newpar_block (used in the newest GUM) should not be treated as newpar

martinpopel · martinpopel · commit ee89d8fbd37d · 2022-02-11T16:27:33.000+01:00
diff --git a/udapi/block/read/conllu.py b/udapi/block/read/conllu.py
@@ -12,7 +12,7 @@
 # This reader accepts also older-style sent_id (until UD v2.0 treebanks are released).
 RE_SENT_ID = re.compile(r'^# sent_id\s*=?\s*(\S+)')
 RE_TEXT = re.compile(r'^# text\s*=\s*(.*)')
-RE_NEWPARDOC = re.compile(r'^# (newpar|newdoc)(?:\s+id\s*=\s*(.+))?')
+RE_NEWPARDOC = re.compile(r'^# (newpar|newdoc)(?:\s+id\s*=\s*(.+))?$')
 RE_JSON = re.compile(r'^# (doc_)?json_([^ =]+)\s*=\s*(.+)')
 RE_GLOBAL_ENTITY = re.compile(r'^# global.Entity\s*=\s*(\S+)')
 
