From 7b37de43371a4df661bc904d34dd2d8a24009d71 Mon Sep 17 00:00:00 2001
From: Chris Little <chrislit@crosswire.org>
Date: Mon, 1 Jul 2013 07:14:12 +0000
Subject: made unhandled tag detection much more lax & greedy, mostly to better
 indicate encoding errors in source docs

git-svn-id: https://www.crosswire.org/svn/sword-tools/trunk@426 07627401-56e2-0310-80f4-f8cd0041bdcd
---
 modules/python/usfm2osis.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'modules/python')

diff --git a/modules/python/usfm2osis.py b/modules/python/usfm2osis.py
index 2d00ef5..e5e499f 100755
--- a/modules/python/usfm2osis.py
+++ b/modules/python/usfm2osis.py
@@ -1350,7 +1350,7 @@ def convertToOsis(sFile):
         osis = osis.replace('<div type="book" osisID="' + sb + '">', '<div type="' + sb.lower() + '">')
 
     if DEBUG:
-        localUnhandledTags = set(re.findall(r'(\\[^\s\*]+?\b\*?)', osis))
+        localUnhandledTags = set(re.findall(r'(\\[^\s]*)', osis))
         if localUnhandledTags:
             print(('Unhandled USFM tags in ' + sFile + ': ' + ', '.join(localUnhandledTags) + ' (' + str(len(localUnhandledTags)) + ' total)'))
 
@@ -1573,7 +1573,7 @@ if __name__ == "__main__":
 
         unhandledTags = set()
         for doc in usfmDocList:
-            unhandledTags |= set(re.findall(r'(\\[^\s\*]+?\b\*?)', osisSegment[doc]))
+            unhandledTags |= set(re.findall(r'(\\[^\s]*)', osisSegment[doc]))
             osisDoc += osisSegment[doc]
 
         osisDoc += '</osisText>\n</osis>\n'
-- 
cgit