Bv2.1 6.2.1: Check that subtitle XML <Language> conforms to RFC 5646.
[libdcp.git] / src / verify.cc
index e20cafb39496389200a2ef1f1f7d6965e91323da..89d84eef97c46fe82d503ce0bf74f75f730fa272 100644 (file)
@@ -1,5 +1,5 @@
 /*
-    Copyright (C) 2018-2019 Carl Hetherington <cth@carlh.net>
+    Copyright (C) 2018-2020 Carl Hetherington <cth@carlh.net>
 
     This file is part of libdcp.
 
 #include "reel.h"
 #include "reel_picture_asset.h"
 #include "reel_sound_asset.h"
+#include "reel_subtitle_asset.h"
+#include "interop_subtitle_asset.h"
+#include "mono_picture_asset.h"
+#include "mono_picture_frame.h"
+#include "stereo_picture_asset.h"
+#include "stereo_picture_frame.h"
 #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>
 #include <xercesc/dom/DOMAttr.hpp>
 #include <xercesc/dom/DOMErrorHandler.hpp>
 #include <xercesc/framework/LocalFileInputSource.hpp>
+#include <xercesc/framework/MemBufInputSource.hpp>
 #include <boost/noncopyable.hpp>
 #include <boost/foreach.hpp>
 #include <boost/algorithm/string.hpp>
-#include <boost/regex.hpp>
 #include <map>
 #include <list>
 #include <vector>
@@ -71,19 +78,15 @@ using std::vector;
 using std::string;
 using std::cout;
 using std::map;
-using boost::shared_ptr;
+using std::max;
+using std::shared_ptr;
 using boost::optional;
 using boost::function;
+using std::dynamic_pointer_cast;
 
 using namespace dcp;
 using namespace xercesc;
 
-enum Result {
-       RESULT_GOOD,
-       RESULT_CPL_PKL_DIFFER,
-       RESULT_BAD
-};
-
 static
 string
 xml_ch_to_string (XMLCh const * a)
@@ -101,6 +104,8 @@ public:
                : _message (xml_ch_to_string(e.getMessage()))
                , _line (e.getLineNumber())
                , _column (e.getColumnNumber())
+               , _public_id (e.getPublicId() ? xml_ch_to_string(e.getPublicId()) : "")
+               , _system_id (e.getSystemId() ? xml_ch_to_string(e.getSystemId()) : "")
        {
 
        }
@@ -117,10 +122,20 @@ public:
                return _column;
        }
 
+       string public_id () const {
+               return _public_id;
+       }
+
+       string system_id () const {
+               return _system_id;
+       }
+
 private:
        string _message;
        uint64_t _line;
        uint64_t _column;
+       string _public_id;
+       string _system_id;
 };
 
 
@@ -155,15 +170,8 @@ private:
        {
                /* XXX: nasty hack */
                if (
-                       e.message() ==
-                       "schema document '/home/carl/src/libdcp/xsd/xml.xsd' has different target namespace "
-                       "from the one specified in instance document 'http://www.w3.org/2001/03/xml.xsd'" ||
-                       e.message() ==
-                       "schema document '/home/carl/src/libdcp/xsd/xmldsig-core-schema.xsd' has different target namespace "
-                       "from the one specified in instance document 'http://www.w3.org/TR/2002/REC-xmldsig-core-20020212/xmldsig-core-schema.xsd'" ||
-                       e.message() ==
-                       "schema document '/home/carl/src/libdcp/xsd/SMPTE-429-8-2006-PKL.xsd' has different target namespace "
-                       "from the one specified in instance document 'http://www.smpte-ra.org/schemas/429-8/2006/PKL'"
+                       e.message().find("schema document") != string::npos &&
+                       e.message().find("has different target namespace from the one specified in instance document") != string::npos
                        ) {
                        return;
                }
@@ -201,19 +209,36 @@ public:
        LocalFileResolver (boost::filesystem::path xsd_dtd_directory)
                : _xsd_dtd_directory (xsd_dtd_directory)
        {
+               /* XXX: I'm not clear on what things need to be in this list; some XSDs are apparently, magically
+                * found without being here.
+                */
                add("http://www.w3.org/2001/XMLSchema.dtd", "XMLSchema.dtd");
                add("http://www.w3.org/2001/03/xml.xsd", "xml.xsd");
                add("http://www.w3.org/TR/2002/REC-xmldsig-core-20020212/xmldsig-core-schema.xsd", "xmldsig-core-schema.xsd");
+               add("http://www.digicine.com/schemas/437-Y/2007/Main-Stereo-Picture-CPL.xsd", "Main-Stereo-Picture-CPL.xsd");
+               add("http://www.digicine.com/PROTO-ASDCP-CPL-20040511.xsd", "PROTO-ASDCP-CPL-20040511.xsd");
+               add("http://www.digicine.com/PROTO-ASDCP-PKL-20040311.xsd", "PROTO-ASDCP-PKL-20040311.xsd");
+               add("http://www.digicine.com/PROTO-ASDCP-AM-20040311.xsd", "PROTO-ASDCP-AM-20040311.xsd");
+               add("http://www.digicine.com/PROTO-ASDCP-CC-CPL-20070926#", "PROTO-ASDCP-CC-CPL-20070926.xsd");
+               add("interop-subs", "DCSubtitle.v1.mattsson.xsd");
+               add("http://www.smpte-ra.org/schemas/428-7/2010/DCST.xsd", "SMPTE-428-7-2010-DCST.xsd");
+               add("http://www.smpte-ra.org/schemas/429-16/2014/CPL-Metadata", "SMPTE-429-16.xsd");
+               add("http://www.dolby.com/schemas/2012/AD", "Dolby-2012-AD.xsd");
+               add("http://www.smpte-ra.org/schemas/429-10/2008/Main-Stereo-Picture-CPL", "SMPTE-429-10-2008.xsd");
        }
 
        InputSource* resolveEntity(XMLCh const *, XMLCh const * system_id)
        {
+               if (!system_id) {
+                       return 0;
+               }
                string system_id_str = xml_ch_to_string (system_id);
+               boost::filesystem::path p = _xsd_dtd_directory;
                if (_files.find(system_id_str) == _files.end()) {
-                       return 0;
+                       p /= system_id_str;
+               } else {
+                       p /= _files[system_id_str];
                }
-
-               boost::filesystem::path p = _xsd_dtd_directory / _files[system_id_str];
                StringToXMLCh ch (p.string());
                return new LocalFileInputSource(ch.get());
        }
@@ -228,9 +253,25 @@ private:
        boost::filesystem::path _xsd_dtd_directory;
 };
 
-static
-list<XMLValidationError>
-validate_xml (boost::filesystem::path xml_file, boost::filesystem::path xsd_dtd_directory)
+
+static void
+parse (XercesDOMParser& parser, boost::filesystem::path xml)
+{
+       parser.parse(xml.string().c_str());
+}
+
+
+static void
+parse (XercesDOMParser& parser, std::string xml)
+{
+       xercesc::MemBufInputSource buf(reinterpret_cast<unsigned char const*>(xml.c_str()), xml.size(), "");
+       parser.parse(buf);
+}
+
+
+template <class T>
+void
+validate_xml (T xml, boost::filesystem::path xsd_dtd_directory, list<VerificationNote>& notes)
 {
        try {
                XMLPlatformUtils::Initialize ();
@@ -247,19 +288,35 @@ validate_xml (boost::filesystem::path xml_file, boost::filesystem::path xsd_dtd_
                parser.setDoNamespaces(true);
                parser.setDoSchema(true);
 
-               map<string, string> schema;
-               schema["http://www.w3.org/2000/09/xmldsig#"] = "xmldsig-core-schema.xsd";
-               schema["http://www.w3.org/TR/2002/REC-xmldsig-core-20020212/xmldsig-core-schema.xsd"] = "xmldsig-core-schema.xsd";
-               schema["http://www.smpte-ra.org/schemas/429-7/2006/CPL"] = "SMPTE-429-7-2006-CPL.xsd";
-               schema["http://www.smpte-ra.org/schemas/429-8/2006/PKL"] = "SMPTE-429-8-2006-PKL.xsd";
-               schema["http://www.w3.org/2001/03/xml.xsd"] = "xml.xsd";
-
+               vector<string> schema;
+               schema.push_back("xml.xsd");
+               schema.push_back("xmldsig-core-schema.xsd");
+               schema.push_back("SMPTE-429-7-2006-CPL.xsd");
+               schema.push_back("SMPTE-429-8-2006-PKL.xsd");
+               schema.push_back("SMPTE-429-9-2007-AM.xsd");
+               schema.push_back("Main-Stereo-Picture-CPL.xsd");
+               schema.push_back("PROTO-ASDCP-CPL-20040511.xsd");
+               schema.push_back("PROTO-ASDCP-PKL-20040311.xsd");
+               schema.push_back("PROTO-ASDCP-AM-20040311.xsd");
+               schema.push_back("DCSubtitle.v1.mattsson.xsd");
+               schema.push_back("DCDMSubtitle-2010.xsd");
+               schema.push_back("PROTO-ASDCP-CC-CPL-20070926.xsd");
+               schema.push_back("SMPTE-429-16.xsd");
+               schema.push_back("Dolby-2012-AD.xsd");
+               schema.push_back("SMPTE-429-10-2008.xsd");
+               schema.push_back("xlink.xsd");
+               schema.push_back("SMPTE-335-2012.xsd");
+               schema.push_back("SMPTE-395-2014-13-1-aaf.xsd");
+               schema.push_back("isdcf-mca.xsd");
+               schema.push_back("SMPTE-429-12-2008.xsd");
+
+               /* XXX: I'm not especially clear what this is for, but it seems to be necessary.
+                * Schemas that are not mentioned in this list are not read, and the things
+                * they describe are not checked.
+                */
                string locations;
-               for (map<string, string>::const_iterator i = schema.begin(); i != schema.end(); ++i) {
-                       locations += i->first;
-                       locations += " ";
-                       boost::filesystem::path p = xsd_dtd_directory / i->second;
-                       locations += p.string() + " ";
+               BOOST_FOREACH (string i, schema) {
+                       locations += String::compose("%1 %1 ", i, i);
                }
 
                parser.setExternalSchemaLocation(locations.c_str());
@@ -271,7 +328,7 @@ validate_xml (boost::filesystem::path xml_file, boost::filesystem::path xsd_dtd_
 
                try {
                        parser.resetDocumentPool();
-                       parser.parse(xml_file.string().c_str());
+                       parse(parser, xml);
                } catch (XMLException& e) {
                        throw MiscError(xml_ch_to_string(e.getMessage()));
                } catch (DOMException& e) {
@@ -283,11 +340,29 @@ validate_xml (boost::filesystem::path xml_file, boost::filesystem::path xsd_dtd_
 
        XMLPlatformUtils::Terminate ();
 
-       return error_handler.errors ();
+       BOOST_FOREACH (XMLValidationError i, error_handler.errors()) {
+               notes.push_back (
+                       VerificationNote(
+                               VerificationNote::VERIFY_ERROR,
+                               VerificationNote::XML_VALIDATION_ERROR,
+                               i.message(),
+                               boost::trim_copy(i.public_id() + " " + i.system_id()),
+                               i.line()
+                               )
+                       );
+       }
 }
 
-static Result
-verify_asset (shared_ptr<DCP> dcp, shared_ptr<ReelMXF> reel_mxf, function<void (float)> progress)
+
+enum VerifyAssetResult {
+       VERIFY_ASSET_RESULT_GOOD,
+       VERIFY_ASSET_RESULT_CPL_PKL_DIFFER,
+       VERIFY_ASSET_RESULT_BAD
+};
+
+
+static VerifyAssetResult
+verify_asset (shared_ptr<const DCP> dcp, shared_ptr<const ReelMXF> reel_mxf, function<void (float)> progress)
 {
        string const actual_hash = reel_mxf->asset_ref()->hash(progress);
 
@@ -309,14 +384,208 @@ verify_asset (shared_ptr<DCP> dcp, shared_ptr<ReelMXF> reel_mxf, function<void (
 
        optional<string> cpl_hash = reel_mxf->hash();
        if (cpl_hash && *cpl_hash != *pkl_hash) {
-               return RESULT_CPL_PKL_DIFFER;
+               return VERIFY_ASSET_RESULT_CPL_PKL_DIFFER;
        }
 
        if (actual_hash != *pkl_hash) {
-               return RESULT_BAD;
+               return VERIFY_ASSET_RESULT_BAD;
+       }
+
+       return VERIFY_ASSET_RESULT_GOOD;
+}
+
+
+void
+verify_language_tag (string tag, list<VerificationNote>& notes)
+{
+       try {
+               dcp::LanguageTag test (tag);
+       } catch (dcp::LanguageTagError &) {
+               notes.push_back (VerificationNote(VerificationNote::VERIFY_BV21_ERROR, VerificationNote::BAD_LANGUAGE, tag));
+       }
+}
+
+
+enum VerifyPictureAssetResult
+{
+       VERIFY_PICTURE_ASSET_RESULT_GOOD,
+       VERIFY_PICTURE_ASSET_RESULT_FRAME_NEARLY_TOO_LARGE,
+       VERIFY_PICTURE_ASSET_RESULT_BAD,
+};
+
+
+int
+biggest_frame_size (shared_ptr<const MonoPictureFrame> frame)
+{
+       return frame->size ();
+}
+
+int
+biggest_frame_size (shared_ptr<const StereoPictureFrame> frame)
+{
+       return max(frame->left()->size(), frame->right()->size());
+}
+
+
+template <class A, class R, class F>
+optional<VerifyPictureAssetResult>
+verify_picture_asset_type (shared_ptr<ReelMXF> reel_mxf, function<void (float)> progress)
+{
+       shared_ptr<A> asset = dynamic_pointer_cast<A>(reel_mxf->asset_ref().asset());
+       if (!asset) {
+               return optional<VerifyPictureAssetResult>();
+       }
+
+       int biggest_frame = 0;
+       shared_ptr<R> reader = asset->start_read ();
+       int64_t const duration = asset->intrinsic_duration ();
+       for (int64_t i = 0; i < duration; ++i) {
+               shared_ptr<const F> frame = reader->get_frame (i);
+               biggest_frame = max(biggest_frame, biggest_frame_size(frame));
+               progress (float(i) / duration);
+       }
+
+       static const int max_frame =   rint(250 * 1000000 / (8 * asset->edit_rate().as_float()));
+       static const int risky_frame = rint(230 * 1000000 / (8 * asset->edit_rate().as_float()));
+       if (biggest_frame > max_frame) {
+               return VERIFY_PICTURE_ASSET_RESULT_BAD;
+       } else if (biggest_frame > risky_frame) {
+               return VERIFY_PICTURE_ASSET_RESULT_FRAME_NEARLY_TOO_LARGE;
+       }
+
+       return VERIFY_PICTURE_ASSET_RESULT_GOOD;
+}
+
+
+static VerifyPictureAssetResult
+verify_picture_asset (shared_ptr<ReelMXF> reel_mxf, function<void (float)> progress)
+{
+       optional<VerifyPictureAssetResult> r = verify_picture_asset_type<MonoPictureAsset, MonoPictureAssetReader, MonoPictureFrame>(reel_mxf, progress);
+       if (!r) {
+               r = verify_picture_asset_type<StereoPictureAsset, StereoPictureAssetReader, StereoPictureFrame>(reel_mxf, progress);
+       }
+
+       DCP_ASSERT (r);
+       return *r;
+}
+
+
+static void
+verify_main_picture_asset (
+       shared_ptr<const DCP> dcp,
+       shared_ptr<const Reel> reel,
+       function<void (string, optional<boost::filesystem::path>)> stage,
+       function<void (float)> progress,
+       list<VerificationNote>& notes
+       )
+{
+       boost::filesystem::path const file = *reel->main_picture()->asset()->file();
+       stage ("Checking picture asset hash", file);
+       VerifyAssetResult const r = verify_asset (dcp, reel->main_picture(), progress);
+       switch (r) {
+               case VERIFY_ASSET_RESULT_BAD:
+                       notes.push_back (
+                               VerificationNote(
+                                       VerificationNote::VERIFY_ERROR, VerificationNote::PICTURE_HASH_INCORRECT, file
+                                       )
+                               );
+                       break;
+               case VERIFY_ASSET_RESULT_CPL_PKL_DIFFER:
+                       notes.push_back (
+                               VerificationNote(
+                                       VerificationNote::VERIFY_ERROR, VerificationNote::PKL_CPL_PICTURE_HASHES_DISAGREE, file
+                                       )
+                               );
+                       break;
+               default:
+                       break;
+       }
+       stage ("Checking picture frame sizes", reel->main_picture()->asset()->file());
+       VerifyPictureAssetResult const pr = verify_picture_asset (reel->main_picture(), progress);
+       switch (pr) {
+               case VERIFY_PICTURE_ASSET_RESULT_BAD:
+                       notes.push_back (
+                               VerificationNote(
+                                       VerificationNote::VERIFY_ERROR, VerificationNote::PICTURE_FRAME_TOO_LARGE, file
+                                       )
+                               );
+                       break;
+               case VERIFY_PICTURE_ASSET_RESULT_FRAME_NEARLY_TOO_LARGE:
+                       notes.push_back (
+                               VerificationNote(
+                                       VerificationNote::VERIFY_WARNING, VerificationNote::PICTURE_FRAME_NEARLY_TOO_LARGE, file
+                                       )
+                               );
+                       break;
+               default:
+                       break;
+       }
+}
+
+
+static void
+verify_main_sound_asset (
+       shared_ptr<const DCP> dcp,
+       shared_ptr<const Reel> reel,
+       function<void (string, optional<boost::filesystem::path>)> stage,
+       function<void (float)> progress,
+       list<VerificationNote>& notes
+       )
+{
+       stage ("Checking sound asset hash", reel->main_sound()->asset()->file());
+       VerifyAssetResult const r = verify_asset (dcp, reel->main_sound(), progress);
+       switch (r) {
+               case VERIFY_ASSET_RESULT_BAD:
+                       notes.push_back (
+                               VerificationNote(
+                                       VerificationNote::VERIFY_ERROR, VerificationNote::SOUND_HASH_INCORRECT, *reel->main_sound()->asset()->file()
+                                       )
+                               );
+                       break;
+               case VERIFY_ASSET_RESULT_CPL_PKL_DIFFER:
+                       notes.push_back (
+                               VerificationNote(
+                                       VerificationNote::VERIFY_ERROR, VerificationNote::PKL_CPL_SOUND_HASHES_DISAGREE, *reel->main_sound()->asset()->file()
+                                       )
+                               );
+                       break;
+               default:
+                       break;
+       }
+}
+
+
+static void
+verify_main_subtitle_reel (shared_ptr<const ReelSubtitleAsset> reel_asset, list<VerificationNote>& notes)
+{
+       /* XXX: is Language compulsory? */
+       if (reel_asset->language()) {
+               verify_language_tag (*reel_asset->language(), notes);
        }
+}
+
 
-       return RESULT_GOOD;
+static void
+verify_main_subtitle_asset (
+       shared_ptr<const Reel> reel,
+       function<void (string, optional<boost::filesystem::path>)> stage,
+       boost::filesystem::path xsd_dtd_directory,
+       list<VerificationNote>& notes
+       )
+{
+       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 (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);
+               }
+       }
 }
 
 
@@ -341,33 +610,44 @@ dcp::verify (
                stage ("Checking DCP", dcp->directory());
                try {
                        dcp->read (&notes);
-               } catch (DCPReadError& e) {
-                       notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::Code::GENERAL_READ, string(e.what())));
+               } catch (ReadError& e) {
+                       notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::GENERAL_READ, string(e.what())));
                } catch (XMLError& e) {
-                       notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::Code::GENERAL_READ, string(e.what())));
+                       notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::GENERAL_READ, string(e.what())));
+               } catch (MXFFileError& e) {
+                       notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::GENERAL_READ, string(e.what())));
+               } catch (cxml::Error& e) {
+                       notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::GENERAL_READ, string(e.what())));
+               }
+
+               if (dcp->standard() != dcp::SMPTE) {
+                       notes.push_back (VerificationNote(VerificationNote::VERIFY_BV21_ERROR, VerificationNote::NOT_SMPTE));
                }
 
                BOOST_FOREACH (shared_ptr<CPL> cpl, dcp->cpls()) {
                        stage ("Checking CPL", cpl->file());
-
-                       list<XMLValidationError> errors = validate_xml (cpl->file().get(), xsd_dtd_directory);
-                       BOOST_FOREACH (XMLValidationError i, errors) {
-                               notes.push_back (VerificationNote(
-                                                        VerificationNote::VERIFY_ERROR, VerificationNote::Code::XML_VALIDATION_ERROR,
-                                                        String::compose("%1 (file %2, line %3)", i.message(), cpl->file()->string(), i.line())
-                                                        ));
-                       }
+                       validate_xml (cpl->file().get(), xsd_dtd_directory, notes);
 
                        /* Check that the CPL's hash corresponds to the PKL */
                        BOOST_FOREACH (shared_ptr<PKL> i, dcp->pkls()) {
                                optional<string> h = i->hash(cpl->id());
-                               if (h && make_digest(Data(*cpl->file())) != *h) {
+                               if (h && make_digest(ArrayData(*cpl->file())) != *h) {
                                        notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::CPL_HASH_INCORRECT));
                                }
                        }
 
                        BOOST_FOREACH (shared_ptr<Reel> reel, cpl->reels()) {
                                stage ("Checking reel", optional<boost::filesystem::path>());
+
+                               BOOST_FOREACH (shared_ptr<ReelAsset> i, reel->assets()) {
+                                       if (i->duration() && (i->duration().get() * i->edit_rate().denominator / i->edit_rate().numerator) < 1) {
+                                               notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::DURATION_TOO_SMALL, i->id()));
+                                       }
+                                       if ((i->intrinsic_duration() * i->edit_rate().denominator / i->edit_rate().numerator) < 1) {
+                                               notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::INTRINSIC_DURATION_TOO_SMALL, i->id()));
+                                       }
+                               }
+
                                if (reel->main_picture()) {
                                        /* Check reel stuff */
                                        Fraction const frame_rate = reel->main_picture()->frame_rate();
@@ -383,40 +663,18 @@ dcp::verify (
                                        }
                                        /* Check asset */
                                        if (reel->main_picture()->asset_ref().resolved()) {
-                                               stage ("Checking picture asset hash", reel->main_picture()->asset()->file());
-                                               Result const r = verify_asset (dcp, reel->main_picture(), progress);
-                                               switch (r) {
-                                               case RESULT_BAD:
-                                                       notes.push_back (
-                                                                       VerificationNote(
-                                                                               VerificationNote::VERIFY_ERROR, VerificationNote::PICTURE_HASH_INCORRECT, *reel->main_picture()->asset()->file()
-                                                                               )
-                                                                       );
-                                                       break;
-                                               case RESULT_CPL_PKL_DIFFER:
-                                                       notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::PKL_CPL_PICTURE_HASHES_DISAGREE));
-                                                       break;
-                                               default:
-                                                       break;
-                                               }
+                                               verify_main_picture_asset (dcp, reel, stage, progress, notes);
                                        }
                                }
+
                                if (reel->main_sound() && reel->main_sound()->asset_ref().resolved()) {
-                                       stage ("Checking sound asset hash", reel->main_sound()->asset()->file());
-                                       Result const r = verify_asset (dcp, reel->main_sound(), progress);
-                                       switch (r) {
-                                       case RESULT_BAD:
-                                               notes.push_back (
-                                                               VerificationNote(
-                                                                       VerificationNote::VERIFY_ERROR, VerificationNote::SOUND_HASH_INCORRECT, *reel->main_sound()->asset()->file()
-                                                                       )
-                                                               );
-                                               break;
-                                       case RESULT_CPL_PKL_DIFFER:
-                                               notes.push_back (VerificationNote (VerificationNote::VERIFY_ERROR, VerificationNote::PKL_CPL_SOUND_HASHES_DISAGREE));
-                                               break;
-                                       default:
-                                               break;
+                                       verify_main_sound_asset (dcp, reel, stage, progress, notes);
+                               }
+
+                               if (reel->main_subtitle()) {
+                                       verify_main_subtitle_reel (reel->main_subtitle(), notes);
+                                       if (reel->main_subtitle()->asset_ref().resolved()) {
+                                               verify_main_subtitle_asset (reel, stage, xsd_dtd_directory, notes);
                                        }
                                }
                        }
@@ -424,14 +682,14 @@ dcp::verify (
 
                BOOST_FOREACH (shared_ptr<PKL> pkl, dcp->pkls()) {
                        stage ("Checking PKL", pkl->file());
+                       validate_xml (pkl->file().get(), xsd_dtd_directory, notes);
+               }
 
-                       list<XMLValidationError> errors = validate_xml (pkl->file().get(), xsd_dtd_directory);
-                       BOOST_FOREACH (XMLValidationError i, errors) {
-                               notes.push_back (VerificationNote(
-                                                        VerificationNote::VERIFY_ERROR, VerificationNote::Code::XML_VALIDATION_ERROR,
-                                                        String::compose("%1 (file %2, line %3)", i.message(), pkl->file()->string(), i.line())
-                                                        ));
-                       }
+               if (dcp->asset_map_path()) {
+                       stage ("Checking ASSETMAP", dcp->asset_map_path().get());
+                       validate_xml (dcp->asset_map_path().get(), xsd_dtd_directory, notes);
+               } else {
+                       notes.push_back (VerificationNote(VerificationNote::VERIFY_ERROR, VerificationNote::MISSING_ASSETMAP));
                }
        }
 
@@ -445,25 +703,41 @@ dcp::note_to_string (dcp::VerificationNote note)
        case dcp::VerificationNote::GENERAL_READ:
                return *note.note();
        case dcp::VerificationNote::CPL_HASH_INCORRECT:
-               return "The hash of the CPL in the PKL does not agree with the CPL file";
+               return "The hash of the CPL in the PKL does not agree with the CPL file.";
        case dcp::VerificationNote::INVALID_PICTURE_FRAME_RATE:
-               return "The picture in a reel has an invalid frame rate";
+               return "The picture in a reel has an invalid frame rate.";
        case dcp::VerificationNote::PICTURE_HASH_INCORRECT:
-               return dcp::String::compose("The hash of the picture asset %1 does not agree with the PKL file", note.file()->filename());
+               return dcp::String::compose("The hash of the picture asset %1 does not agree with the PKL file.", note.file()->filename());
        case dcp::VerificationNote::PKL_CPL_PICTURE_HASHES_DISAGREE:
-               return "The PKL and CPL hashes disagree for a picture asset.";
+               return dcp::String::compose("The PKL and CPL hashes disagree for the picture asset %1.", note.file()->filename());
        case dcp::VerificationNote::SOUND_HASH_INCORRECT:
-               return dcp::String::compose("The hash of the sound asset %1 does not agree with the PKL file", note.file()->filename());
+               return dcp::String::compose("The hash of the sound asset %1 does not agree with the PKL file.", note.file()->filename());
        case dcp::VerificationNote::PKL_CPL_SOUND_HASHES_DISAGREE:
-               return "The PKL and CPL hashes disagree for a sound asset.";
+               return dcp::String::compose("The PKL and CPL hashes disagree for the sound asset %1.", note.file()->filename());
        case dcp::VerificationNote::EMPTY_ASSET_PATH:
                return "The asset map contains an empty asset path.";
        case dcp::VerificationNote::MISSING_ASSET:
-               return "The file for an asset in the asset map cannot be found.";
+               return String::compose("The file for an asset in the asset map cannot be found; missing file is %1.", note.file()->filename());
        case dcp::VerificationNote::MISMATCHED_STANDARD:
                return "The DCP contains both SMPTE and Interop parts.";
        case dcp::VerificationNote::XML_VALIDATION_ERROR:
-               return "An XML file is badly formed.";
+               return String::compose("An XML file is badly formed: %1 (%2:%3)", note.note().get(), note.file()->filename(), note.line().get());
+       case dcp::VerificationNote::MISSING_ASSETMAP:
+               return "No ASSETMAP or ASSETMAP.xml was found.";
+       case dcp::VerificationNote::INTRINSIC_DURATION_TOO_SMALL:
+               return String::compose("The intrinsic duration of an asset is less than 1 second long: %1", note.note().get());
+       case dcp::VerificationNote::DURATION_TOO_SMALL:
+               return String::compose("The duration of an asset is less than 1 second long: %1", note.note().get());
+       case dcp::VerificationNote::PICTURE_FRAME_TOO_LARGE:
+               return String::compose("The instantaneous bit rate of the picture asset %1 is larger than the limit of 250Mbit/s in at least one place.", note.file()->filename());
+       case dcp::VerificationNote::PICTURE_FRAME_NEARLY_TOO_LARGE:
+               return String::compose("The instantaneous bit rate of the picture asset %1 is close to the limit of 250Mbit/s in at least one place.", note.file()->filename());
+       case dcp::VerificationNote::EXTERNAL_ASSET:
+               return String::compose("An asset that this DCP refers to is not included in the DCP.  It may be a VF.  Missing asset is %1.", note.note().get());
+       case dcp::VerificationNote::NOT_SMPTE:
+               return "This DCP does not use the SMPTE standard, which is required for Bv2.1 compliance.";
+       case dcp::VerificationNote::BAD_LANGUAGE:
+               return String::compose("The DCP specifies a language '%1' which does not conform to the RFC 5646 standard.", note.note().get());
        }
 
        return "";