debug

alSentenceSplitting.py

@@ -90,7 +90,7 @@ def altok_split_sentences(s, lang, weak_sbound = 0, less_lists = 0, noxml = 0, a
 			s = re.sub(r" ([：；:;]|)( +)", r" \1</s>\2<s>", s)
 	elif re.search(r"^km(_|$)", lang):
 		s = re.sub(r" ([។៕])( +)", r" \1</s>\2<s>", s)
-	elif re.search(r"^th(_|$)", lang)::
+	elif re.search(r"^th(_|$)", lang):
 		pass
 	else:
 		s = re.sub(rf"([…\.:;\?\!])(  +)([\"“”\˝] {maj.pattern}[^\"“”\˝<>]*[\.:;\?\!] [\"“”\˝])(  +)({maj.pattern})", r"\1</s>\2<s>\3</s>\4<s>\5", s) # detection of sentences entirely surrounded by double quotes