Strip Unicode U+202B (right-to-left-embedding) code; it looks like DoM does RTL (at least) partially correctly without this.

1 files changed, 6 insertions, 0 deletions

diff --git a/src/subrip_reader.cc b/src/subrip_reader.cc
index 02ee20a..b5d0446 100644
--- a/src/subrip_reader.cc
+++ b/src/subrip_reader.cc
@@ -42,6 +42,7 @@ using boost::lexical_cast;
 using boost::to_upper;
 using boost::optional;
 using boost::function;
+using boost::algorithm::replace_all;
 using namespace sub;
 
 /** @param s Subtitle string encoded in UTF-8 */
@@ -233,6 +234,11 @@ SubripReader::convert_line (string t, RawSubtitle& p)
 		}
 	}
 
+	/* Strip Unicode U+202B (right-to-left embedding) as sometimes it is rendered
+	   as a missing character.  This may be a hack.
+	*/
+	replace_all (p.text, "\xe2\x80\xab", "");
+
 	maybe_content (p);
 }