yes']})
         # PROPER NOUNS #########################################################
         elif node.upos == 'PROPN':
             self.check_required_features(node, ['Gender', 'Number', 'Case'])
@@ -64,14 +67,16 @@ def process_node(self, node):
                     'Number': ['Sing', 'Dual', 'Plur'],
                     'Case': ['Nom', 'Gen', 'Dat', 'Acc', 'Voc', 'Loc', 'Ins'],
                     'NameType': ['Giv', 'Sur', 'Geo', 'Nat'],
-                    'Foreign': ['Yes']})
+                    'Foreign': ['Yes'],
+                    'Abbr': ['Yes']})
             else:
                 self.check_allowed_features(node, {
                     'Gender': ['Masc', 'Fem', 'Neut'],
                     'Number': ['Sing', 'Dual', 'Plur'],
                     'Case': ['Nom', 'Gen', 'Dat', 'Acc', 'Voc', 'Loc', 'Ins'],
                     'NameType': ['Giv', 'Sur', 'Geo', 'Nat'],
-                    'Foreign': ['Yes']})
+                    'Foreign': ['Yes'],
+                    'Abbr': ['Yes']})
         # ADJECTIVES ###########################################################
         elif node.upos == 'ADJ':
             if node.feats['Poss'] == 'Yes': # possessive adjectives

udapi/block/ud/fixadvmodbyupos.py

@@ -51,6 +51,8 @@ def process_node(self, node):
         elif node.udeprel == 'mark':
             if node.upos in ['PRON', 'DET']:
                 node.deprel = 'nsubj' # it could be also obj, iobj, obl or nmod; just guessing what might be more probable
+            elif node.upos == 'NOUN':
+                node.deprel = 'obl'
             elif node.upos == 'INTJ':
                 node.deprel = 'discourse'
         elif node.udeprel == 'cc':
@@ -71,6 +73,8 @@ def process_node(self, node):
                 node.deprel = 'aux'
             elif node.upos == 'VERB':
                 node.deprel = 'dep'
+            elif node.upos == 'SCONJ':
+                node.deprel = 'mark'
             elif node.upos == 'X':
                 node.deprel = 'dep'
         elif node.udeprel == 'nummod':