retrace bug fix

TalkBank · Apr 20, 2024 · c390b42 · c390b42
1 parent 179555b
commit c390b42
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 7 deletions.
diff --git a/batchalign/document.py b/batchalign/document.py
@@ -237,7 +237,7 @@ def _detokenize(self):
                             self.content[indx+2].text != i.text and
                             self.content[indx+2].type == TokenType.RETRACE):
                             result.append("<")
-                elif self.content[indx+1].type == TokenType.REGULAR:
+                elif self.content[indx+1].type != TokenType.RETRACE:
                     if indx > 0 and self.content[indx-1].type == TokenType.RETRACE and self.content[indx-1].text != i.text:
                         result.append(i.text+">")
                         result.append("[/]")

diff --git a/batchalign/version b/batchalign/version
@@ -1,3 +1,3 @@
-0.7.1-beta.4
+0.7.1-beta.5
 Apr 20th, 2024
-more alignment bug patches
+minor retrace bug fix
diff --git a/scratchpad.py b/scratchpad.py
@@ -20,11 +20,15 @@
 ########### The Batchalign Core Test Harness ###########
 # from batchalign.formats.chat.parser import chat_parse_utterance
 
-sue = StanzaUtteranceEngine()
-doc = Document.new("john is a chicken and robert is a chicken", lang="eng")
+# ng = NgramRetraceEngine()
+# disf = DisfluencyReplacementEngine()
+# doc = Document.new("um I'm seeing I'm seeing light and dark to create uh uh to create uh to create uh time", lang="eng")
+# pipe = BatchalignPipeline(ng, disf)
+# tmp = pipe(doc)
+# tmp
+
 
-tmp = sue(doc)
-tmp
+# tmp[0].content
 
 
 # import stanza