patch minor UD bugs

batchalign/document.py

@@ -193,9 +193,12 @@ def __len__(self):
 
     def __str__(self):
         if self.text != None:
-            return self.text
+            t = self.text
         else:
-            return self._detokenize()
+            t = self._detokenize()
+
+        t = t.replace(". . .", "+...")
+        return t
 
     def __repr__(self):
         return str(self)

batchalign/pipelines/asr/utils.py

@@ -188,7 +188,7 @@ def process_generation(output, lang="eng", utterance_engine=None):
 
         final_utterances.append(Utterance(
             tier=participant,
-            content = words
+            content=words
         ))
 
     doc.content = final_utterances

batchalign/pipelines/morphosyntax/ud.py

@@ -848,7 +848,10 @@ def morphoanalyze(doc: Document, retokenize:bool, status_hook:callable = None, *
                                     for j in i]
 
                 retokenized_ut = " ".join(i for i in chunks_backplate if i.strip() not in ["(", ")"])
+                retokenized_ut = retokenized_ut.replace("^", "")
                 retokenized_ut = re.sub(r" +", " ", retokenized_ut)
+                retokenized_ut = retokenized_ut.replace("+ \"", "+\"")
+                retokenized_ut = retokenized_ut.replace(" >", ">")
                 # pray to everyone that it works---this will simply crash and ignore
                 # the utterance if it didn't work, so we are doing this as a sanity
                 # check rather than needing the parsed result

batchalign/version

@@ -1,3 +1,3 @@
-0.7.1-beta.9
+0.7.1-beta.10
 May 21st, 2024
-minute %umor implementation changes
+patch minor ud bugs