From 0d7a7b85095a5accf5f3666bc1d0a037a5455d0b Mon Sep 17 00:00:00 2001
From: Carl Hetherington <cth@carlh.net>
Date: Wed, 9 Dec 2020 10:21:57 +0100
Subject: [PATCH 1/1]  Bv2.1 6.2.1: Check that subtitle XML <Language> conforms
 to RFC 5646.

---
 BRANCH                     |  2 +-
 src/smpte_subtitle_asset.h |  3 +++
 src/verify.cc              | 14 +++++++++++---
 test/verify_test.cc        | 13 ++++++++++---
 4 files changed, 25 insertions(+), 7 deletions(-)

diff --git a/BRANCH b/BRANCH
index 43445112..5413cfb4 100644
--- a/BRANCH
+++ b/BRANCH
@@ -9,7 +9,7 @@ Mark things with [Bv2.1_paragraph]
     immersive audio DCPs must comply with 429-18, 429-19 [*] don't have these standards
 6.2 Language and Territory must comply with RFC 5646
     - subtitle reel <Language> /
-    - subtitle XML <Language>
+    - subtitle XML <Language> /
     - sound reel <Language>
     - ccap reel <Language>
     - MainSubtitleLanguageList
diff --git a/src/smpte_subtitle_asset.h b/src/smpte_subtitle_asset.h
index b66b1ea6..28aecefa 100644
--- a/src/smpte_subtitle_asset.h
+++ b/src/smpte_subtitle_asset.h
@@ -48,6 +48,8 @@ namespace ASDCP {
 	}
 }
 
+struct verify_test26;
+
 namespace dcp {
 
 class SMPTELoadFontNode;
@@ -172,6 +174,7 @@ protected:
 private:
 	friend struct ::write_smpte_subtitle_test;
 	friend struct ::write_smpte_subtitle_test2;
+	friend struct ::verify_test26;
 
 	void read_fonts (std::shared_ptr<ASDCP::TimedText::MXFReader>);
 	void parse_xml (std::shared_ptr<cxml::Document> xml);
diff --git a/src/verify.cc b/src/verify.cc
index b4408946..89d84eef 100644
--- a/src/verify.cc
+++ b/src/verify.cc
@@ -46,6 +46,7 @@
 #include "exceptions.h"
 #include "compose.hpp"
 #include "raw_convert.h"
+#include "smpte_subtitle_asset.h"
 #include <xercesc/util/PlatformUtils.hpp>
 #include <xercesc/parsers/XercesDOMParser.hpp>
 #include <xercesc/parsers/AbstractDOMParser.hpp>
@@ -572,12 +573,19 @@ verify_main_subtitle_asset (
 	list<VerificationNote>& notes
 	)
 {
-	shared_ptr<ReelSubtitleAsset> reel_asset = reel->main_subtitle ();
-	stage ("Checking subtitle XML", reel->main_subtitle()->asset()->file());
+	shared_ptr<SubtitleAsset> asset = reel->main_subtitle()->asset();
+	stage ("Checking subtitle XML", asset->file());
 	/* Note: we must not use SubtitleAsset::xml_as_string() here as that will mean the data on disk
 	 * gets passed through libdcp which may clean up and therefore hide errors.
 	 */
-	validate_xml (reel->main_subtitle()->asset()->raw_xml(), xsd_dtd_directory, notes);
+	validate_xml (asset->raw_xml(), xsd_dtd_directory, notes);
+
+	shared_ptr<SMPTESubtitleAsset> smpte = dynamic_pointer_cast<SMPTESubtitleAsset>(asset);
+	if (smpte) {
+		if (smpte->language()) {
+			verify_language_tag (*smpte->language(), notes);
+		}
+	}
 }
 
 
diff --git a/test/verify_test.cc b/test/verify_test.cc
index 3e983e4d..62b5c6c4 100644
--- a/test/verify_test.cc
+++ b/test/verify_test.cc
@@ -874,13 +874,15 @@ BOOST_AUTO_TEST_CASE (verify_test25)
 }
 
 
-/* SMPTE DCP with invalid <Language> in the MainSubtitle */
+/* SMPTE DCP with invalid <Language> in the MainSubtitle reel and also in the XML within the MXF */
 BOOST_AUTO_TEST_CASE (verify_test26)
 {
 	boost::filesystem::path const dir("build/test/verify_test26");
 	prepare_directory (dir);
 	boost::filesystem::copy_file ("test/data/subs.mxf", dir / "subs.mxf");
 	shared_ptr<dcp::SMPTESubtitleAsset> asset(new dcp::SMPTESubtitleAsset(dir / "subs.mxf"));
+	asset->_language = "wrong-andbad";
+	asset->write (dir / "subs.mxf");
 	shared_ptr<dcp::ReelSubtitleAsset> reel_asset(new dcp::ReelSubtitleAsset(asset, dcp::Fraction(24, 1), 16 * 24, 0));
 	reel_asset->_language = "badlang";
 	shared_ptr<dcp::Reel> reel(new dcp::Reel());
@@ -894,9 +896,14 @@ BOOST_AUTO_TEST_CASE (verify_test26)
 	vector<boost::filesystem::path> dirs;
 	dirs.push_back (dir);
 	list<dcp::VerificationNote> notes = dcp::verify (dirs, &stage, &progress, xsd_test);
-	BOOST_REQUIRE_EQUAL (notes.size(), 1U);
+	BOOST_REQUIRE_EQUAL (notes.size(), 2U);
 	list<dcp::VerificationNote>::const_iterator i = notes.begin ();
 	BOOST_CHECK_EQUAL (i->code(), dcp::VerificationNote::BAD_LANGUAGE);
-
+	BOOST_REQUIRE (i->note());
+	BOOST_CHECK_EQUAL (*i->note(), "badlang");
+	i++;
+	BOOST_CHECK_EQUAL (i->code(), dcp::VerificationNote::BAD_LANGUAGE);
+	BOOST_REQUIRE (i->note());
+	BOOST_CHECK_EQUAL (*i->note(), "wrong-andbad");
 }
 
-- 
2.30.2