Cleanup: remove some unpleasant (and I think unnecessary) casting.
[libdcp.git] / src / verify.cc
1 /*
2     Copyright (C) 2018-2021 Carl Hetherington <cth@carlh.net>
3
4     This file is part of libdcp.
5
6     libdcp is free software; you can redistribute it and/or modify
7     it under the terms of the GNU General Public License as published by
8     the Free Software Foundation; either version 2 of the License, or
9     (at your option) any later version.
10
11     libdcp is distributed in the hope that it will be useful,
12     but WITHOUT ANY WARRANTY; without even the implied warranty of
13     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
14     GNU General Public License for more details.
15
16     You should have received a copy of the GNU General Public License
17     along with libdcp.  If not, see <http://www.gnu.org/licenses/>.
18
19     In addition, as a special exception, the copyright holders give
20     permission to link the code of portions of this program with the
21     OpenSSL library under certain conditions as described in each
22     individual source file, and distribute linked combinations
23     including the two.
24
25     You must obey the GNU General Public License in all respects
26     for all of the code used other than OpenSSL.  If you modify
27     file(s) with this exception, you may extend this exception to your
28     version of the file(s), but you are not obligated to do so.  If you
29     do not wish to do so, delete this exception statement from your
30     version.  If you delete this exception statement from all source
31     files in the program, then also delete it here.
32 */
33
34
35 /** @file  src/verify.cc
36  *  @brief dcp::verify() method and associated code
37  */
38
39
40 #include "compose.hpp"
41 #include "cpl.h"
42 #include "dcp.h"
43 #include "exceptions.h"
44 #include "interop_subtitle_asset.h"
45 #include "mono_picture_asset.h"
46 #include "mono_picture_frame.h"
47 #include "raw_convert.h"
48 #include "reel.h"
49 #include "reel_closed_caption_asset.h"
50 #include "reel_interop_subtitle_asset.h"
51 #include "reel_markers_asset.h"
52 #include "reel_picture_asset.h"
53 #include "reel_sound_asset.h"
54 #include "reel_smpte_subtitle_asset.h"
55 #include "reel_subtitle_asset.h"
56 #include "smpte_subtitle_asset.h"
57 #include "stereo_picture_asset.h"
58 #include "stereo_picture_frame.h"
59 #include "verify.h"
60 #include "verify_j2k.h"
61 #include <libxml/parserInternals.h>
62 #include <xercesc/dom/DOMAttr.hpp>
63 #include <xercesc/dom/DOMDocument.hpp>
64 #include <xercesc/dom/DOMError.hpp>
65 #include <xercesc/dom/DOMErrorHandler.hpp>
66 #include <xercesc/dom/DOMException.hpp>
67 #include <xercesc/dom/DOMImplementation.hpp>
68 #include <xercesc/dom/DOMImplementationLS.hpp>
69 #include <xercesc/dom/DOMImplementationRegistry.hpp>
70 #include <xercesc/dom/DOMLSParser.hpp>
71 #include <xercesc/dom/DOMLocator.hpp>
72 #include <xercesc/dom/DOMNamedNodeMap.hpp>
73 #include <xercesc/dom/DOMNodeList.hpp>
74 #include <xercesc/framework/LocalFileInputSource.hpp>
75 #include <xercesc/framework/MemBufInputSource.hpp>
76 #include <xercesc/parsers/AbstractDOMParser.hpp>
77 #include <xercesc/parsers/XercesDOMParser.hpp>
78 #include <xercesc/sax/HandlerBase.hpp>
79 #include <xercesc/util/PlatformUtils.hpp>
80 #include <boost/algorithm/string.hpp>
81 #include <iostream>
82 #include <map>
83 #include <regex>
84 #include <set>
85 #include <vector>
86
87
88 using std::cout;
89 using std::dynamic_pointer_cast;
90 using std::list;
91 using std::make_shared;
92 using std::map;
93 using std::max;
94 using std::set;
95 using std::shared_ptr;
96 using std::string;
97 using std::vector;
98 using boost::optional;
99 using boost::function;
100
101
102 using namespace dcp;
103 using namespace xercesc;
104
105
106 static
107 string
108 xml_ch_to_string (XMLCh const * a)
109 {
110         char* x = XMLString::transcode(a);
111         string const o(x);
112         XMLString::release(&x);
113         return o;
114 }
115
116
117 class XMLValidationError
118 {
119 public:
120         XMLValidationError (SAXParseException const & e)
121                 : _message (xml_ch_to_string(e.getMessage()))
122                 , _line (e.getLineNumber())
123                 , _column (e.getColumnNumber())
124                 , _public_id (e.getPublicId() ? xml_ch_to_string(e.getPublicId()) : "")
125                 , _system_id (e.getSystemId() ? xml_ch_to_string(e.getSystemId()) : "")
126         {
127
128         }
129
130         string message () const {
131                 return _message;
132         }
133
134         uint64_t line () const {
135                 return _line;
136         }
137
138         uint64_t column () const {
139                 return _column;
140         }
141
142         string public_id () const {
143                 return _public_id;
144         }
145
146         string system_id () const {
147                 return _system_id;
148         }
149
150 private:
151         string _message;
152         uint64_t _line;
153         uint64_t _column;
154         string _public_id;
155         string _system_id;
156 };
157
158
159 class DCPErrorHandler : public ErrorHandler
160 {
161 public:
162         void warning(const SAXParseException& e) override
163         {
164                 maybe_add (XMLValidationError(e));
165         }
166
167         void error(const SAXParseException& e) override
168         {
169                 maybe_add (XMLValidationError(e));
170         }
171
172         void fatalError(const SAXParseException& e) override
173         {
174                 maybe_add (XMLValidationError(e));
175         }
176
177         void resetErrors() override {
178                 _errors.clear ();
179         }
180
181         list<XMLValidationError> errors () const {
182                 return _errors;
183         }
184
185 private:
186         void maybe_add (XMLValidationError e)
187         {
188                 /* XXX: nasty hack */
189                 if (
190                         e.message().find("schema document") != string::npos &&
191                         e.message().find("has different target namespace from the one specified in instance document") != string::npos
192                         ) {
193                         return;
194                 }
195
196                 _errors.push_back (e);
197         }
198
199         list<XMLValidationError> _errors;
200 };
201
202
203 class StringToXMLCh
204 {
205 public:
206         StringToXMLCh (string a)
207         {
208                 _buffer = XMLString::transcode(a.c_str());
209         }
210
211         StringToXMLCh (StringToXMLCh const&) = delete;
212         StringToXMLCh& operator= (StringToXMLCh const&) = delete;
213
214         ~StringToXMLCh ()
215         {
216                 XMLString::release (&_buffer);
217         }
218
219         XMLCh const * get () const {
220                 return _buffer;
221         }
222
223 private:
224         XMLCh* _buffer;
225 };
226
227
228 class LocalFileResolver : public EntityResolver
229 {
230 public:
231         LocalFileResolver (boost::filesystem::path xsd_dtd_directory)
232                 : _xsd_dtd_directory (xsd_dtd_directory)
233         {
234                 /* XXX: I'm not clear on what things need to be in this list; some XSDs are apparently, magically
235                  * found without being here.
236                  */
237                 add("http://www.w3.org/2001/XMLSchema.dtd", "XMLSchema.dtd");
238                 add("http://www.w3.org/2001/03/xml.xsd", "xml.xsd");
239                 add("http://www.w3.org/TR/2002/REC-xmldsig-core-20020212/xmldsig-core-schema.xsd", "xmldsig-core-schema.xsd");
240                 add("http://www.digicine.com/schemas/437-Y/2007/Main-Stereo-Picture-CPL.xsd", "Main-Stereo-Picture-CPL.xsd");
241                 add("http://www.digicine.com/PROTO-ASDCP-CPL-20040511.xsd", "PROTO-ASDCP-CPL-20040511.xsd");
242                 add("http://www.digicine.com/PROTO-ASDCP-PKL-20040311.xsd", "PROTO-ASDCP-PKL-20040311.xsd");
243                 add("http://www.digicine.com/PROTO-ASDCP-AM-20040311.xsd", "PROTO-ASDCP-AM-20040311.xsd");
244                 add("http://www.digicine.com/PROTO-ASDCP-CC-CPL-20070926#", "PROTO-ASDCP-CC-CPL-20070926.xsd");
245                 add("interop-subs", "DCSubtitle.v1.mattsson.xsd");
246                 add("http://www.smpte-ra.org/schemas/428-7/2010/DCST.xsd", "DCDMSubtitle-2010.xsd");
247                 add("http://www.smpte-ra.org/schemas/428-7/2014/DCST.xsd", "DCDMSubtitle-2014.xsd");
248                 add("http://www.smpte-ra.org/schemas/429-16/2014/CPL-Metadata", "SMPTE-429-16.xsd");
249                 add("http://www.dolby.com/schemas/2012/AD", "Dolby-2012-AD.xsd");
250                 add("http://www.smpte-ra.org/schemas/429-10/2008/Main-Stereo-Picture-CPL", "SMPTE-429-10-2008.xsd");
251         }
252
253         InputSource* resolveEntity(XMLCh const *, XMLCh const * system_id) override
254         {
255                 if (!system_id) {
256                         return 0;
257                 }
258                 auto system_id_str = xml_ch_to_string (system_id);
259                 auto p = _xsd_dtd_directory;
260                 if (_files.find(system_id_str) == _files.end()) {
261                         p /= system_id_str;
262                 } else {
263                         p /= _files[system_id_str];
264                 }
265                 StringToXMLCh ch (p.string());
266                 return new LocalFileInputSource(ch.get());
267         }
268
269 private:
270         void add (string uri, string file)
271         {
272                 _files[uri] = file;
273         }
274
275         std::map<string, string> _files;
276         boost::filesystem::path _xsd_dtd_directory;
277 };
278
279
280 static void
281 parse (XercesDOMParser& parser, boost::filesystem::path xml)
282 {
283         parser.parse(xml.c_str());
284 }
285
286
287 static void
288 parse (XercesDOMParser& parser, string xml)
289 {
290         xercesc::MemBufInputSource buf(reinterpret_cast<unsigned char const*>(xml.c_str()), xml.size(), "");
291         parser.parse(buf);
292 }
293
294
295 template <class T>
296 void
297 validate_xml (T xml, boost::filesystem::path xsd_dtd_directory, vector<VerificationNote>& notes)
298 {
299         try {
300                 XMLPlatformUtils::Initialize ();
301         } catch (XMLException& e) {
302                 throw MiscError ("Failed to initialise xerces library");
303         }
304
305         DCPErrorHandler error_handler;
306
307         /* All the xerces objects in this scope must be destroyed before XMLPlatformUtils::Terminate() is called */
308         {
309                 XercesDOMParser parser;
310                 parser.setValidationScheme(XercesDOMParser::Val_Always);
311                 parser.setDoNamespaces(true);
312                 parser.setDoSchema(true);
313
314                 vector<string> schema;
315                 schema.push_back("xml.xsd");
316                 schema.push_back("xmldsig-core-schema.xsd");
317                 schema.push_back("SMPTE-429-7-2006-CPL.xsd");
318                 schema.push_back("SMPTE-429-8-2006-PKL.xsd");
319                 schema.push_back("SMPTE-429-9-2007-AM.xsd");
320                 schema.push_back("Main-Stereo-Picture-CPL.xsd");
321                 schema.push_back("PROTO-ASDCP-CPL-20040511.xsd");
322                 schema.push_back("PROTO-ASDCP-PKL-20040311.xsd");
323                 schema.push_back("PROTO-ASDCP-AM-20040311.xsd");
324                 schema.push_back("DCSubtitle.v1.mattsson.xsd");
325                 schema.push_back("DCDMSubtitle-2010.xsd");
326                 schema.push_back("DCDMSubtitle-2014.xsd");
327                 schema.push_back("PROTO-ASDCP-CC-CPL-20070926.xsd");
328                 schema.push_back("SMPTE-429-16.xsd");
329                 schema.push_back("Dolby-2012-AD.xsd");
330                 schema.push_back("SMPTE-429-10-2008.xsd");
331                 schema.push_back("xlink.xsd");
332                 schema.push_back("SMPTE-335-2012.xsd");
333                 schema.push_back("SMPTE-395-2014-13-1-aaf.xsd");
334                 schema.push_back("isdcf-mca.xsd");
335                 schema.push_back("SMPTE-429-12-2008.xsd");
336
337                 /* XXX: I'm not especially clear what this is for, but it seems to be necessary.
338                  * Schemas that are not mentioned in this list are not read, and the things
339                  * they describe are not checked.
340                  */
341                 string locations;
342                 for (auto i: schema) {
343                         locations += String::compose("%1 %1 ", i, i);
344                 }
345
346                 parser.setExternalSchemaLocation(locations.c_str());
347                 parser.setValidationSchemaFullChecking(true);
348                 parser.setErrorHandler(&error_handler);
349
350                 LocalFileResolver resolver (xsd_dtd_directory);
351                 parser.setEntityResolver(&resolver);
352
353                 try {
354                         parser.resetDocumentPool();
355                         parse(parser, xml);
356                 } catch (XMLException& e) {
357                         throw MiscError(xml_ch_to_string(e.getMessage()));
358                 } catch (DOMException& e) {
359                         throw MiscError(xml_ch_to_string(e.getMessage()));
360                 } catch (...) {
361                         throw MiscError("Unknown exception from xerces");
362                 }
363         }
364
365         XMLPlatformUtils::Terminate ();
366
367         for (auto i: error_handler.errors()) {
368                 notes.push_back ({
369                         VerificationNote::Type::ERROR,
370                         VerificationNote::Code::INVALID_XML,
371                         i.message(),
372                         boost::trim_copy(i.public_id() + " " + i.system_id()),
373                         i.line()
374                 });
375         }
376 }
377
378
379 enum class VerifyAssetResult {
380         GOOD,
381         CPL_PKL_DIFFER,
382         BAD
383 };
384
385
386 static VerifyAssetResult
387 verify_asset (shared_ptr<const DCP> dcp, shared_ptr<const ReelFileAsset> reel_file_asset, function<void (float)> progress)
388 {
389         auto const actual_hash = reel_file_asset->asset_ref()->hash(progress);
390
391         auto pkls = dcp->pkls();
392         /* We've read this DCP in so it must have at least one PKL */
393         DCP_ASSERT (!pkls.empty());
394
395         auto asset = reel_file_asset->asset_ref().asset();
396
397         optional<string> pkl_hash;
398         for (auto i: pkls) {
399                 pkl_hash = i->hash (reel_file_asset->asset_ref()->id());
400                 if (pkl_hash) {
401                         break;
402                 }
403         }
404
405         DCP_ASSERT (pkl_hash);
406
407         auto cpl_hash = reel_file_asset->hash();
408         if (cpl_hash && *cpl_hash != *pkl_hash) {
409                 return VerifyAssetResult::CPL_PKL_DIFFER;
410         }
411
412         if (actual_hash != *pkl_hash) {
413                 return VerifyAssetResult::BAD;
414         }
415
416         return VerifyAssetResult::GOOD;
417 }
418
419
420 void
421 verify_language_tag (string tag, vector<VerificationNote>& notes)
422 {
423         try {
424                 LanguageTag test (tag);
425         } catch (LanguageTagError &) {
426                 notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_LANGUAGE, tag});
427         }
428 }
429
430
431 static void
432 verify_picture_asset (shared_ptr<const ReelFileAsset> reel_file_asset, boost::filesystem::path file, vector<VerificationNote>& notes, function<void (float)> progress)
433 {
434         int biggest_frame = 0;
435         auto asset = dynamic_pointer_cast<PictureAsset>(reel_file_asset->asset_ref().asset());
436         auto const duration = asset->intrinsic_duration ();
437
438         auto check_and_add = [&notes](vector<VerificationNote> const& j2k_notes) {
439                 for (auto i: j2k_notes) {
440                         if (find(notes.begin(), notes.end(), i) == notes.end()) {
441                                 notes.push_back (i);
442                         }
443                 }
444         };
445
446         if (auto mono_asset = dynamic_pointer_cast<MonoPictureAsset>(reel_file_asset->asset_ref().asset())) {
447                 auto reader = mono_asset->start_read ();
448                 for (int64_t i = 0; i < duration; ++i) {
449                         auto frame = reader->get_frame (i);
450                         biggest_frame = max(biggest_frame, frame->size());
451                         if (!mono_asset->encrypted() || mono_asset->key()) {
452                                 vector<VerificationNote> j2k_notes;
453                                 verify_j2k(frame, i, mono_asset->frame_rate().numerator, j2k_notes);
454                                 check_and_add (j2k_notes);
455                         }
456                         progress (float(i) / duration);
457                 }
458         } else if (auto stereo_asset = dynamic_pointer_cast<StereoPictureAsset>(asset)) {
459                 auto reader = stereo_asset->start_read ();
460                 for (int64_t i = 0; i < duration; ++i) {
461                         auto frame = reader->get_frame (i);
462                         biggest_frame = max(biggest_frame, max(frame->left()->size(), frame->right()->size()));
463                         if (!stereo_asset->encrypted() || stereo_asset->key()) {
464                                 vector<VerificationNote> j2k_notes;
465                                 verify_j2k(frame->left(), i, stereo_asset->frame_rate().numerator, j2k_notes);
466                                 verify_j2k(frame->right(), i, stereo_asset->frame_rate().numerator, j2k_notes);
467                                 check_and_add (j2k_notes);
468                         }
469                         progress (float(i) / duration);
470                 }
471
472         }
473
474         static const int max_frame =   rint(250 * 1000000 / (8 * asset->edit_rate().as_float()));
475         static const int risky_frame = rint(230 * 1000000 / (8 * asset->edit_rate().as_float()));
476         if (biggest_frame > max_frame) {
477                 notes.push_back ({
478                         VerificationNote::Type::ERROR, VerificationNote::Code::INVALID_PICTURE_FRAME_SIZE_IN_BYTES, file
479                 });
480         } else if (biggest_frame > risky_frame) {
481                 notes.push_back ({
482                         VerificationNote::Type::WARNING, VerificationNote::Code::NEARLY_INVALID_PICTURE_FRAME_SIZE_IN_BYTES, file
483                 });
484         }
485 }
486
487
488 static void
489 verify_main_picture_asset (
490         shared_ptr<const DCP> dcp,
491         shared_ptr<const ReelPictureAsset> reel_asset,
492         function<void (string, optional<boost::filesystem::path>)> stage,
493         function<void (float)> progress,
494         VerificationOptions options,
495         vector<VerificationNote>& notes
496         )
497 {
498         auto asset = reel_asset->asset();
499         auto const file = *asset->file();
500
501         if (options.check_asset_hashes && (!options.maximum_asset_size_for_hash_check || boost::filesystem::file_size(file) < *options.maximum_asset_size_for_hash_check)) {
502                 stage ("Checking picture asset hash", file);
503                 auto const r = verify_asset (dcp, reel_asset, progress);
504                 switch (r) {
505                         case VerifyAssetResult::BAD:
506                                 notes.push_back ({
507                                         VerificationNote::Type::ERROR, VerificationNote::Code::INCORRECT_PICTURE_HASH, file
508                                 });
509                                 break;
510                         case VerifyAssetResult::CPL_PKL_DIFFER:
511                                 notes.push_back ({
512                                         VerificationNote::Type::ERROR, VerificationNote::Code::MISMATCHED_PICTURE_HASHES, file
513                                 });
514                                 break;
515                         default:
516                                 break;
517                 }
518         }
519
520         stage ("Checking picture frame sizes", asset->file());
521         verify_picture_asset (reel_asset, file, notes, progress);
522
523         /* Only flat/scope allowed by Bv2.1 */
524         if (
525                 asset->size() != Size(2048, 858) &&
526                 asset->size() != Size(1998, 1080) &&
527                 asset->size() != Size(4096, 1716) &&
528                 asset->size() != Size(3996, 2160)) {
529                 notes.push_back({
530                         VerificationNote::Type::BV21_ERROR,
531                         VerificationNote::Code::INVALID_PICTURE_SIZE_IN_PIXELS,
532                         String::compose("%1x%2", asset->size().width, asset->size().height),
533                         file
534                 });
535         }
536
537         /* Only 24, 25, 48fps allowed for 2K */
538         if (
539                 (asset->size() == Size(2048, 858) || asset->size() == Size(1998, 1080)) &&
540                 (asset->edit_rate() != Fraction(24, 1) && asset->edit_rate() != Fraction(25, 1) && asset->edit_rate() != Fraction(48, 1))
541            ) {
542                 notes.push_back({
543                         VerificationNote::Type::BV21_ERROR,
544                         VerificationNote::Code::INVALID_PICTURE_FRAME_RATE_FOR_2K,
545                         String::compose("%1/%2", asset->edit_rate().numerator, asset->edit_rate().denominator),
546                         file
547                 });
548         }
549
550         if (asset->size() == Size(4096, 1716) || asset->size() == Size(3996, 2160)) {
551                 /* Only 24fps allowed for 4K */
552                 if (asset->edit_rate() != Fraction(24, 1)) {
553                         notes.push_back({
554                                 VerificationNote::Type::BV21_ERROR,
555                                 VerificationNote::Code::INVALID_PICTURE_FRAME_RATE_FOR_4K,
556                                 String::compose("%1/%2", asset->edit_rate().numerator, asset->edit_rate().denominator),
557                                 file
558                         });
559                 }
560
561                 /* Only 2D allowed for 4K */
562                 if (dynamic_pointer_cast<const StereoPictureAsset>(asset)) {
563                         notes.push_back({
564                                 VerificationNote::Type::BV21_ERROR,
565                                 VerificationNote::Code::INVALID_PICTURE_ASSET_RESOLUTION_FOR_3D,
566                                 String::compose("%1/%2", asset->edit_rate().numerator, asset->edit_rate().denominator),
567                                 file
568                         });
569
570                 }
571         }
572
573 }
574
575
576 struct State
577 {
578         boost::optional<string> subtitle_language;
579         boost::optional<int> audio_channels;
580 };
581
582
583 static void
584 verify_main_sound_asset (
585         shared_ptr<const DCP> dcp,
586         shared_ptr<const ReelSoundAsset> reel_asset,
587         function<void (string, optional<boost::filesystem::path>)> stage,
588         function<void (float)> progress,
589         VerificationOptions options,
590         vector<VerificationNote>& notes,
591         State& state
592         )
593 {
594         auto asset = reel_asset->asset();
595         auto const file = *asset->file();
596
597         if (options.check_asset_hashes && (!options.maximum_asset_size_for_hash_check || boost::filesystem::file_size(file) < *options.maximum_asset_size_for_hash_check)) {
598                 stage("Checking sound asset hash", file);
599                 auto const r = verify_asset (dcp, reel_asset, progress);
600                 switch (r) {
601                         case VerifyAssetResult::BAD:
602                                 notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::INCORRECT_SOUND_HASH, file});
603                                 break;
604                         case VerifyAssetResult::CPL_PKL_DIFFER:
605                                 notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::MISMATCHED_SOUND_HASHES, file});
606                                 break;
607                         default:
608                                 break;
609                 }
610         }
611
612         if (!state.audio_channels) {
613                 state.audio_channels = asset->channels();
614         } else if (*state.audio_channels != asset->channels()) {
615                 notes.push_back({ VerificationNote::Type::ERROR, VerificationNote::Code::MISMATCHED_SOUND_CHANNEL_COUNTS, file });
616         }
617
618         stage ("Checking sound asset metadata", file);
619
620         if (auto lang = asset->language()) {
621                 verify_language_tag (*lang, notes);
622         }
623         if (asset->sampling_rate() != 48000) {
624                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_SOUND_FRAME_RATE, raw_convert<string>(asset->sampling_rate()), file});
625         }
626 }
627
628
629 static void
630 verify_main_subtitle_reel (shared_ptr<const ReelSubtitleAsset> reel_asset, vector<VerificationNote>& notes)
631 {
632         /* XXX: is Language compulsory? */
633         if (reel_asset->language()) {
634                 verify_language_tag (*reel_asset->language(), notes);
635         }
636
637         if (!reel_asset->entry_point()) {
638                 notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_SUBTITLE_ENTRY_POINT, reel_asset->id() });
639         } else if (reel_asset->entry_point().get()) {
640                 notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INCORRECT_SUBTITLE_ENTRY_POINT, reel_asset->id() });
641         }
642 }
643
644
645 static void
646 verify_closed_caption_reel (shared_ptr<const ReelClosedCaptionAsset> reel_asset, vector<VerificationNote>& notes)
647 {
648         /* XXX: is Language compulsory? */
649         if (reel_asset->language()) {
650                 verify_language_tag (*reel_asset->language(), notes);
651         }
652
653         if (!reel_asset->entry_point()) {
654                 notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_CLOSED_CAPTION_ENTRY_POINT, reel_asset->id() });
655         } else if (reel_asset->entry_point().get()) {
656                 notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INCORRECT_CLOSED_CAPTION_ENTRY_POINT, reel_asset->id() });
657         }
658 }
659
660
661 /** Verify stuff that is common to both subtitles and closed captions */
662 void
663 verify_smpte_timed_text_asset (
664         shared_ptr<const SMPTESubtitleAsset> asset,
665         optional<int64_t> reel_asset_duration,
666         vector<VerificationNote>& notes
667         )
668 {
669         if (asset->language()) {
670                 verify_language_tag (*asset->language(), notes);
671         } else {
672                 notes.push_back ({ VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_SUBTITLE_LANGUAGE, *asset->file() });
673         }
674
675         auto const size = boost::filesystem::file_size(asset->file().get());
676         if (size > 115 * 1024 * 1024) {
677                 notes.push_back (
678                         { VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_TIMED_TEXT_SIZE_IN_BYTES, raw_convert<string>(size), *asset->file() }
679                         );
680         }
681
682         /* XXX: I'm not sure what Bv2.1_7.2.1 means when it says "the font resource shall not be larger than 10MB"
683          * but I'm hoping that checking for the total size of all fonts being <= 10MB will do.
684          */
685         auto fonts = asset->font_data ();
686         int total_size = 0;
687         for (auto i: fonts) {
688                 total_size += i.second.size();
689         }
690         if (total_size > 10 * 1024 * 1024) {
691                 notes.push_back ({ VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_TIMED_TEXT_FONT_SIZE_IN_BYTES, raw_convert<string>(total_size), asset->file().get() });
692         }
693
694         if (!asset->start_time()) {
695                 notes.push_back ({ VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_SUBTITLE_START_TIME, asset->file().get() });
696         } else if (asset->start_time() != Time()) {
697                 notes.push_back ({ VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_SUBTITLE_START_TIME, asset->file().get() });
698         }
699
700         if (reel_asset_duration && *reel_asset_duration != asset->intrinsic_duration()) {
701                 notes.push_back (
702                         {
703                                 VerificationNote::Type::BV21_ERROR,
704                                 VerificationNote::Code::MISMATCHED_TIMED_TEXT_DURATION,
705                                 String::compose("%1 %2", *reel_asset_duration, asset->intrinsic_duration()),
706                                 asset->file().get()
707                         });
708         }
709 }
710
711
712 /** Verify Interop subtitle-only stuff */
713 void
714 verify_interop_subtitle_asset(shared_ptr<const InteropSubtitleAsset> asset, vector<VerificationNote>& notes)
715 {
716         if (asset->subtitles().empty()) {
717                 notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::MISSING_SUBTITLE, asset->id(), asset->file().get() });
718         }
719         auto const unresolved = asset->unresolved_fonts();
720         if (!unresolved.empty()) {
721                 notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::MISSING_FONT, unresolved.front() });
722         }
723 }
724
725
726 /** Verify SMPTE subtitle-only stuff */
727 void
728 verify_smpte_subtitle_asset (
729         shared_ptr<const SMPTESubtitleAsset> asset,
730         vector<VerificationNote>& notes,
731         State& state
732         )
733 {
734         if (asset->language()) {
735                 if (!state.subtitle_language) {
736                         state.subtitle_language = *asset->language();
737                 } else if (state.subtitle_language != *asset->language()) {
738                         notes.push_back ({ VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISMATCHED_SUBTITLE_LANGUAGES });
739                 }
740         }
741
742         DCP_ASSERT (asset->resource_id());
743         auto xml_id = asset->xml_id();
744         if (xml_id) {
745                 if (asset->resource_id().get() != xml_id) {
746                         notes.push_back ({ VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISMATCHED_TIMED_TEXT_RESOURCE_ID });
747                 }
748
749                 if (asset->id() == asset->resource_id().get() || asset->id() == xml_id) {
750                         notes.push_back ({ VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INCORRECT_TIMED_TEXT_ASSET_ID });
751                 }
752         } else {
753                 notes.push_back ({VerificationNote::Type::WARNING, VerificationNote::Code::MISSED_CHECK_OF_ENCRYPTED});
754         }
755
756         if (asset->raw_xml()) {
757                 /* Deluxe require this in their QC even if it seems never to be mentioned in any standard */
758                 cxml::Document doc("SubtitleReel");
759                 doc.read_string(*asset->raw_xml());
760                 auto issue_date = doc.string_child("IssueDate");
761                 std::regex reg("^\\d\\d\\d\\d-\\d\\d-\\d\\dT\\d\\d:\\d\\d:\\d\\d$");
762                 if (!std::regex_match(issue_date, reg)) {
763                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_ISSUE_DATE, issue_date});
764                 }
765         }
766 }
767
768
769 /** Verify all subtitle stuff */
770 static void
771 verify_subtitle_asset (
772         shared_ptr<const SubtitleAsset> asset,
773         optional<int64_t> reel_asset_duration,
774         function<void (string, optional<boost::filesystem::path>)> stage,
775         boost::filesystem::path xsd_dtd_directory,
776         vector<VerificationNote>& notes,
777         State& state
778         )
779 {
780         stage ("Checking subtitle XML", asset->file());
781         /* Note: we must not use SubtitleAsset::xml_as_string() here as that will mean the data on disk
782          * gets passed through libdcp which may clean up and therefore hide errors.
783          */
784         if (asset->raw_xml()) {
785                 validate_xml (asset->raw_xml().get(), xsd_dtd_directory, notes);
786         } else {
787                 notes.push_back ({VerificationNote::Type::WARNING, VerificationNote::Code::MISSED_CHECK_OF_ENCRYPTED});
788         }
789
790         auto namespace_count = [](shared_ptr<const SubtitleAsset> asset, string root_node) {
791                 cxml::Document doc(root_node);
792                 doc.read_string(asset->raw_xml().get());
793                 auto root = dynamic_cast<xmlpp::Element*>(doc.node())->cobj();
794                 int count = 0;
795                 for (auto ns = root->nsDef; ns != nullptr; ns = ns->next) {
796                         ++count;
797                 }
798                 return count;
799         };
800
801         auto interop = dynamic_pointer_cast<const InteropSubtitleAsset>(asset);
802         if (interop) {
803                 verify_interop_subtitle_asset(interop, notes);
804                 if (namespace_count(asset, "DCSubtitle") > 1) {
805                         notes.push_back({ VerificationNote::Type::WARNING, VerificationNote::Code::INCORRECT_SUBTITLE_NAMESPACE_COUNT, asset->id() });
806                 }
807         }
808
809         auto smpte = dynamic_pointer_cast<const SMPTESubtitleAsset>(asset);
810         if (smpte) {
811                 verify_smpte_timed_text_asset (smpte, reel_asset_duration, notes);
812                 verify_smpte_subtitle_asset (smpte, notes, state);
813                 /* This asset may be encrypted and in that case we'll have no raw_xml() */
814                 if (asset->raw_xml() && namespace_count(asset, "SubtitleReel") > 1) {
815                         notes.push_back({ VerificationNote::Type::WARNING, VerificationNote::Code::INCORRECT_SUBTITLE_NAMESPACE_COUNT, asset->id()});
816                 }
817         }
818 }
819
820
821 /** Verify all closed caption stuff */
822 static void
823 verify_closed_caption_asset (
824         shared_ptr<const SubtitleAsset> asset,
825         optional<int64_t> reel_asset_duration,
826         function<void (string, optional<boost::filesystem::path>)> stage,
827         boost::filesystem::path xsd_dtd_directory,
828         vector<VerificationNote>& notes
829         )
830 {
831         stage ("Checking closed caption XML", asset->file());
832         /* Note: we must not use SubtitleAsset::xml_as_string() here as that will mean the data on disk
833          * gets passed through libdcp which may clean up and therefore hide errors.
834          */
835         auto raw_xml = asset->raw_xml();
836         if (raw_xml) {
837                 validate_xml (*raw_xml, xsd_dtd_directory, notes);
838                 if (raw_xml->size() > 256 * 1024) {
839                         notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_CLOSED_CAPTION_XML_SIZE_IN_BYTES, raw_convert<string>(raw_xml->size()), *asset->file()});
840                 }
841         } else {
842                 notes.push_back ({VerificationNote::Type::WARNING, VerificationNote::Code::MISSED_CHECK_OF_ENCRYPTED});
843         }
844
845         auto smpte = dynamic_pointer_cast<const SMPTESubtitleAsset>(asset);
846         if (smpte) {
847                 verify_smpte_timed_text_asset (smpte, reel_asset_duration, notes);
848         }
849 }
850
851
852 /** Check the timing of the individual subtitles and make sure there are no empty <Text> nodes etc. */
853 static
854 void
855 verify_text_details (
856         vector<shared_ptr<Reel>> reels,
857         int edit_rate,
858         vector<VerificationNote>& notes,
859         std::function<bool (shared_ptr<Reel>)> check,
860         std::function<optional<string> (shared_ptr<Reel>)> xml,
861         std::function<int64_t (shared_ptr<Reel>)> duration
862         )
863 {
864         /* end of last subtitle (in editable units) */
865         optional<int64_t> last_out;
866         auto too_short = false;
867         auto too_close = false;
868         auto too_early = false;
869         auto reel_overlap = false;
870         auto empty_text = false;
871         /* current reel start time (in editable units) */
872         int64_t reel_offset = 0;
873         vector<string> font_ids;
874         optional<string> missing_load_font_id;
875
876         std::function<void (cxml::ConstNodePtr, optional<int>, optional<Time>, int, bool)> parse;
877         parse = [&parse, &last_out, &too_short, &too_close, &too_early, &empty_text, &reel_offset, &font_ids, &missing_load_font_id](cxml::ConstNodePtr node, optional<int> tcr, optional<Time> start_time, int er, bool first_reel) {
878                 if (node->name() == "Subtitle") {
879                         Time in (node->string_attribute("TimeIn"), tcr);
880                         if (start_time) {
881                                 in -= *start_time;
882                         }
883                         Time out (node->string_attribute("TimeOut"), tcr);
884                         if (start_time) {
885                                 out -= *start_time;
886                         }
887                         if (first_reel && tcr && in < Time(0, 0, 4, 0, *tcr)) {
888                                 too_early = true;
889                         }
890                         auto length = out - in;
891                         if (length.as_editable_units_ceil(er) < 15) {
892                                 too_short = true;
893                         }
894                         if (last_out) {
895                                 /* XXX: this feels dubious - is it really what Bv2.1 means? */
896                                 auto distance = reel_offset + in.as_editable_units_ceil(er) - *last_out;
897                                 if (distance >= 0 && distance < 2) {
898                                         too_close = true;
899                                 }
900                         }
901                         last_out = reel_offset + out.as_editable_units_floor(er);
902                 } else if (node->name() == "Text") {
903                         std::function<bool (cxml::ConstNodePtr)> node_has_content = [&](cxml::ConstNodePtr node) {
904                                 if (!node->content().empty()) {
905                                         return true;
906                                 }
907                                 for (auto i: node->node_children()) {
908                                         if (node_has_content(i)) {
909                                                 return true;
910                                         }
911                                 }
912                                 return false;
913                         };
914                         if (!node_has_content(node)) {
915                                 empty_text = true;
916                         }
917                 } else if (node->name() == "LoadFont") {
918                         if (auto const id = node->optional_string_attribute("Id")) {
919                                 font_ids.push_back(*id);
920                         }
921                 } else if (node->name() == "Font") {
922                         if (auto const font_id = node->optional_string_attribute("Id")) {
923                                 if (std::find_if(font_ids.begin(), font_ids.end(), [font_id](string const& id) { return id == font_id; }) == font_ids.end()) {
924                                         missing_load_font_id = font_id;
925                                 }
926                         }
927                 }
928                 for (auto i: node->node_children()) {
929                         parse(i, tcr, start_time, er, first_reel);
930                 }
931         };
932
933         for (auto i = 0U; i < reels.size(); ++i) {
934                 if (!check(reels[i])) {
935                         continue;
936                 }
937
938                 auto reel_xml = xml(reels[i]);
939                 if (!reel_xml) {
940                         notes.push_back ({VerificationNote::Type::WARNING, VerificationNote::Code::MISSED_CHECK_OF_ENCRYPTED});
941                         continue;
942                 }
943
944                 /* We need to look at <Subtitle> instances in the XML being checked, so we can't use the subtitles
945                  * read in by libdcp's parser.
946                  */
947
948                 shared_ptr<cxml::Document> doc;
949                 optional<int> tcr;
950                 optional<Time> start_time;
951                 try {
952                         doc = make_shared<cxml::Document>("SubtitleReel");
953                         doc->read_string (*reel_xml);
954                         tcr = doc->number_child<int>("TimeCodeRate");
955                         auto start_time_string = doc->optional_string_child("StartTime");
956                         if (start_time_string) {
957                                 start_time = Time(*start_time_string, tcr);
958                         }
959                 } catch (...) {
960                         doc = make_shared<cxml::Document>("DCSubtitle");
961                         doc->read_string (*reel_xml);
962                 }
963                 parse (doc, tcr, start_time, edit_rate, i == 0);
964                 auto end = reel_offset + duration(reels[i]);
965                 if (last_out && *last_out > end) {
966                         reel_overlap = true;
967                 }
968                 reel_offset = end;
969         }
970
971         if (last_out && *last_out > reel_offset) {
972                 reel_overlap = true;
973         }
974
975         if (too_early) {
976                 notes.push_back({
977                         VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_FIRST_TEXT_TIME
978                 });
979         }
980
981         if (too_short) {
982                 notes.push_back ({
983                         VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_DURATION
984                 });
985         }
986
987         if (too_close) {
988                 notes.push_back ({
989                         VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_SPACING
990                 });
991         }
992
993         if (reel_overlap) {
994                 notes.push_back ({
995                         VerificationNote::Type::ERROR, VerificationNote::Code::SUBTITLE_OVERLAPS_REEL_BOUNDARY
996                 });
997         }
998
999         if (empty_text) {
1000                 notes.push_back ({
1001                         VerificationNote::Type::WARNING, VerificationNote::Code::EMPTY_TEXT
1002                 });
1003         }
1004
1005         if (missing_load_font_id) {
1006                 notes.push_back(dcp::VerificationNote(VerificationNote::Type::ERROR, VerificationNote::Code::MISSING_LOAD_FONT_FOR_FONT).set_id(*missing_load_font_id));
1007         }
1008 }
1009
1010
1011 static
1012 void
1013 verify_closed_caption_details (
1014         vector<shared_ptr<Reel>> reels,
1015         vector<VerificationNote>& notes
1016         )
1017 {
1018         std::function<void (cxml::ConstNodePtr node, std::vector<cxml::ConstNodePtr>& text_or_image)> find_text_or_image;
1019         find_text_or_image = [&find_text_or_image](cxml::ConstNodePtr node, std::vector<cxml::ConstNodePtr>& text_or_image) {
1020                 for (auto i: node->node_children()) {
1021                         if (i->name() == "Text") {
1022                                 text_or_image.push_back (i);
1023                         } else {
1024                                 find_text_or_image (i, text_or_image);
1025                         }
1026                 }
1027         };
1028
1029         auto mismatched_valign = false;
1030         auto incorrect_order = false;
1031
1032         std::function<void (cxml::ConstNodePtr)> parse;
1033         parse = [&parse, &find_text_or_image, &mismatched_valign, &incorrect_order](cxml::ConstNodePtr node) {
1034                 if (node->name() == "Subtitle") {
1035                         vector<cxml::ConstNodePtr> text_or_image;
1036                         find_text_or_image (node, text_or_image);
1037                         optional<string> last_valign;
1038                         optional<float> last_vpos;
1039                         for (auto i: text_or_image) {
1040                                 auto valign = i->optional_string_attribute("VAlign");
1041                                 if (!valign) {
1042                                         valign = i->optional_string_attribute("Valign").get_value_or("center");
1043                                 }
1044                                 auto vpos = i->optional_number_attribute<float>("VPosition");
1045                                 if (!vpos) {
1046                                         vpos = i->optional_number_attribute<float>("Vposition").get_value_or(50);
1047                                 }
1048
1049                                 if (last_valign) {
1050                                         if (*last_valign != valign) {
1051                                                 mismatched_valign = true;
1052                                         }
1053                                 }
1054                                 last_valign = valign;
1055
1056                                 if (!mismatched_valign) {
1057                                         if (last_vpos) {
1058                                                 if (*last_valign == "top" || *last_valign == "center") {
1059                                                         if (*vpos < *last_vpos) {
1060                                                                 incorrect_order = true;
1061                                                         }
1062                                                 } else {
1063                                                         if (*vpos > *last_vpos) {
1064                                                                 incorrect_order = true;
1065                                                         }
1066                                                 }
1067                                         }
1068                                         last_vpos = vpos;
1069                                 }
1070                         }
1071                 }
1072
1073                 for (auto i: node->node_children()) {
1074                         parse(i);
1075                 }
1076         };
1077
1078         for (auto reel: reels) {
1079                 for (auto ccap: reel->closed_captions()) {
1080                         auto reel_xml = ccap->asset()->raw_xml();
1081                         if (!reel_xml) {
1082                                 notes.push_back ({VerificationNote::Type::WARNING, VerificationNote::Code::MISSED_CHECK_OF_ENCRYPTED});
1083                                 continue;
1084                         }
1085
1086                         /* We need to look at <Subtitle> instances in the XML being checked, so we can't use the subtitles
1087                          * read in by libdcp's parser.
1088                          */
1089
1090                         shared_ptr<cxml::Document> doc;
1091                         optional<int> tcr;
1092                         optional<Time> start_time;
1093                         try {
1094                                 doc = make_shared<cxml::Document>("SubtitleReel");
1095                                 doc->read_string (*reel_xml);
1096                         } catch (...) {
1097                                 doc = make_shared<cxml::Document>("DCSubtitle");
1098                                 doc->read_string (*reel_xml);
1099                         }
1100                         parse (doc);
1101                 }
1102         }
1103
1104         if (mismatched_valign) {
1105                 notes.push_back ({
1106                         VerificationNote::Type::ERROR, VerificationNote::Code::MISMATCHED_CLOSED_CAPTION_VALIGN,
1107                 });
1108         }
1109
1110         if (incorrect_order) {
1111                 notes.push_back ({
1112                         VerificationNote::Type::ERROR, VerificationNote::Code::INCORRECT_CLOSED_CAPTION_ORDERING,
1113                 });
1114         }
1115 }
1116
1117
1118 struct LinesCharactersResult
1119 {
1120         bool warning_length_exceeded = false;
1121         bool error_length_exceeded = false;
1122         bool line_count_exceeded = false;
1123 };
1124
1125
1126 static
1127 void
1128 verify_text_lines_and_characters (
1129         shared_ptr<SubtitleAsset> asset,
1130         int warning_length,
1131         int error_length,
1132         LinesCharactersResult* result
1133         )
1134 {
1135         class Event
1136         {
1137         public:
1138                 Event (Time time_, float position_, int characters_)
1139                         : time (time_)
1140                         , position (position_)
1141                         , characters (characters_)
1142                 {}
1143
1144                 Event (Time time_, shared_ptr<Event> start_)
1145                         : time (time_)
1146                         , start (start_)
1147                 {}
1148
1149                 Time time;
1150                 int position; //< position from 0 at top of screen to 100 at bottom
1151                 int characters;
1152                 shared_ptr<Event> start;
1153         };
1154
1155         vector<shared_ptr<Event>> events;
1156
1157         auto position = [](shared_ptr<const SubtitleString> sub) {
1158                 switch (sub->v_align()) {
1159                 case VAlign::TOP:
1160                         return lrintf(sub->v_position() * 100);
1161                 case VAlign::CENTER:
1162                         return lrintf((0.5f + sub->v_position()) * 100);
1163                 case VAlign::BOTTOM:
1164                         return lrintf((1.0f - sub->v_position()) * 100);
1165                 }
1166
1167                 return 0L;
1168         };
1169
1170         for (auto j: asset->subtitles()) {
1171                 auto text = dynamic_pointer_cast<const SubtitleString>(j);
1172                 if (text) {
1173                         auto in = make_shared<Event>(text->in(), position(text), text->text().length());
1174                         events.push_back(in);
1175                         events.push_back(make_shared<Event>(text->out(), in));
1176                 }
1177         }
1178
1179         std::sort(events.begin(), events.end(), [](shared_ptr<Event> const& a, shared_ptr<Event>const& b) {
1180                 return a->time < b->time;
1181         });
1182
1183         map<int, int> current;
1184         for (auto i: events) {
1185                 if (current.size() > 3) {
1186                         result->line_count_exceeded = true;
1187                 }
1188                 for (auto j: current) {
1189                         if (j.second > warning_length) {
1190                                 result->warning_length_exceeded = true;
1191                         }
1192                         if (j.second > error_length) {
1193                                 result->error_length_exceeded = true;
1194                         }
1195                 }
1196
1197                 if (i->start) {
1198                         /* end of a subtitle */
1199                         DCP_ASSERT (current.find(i->start->position) != current.end());
1200                         if (current[i->start->position] == i->start->characters) {
1201                                 current.erase(i->start->position);
1202                         } else {
1203                                 current[i->start->position] -= i->start->characters;
1204                         }
1205                 } else {
1206                         /* start of a subtitle */
1207                         if (current.find(i->position) == current.end()) {
1208                                 current[i->position] = i->characters;
1209                         } else {
1210                                 current[i->position] += i->characters;
1211                         }
1212                 }
1213         }
1214 }
1215
1216
1217 static
1218 void
1219 verify_text_details (vector<shared_ptr<Reel>> reels, vector<VerificationNote>& notes)
1220 {
1221         if (reels.empty()) {
1222                 return;
1223         }
1224
1225         if (reels[0]->main_subtitle()) {
1226                 verify_text_details (reels, reels[0]->main_subtitle()->edit_rate().numerator, notes,
1227                         [](shared_ptr<Reel> reel) {
1228                                 return static_cast<bool>(reel->main_subtitle());
1229                         },
1230                         [](shared_ptr<Reel> reel) {
1231                                 return reel->main_subtitle()->asset()->raw_xml();
1232                         },
1233                         [](shared_ptr<Reel> reel) {
1234                                 return reel->main_subtitle()->actual_duration();
1235                         }
1236                 );
1237         }
1238
1239         for (auto i = 0U; i < reels[0]->closed_captions().size(); ++i) {
1240                 verify_text_details (reels, reels[0]->closed_captions()[i]->edit_rate().numerator, notes,
1241                         [i](shared_ptr<Reel> reel) {
1242                                 return i < reel->closed_captions().size();
1243                         },
1244                         [i](shared_ptr<Reel> reel) {
1245                                 return reel->closed_captions()[i]->asset()->raw_xml();
1246                         },
1247                         [i](shared_ptr<Reel> reel) {
1248                                 return reel->closed_captions()[i]->actual_duration();
1249                         }
1250                 );
1251         }
1252
1253         verify_closed_caption_details (reels, notes);
1254 }
1255
1256
1257 void
1258 verify_extension_metadata(shared_ptr<const CPL> cpl, vector<VerificationNote>& notes)
1259 {
1260         DCP_ASSERT (cpl->file());
1261         cxml::Document doc ("CompositionPlaylist");
1262         doc.read_file (cpl->file().get());
1263
1264         auto missing = false;
1265         string malformed;
1266
1267         if (auto reel_list = doc.node_child("ReelList")) {
1268                 auto reels = reel_list->node_children("Reel");
1269                 if (!reels.empty()) {
1270                         if (auto asset_list = reels[0]->optional_node_child("AssetList")) {
1271                                 if (auto metadata = asset_list->optional_node_child("CompositionMetadataAsset")) {
1272                                         if (auto extension_list = metadata->optional_node_child("ExtensionMetadataList")) {
1273                                                 missing = true;
1274                                                 for (auto extension: extension_list->node_children("ExtensionMetadata")) {
1275                                                         if (extension->optional_string_attribute("scope").get_value_or("") != "http://isdcf.com/ns/cplmd/app") {
1276                                                                 continue;
1277                                                         }
1278                                                         missing = false;
1279                                                         if (auto name = extension->optional_node_child("Name")) {
1280                                                                 if (name->content() != "Application") {
1281                                                                         malformed = "<Name> should be 'Application'";
1282                                                                 }
1283                                                         }
1284                                                         if (auto property_list = extension->optional_node_child("PropertyList")) {
1285                                                                 if (auto property = property_list->optional_node_child("Property")) {
1286                                                                         if (auto name = property->optional_node_child("Name")) {
1287                                                                                 if (name->content() != "DCP Constraints Profile") {
1288                                                                                         malformed = "<Name> property should be 'DCP Constraints Profile'";
1289                                                                                 }
1290                                                                         }
1291                                                                         if (auto value = property->optional_node_child("Value")) {
1292                                                                                 if (value->content() != "SMPTE-RDD-52:2020-Bv2.1") {
1293                                                                                         malformed = "<Value> property should be 'SMPTE-RDD-52:2020-Bv2.1'";
1294                                                                                 }
1295                                                                         }
1296                                                                 }
1297                                                         }
1298                                                 }
1299                                         } else {
1300                                                 missing = true;
1301                                         }
1302                                 }
1303                         }
1304                 }
1305         }
1306
1307         if (missing) {
1308                 notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_EXTENSION_METADATA, cpl->id(), cpl->file().get()});
1309         } else if (!malformed.empty()) {
1310                 notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_EXTENSION_METADATA, malformed, cpl->file().get()});
1311         }
1312 }
1313
1314
1315 bool
1316 pkl_has_encrypted_assets(shared_ptr<const DCP> dcp, shared_ptr<const PKL> pkl)
1317 {
1318         vector<string> encrypted;
1319         for (auto i: dcp->cpls()) {
1320                 for (auto j: i->reel_file_assets()) {
1321                         if (j->asset_ref().resolved()) {
1322                                 auto mxf = dynamic_pointer_cast<MXF>(j->asset_ref().asset());
1323                                 if (mxf && mxf->encrypted()) {
1324                                         encrypted.push_back(j->asset_ref().id());
1325                                 }
1326                         }
1327                 }
1328         }
1329
1330         for (auto i: pkl->assets()) {
1331                 if (find(encrypted.begin(), encrypted.end(), i->id()) != encrypted.end()) {
1332                         return true;
1333                 }
1334         }
1335
1336         return false;
1337 }
1338
1339
1340 static
1341 void
1342 verify_reel(
1343         shared_ptr<const DCP> dcp,
1344         shared_ptr<const CPL> cpl,
1345         shared_ptr<const Reel> reel,
1346         optional<dcp::Size> main_picture_active_area,
1347         function<void (string, optional<boost::filesystem::path>)> stage,
1348         boost::filesystem::path xsd_dtd_directory,
1349         function<void (float)> progress,
1350         VerificationOptions options,
1351         vector<VerificationNote>& notes,
1352         State& state,
1353         bool* have_main_subtitle,
1354         bool* have_no_main_subtitle,
1355         size_t* most_closed_captions,
1356         size_t* fewest_closed_captions,
1357         map<Marker, Time>* markers_seen
1358         )
1359 {
1360         for (auto i: reel->assets()) {
1361                 if (i->duration() && (i->duration().get() * i->edit_rate().denominator / i->edit_rate().numerator) < 1) {
1362                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::INVALID_DURATION, i->id()});
1363                 }
1364                 if ((i->intrinsic_duration() * i->edit_rate().denominator / i->edit_rate().numerator) < 1) {
1365                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::INVALID_INTRINSIC_DURATION, i->id()});
1366                 }
1367                 auto file_asset = dynamic_pointer_cast<ReelFileAsset>(i);
1368                 if (i->encryptable() && !file_asset->hash()) {
1369                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_HASH, i->id()});
1370                 }
1371         }
1372
1373         if (dcp->standard() == Standard::SMPTE) {
1374                 boost::optional<int64_t> duration;
1375                 for (auto i: reel->assets()) {
1376                         if (!duration) {
1377                                 duration = i->actual_duration();
1378                         } else if (*duration != i->actual_duration()) {
1379                                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISMATCHED_ASSET_DURATION});
1380                                 break;
1381                         }
1382                 }
1383         }
1384
1385         if (reel->main_picture()) {
1386                 /* Check reel stuff */
1387                 auto const frame_rate = reel->main_picture()->frame_rate();
1388                 if (frame_rate.denominator != 1 ||
1389                     (frame_rate.numerator != 24 &&
1390                      frame_rate.numerator != 25 &&
1391                      frame_rate.numerator != 30 &&
1392                      frame_rate.numerator != 48 &&
1393                      frame_rate.numerator != 50 &&
1394                      frame_rate.numerator != 60 &&
1395                      frame_rate.numerator != 96)) {
1396                         notes.push_back({
1397                                 VerificationNote::Type::ERROR,
1398                                 VerificationNote::Code::INVALID_PICTURE_FRAME_RATE,
1399                                 String::compose("%1/%2", frame_rate.numerator, frame_rate.denominator)
1400                         });
1401                 }
1402                 /* Check asset */
1403                 if (reel->main_picture()->asset_ref().resolved()) {
1404                         verify_main_picture_asset(dcp, reel->main_picture(), stage, progress, options, notes);
1405                         auto const asset_size = reel->main_picture()->asset()->size();
1406                         if (main_picture_active_area) {
1407                                 if (main_picture_active_area->width > asset_size.width) {
1408                                         notes.push_back({
1409                                                         VerificationNote::Type::ERROR,
1410                                                         VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA,
1411                                                         String::compose("width %1 is bigger than the asset width %2", main_picture_active_area->width, asset_size.width),
1412                                                         cpl->file().get()
1413                                                         });
1414                                 }
1415                                 if (main_picture_active_area->height > asset_size.height) {
1416                                         notes.push_back({
1417                                                         VerificationNote::Type::ERROR,
1418                                                         VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA,
1419                                                         String::compose("height %1 is bigger than the asset height %2", main_picture_active_area->height, asset_size.height),
1420                                                         cpl->file().get()
1421                                                         });
1422                                 }
1423                         }
1424                 }
1425         }
1426
1427         if (reel->main_sound() && reel->main_sound()->asset_ref().resolved()) {
1428                 verify_main_sound_asset(dcp, reel->main_sound(), stage, progress, options, notes, state);
1429         }
1430
1431         if (reel->main_subtitle()) {
1432                 verify_main_subtitle_reel(reel->main_subtitle(), notes);
1433                 if (reel->main_subtitle()->asset_ref().resolved()) {
1434                         verify_subtitle_asset(reel->main_subtitle()->asset(), reel->main_subtitle()->duration(), stage, xsd_dtd_directory, notes, state);
1435                 }
1436                 *have_main_subtitle = true;
1437         } else {
1438                 *have_no_main_subtitle = true;
1439         }
1440
1441         for (auto i: reel->closed_captions()) {
1442                 verify_closed_caption_reel(i, notes);
1443                 if (i->asset_ref().resolved()) {
1444                         verify_closed_caption_asset(i->asset(), i->duration(), stage, xsd_dtd_directory, notes);
1445                 }
1446         }
1447
1448         if (reel->main_markers()) {
1449                 for (auto const& i: reel->main_markers()->get()) {
1450                         markers_seen->insert(i);
1451                 }
1452                 if (reel->main_markers()->entry_point()) {
1453                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::UNEXPECTED_ENTRY_POINT});
1454                 }
1455                 if (reel->main_markers()->duration()) {
1456                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::UNEXPECTED_DURATION});
1457                 }
1458         }
1459
1460         *fewest_closed_captions = std::min(*fewest_closed_captions, reel->closed_captions().size());
1461         *most_closed_captions = std::max(*most_closed_captions, reel->closed_captions().size());
1462
1463 }
1464
1465
1466 static
1467 void
1468 verify_cpl(
1469         shared_ptr<const DCP> dcp,
1470         shared_ptr<const CPL> cpl,
1471         function<void (string, optional<boost::filesystem::path>)> stage,
1472         boost::filesystem::path xsd_dtd_directory,
1473         function<void (float)> progress,
1474         VerificationOptions options,
1475         vector<VerificationNote>& notes,
1476         State& state
1477         )
1478 {
1479         stage("Checking CPL", cpl->file());
1480         validate_xml(cpl->file().get(), xsd_dtd_directory, notes);
1481
1482         if (cpl->any_encrypted() && !cpl->all_encrypted()) {
1483                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::PARTIALLY_ENCRYPTED});
1484         }
1485
1486         for (auto const& i: cpl->additional_subtitle_languages()) {
1487                 verify_language_tag(i, notes);
1488         }
1489
1490         if (!cpl->content_kind().scope() || *cpl->content_kind().scope() == "http://www.smpte-ra.org/schemas/429-7/2006/CPL#standard-content") {
1491                 /* This is a content kind from http://www.smpte-ra.org/schemas/429-7/2006/CPL#standard-content; make sure it's one
1492                  * of the approved ones.
1493                  */
1494                 auto all = ContentKind::all();
1495                 auto name = cpl->content_kind().name();
1496                 transform(name.begin(), name.end(), name.begin(), ::tolower);
1497                 auto iter = std::find_if(all.begin(), all.end(), [name](ContentKind const& k) { return !k.scope() && k.name() == name; });
1498                 if (iter == all.end()) {
1499                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::INVALID_CONTENT_KIND, cpl->content_kind().name()});
1500                 }
1501         }
1502
1503         if (cpl->release_territory()) {
1504                 if (!cpl->release_territory_scope() || cpl->release_territory_scope().get() != "http://www.smpte-ra.org/schemas/429-16/2014/CPL-Metadata#scope/release-territory/UNM49") {
1505                         auto terr = cpl->release_territory().get();
1506                         /* Must be a valid region tag, or "001" */
1507                         try {
1508                                 LanguageTag::RegionSubtag test(terr);
1509                         } catch (...) {
1510                                 if (terr != "001") {
1511                                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_LANGUAGE, terr});
1512                                 }
1513                         }
1514                 }
1515         }
1516
1517         if (dcp->standard() == Standard::SMPTE) {
1518                 if (!cpl->annotation_text()) {
1519                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_CPL_ANNOTATION_TEXT, cpl->id(), cpl->file().get()});
1520                 } else if (cpl->annotation_text().get() != cpl->content_title_text()) {
1521                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::MISMATCHED_CPL_ANNOTATION_TEXT, cpl->id(), cpl->file().get()});
1522                 }
1523         }
1524
1525         for (auto i: dcp->pkls()) {
1526                 /* Check that the CPL's hash corresponds to the PKL */
1527                 optional<string> h = i->hash(cpl->id());
1528                 if (h && make_digest(ArrayData(*cpl->file())) != *h) {
1529                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::MISMATCHED_CPL_HASHES, cpl->id(), cpl->file().get()});
1530                 }
1531
1532                 /* Check that any PKL with a single CPL has its AnnotationText the same as the CPL's ContentTitleText */
1533                 optional<string> required_annotation_text;
1534                 for (auto j: i->assets()) {
1535                         /* See if this is a CPL */
1536                         for (auto k: dcp->cpls()) {
1537                                 if (j->id() == k->id()) {
1538                                         if (!required_annotation_text) {
1539                                                 /* First CPL we have found; this is the required AnnotationText unless we find another */
1540                                                 required_annotation_text = cpl->content_title_text();
1541                                         } else {
1542                                                 /* There's more than one CPL so we don't care what the PKL's AnnotationText is */
1543                                                 required_annotation_text = boost::none;
1544                                         }
1545                                 }
1546                         }
1547                 }
1548
1549                 if (required_annotation_text && i->annotation_text() != required_annotation_text) {
1550                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISMATCHED_PKL_ANNOTATION_TEXT_WITH_CPL, i->id(), i->file().get()});
1551                 }
1552         }
1553
1554         /* set to true if any reel has a MainSubtitle */
1555         auto have_main_subtitle = false;
1556         /* set to true if any reel has no MainSubtitle */
1557         auto have_no_main_subtitle = false;
1558         /* fewest number of closed caption assets seen in a reel */
1559         size_t fewest_closed_captions = SIZE_MAX;
1560         /* most number of closed caption assets seen in a reel */
1561         size_t most_closed_captions = 0;
1562         map<Marker, Time> markers_seen;
1563
1564         auto const main_picture_active_area = cpl->main_picture_active_area();
1565         if (main_picture_active_area && (main_picture_active_area->width % 2)) {
1566                 notes.push_back({
1567                                 VerificationNote::Type::ERROR,
1568                                 VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA,
1569                                 String::compose("width %1 is not a multiple of 2", main_picture_active_area->width),
1570                                 cpl->file().get()
1571                         });
1572         }
1573         if (main_picture_active_area && (main_picture_active_area->height % 2)) {
1574                 notes.push_back({
1575                                 VerificationNote::Type::ERROR,
1576                                 VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA,
1577                                 String::compose("height %1 is not a multiple of 2", main_picture_active_area->height),
1578                                 cpl->file().get()
1579                         });
1580         }
1581
1582         for (auto reel: cpl->reels()) {
1583                 stage("Checking reel", optional<boost::filesystem::path>());
1584                 verify_reel(
1585                         dcp,
1586                         cpl,
1587                         reel,
1588                         main_picture_active_area,
1589                         stage,
1590                         xsd_dtd_directory,
1591                         progress,
1592                         options,
1593                         notes,
1594                         state,
1595                         &have_main_subtitle,
1596                         &have_no_main_subtitle,
1597                         &most_closed_captions,
1598                         &fewest_closed_captions,
1599                         &markers_seen
1600                         );
1601         }
1602
1603         verify_text_details(cpl->reels(), notes);
1604
1605         if (dcp->standard() == Standard::SMPTE) {
1606                 if (auto msc = cpl->main_sound_configuration()) {
1607                         if (state.audio_channels && msc->channels() != *state.audio_channels) {
1608                                 notes.push_back({
1609                                                 VerificationNote::Type::ERROR,
1610                                                 VerificationNote::Code::INVALID_MAIN_SOUND_CONFIGURATION,
1611                                                 String::compose("MainSoundConfiguration has %1 channels but sound assets have %2", msc->channels(), *state.audio_channels),
1612                                                 cpl->file().get()
1613                                         });
1614                         }
1615                 }
1616
1617                 if (have_main_subtitle && have_no_main_subtitle) {
1618                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_MAIN_SUBTITLE_FROM_SOME_REELS});
1619                 }
1620
1621                 if (fewest_closed_captions != most_closed_captions) {
1622                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISMATCHED_CLOSED_CAPTION_ASSET_COUNTS});
1623                 }
1624
1625                 if (cpl->content_kind() == ContentKind::FEATURE) {
1626                         if (markers_seen.find(Marker::FFEC) == markers_seen.end()) {
1627                                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_FFEC_IN_FEATURE});
1628                         }
1629                         if (markers_seen.find(Marker::FFMC) == markers_seen.end()) {
1630                                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_FFMC_IN_FEATURE});
1631                         }
1632                 }
1633
1634                 auto ffoc = markers_seen.find(Marker::FFOC);
1635                 if (ffoc == markers_seen.end()) {
1636                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::MISSING_FFOC});
1637                 } else if (ffoc->second.e != 1) {
1638                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::INCORRECT_FFOC, raw_convert<string>(ffoc->second.e)});
1639                 }
1640
1641                 auto lfoc = markers_seen.find(Marker::LFOC);
1642                 if (lfoc == markers_seen.end()) {
1643                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::MISSING_LFOC});
1644                 } else {
1645                         auto lfoc_time = lfoc->second.as_editable_units_ceil(lfoc->second.tcr);
1646                         if (lfoc_time != (cpl->reels().back()->duration() - 1)) {
1647                                 notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::INCORRECT_LFOC, raw_convert<string>(lfoc_time)});
1648                         }
1649                 }
1650
1651                 LinesCharactersResult result;
1652                 for (auto reel: cpl->reels()) {
1653                         if (reel->main_subtitle() && reel->main_subtitle()->asset()) {
1654                                 verify_text_lines_and_characters(reel->main_subtitle()->asset(), 52, 79, &result);
1655                         }
1656                 }
1657
1658                 if (result.line_count_exceeded) {
1659                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_LINE_COUNT});
1660                 }
1661                 if (result.error_length_exceeded) {
1662                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_LINE_LENGTH});
1663                 } else if (result.warning_length_exceeded) {
1664                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::NEARLY_INVALID_SUBTITLE_LINE_LENGTH});
1665                 }
1666
1667                 result = LinesCharactersResult();
1668                 for (auto reel: cpl->reels()) {
1669                         for (auto i: reel->closed_captions()) {
1670                                 if (i->asset()) {
1671                                         verify_text_lines_and_characters(i->asset(), 32, 32, &result);
1672                                 }
1673                         }
1674                 }
1675
1676                 if (result.line_count_exceeded) {
1677                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_CLOSED_CAPTION_LINE_COUNT});
1678                 }
1679                 if (result.error_length_exceeded) {
1680                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_CLOSED_CAPTION_LINE_LENGTH});
1681                 }
1682
1683                 if (!cpl->read_composition_metadata()) {
1684                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_CPL_METADATA, cpl->id(), cpl->file().get()});
1685                 } else if (!cpl->version_number()) {
1686                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_CPL_METADATA_VERSION_NUMBER, cpl->id(), cpl->file().get()});
1687                 }
1688
1689                 verify_extension_metadata(cpl, notes);
1690
1691                 if (cpl->any_encrypted()) {
1692                         cxml::Document doc("CompositionPlaylist");
1693                         DCP_ASSERT(cpl->file());
1694                         doc.read_file(cpl->file().get());
1695                         if (!doc.optional_node_child("Signature")) {
1696                                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::UNSIGNED_CPL_WITH_ENCRYPTED_CONTENT, cpl->id(), cpl->file().get()});
1697                         }
1698                 }
1699         }
1700 }
1701
1702
1703 static
1704 void
1705 verify_pkl(
1706         shared_ptr<const DCP> dcp,
1707         shared_ptr<const PKL> pkl,
1708         boost::filesystem::path xsd_dtd_directory,
1709         vector<VerificationNote>& notes
1710         )
1711 {
1712         validate_xml(pkl->file().get(), xsd_dtd_directory, notes);
1713
1714         if (pkl_has_encrypted_assets(dcp, pkl)) {
1715                 cxml::Document doc("PackingList");
1716                 doc.read_file(pkl->file().get());
1717                 if (!doc.optional_node_child("Signature")) {
1718                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::UNSIGNED_PKL_WITH_ENCRYPTED_CONTENT, pkl->id(), pkl->file().get()});
1719                 }
1720         }
1721
1722         set<string> uuid_set;
1723         for (auto asset: pkl->assets()) {
1724                 if (!uuid_set.insert(asset->id()).second) {
1725                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::DUPLICATE_ASSET_ID_IN_PKL, pkl->id(), pkl->file().get()});
1726                         break;
1727                 }
1728         }
1729 }
1730
1731
1732
1733 static
1734 void
1735 verify_assetmap(
1736         shared_ptr<const DCP> dcp,
1737         boost::filesystem::path xsd_dtd_directory,
1738         vector<VerificationNote>& notes
1739         )
1740 {
1741         auto asset_map = dcp->asset_map();
1742         DCP_ASSERT(asset_map);
1743
1744         validate_xml(asset_map->file().get(), xsd_dtd_directory, notes);
1745
1746         set<string> uuid_set;
1747         for (auto const& asset: asset_map->assets()) {
1748                 if (!uuid_set.insert(asset.id()).second) {
1749                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::DUPLICATE_ASSET_ID_IN_ASSETMAP, asset_map->id(), asset_map->file().get()});
1750                         break;
1751                 }
1752         }
1753 }
1754
1755
1756 vector<VerificationNote>
1757 dcp::verify (
1758         vector<boost::filesystem::path> directories,
1759         function<void (string, optional<boost::filesystem::path>)> stage,
1760         function<void (float)> progress,
1761         VerificationOptions options,
1762         optional<boost::filesystem::path> xsd_dtd_directory
1763         )
1764 {
1765         if (!xsd_dtd_directory) {
1766                 xsd_dtd_directory = resources_directory() / "xsd";
1767         }
1768         *xsd_dtd_directory = boost::filesystem::canonical (*xsd_dtd_directory);
1769
1770         vector<VerificationNote> notes;
1771         State state{};
1772
1773         vector<shared_ptr<DCP>> dcps;
1774         for (auto i: directories) {
1775                 dcps.push_back (make_shared<DCP>(i));
1776         }
1777
1778         for (auto dcp: dcps) {
1779                 stage ("Checking DCP", dcp->directory());
1780                 bool carry_on = true;
1781                 try {
1782                         dcp->read (&notes, true);
1783                 } catch (MissingAssetmapError& e) {
1784                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1785                         carry_on = false;
1786                 } catch (ReadError& e) {
1787                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1788                 } catch (XMLError& e) {
1789                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1790                 } catch (MXFFileError& e) {
1791                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1792                 } catch (BadURNUUIDError& e) {
1793                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1794                 } catch (cxml::Error& e) {
1795                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1796                 }
1797
1798                 if (!carry_on) {
1799                         continue;
1800                 }
1801
1802                 if (dcp->standard() != Standard::SMPTE) {
1803                         notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_STANDARD});
1804                 }
1805
1806                 for (auto cpl: dcp->cpls()) {
1807                         verify_cpl(
1808                                 dcp,
1809                                 cpl,
1810                                 stage,
1811                                 *xsd_dtd_directory,
1812                                 progress,
1813                                 options,
1814                                 notes,
1815                                 state
1816                                 );
1817                 }
1818
1819                 for (auto pkl: dcp->pkls()) {
1820                         stage("Checking PKL", pkl->file());
1821                         verify_pkl(dcp, pkl, *xsd_dtd_directory, notes);
1822                 }
1823
1824                 if (dcp->asset_map_file()) {
1825                         stage("Checking ASSETMAP", dcp->asset_map_file().get());
1826                         verify_assetmap(dcp, *xsd_dtd_directory, notes);
1827                 } else {
1828                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::MISSING_ASSETMAP});
1829                 }
1830         }
1831
1832         return notes;
1833 }
1834
1835
1836 string
1837 dcp::note_to_string (VerificationNote note)
1838 {
1839         /** These strings should say what is wrong, incorporating any extra details (ID, filenames etc.).
1840          *
1841          *  e.g. "ClosedCaption asset has no <EntryPoint> tag.",
1842          *  not "ClosedCaption assets must have an <EntryPoint> tag."
1843          *
1844          *  It's OK to use XML tag names where they are clear.
1845          *  If both ID and filename are available, use only the ID.
1846          *  End messages with a full stop.
1847          *  Messages should not mention whether or not their errors are a part of Bv2.1.
1848          */
1849         switch (note.code()) {
1850         case VerificationNote::Code::FAILED_READ:
1851                 return *note.note();
1852         case VerificationNote::Code::MISMATCHED_CPL_HASHES:
1853                 return String::compose("The hash of the CPL %1 in the PKL does not agree with the CPL file.", note.note().get());
1854         case VerificationNote::Code::INVALID_PICTURE_FRAME_RATE:
1855                 return String::compose("The picture in a reel has an invalid frame rate %1.", note.note().get());
1856         case VerificationNote::Code::INCORRECT_PICTURE_HASH:
1857                 return String::compose("The hash of the picture asset %1 does not agree with the PKL file.", note.file()->filename());
1858         case VerificationNote::Code::MISMATCHED_PICTURE_HASHES:
1859                 return String::compose("The PKL and CPL hashes differ for the picture asset %1.", note.file()->filename());
1860         case VerificationNote::Code::INCORRECT_SOUND_HASH:
1861                 return String::compose("The hash of the sound asset %1 does not agree with the PKL file.", note.file()->filename());
1862         case VerificationNote::Code::MISMATCHED_SOUND_HASHES:
1863                 return String::compose("The PKL and CPL hashes differ for the sound asset %1.", note.file()->filename());
1864         case VerificationNote::Code::EMPTY_ASSET_PATH:
1865                 return "The asset map contains an empty asset path.";
1866         case VerificationNote::Code::MISSING_ASSET:
1867                 return String::compose("The file %1 for an asset in the asset map cannot be found.", note.file()->filename());
1868         case VerificationNote::Code::MISMATCHED_STANDARD:
1869                 return "The DCP contains both SMPTE and Interop parts.";
1870         case VerificationNote::Code::INVALID_XML:
1871                 return String::compose("An XML file is badly formed: %1 (%2:%3)", note.note().get(), note.file()->filename(), note.line().get());
1872         case VerificationNote::Code::MISSING_ASSETMAP:
1873                 return "No valid ASSETMAP or ASSETMAP.xml was found.";
1874         case VerificationNote::Code::INVALID_INTRINSIC_DURATION:
1875                 return String::compose("The intrinsic duration of the asset %1 is less than 1 second.", note.note().get());
1876         case VerificationNote::Code::INVALID_DURATION:
1877                 return String::compose("The duration of the asset %1 is less than 1 second.", note.note().get());
1878         case VerificationNote::Code::INVALID_PICTURE_FRAME_SIZE_IN_BYTES:
1879                 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());
1880         case VerificationNote::Code::NEARLY_INVALID_PICTURE_FRAME_SIZE_IN_BYTES:
1881                 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());
1882         case VerificationNote::Code::EXTERNAL_ASSET:
1883                 return String::compose("The asset %1 that this DCP refers to is not included in the DCP.  It may be a VF.", note.note().get());
1884         case VerificationNote::Code::THREED_ASSET_MARKED_AS_TWOD:
1885                 return String::compose("The asset %1 is 3D but its MXF is marked as 2D.", note.file()->filename());
1886         case VerificationNote::Code::INVALID_STANDARD:
1887                 return "This DCP does not use the SMPTE standard.";
1888         case VerificationNote::Code::INVALID_LANGUAGE:
1889                 return String::compose("The DCP specifies a language '%1' which does not conform to the RFC 5646 standard.", note.note().get());
1890         case VerificationNote::Code::INVALID_PICTURE_SIZE_IN_PIXELS:
1891                 return String::compose("The size %1 of picture asset %2 is not allowed.", note.note().get(), note.file()->filename());
1892         case VerificationNote::Code::INVALID_PICTURE_FRAME_RATE_FOR_2K:
1893                 return String::compose("The frame rate %1 of picture asset %2 is not allowed for 2K DCPs.", note.note().get(), note.file()->filename());
1894         case VerificationNote::Code::INVALID_PICTURE_FRAME_RATE_FOR_4K:
1895                 return String::compose("The frame rate %1 of picture asset %2 is not allowed for 4K DCPs.", note.note().get(), note.file()->filename());
1896         case VerificationNote::Code::INVALID_PICTURE_ASSET_RESOLUTION_FOR_3D:
1897                 return "3D 4K DCPs are not allowed.";
1898         case VerificationNote::Code::INVALID_CLOSED_CAPTION_XML_SIZE_IN_BYTES:
1899                 return String::compose("The size %1 of the closed caption asset %2 is larger than the 256KB maximum.", note.note().get(), note.file()->filename());
1900         case VerificationNote::Code::INVALID_TIMED_TEXT_SIZE_IN_BYTES:
1901                 return String::compose("The size %1 of the timed text asset %2 is larger than the 115MB maximum.", note.note().get(), note.file()->filename());
1902         case VerificationNote::Code::INVALID_TIMED_TEXT_FONT_SIZE_IN_BYTES:
1903                 return String::compose("The size %1 of the fonts in timed text asset %2 is larger than the 10MB maximum.", note.note().get(), note.file()->filename());
1904         case VerificationNote::Code::MISSING_SUBTITLE_LANGUAGE:
1905                 return String::compose("The XML for the SMPTE subtitle asset %1 has no <Language> tag.", note.file()->filename());
1906         case VerificationNote::Code::MISMATCHED_SUBTITLE_LANGUAGES:
1907                 return "Some subtitle assets have different <Language> tags than others";
1908         case VerificationNote::Code::MISSING_SUBTITLE_START_TIME:
1909                 return String::compose("The XML for the SMPTE subtitle asset %1 has no <StartTime> tag.", note.file()->filename());
1910         case VerificationNote::Code::INVALID_SUBTITLE_START_TIME:
1911                 return String::compose("The XML for a SMPTE subtitle asset %1 has a non-zero <StartTime> tag.", note.file()->filename());
1912         case VerificationNote::Code::INVALID_SUBTITLE_FIRST_TEXT_TIME:
1913                 return "The first subtitle or closed caption is less than 4 seconds from the start of the DCP.";
1914         case VerificationNote::Code::INVALID_SUBTITLE_DURATION:
1915                 return "At least one subtitle lasts less than 15 frames.";
1916         case VerificationNote::Code::INVALID_SUBTITLE_SPACING:
1917                 return "At least one pair of subtitles is separated by less than 2 frames.";
1918         case VerificationNote::Code::SUBTITLE_OVERLAPS_REEL_BOUNDARY:
1919                 return "At least one subtitle extends outside of its reel.";
1920         case VerificationNote::Code::INVALID_SUBTITLE_LINE_COUNT:
1921                 return "There are more than 3 subtitle lines in at least one place in the DCP.";
1922         case VerificationNote::Code::NEARLY_INVALID_SUBTITLE_LINE_LENGTH:
1923                 return "There are more than 52 characters in at least one subtitle line.";
1924         case VerificationNote::Code::INVALID_SUBTITLE_LINE_LENGTH:
1925                 return "There are more than 79 characters in at least one subtitle line.";
1926         case VerificationNote::Code::INVALID_CLOSED_CAPTION_LINE_COUNT:
1927                 return "There are more than 3 closed caption lines in at least one place.";
1928         case VerificationNote::Code::INVALID_CLOSED_CAPTION_LINE_LENGTH:
1929                 return "There are more than 32 characters in at least one closed caption line.";
1930         case VerificationNote::Code::INVALID_SOUND_FRAME_RATE:
1931                 return String::compose("The sound asset %1 has a sampling rate of %2", note.file()->filename(), note.note().get());
1932         case VerificationNote::Code::MISSING_CPL_ANNOTATION_TEXT:
1933                 return String::compose("The CPL %1 has no <AnnotationText> tag.", note.note().get());
1934         case VerificationNote::Code::MISMATCHED_CPL_ANNOTATION_TEXT:
1935                 return String::compose("The CPL %1 has an <AnnotationText> which differs from its <ContentTitleText>.", note.note().get());
1936         case VerificationNote::Code::MISMATCHED_ASSET_DURATION:
1937                 return "All assets in a reel do not have the same duration.";
1938         case VerificationNote::Code::MISSING_MAIN_SUBTITLE_FROM_SOME_REELS:
1939                 return "At least one reel contains a subtitle asset, but some reel(s) do not.";
1940         case VerificationNote::Code::MISMATCHED_CLOSED_CAPTION_ASSET_COUNTS:
1941                 return "At least one reel has closed captions, but reels have different numbers of closed caption assets.";
1942         case VerificationNote::Code::MISSING_SUBTITLE_ENTRY_POINT:
1943                 return String::compose("The subtitle asset %1 has no <EntryPoint> tag.", note.note().get());
1944         case VerificationNote::Code::INCORRECT_SUBTITLE_ENTRY_POINT:
1945                 return String::compose("The subtitle asset %1 has an <EntryPoint> other than 0.", note.note().get());
1946         case VerificationNote::Code::MISSING_CLOSED_CAPTION_ENTRY_POINT:
1947                 return String::compose("The closed caption asset %1 has no <EntryPoint> tag.", note.note().get());
1948         case VerificationNote::Code::INCORRECT_CLOSED_CAPTION_ENTRY_POINT:
1949                 return String::compose("The closed caption asset %1 has an <EntryPoint> other than 0.", note.note().get());
1950         case VerificationNote::Code::MISSING_HASH:
1951                 return String::compose("The asset %1 has no <Hash> tag in the CPL.", note.note().get());
1952         case VerificationNote::Code::MISSING_FFEC_IN_FEATURE:
1953                 return "The DCP is marked as a Feature but there is no FFEC (first frame of end credits) marker.";
1954         case VerificationNote::Code::MISSING_FFMC_IN_FEATURE:
1955                 return "The DCP is marked as a Feature but there is no FFMC (first frame of moving credits) marker.";
1956         case VerificationNote::Code::MISSING_FFOC:
1957                 return "There should be a FFOC (first frame of content) marker.";
1958         case VerificationNote::Code::MISSING_LFOC:
1959                 return "There should be a LFOC (last frame of content) marker.";
1960         case VerificationNote::Code::INCORRECT_FFOC:
1961                 return String::compose("The FFOC marker is %1 instead of 1", note.note().get());
1962         case VerificationNote::Code::INCORRECT_LFOC:
1963                 return String::compose("The LFOC marker is %1 instead of 1 less than the duration of the last reel.", note.note().get());
1964         case VerificationNote::Code::MISSING_CPL_METADATA:
1965                 return String::compose("The CPL %1 has no <CompositionMetadataAsset> tag.", note.note().get());
1966         case VerificationNote::Code::MISSING_CPL_METADATA_VERSION_NUMBER:
1967                 return String::compose("The CPL %1 has no <VersionNumber> in its <CompositionMetadataAsset>.", note.note().get());
1968         case VerificationNote::Code::MISSING_EXTENSION_METADATA:
1969                 return String::compose("The CPL %1 has no <ExtensionMetadata> in its <CompositionMetadataAsset>.", note.note().get());
1970         case VerificationNote::Code::INVALID_EXTENSION_METADATA:
1971                 return String::compose("The CPL %1 has a malformed <ExtensionMetadata> (%2).", note.file()->filename(), note.note().get());
1972         case VerificationNote::Code::UNSIGNED_CPL_WITH_ENCRYPTED_CONTENT:
1973                 return String::compose("The CPL %1, which has encrypted content, is not signed.", note.note().get());
1974         case VerificationNote::Code::UNSIGNED_PKL_WITH_ENCRYPTED_CONTENT:
1975                 return String::compose("The PKL %1, which has encrypted content, is not signed.", note.note().get());
1976         case VerificationNote::Code::MISMATCHED_PKL_ANNOTATION_TEXT_WITH_CPL:
1977                 return String::compose("The PKL %1 has only one CPL but its <AnnotationText> does not match the CPL's <ContentTitleText>.", note.note().get());
1978         case VerificationNote::Code::PARTIALLY_ENCRYPTED:
1979                 return "Some assets are encrypted but some are not.";
1980         case VerificationNote::Code::INVALID_JPEG2000_CODESTREAM:
1981                 return String::compose("The JPEG2000 codestream for at least one frame is invalid (%1).", note.note().get());
1982         case VerificationNote::Code::INVALID_JPEG2000_GUARD_BITS_FOR_2K:
1983                 return String::compose("The JPEG2000 codestream uses %1 guard bits in a 2K image instead of 1.", note.note().get());
1984         case VerificationNote::Code::INVALID_JPEG2000_GUARD_BITS_FOR_4K:
1985                 return String::compose("The JPEG2000 codestream uses %1 guard bits in a 4K image instead of 2.", note.note().get());
1986         case VerificationNote::Code::INVALID_JPEG2000_TILE_SIZE:
1987                 return "The JPEG2000 tile size is not the same as the image size.";
1988         case VerificationNote::Code::INVALID_JPEG2000_CODE_BLOCK_WIDTH:
1989                 return String::compose("The JPEG2000 codestream uses a code block width of %1 instead of 32.", note.note().get());
1990         case VerificationNote::Code::INVALID_JPEG2000_CODE_BLOCK_HEIGHT:
1991                 return String::compose("The JPEG2000 codestream uses a code block height of %1 instead of 32.", note.note().get());
1992         case VerificationNote::Code::INCORRECT_JPEG2000_POC_MARKER_COUNT_FOR_2K:
1993                 return String::compose("%1 POC markers found in 2K JPEG2000 codestream instead of 0.", note.note().get());
1994         case VerificationNote::Code::INCORRECT_JPEG2000_POC_MARKER_COUNT_FOR_4K:
1995                 return String::compose("%1 POC markers found in 4K JPEG2000 codestream instead of 1.", note.note().get());
1996         case VerificationNote::Code::INCORRECT_JPEG2000_POC_MARKER:
1997                 return String::compose("Incorrect POC marker content found (%1).", note.note().get());
1998         case VerificationNote::Code::INVALID_JPEG2000_POC_MARKER_LOCATION:
1999                 return "POC marker found outside main header.";
2000         case VerificationNote::Code::INVALID_JPEG2000_TILE_PARTS_FOR_2K:
2001                 return String::compose("The JPEG2000 codestream has %1 tile parts in a 2K image instead of 3.", note.note().get());
2002         case VerificationNote::Code::INVALID_JPEG2000_TILE_PARTS_FOR_4K:
2003                 return String::compose("The JPEG2000 codestream has %1 tile parts in a 4K image instead of 6.", note.note().get());
2004         case VerificationNote::Code::MISSING_JPEG200_TLM_MARKER:
2005                 return "No TLM marker was found in a JPEG2000 codestream.";
2006         case VerificationNote::Code::MISMATCHED_TIMED_TEXT_RESOURCE_ID:
2007                 return "The Resource ID in a timed text MXF did not match the ID of the contained XML.";
2008         case VerificationNote::Code::INCORRECT_TIMED_TEXT_ASSET_ID:
2009                 return "The Asset ID in a timed text MXF is the same as the Resource ID or that of the contained XML.";
2010         case VerificationNote::Code::MISMATCHED_TIMED_TEXT_DURATION:
2011         {
2012                 vector<string> parts;
2013                 boost::split (parts, note.note().get(), boost::is_any_of(" "));
2014                 DCP_ASSERT (parts.size() == 2);
2015                 return String::compose("The reel duration of some timed text (%1) is not the same as the ContainerDuration of its MXF (%2).", parts[0], parts[1]);
2016         }
2017         case VerificationNote::Code::MISSED_CHECK_OF_ENCRYPTED:
2018                 return "Some aspect of this DCP could not be checked because it is encrypted.";
2019         case VerificationNote::Code::EMPTY_TEXT:
2020                 return "There is an empty <Text> node in a subtitle or closed caption.";
2021         case VerificationNote::Code::MISMATCHED_CLOSED_CAPTION_VALIGN:
2022                 return "Some closed <Text> or <Image> nodes have different vertical alignments within a <Subtitle>.";
2023         case VerificationNote::Code::INCORRECT_CLOSED_CAPTION_ORDERING:
2024                 return "Some closed captions are not listed in the order of their vertical position.";
2025         case VerificationNote::Code::UNEXPECTED_ENTRY_POINT:
2026                 return "There is an <EntryPoint> node inside a <MainMarkers>.";
2027         case VerificationNote::Code::UNEXPECTED_DURATION:
2028                 return "There is an <Duration> node inside a <MainMarkers>.";
2029         case VerificationNote::Code::INVALID_CONTENT_KIND:
2030                 return String::compose("<ContentKind> has an invalid value %1.", note.note().get());
2031         case VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA:
2032                 return String::compose("<MainPictureActiveaArea> has an invalid value: %1", note.note().get());
2033         case VerificationNote::Code::DUPLICATE_ASSET_ID_IN_PKL:
2034                 return String::compose("The PKL %1 has more than one asset with the same ID.", note.note().get());
2035         case VerificationNote::Code::DUPLICATE_ASSET_ID_IN_ASSETMAP:
2036                 return String::compose("The ASSETMAP %1 has more than one asset with the same ID.", note.note().get());
2037         case VerificationNote::Code::MISSING_SUBTITLE:
2038                 return String::compose("The subtitle asset %1 has no subtitles.", note.note().get());
2039         case VerificationNote::Code::INVALID_SUBTITLE_ISSUE_DATE:
2040                 return String::compose("<IssueDate> has an invalid value: %1", note.note().get());
2041         case VerificationNote::Code::MISMATCHED_SOUND_CHANNEL_COUNTS:
2042                 return String::compose("The sound assets do not all have the same channel count; the first to differ is %1", note.file()->filename());
2043         case VerificationNote::Code::INVALID_MAIN_SOUND_CONFIGURATION:
2044                 return String::compose("<MainSoundConfiguration> has an invalid value: %1", note.note().get());
2045         case VerificationNote::Code::MISSING_FONT:
2046                 return String::compose("The font file for font ID \"%1\" was not found, or was not referred to in the ASSETMAP.", note.note().get());
2047         case VerificationNote::Code::INVALID_JPEG2000_TILE_PART_SIZE:
2048                 return String::compose(
2049                         "Frame %1 has an image component that is too large (component %2 is %3 bytes in size).",
2050                         note.frame().get(), note.component().get(), note.size().get()
2051                         );
2052         case VerificationNote::Code::INCORRECT_SUBTITLE_NAMESPACE_COUNT:
2053                 return String::compose("The XML in the subtitle asset %1 has more than one namespace declaration.", note.note().get());
2054         case VerificationNote::Code::MISSING_LOAD_FONT_FOR_FONT:
2055                 return String::compose("A subtitle or closed caption refers to a font with ID %1 that does not have a corresponding <LoadFont> node", note.id().get());
2056         }
2057
2058         return "";
2059 }
2060
2061
2062 bool
2063 dcp::operator== (dcp::VerificationNote const& a, dcp::VerificationNote const& b)
2064 {
2065         return a.type() == b.type() && a.code() == b.code() && a.note() == b.note() && a.file() == b.file() && a.line() == b.line();
2066 }
2067
2068
2069 bool
2070 dcp::operator< (dcp::VerificationNote const& a, dcp::VerificationNote const& b)
2071 {
2072         if (a.type() != b.type()) {
2073                 return a.type() < b.type();
2074         }
2075
2076         if (a.code() != b.code()) {
2077                 return a.code() < b.code();
2078         }
2079
2080         if (a.note() != b.note()) {
2081                 return a.note().get_value_or("") < b.note().get_value_or("");
2082         }
2083
2084         if (a.file() != b.file()) {
2085                 return a.file().get_value_or("") < b.file().get_value_or("");
2086         }
2087
2088         return a.line().get_value_or(0) < b.line().get_value_or(0);
2089 }
2090
2091
2092 std::ostream&
2093 dcp::operator<< (std::ostream& s, dcp::VerificationNote const& note)
2094 {
2095         s << note_to_string (note);
2096         if (note.note()) {
2097                 s << " [" << note.note().get() << "]";
2098         }
2099         if (note.file()) {
2100                 s << " [" << note.file().get() << "]";
2101         }
2102         if (note.line()) {
2103                 s << " [" << note.line().get() << "]";
2104         }
2105         return s;
2106 }
2107