From f5f054c43039bc59e02930913464eec958232edd Mon Sep 17 00:00:00 2001
From: ssubraveti <ssubraveti@ancestry.com>
Date: Wed, 23 Sep 2020 10:21:04 -0700
Subject: [PATCH] Fix case where multiple spans are covered by a single entity

---
 tools/anntoconll.py | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/tools/anntoconll.py b/tools/anntoconll.py
index adcd334c3..d74153a1c 100644
--- a/tools/anntoconll.py
+++ b/tools/anntoconll.py
@@ -262,10 +262,11 @@ def parse_textbounds(f):
             continue
 
         id_, type_offsets, text = l.split('\t')
-        type_, start, end = type_offsets.split()
-        start, end = int(start), int(end)
-
-        textbounds.append(Textbound(start, end, type_, text))
+        offsets = ' '.join(type_offsets.split()[1:]).split(';')
+        for offset in offsets:
+            start = int(offset.split()[0])
+            end = int(offset.split()[1])
+            textbounds.append(Textbound(start, end, type_, text))
 
     return textbounds