Handle subtitles which contain Pango markup.

3 files changed, 33 insertions, 1 deletions

diff --git a/src/lib/subtitle_decoder.cc b/src/lib/subtitle_decoder.cc
index ba6fe4600..bc4a75ca8 100644
--- a/src/lib/subtitle_decoder.cc
+++ b/src/lib/subtitle_decoder.cc
@@ -24,6 +24,7 @@
 #include <sub/subtitle.h>
 #include <boost/shared_ptr.hpp>
 #include <boost/foreach.hpp>
+#include <boost/algorithm/string.hpp>
 #include <iostream>
 
 using std::list;
@@ -64,6 +65,16 @@ SubtitleDecoder::give_image (ContentTimePeriod period, shared_ptr<Image> image,
 void
 SubtitleDecoder::give_text (ContentTimePeriod period, list<dcp::SubtitleString> s)
 {
+	/* We must escape < and > in strings, otherwise they might confuse our subtitle
+	   renderer (which uses some HTML-esque markup to do bold/italic etc.)
+	*/
+	BOOST_FOREACH (dcp::SubtitleString& i, s) {
+		string t = i.text ();
+		boost::algorithm::replace_all (t, "<", "&lt;");
+		boost::algorithm::replace_all (t, ">", "&gt;");
+		i.set_text (t);
+	}
+
 	_decoded_text.push_back (ContentTextSubtitle (period, s));
 }
 
diff --git a/test/data b/test/data
-Subproject c3555ace037d5209c9a7e3a4424180fdf8b4d9b
+Subproject a6a8263ed232d8a9a2dc478b073f835d256d05b
diff --git a/test/dcp_subtitle_test.cc b/test/dcp_subtitle_test.cc
index 27dd87bbf..34059a680 100644
--- a/test/dcp_subtitle_test.cc
+++ b/test/dcp_subtitle_test.cc
@@ -1,5 +1,5 @@
 /*
-    Copyright (C) 2014 Carl Hetherington <cth@carlh.net>
+    Copyright (C) 2014-2016 Carl Hetherington <cth@carlh.net>
 
     This file is part of DCP-o-matic.
 
@@ -29,6 +29,7 @@
 #include "lib/ratio.h"
 #include "lib/dcp_decoder.h"
 #include "lib/dcp_content_type.h"
+#include "lib/dcp_subtitle_decoder.h"
 #include "lib/subtitle_content.h"
 #include "lib/content_subtitle.h"
 #include "lib/subtitle_decoder.h"
@@ -101,3 +102,23 @@ BOOST_AUTO_TEST_CASE (dcp_subtitle_within_dcp_test)
 	BOOST_CHECK_EQUAL (subs.front().subs.front().text(), "Noch mal.");
 	BOOST_CHECK_EQUAL (subs.front().subs.back().text(), "Encore une fois.");
 }
+
+/** Test subtitles whose text includes things like &lt;b&gt; */
+BOOST_AUTO_TEST_CASE (dcp_subtitle_test2)
+{
+	shared_ptr<Film> film = new_test_film ("dcp_subtitle_test2");
+	film->set_container (Ratio::from_id ("185"));
+	film->set_dcp_content_type (DCPContentType::from_isdcf_name ("TLR"));
+	film->set_name ("frobozz");
+	shared_ptr<DCPSubtitleContent> content (new DCPSubtitleContent (film, "test/data/dcp_sub2.xml"));
+	film->examine_and_add_content (content);
+	wait_for_jobs ();
+
+	shared_ptr<DCPSubtitleDecoder> decoder (new DCPSubtitleDecoder (content));
+	list<ContentTextSubtitle> sub = decoder->subtitle->get_text (
+		ContentTimePeriod (ContentTime::from_seconds(0), ContentTime::from_seconds(2)), true, true
+		);
+	BOOST_REQUIRE_EQUAL (sub.size(), 1);
+	BOOST_REQUIRE_EQUAL (sub.front().subs.size(), 1);
+	BOOST_CHECK_EQUAL (sub.front().subs.front().text(), "&lt;b&gt;Hello world!&lt;/b&gt;");
+}