Add Spacy utils

orangecontrib/text/corpus.py

@@ -406,6 +406,10 @@ def has_tokens(self):
         """ Return whether corpus is preprocessed or not. """
         return self._tokens is not None
 
+    def has_tags(self):
+        """ Return whether corpus is POS tagged or not. """
+        return self._pos_tags is not None
+
     def _base_tokens(self):
         from orangecontrib.text.preprocess import BASE_TRANSFORMER, \
             BASE_TOKENIZER, PreprocessorList

orangecontrib/text/language.py

@@ -101,6 +101,8 @@
     "vi": "Vietnamese",
     "zh": "Chinese",
     "zh_char": "Chinese - Chinese Characters",
+    # Spacy code for multi-language model
+    "xx": "Multi-language",
     None: None,
 }
 LANG2ISO = {lang: code for code, lang in ISO2LANG.items()}

requirements.txt

@@ -25,6 +25,7 @@ serverfiles
 simhash >=1.11
 shapely >=2.0
 six
+spacy
 tweepy >=4.0.0
 ufal.udpipe >=1.2.0.3
 trimesh >=3.9.8  # required by alphashape