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 */
 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
 874         std::function<void (cxml::ConstNodePtr, optional<int>, optional<Time>, int, bool)> parse;
 875         parse = [&parse, &last_out, &too_short, &too_close, &too_early, &empty_text, &reel_offset](cxml::ConstNodePtr node, optional<int> tcr, optional<Time> start_time, int er, bool first_reel) {
 876                 if (node->name() == "Subtitle") {
 877                         Time in (node->string_attribute("TimeIn"), tcr);
 878                         if (start_time) {
 879                                 in -= *start_time;
 880                         }
 881                         Time out (node->string_attribute("TimeOut"), tcr);
 882                         if (start_time) {
 883                                 out -= *start_time;
 884                         }
 885                         if (first_reel && tcr && in < Time(0, 0, 4, 0, *tcr)) {
 886                                 too_early = true;
 887                         }
 888                         auto length = out - in;
 889                         if (length.as_editable_units_ceil(er) < 15) {
 890                                 too_short = true;
 891                         }
 892                         if (last_out) {
 893                                 /* XXX: this feels dubious - is it really what Bv2.1 means? */
 894                                 auto distance = reel_offset + in.as_editable_units_ceil(er) - *last_out;
 895                                 if (distance >= 0 && distance < 2) {
 896                                         too_close = true;
 897                                 }
 898                         }
 899                         last_out = reel_offset + out.as_editable_units_floor(er);
 900                 } else if (node->name() == "Text") {
 901                         std::function<bool (cxml::ConstNodePtr)> node_has_content = [&](cxml::ConstNodePtr node) {
 902                                 if (!node->content().empty()) {
 903                                         return true;
 904                                 }
 905                                 for (auto i: node->node_children()) {
 906                                         if (node_has_content(i)) {
 907                                                 return true;
 908                                         }
 909                                 }
 910                                 return false;
 911                         };
 912                         if (!node_has_content(node)) {
 913                                 empty_text = true;
 914                         }
 915                 }
 916
 917                 for (auto i: node->node_children()) {
 918                         parse(i, tcr, start_time, er, first_reel);
 919                 }
 920         };
 921
 922         for (auto i = 0U; i < reels.size(); ++i) {
 923                 if (!check(reels[i])) {
 924                         continue;
 925                 }
 926
 927                 auto reel_xml = xml(reels[i]);
 928                 if (!reel_xml) {
 929                         notes.push_back ({VerificationNote::Type::WARNING, VerificationNote::Code::MISSED_CHECK_OF_ENCRYPTED});
 930                         continue;
 931                 }
 932
 933                 /* We need to look at <Subtitle> instances in the XML being checked, so we can't use the subtitles
 934                  * read in by libdcp's parser.
 935                  */
 936
 937                 shared_ptr<cxml::Document> doc;
 938                 optional<int> tcr;
 939                 optional<Time> start_time;
 940                 try {
 941                         doc = make_shared<cxml::Document>("SubtitleReel");
 942                         doc->read_string (*reel_xml);
 943                         tcr = doc->number_child<int>("TimeCodeRate");
 944                         auto start_time_string = doc->optional_string_child("StartTime");
 945                         if (start_time_string) {
 946                                 start_time = Time(*start_time_string, tcr);
 947                         }
 948                 } catch (...) {
 949                         doc = make_shared<cxml::Document>("DCSubtitle");
 950                         doc->read_string (*reel_xml);
 951                 }
 952                 parse (doc, tcr, start_time, edit_rate, i == 0);
 953                 auto end = reel_offset + duration(reels[i]);
 954                 if (last_out && *last_out > end) {
 955                         reel_overlap = true;
 956                 }
 957                 reel_offset = end;
 958         }
 959
 960         if (last_out && *last_out > reel_offset) {
 961                 reel_overlap = true;
 962         }
 963
 964         if (too_early) {
 965                 notes.push_back({
 966                         VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_FIRST_TEXT_TIME
 967                 });
 968         }
 969
 970         if (too_short) {
 971                 notes.push_back ({
 972                         VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_DURATION
 973                 });
 974         }
 975
 976         if (too_close) {
 977                 notes.push_back ({
 978                         VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_SPACING
 979                 });
 980         }
 981
 982         if (reel_overlap) {
 983                 notes.push_back ({
 984                         VerificationNote::Type::ERROR, VerificationNote::Code::SUBTITLE_OVERLAPS_REEL_BOUNDARY
 985                 });
 986         }
 987
 988         if (empty_text) {
 989                 notes.push_back ({
 990                         VerificationNote::Type::WARNING, VerificationNote::Code::EMPTY_TEXT
 991                 });
 992         }
 993 }
 994
 995
 996 static
 997 void
 998 verify_closed_caption_details (
 999         vector<shared_ptr<Reel>> reels,
1000         vector<VerificationNote>& notes
1001         )
1002 {
1003         std::function<void (cxml::ConstNodePtr node, std::vector<cxml::ConstNodePtr>& text_or_image)> find_text_or_image;
1004         find_text_or_image = [&find_text_or_image](cxml::ConstNodePtr node, std::vector<cxml::ConstNodePtr>& text_or_image) {
1005                 for (auto i: node->node_children()) {
1006                         if (i->name() == "Text") {
1007                                 text_or_image.push_back (i);
1008                         } else {
1009                                 find_text_or_image (i, text_or_image);
1010                         }
1011                 }
1012         };
1013
1014         auto mismatched_valign = false;
1015         auto incorrect_order = false;
1016
1017         std::function<void (cxml::ConstNodePtr)> parse;
1018         parse = [&parse, &find_text_or_image, &mismatched_valign, &incorrect_order](cxml::ConstNodePtr node) {
1019                 if (node->name() == "Subtitle") {
1020                         vector<cxml::ConstNodePtr> text_or_image;
1021                         find_text_or_image (node, text_or_image);
1022                         optional<string> last_valign;
1023                         optional<float> last_vpos;
1024                         for (auto i: text_or_image) {
1025                                 auto valign = i->optional_string_attribute("VAlign");
1026                                 if (!valign) {
1027                                         valign = i->optional_string_attribute("Valign").get_value_or("center");
1028                                 }
1029                                 auto vpos = i->optional_number_attribute<float>("VPosition");
1030                                 if (!vpos) {
1031                                         vpos = i->optional_number_attribute<float>("Vposition").get_value_or(50);
1032                                 }
1033
1034                                 if (last_valign) {
1035                                         if (*last_valign != valign) {
1036                                                 mismatched_valign = true;
1037                                         }
1038                                 }
1039                                 last_valign = valign;
1040
1041                                 if (!mismatched_valign) {
1042                                         if (last_vpos) {
1043                                                 if (*last_valign == "top" || *last_valign == "center") {
1044                                                         if (*vpos < *last_vpos) {
1045                                                                 incorrect_order = true;
1046                                                         }
1047                                                 } else {
1048                                                         if (*vpos > *last_vpos) {
1049                                                                 incorrect_order = true;
1050                                                         }
1051                                                 }
1052                                         }
1053                                         last_vpos = vpos;
1054                                 }
1055                         }
1056                 }
1057
1058                 for (auto i: node->node_children()) {
1059                         parse(i);
1060                 }
1061         };
1062
1063         for (auto reel: reels) {
1064                 for (auto ccap: reel->closed_captions()) {
1065                         auto reel_xml = ccap->asset()->raw_xml();
1066                         if (!reel_xml) {
1067                                 notes.push_back ({VerificationNote::Type::WARNING, VerificationNote::Code::MISSED_CHECK_OF_ENCRYPTED});
1068                                 continue;
1069                         }
1070
1071                         /* We need to look at <Subtitle> instances in the XML being checked, so we can't use the subtitles
1072                          * read in by libdcp's parser.
1073                          */
1074
1075                         shared_ptr<cxml::Document> doc;
1076                         optional<int> tcr;
1077                         optional<Time> start_time;
1078                         try {
1079                                 doc = make_shared<cxml::Document>("SubtitleReel");
1080                                 doc->read_string (*reel_xml);
1081                         } catch (...) {
1082                                 doc = make_shared<cxml::Document>("DCSubtitle");
1083                                 doc->read_string (*reel_xml);
1084                         }
1085                         parse (doc);
1086                 }
1087         }
1088
1089         if (mismatched_valign) {
1090                 notes.push_back ({
1091                         VerificationNote::Type::ERROR, VerificationNote::Code::MISMATCHED_CLOSED_CAPTION_VALIGN,
1092                 });
1093         }
1094
1095         if (incorrect_order) {
1096                 notes.push_back ({
1097                         VerificationNote::Type::ERROR, VerificationNote::Code::INCORRECT_CLOSED_CAPTION_ORDERING,
1098                 });
1099         }
1100 }
1101
1102
1103 struct LinesCharactersResult
1104 {
1105         bool warning_length_exceeded = false;
1106         bool error_length_exceeded = false;
1107         bool line_count_exceeded = false;
1108 };
1109
1110
1111 static
1112 void
1113 verify_text_lines_and_characters (
1114         shared_ptr<SubtitleAsset> asset,
1115         int warning_length,
1116         int error_length,
1117         LinesCharactersResult* result
1118         )
1119 {
1120         class Event
1121         {
1122         public:
1123                 Event (Time time_, float position_, int characters_)
1124                         : time (time_)
1125                         , position (position_)
1126                         , characters (characters_)
1127                 {}
1128
1129                 Event (Time time_, shared_ptr<Event> start_)
1130                         : time (time_)
1131                         , start (start_)
1132                 {}
1133
1134                 Time time;
1135                 int position; //< position from 0 at top of screen to 100 at bottom
1136                 int characters;
1137                 shared_ptr<Event> start;
1138         };
1139
1140         vector<shared_ptr<Event>> events;
1141
1142         auto position = [](shared_ptr<const SubtitleString> sub) {
1143                 switch (sub->v_align()) {
1144                 case VAlign::TOP:
1145                         return lrintf(sub->v_position() * 100);
1146                 case VAlign::CENTER:
1147                         return lrintf((0.5f + sub->v_position()) * 100);
1148                 case VAlign::BOTTOM:
1149                         return lrintf((1.0f - sub->v_position()) * 100);
1150                 }
1151
1152                 return 0L;
1153         };
1154
1155         for (auto j: asset->subtitles()) {
1156                 auto text = dynamic_pointer_cast<const SubtitleString>(j);
1157                 if (text) {
1158                         auto in = make_shared<Event>(text->in(), position(text), text->text().length());
1159                         events.push_back(in);
1160                         events.push_back(make_shared<Event>(text->out(), in));
1161                 }
1162         }
1163
1164         std::sort(events.begin(), events.end(), [](shared_ptr<Event> const& a, shared_ptr<Event>const& b) {
1165                 return a->time < b->time;
1166         });
1167
1168         map<int, int> current;
1169         for (auto i: events) {
1170                 if (current.size() > 3) {
1171                         result->line_count_exceeded = true;
1172                 }
1173                 for (auto j: current) {
1174                         if (j.second > warning_length) {
1175                                 result->warning_length_exceeded = true;
1176                         }
1177                         if (j.second > error_length) {
1178                                 result->error_length_exceeded = true;
1179                         }
1180                 }
1181
1182                 if (i->start) {
1183                         /* end of a subtitle */
1184                         DCP_ASSERT (current.find(i->start->position) != current.end());
1185                         if (current[i->start->position] == i->start->characters) {
1186                                 current.erase(i->start->position);
1187                         } else {
1188                                 current[i->start->position] -= i->start->characters;
1189                         }
1190                 } else {
1191                         /* start of a subtitle */
1192                         if (current.find(i->position) == current.end()) {
1193                                 current[i->position] = i->characters;
1194                         } else {
1195                                 current[i->position] += i->characters;
1196                         }
1197                 }
1198         }
1199 }
1200
1201
1202 static
1203 void
1204 verify_text_details (vector<shared_ptr<Reel>> reels, vector<VerificationNote>& notes)
1205 {
1206         if (reels.empty()) {
1207                 return;
1208         }
1209
1210         if (reels[0]->main_subtitle()) {
1211                 verify_text_details (reels, reels[0]->main_subtitle()->edit_rate().numerator, notes,
1212                         [](shared_ptr<Reel> reel) {
1213                                 return static_cast<bool>(reel->main_subtitle());
1214                         },
1215                         [](shared_ptr<Reel> reel) {
1216                                 auto interop = dynamic_pointer_cast<ReelInteropSubtitleAsset>(reel->main_subtitle());
1217                                 if (interop) {
1218                                         return interop->asset()->raw_xml();
1219                                 }
1220                                 auto smpte = dynamic_pointer_cast<ReelSMPTESubtitleAsset>(reel->main_subtitle());
1221                                 DCP_ASSERT (smpte);
1222                                 return smpte->asset()->raw_xml();
1223                         },
1224                         [](shared_ptr<Reel> reel) {
1225                                 return reel->main_subtitle()->actual_duration();
1226                         }
1227                 );
1228         }
1229
1230         for (auto i = 0U; i < reels[0]->closed_captions().size(); ++i) {
1231                 verify_text_details (reels, reels[0]->closed_captions()[i]->edit_rate().numerator, notes,
1232                         [i](shared_ptr<Reel> reel) {
1233                                 return i < reel->closed_captions().size();
1234                         },
1235                         [i](shared_ptr<Reel> reel) {
1236                                 return reel->closed_captions()[i]->asset()->raw_xml();
1237                         },
1238                         [i](shared_ptr<Reel> reel) {
1239                                 return reel->closed_captions()[i]->actual_duration();
1240                         }
1241                 );
1242         }
1243
1244         verify_closed_caption_details (reels, notes);
1245 }
1246
1247
1248 void
1249 verify_extension_metadata(shared_ptr<const CPL> cpl, vector<VerificationNote>& notes)
1250 {
1251         DCP_ASSERT (cpl->file());
1252         cxml::Document doc ("CompositionPlaylist");
1253         doc.read_file (cpl->file().get());
1254
1255         auto missing = false;
1256         string malformed;
1257
1258         if (auto reel_list = doc.node_child("ReelList")) {
1259                 auto reels = reel_list->node_children("Reel");
1260                 if (!reels.empty()) {
1261                         if (auto asset_list = reels[0]->optional_node_child("AssetList")) {
1262                                 if (auto metadata = asset_list->optional_node_child("CompositionMetadataAsset")) {
1263                                         if (auto extension_list = metadata->optional_node_child("ExtensionMetadataList")) {
1264                                                 missing = true;
1265                                                 for (auto extension: extension_list->node_children("ExtensionMetadata")) {
1266                                                         if (extension->optional_string_attribute("scope").get_value_or("") != "http://isdcf.com/ns/cplmd/app") {
1267                                                                 continue;
1268                                                         }
1269                                                         missing = false;
1270                                                         if (auto name = extension->optional_node_child("Name")) {
1271                                                                 if (name->content() != "Application") {
1272                                                                         malformed = "<Name> should be 'Application'";
1273                                                                 }
1274                                                         }
1275                                                         if (auto property_list = extension->optional_node_child("PropertyList")) {
1276                                                                 if (auto property = property_list->optional_node_child("Property")) {
1277                                                                         if (auto name = property->optional_node_child("Name")) {
1278                                                                                 if (name->content() != "DCP Constraints Profile") {
1279                                                                                         malformed = "<Name> property should be 'DCP Constraints Profile'";
1280                                                                                 }
1281                                                                         }
1282                                                                         if (auto value = property->optional_node_child("Value")) {
1283                                                                                 if (value->content() != "SMPTE-RDD-52:2020-Bv2.1") {
1284                                                                                         malformed = "<Value> property should be 'SMPTE-RDD-52:2020-Bv2.1'";
1285                                                                                 }
1286                                                                         }
1287                                                                 }
1288                                                         }
1289                                                 }
1290                                         } else {
1291                                                 missing = true;
1292                                         }
1293                                 }
1294                         }
1295                 }
1296         }
1297
1298         if (missing) {
1299                 notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_EXTENSION_METADATA, cpl->id(), cpl->file().get()});
1300         } else if (!malformed.empty()) {
1301                 notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_EXTENSION_METADATA, malformed, cpl->file().get()});
1302         }
1303 }
1304
1305
1306 bool
1307 pkl_has_encrypted_assets(shared_ptr<const DCP> dcp, shared_ptr<const PKL> pkl)
1308 {
1309         vector<string> encrypted;
1310         for (auto i: dcp->cpls()) {
1311                 for (auto j: i->reel_file_assets()) {
1312                         if (j->asset_ref().resolved()) {
1313                                 auto mxf = dynamic_pointer_cast<MXF>(j->asset_ref().asset());
1314                                 if (mxf && mxf->encrypted()) {
1315                                         encrypted.push_back(j->asset_ref().id());
1316                                 }
1317                         }
1318                 }
1319         }
1320
1321         for (auto i: pkl->assets()) {
1322                 if (find(encrypted.begin(), encrypted.end(), i->id()) != encrypted.end()) {
1323                         return true;
1324                 }
1325         }
1326
1327         return false;
1328 }
1329
1330
1331 static
1332 void
1333 verify_reel(
1334         shared_ptr<const DCP> dcp,
1335         shared_ptr<const CPL> cpl,
1336         shared_ptr<const Reel> reel,
1337         optional<dcp::Size> main_picture_active_area,
1338         function<void (string, optional<boost::filesystem::path>)> stage,
1339         boost::filesystem::path xsd_dtd_directory,
1340         function<void (float)> progress,
1341         VerificationOptions options,
1342         vector<VerificationNote>& notes,
1343         State& state,
1344         bool* have_main_subtitle,
1345         bool* have_no_main_subtitle,
1346         size_t* most_closed_captions,
1347         size_t* fewest_closed_captions,
1348         map<Marker, Time>* markers_seen
1349         )
1350 {
1351         for (auto i: reel->assets()) {
1352                 if (i->duration() && (i->duration().get() * i->edit_rate().denominator / i->edit_rate().numerator) < 1) {
1353                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::INVALID_DURATION, i->id()});
1354                 }
1355                 if ((i->intrinsic_duration() * i->edit_rate().denominator / i->edit_rate().numerator) < 1) {
1356                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::INVALID_INTRINSIC_DURATION, i->id()});
1357                 }
1358                 auto file_asset = dynamic_pointer_cast<ReelFileAsset>(i);
1359                 if (i->encryptable() && !file_asset->hash()) {
1360                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_HASH, i->id()});
1361                 }
1362         }
1363
1364         if (dcp->standard() == Standard::SMPTE) {
1365                 boost::optional<int64_t> duration;
1366                 for (auto i: reel->assets()) {
1367                         if (!duration) {
1368                                 duration = i->actual_duration();
1369                         } else if (*duration != i->actual_duration()) {
1370                                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISMATCHED_ASSET_DURATION});
1371                                 break;
1372                         }
1373                 }
1374         }
1375
1376         if (reel->main_picture()) {
1377                 /* Check reel stuff */
1378                 auto const frame_rate = reel->main_picture()->frame_rate();
1379                 if (frame_rate.denominator != 1 ||
1380                     (frame_rate.numerator != 24 &&
1381                      frame_rate.numerator != 25 &&
1382                      frame_rate.numerator != 30 &&
1383                      frame_rate.numerator != 48 &&
1384                      frame_rate.numerator != 50 &&
1385                      frame_rate.numerator != 60 &&
1386                      frame_rate.numerator != 96)) {
1387                         notes.push_back({
1388                                 VerificationNote::Type::ERROR,
1389                                 VerificationNote::Code::INVALID_PICTURE_FRAME_RATE,
1390                                 String::compose("%1/%2", frame_rate.numerator, frame_rate.denominator)
1391                         });
1392                 }
1393                 /* Check asset */
1394                 if (reel->main_picture()->asset_ref().resolved()) {
1395                         verify_main_picture_asset(dcp, reel->main_picture(), stage, progress, options, notes);
1396                         auto const asset_size = reel->main_picture()->asset()->size();
1397                         if (main_picture_active_area) {
1398                                 if (main_picture_active_area->width > asset_size.width) {
1399                                         notes.push_back({
1400                                                         VerificationNote::Type::ERROR,
1401                                                         VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA,
1402                                                         String::compose("width %1 is bigger than the asset width %2", main_picture_active_area->width, asset_size.width),
1403                                                         cpl->file().get()
1404                                                         });
1405                                 }
1406                                 if (main_picture_active_area->height > asset_size.height) {
1407                                         notes.push_back({
1408                                                         VerificationNote::Type::ERROR,
1409                                                         VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA,
1410                                                         String::compose("height %1 is bigger than the asset height %2", main_picture_active_area->height, asset_size.height),
1411                                                         cpl->file().get()
1412                                                         });
1413                                 }
1414                         }
1415                 }
1416         }
1417
1418         if (reel->main_sound() && reel->main_sound()->asset_ref().resolved()) {
1419                 verify_main_sound_asset(dcp, reel->main_sound(), stage, progress, options, notes, state);
1420         }
1421
1422         if (reel->main_subtitle()) {
1423                 verify_main_subtitle_reel(reel->main_subtitle(), notes);
1424                 if (reel->main_subtitle()->asset_ref().resolved()) {
1425                         verify_subtitle_asset(reel->main_subtitle()->asset(), reel->main_subtitle()->duration(), stage, xsd_dtd_directory, notes, state);
1426                 }
1427                 *have_main_subtitle = true;
1428         } else {
1429                 *have_no_main_subtitle = true;
1430         }
1431
1432         for (auto i: reel->closed_captions()) {
1433                 verify_closed_caption_reel(i, notes);
1434                 if (i->asset_ref().resolved()) {
1435                         verify_closed_caption_asset(i->asset(), i->duration(), stage, xsd_dtd_directory, notes);
1436                 }
1437         }
1438
1439         if (reel->main_markers()) {
1440                 for (auto const& i: reel->main_markers()->get()) {
1441                         markers_seen->insert(i);
1442                 }
1443                 if (reel->main_markers()->entry_point()) {
1444                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::UNEXPECTED_ENTRY_POINT});
1445                 }
1446                 if (reel->main_markers()->duration()) {
1447                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::UNEXPECTED_DURATION});
1448                 }
1449         }
1450
1451         *fewest_closed_captions = std::min(*fewest_closed_captions, reel->closed_captions().size());
1452         *most_closed_captions = std::max(*most_closed_captions, reel->closed_captions().size());
1453
1454 }
1455
1456
1457 static
1458 void
1459 verify_cpl(
1460         shared_ptr<const DCP> dcp,
1461         shared_ptr<const CPL> cpl,
1462         function<void (string, optional<boost::filesystem::path>)> stage,
1463         boost::filesystem::path xsd_dtd_directory,
1464         function<void (float)> progress,
1465         VerificationOptions options,
1466         vector<VerificationNote>& notes,
1467         State& state
1468         )
1469 {
1470         stage("Checking CPL", cpl->file());
1471         validate_xml(cpl->file().get(), xsd_dtd_directory, notes);
1472
1473         if (cpl->any_encrypted() && !cpl->all_encrypted()) {
1474                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::PARTIALLY_ENCRYPTED});
1475         }
1476
1477         for (auto const& i: cpl->additional_subtitle_languages()) {
1478                 verify_language_tag(i, notes);
1479         }
1480
1481         if (!cpl->content_kind().scope() || *cpl->content_kind().scope() == "http://www.smpte-ra.org/schemas/429-7/2006/CPL#standard-content") {
1482                 /* This is a content kind from http://www.smpte-ra.org/schemas/429-7/2006/CPL#standard-content; make sure it's one
1483                  * of the approved ones.
1484                  */
1485                 auto all = ContentKind::all();
1486                 auto name = cpl->content_kind().name();
1487                 transform(name.begin(), name.end(), name.begin(), ::tolower);
1488                 auto iter = std::find_if(all.begin(), all.end(), [name](ContentKind const& k) { return !k.scope() && k.name() == name; });
1489                 if (iter == all.end()) {
1490                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::INVALID_CONTENT_KIND, cpl->content_kind().name()});
1491                 }
1492         }
1493
1494         if (cpl->release_territory()) {
1495                 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") {
1496                         auto terr = cpl->release_territory().get();
1497                         /* Must be a valid region tag, or "001" */
1498                         try {
1499                                 LanguageTag::RegionSubtag test(terr);
1500                         } catch (...) {
1501                                 if (terr != "001") {
1502                                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_LANGUAGE, terr});
1503                                 }
1504                         }
1505                 }
1506         }
1507
1508         if (dcp->standard() == Standard::SMPTE) {
1509                 if (!cpl->annotation_text()) {
1510                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_CPL_ANNOTATION_TEXT, cpl->id(), cpl->file().get()});
1511                 } else if (cpl->annotation_text().get() != cpl->content_title_text()) {
1512                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::MISMATCHED_CPL_ANNOTATION_TEXT, cpl->id(), cpl->file().get()});
1513                 }
1514         }
1515
1516         for (auto i: dcp->pkls()) {
1517                 /* Check that the CPL's hash corresponds to the PKL */
1518                 optional<string> h = i->hash(cpl->id());
1519                 if (h && make_digest(ArrayData(*cpl->file())) != *h) {
1520                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::MISMATCHED_CPL_HASHES, cpl->id(), cpl->file().get()});
1521                 }
1522
1523                 /* Check that any PKL with a single CPL has its AnnotationText the same as the CPL's ContentTitleText */
1524                 optional<string> required_annotation_text;
1525                 for (auto j: i->assets()) {
1526                         /* See if this is a CPL */
1527                         for (auto k: dcp->cpls()) {
1528                                 if (j->id() == k->id()) {
1529                                         if (!required_annotation_text) {
1530                                                 /* First CPL we have found; this is the required AnnotationText unless we find another */
1531                                                 required_annotation_text = cpl->content_title_text();
1532                                         } else {
1533                                                 /* There's more than one CPL so we don't care what the PKL's AnnotationText is */
1534                                                 required_annotation_text = boost::none;
1535                                         }
1536                                 }
1537                         }
1538                 }
1539
1540                 if (required_annotation_text && i->annotation_text() != required_annotation_text) {
1541                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISMATCHED_PKL_ANNOTATION_TEXT_WITH_CPL, i->id(), i->file().get()});
1542                 }
1543         }
1544
1545         /* set to true if any reel has a MainSubtitle */
1546         auto have_main_subtitle = false;
1547         /* set to true if any reel has no MainSubtitle */
1548         auto have_no_main_subtitle = false;
1549         /* fewest number of closed caption assets seen in a reel */
1550         size_t fewest_closed_captions = SIZE_MAX;
1551         /* most number of closed caption assets seen in a reel */
1552         size_t most_closed_captions = 0;
1553         map<Marker, Time> markers_seen;
1554
1555         auto const main_picture_active_area = cpl->main_picture_active_area();
1556         if (main_picture_active_area && (main_picture_active_area->width % 2)) {
1557                 notes.push_back({
1558                                 VerificationNote::Type::ERROR,
1559                                 VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA,
1560                                 String::compose("width %1 is not a multiple of 2", main_picture_active_area->width),
1561                                 cpl->file().get()
1562                         });
1563         }
1564         if (main_picture_active_area && (main_picture_active_area->height % 2)) {
1565                 notes.push_back({
1566                                 VerificationNote::Type::ERROR,
1567                                 VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA,
1568                                 String::compose("height %1 is not a multiple of 2", main_picture_active_area->height),
1569                                 cpl->file().get()
1570                         });
1571         }
1572
1573         for (auto reel: cpl->reels()) {
1574                 stage("Checking reel", optional<boost::filesystem::path>());
1575                 verify_reel(
1576                         dcp,
1577                         cpl,
1578                         reel,
1579                         main_picture_active_area,
1580                         stage,
1581                         xsd_dtd_directory,
1582                         progress,
1583                         options,
1584                         notes,
1585                         state,
1586                         &have_main_subtitle,
1587                         &have_no_main_subtitle,
1588                         &most_closed_captions,
1589                         &fewest_closed_captions,
1590                         &markers_seen
1591                         );
1592         }
1593
1594         verify_text_details(cpl->reels(), notes);
1595
1596         if (dcp->standard() == Standard::SMPTE) {
1597                 if (auto msc = cpl->main_sound_configuration()) {
1598                         if (state.audio_channels && msc->channels() != *state.audio_channels) {
1599                                 notes.push_back({
1600                                                 VerificationNote::Type::ERROR,
1601                                                 VerificationNote::Code::INVALID_MAIN_SOUND_CONFIGURATION,
1602                                                 String::compose("MainSoundConfiguration has %1 channels but sound assets have %2", msc->channels(), *state.audio_channels),
1603                                                 cpl->file().get()
1604                                         });
1605                         }
1606                 }
1607
1608                 if (have_main_subtitle && have_no_main_subtitle) {
1609                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_MAIN_SUBTITLE_FROM_SOME_REELS});
1610                 }
1611
1612                 if (fewest_closed_captions != most_closed_captions) {
1613                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISMATCHED_CLOSED_CAPTION_ASSET_COUNTS});
1614                 }
1615
1616                 if (cpl->content_kind() == ContentKind::FEATURE) {
1617                         if (markers_seen.find(Marker::FFEC) == markers_seen.end()) {
1618                                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_FFEC_IN_FEATURE});
1619                         }
1620                         if (markers_seen.find(Marker::FFMC) == markers_seen.end()) {
1621                                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_FFMC_IN_FEATURE});
1622                         }
1623                 }
1624
1625                 auto ffoc = markers_seen.find(Marker::FFOC);
1626                 if (ffoc == markers_seen.end()) {
1627                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::MISSING_FFOC});
1628                 } else if (ffoc->second.e != 1) {
1629                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::INCORRECT_FFOC, raw_convert<string>(ffoc->second.e)});
1630                 }
1631
1632                 auto lfoc = markers_seen.find(Marker::LFOC);
1633                 if (lfoc == markers_seen.end()) {
1634                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::MISSING_LFOC});
1635                 } else {
1636                         auto lfoc_time = lfoc->second.as_editable_units_ceil(lfoc->second.tcr);
1637                         if (lfoc_time != (cpl->reels().back()->duration() - 1)) {
1638                                 notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::INCORRECT_LFOC, raw_convert<string>(lfoc_time)});
1639                         }
1640                 }
1641
1642                 LinesCharactersResult result;
1643                 for (auto reel: cpl->reels()) {
1644                         if (reel->main_subtitle() && reel->main_subtitle()->asset()) {
1645                                 verify_text_lines_and_characters(reel->main_subtitle()->asset(), 52, 79, &result);
1646                         }
1647                 }
1648
1649                 if (result.line_count_exceeded) {
1650                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_LINE_COUNT});
1651                 }
1652                 if (result.error_length_exceeded) {
1653                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::INVALID_SUBTITLE_LINE_LENGTH});
1654                 } else if (result.warning_length_exceeded) {
1655                         notes.push_back({VerificationNote::Type::WARNING, VerificationNote::Code::NEARLY_INVALID_SUBTITLE_LINE_LENGTH});
1656                 }
1657
1658                 result = LinesCharactersResult();
1659                 for (auto reel: cpl->reels()) {
1660                         for (auto i: reel->closed_captions()) {
1661                                 if (i->asset()) {
1662                                         verify_text_lines_and_characters(i->asset(), 32, 32, &result);
1663                                 }
1664                         }
1665                 }
1666
1667                 if (result.line_count_exceeded) {
1668                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_CLOSED_CAPTION_LINE_COUNT});
1669                 }
1670                 if (result.error_length_exceeded) {
1671                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_CLOSED_CAPTION_LINE_LENGTH});
1672                 }
1673
1674                 if (!cpl->read_composition_metadata()) {
1675                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_CPL_METADATA, cpl->id(), cpl->file().get()});
1676                 } else if (!cpl->version_number()) {
1677                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::MISSING_CPL_METADATA_VERSION_NUMBER, cpl->id(), cpl->file().get()});
1678                 }
1679
1680                 verify_extension_metadata(cpl, notes);
1681
1682                 if (cpl->any_encrypted()) {
1683                         cxml::Document doc("CompositionPlaylist");
1684                         DCP_ASSERT(cpl->file());
1685                         doc.read_file(cpl->file().get());
1686                         if (!doc.optional_node_child("Signature")) {
1687                                 notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::UNSIGNED_CPL_WITH_ENCRYPTED_CONTENT, cpl->id(), cpl->file().get()});
1688                         }
1689                 }
1690         }
1691 }
1692
1693
1694 static
1695 void
1696 verify_pkl(
1697         shared_ptr<const DCP> dcp,
1698         shared_ptr<const PKL> pkl,
1699         boost::filesystem::path xsd_dtd_directory,
1700         vector<VerificationNote>& notes
1701         )
1702 {
1703         validate_xml(pkl->file().get(), xsd_dtd_directory, notes);
1704
1705         if (pkl_has_encrypted_assets(dcp, pkl)) {
1706                 cxml::Document doc("PackingList");
1707                 doc.read_file(pkl->file().get());
1708                 if (!doc.optional_node_child("Signature")) {
1709                         notes.push_back({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::UNSIGNED_PKL_WITH_ENCRYPTED_CONTENT, pkl->id(), pkl->file().get()});
1710                 }
1711         }
1712
1713         set<string> uuid_set;
1714         for (auto asset: pkl->assets()) {
1715                 if (!uuid_set.insert(asset->id()).second) {
1716                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::DUPLICATE_ASSET_ID_IN_PKL, pkl->id(), pkl->file().get()});
1717                         break;
1718                 }
1719         }
1720 }
1721
1722
1723
1724 static
1725 void
1726 verify_assetmap(
1727         shared_ptr<const DCP> dcp,
1728         boost::filesystem::path xsd_dtd_directory,
1729         vector<VerificationNote>& notes
1730         )
1731 {
1732         auto asset_map = dcp->asset_map();
1733         DCP_ASSERT(asset_map);
1734
1735         validate_xml(asset_map->file().get(), xsd_dtd_directory, notes);
1736
1737         set<string> uuid_set;
1738         for (auto const& asset: asset_map->assets()) {
1739                 if (!uuid_set.insert(asset.id()).second) {
1740                         notes.push_back({VerificationNote::Type::ERROR, VerificationNote::Code::DUPLICATE_ASSET_ID_IN_ASSETMAP, asset_map->id(), asset_map->file().get()});
1741                         break;
1742                 }
1743         }
1744 }
1745
1746
1747 vector<VerificationNote>
1748 dcp::verify (
1749         vector<boost::filesystem::path> directories,
1750         function<void (string, optional<boost::filesystem::path>)> stage,
1751         function<void (float)> progress,
1752         VerificationOptions options,
1753         optional<boost::filesystem::path> xsd_dtd_directory
1754         )
1755 {
1756         if (!xsd_dtd_directory) {
1757                 xsd_dtd_directory = resources_directory() / "xsd";
1758         }
1759         *xsd_dtd_directory = boost::filesystem::canonical (*xsd_dtd_directory);
1760
1761         vector<VerificationNote> notes;
1762         State state{};
1763
1764         vector<shared_ptr<DCP>> dcps;
1765         for (auto i: directories) {
1766                 dcps.push_back (make_shared<DCP>(i));
1767         }
1768
1769         for (auto dcp: dcps) {
1770                 stage ("Checking DCP", dcp->directory());
1771                 bool carry_on = true;
1772                 try {
1773                         dcp->read (&notes, true);
1774                 } catch (MissingAssetmapError& e) {
1775                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1776                         carry_on = false;
1777                 } catch (ReadError& e) {
1778                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1779                 } catch (XMLError& e) {
1780                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1781                 } catch (MXFFileError& e) {
1782                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1783                 } catch (cxml::Error& e) {
1784                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::FAILED_READ, string(e.what())});
1785                 }
1786
1787                 if (!carry_on) {
1788                         continue;
1789                 }
1790
1791                 if (dcp->standard() != Standard::SMPTE) {
1792                         notes.push_back ({VerificationNote::Type::BV21_ERROR, VerificationNote::Code::INVALID_STANDARD});
1793                 }
1794
1795                 for (auto cpl: dcp->cpls()) {
1796                         verify_cpl(
1797                                 dcp,
1798                                 cpl,
1799                                 stage,
1800                                 *xsd_dtd_directory,
1801                                 progress,
1802                                 options,
1803                                 notes,
1804                                 state
1805                                 );
1806                 }
1807
1808                 for (auto pkl: dcp->pkls()) {
1809                         stage("Checking PKL", pkl->file());
1810                         verify_pkl(dcp, pkl, *xsd_dtd_directory, notes);
1811                 }
1812
1813                 if (dcp->asset_map_file()) {
1814                         stage("Checking ASSETMAP", dcp->asset_map_file().get());
1815                         verify_assetmap(dcp, *xsd_dtd_directory, notes);
1816                 } else {
1817                         notes.push_back ({VerificationNote::Type::ERROR, VerificationNote::Code::MISSING_ASSETMAP});
1818                 }
1819         }
1820
1821         return notes;
1822 }
1823
1824
1825 string
1826 dcp::note_to_string (VerificationNote note)
1827 {
1828         /** These strings should say what is wrong, incorporating any extra details (ID, filenames etc.).
1829          *
1830          *  e.g. "ClosedCaption asset has no <EntryPoint> tag.",
1831          *  not "ClosedCaption assets must have an <EntryPoint> tag."
1832          *
1833          *  It's OK to use XML tag names where they are clear.
1834          *  If both ID and filename are available, use only the ID.
1835          *  End messages with a full stop.
1836          *  Messages should not mention whether or not their errors are a part of Bv2.1.
1837          */
1838         switch (note.code()) {
1839         case VerificationNote::Code::FAILED_READ:
1840                 return *note.note();
1841         case VerificationNote::Code::MISMATCHED_CPL_HASHES:
1842                 return String::compose("The hash of the CPL %1 in the PKL does not agree with the CPL file.", note.note().get());
1843         case VerificationNote::Code::INVALID_PICTURE_FRAME_RATE:
1844                 return String::compose("The picture in a reel has an invalid frame rate %1.", note.note().get());
1845         case VerificationNote::Code::INCORRECT_PICTURE_HASH:
1846                 return String::compose("The hash of the picture asset %1 does not agree with the PKL file.", note.file()->filename());
1847         case VerificationNote::Code::MISMATCHED_PICTURE_HASHES:
1848                 return String::compose("The PKL and CPL hashes differ for the picture asset %1.", note.file()->filename());
1849         case VerificationNote::Code::INCORRECT_SOUND_HASH:
1850                 return String::compose("The hash of the sound asset %1 does not agree with the PKL file.", note.file()->filename());
1851         case VerificationNote::Code::MISMATCHED_SOUND_HASHES:
1852                 return String::compose("The PKL and CPL hashes differ for the sound asset %1.", note.file()->filename());
1853         case VerificationNote::Code::EMPTY_ASSET_PATH:
1854                 return "The asset map contains an empty asset path.";
1855         case VerificationNote::Code::MISSING_ASSET:
1856                 return String::compose("The file %1 for an asset in the asset map cannot be found.", note.file()->filename());
1857         case VerificationNote::Code::MISMATCHED_STANDARD:
1858                 return "The DCP contains both SMPTE and Interop parts.";
1859         case VerificationNote::Code::INVALID_XML:
1860                 return String::compose("An XML file is badly formed: %1 (%2:%3)", note.note().get(), note.file()->filename(), note.line().get());
1861         case VerificationNote::Code::MISSING_ASSETMAP:
1862                 return "No ASSETMAP or ASSETMAP.xml was found.";
1863         case VerificationNote::Code::INVALID_INTRINSIC_DURATION:
1864                 return String::compose("The intrinsic duration of the asset %1 is less than 1 second.", note.note().get());
1865         case VerificationNote::Code::INVALID_DURATION:
1866                 return String::compose("The duration of the asset %1 is less than 1 second.", note.note().get());
1867         case VerificationNote::Code::INVALID_PICTURE_FRAME_SIZE_IN_BYTES:
1868                 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());
1869         case VerificationNote::Code::NEARLY_INVALID_PICTURE_FRAME_SIZE_IN_BYTES:
1870                 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());
1871         case VerificationNote::Code::EXTERNAL_ASSET:
1872                 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());
1873         case VerificationNote::Code::THREED_ASSET_MARKED_AS_TWOD:
1874                 return String::compose("The asset %1 is 3D but its MXF is marked as 2D.", note.file()->filename());
1875         case VerificationNote::Code::INVALID_STANDARD:
1876                 return "This DCP does not use the SMPTE standard.";
1877         case VerificationNote::Code::INVALID_LANGUAGE:
1878                 return String::compose("The DCP specifies a language '%1' which does not conform to the RFC 5646 standard.", note.note().get());
1879         case VerificationNote::Code::INVALID_PICTURE_SIZE_IN_PIXELS:
1880                 return String::compose("The size %1 of picture asset %2 is not allowed.", note.note().get(), note.file()->filename());
1881         case VerificationNote::Code::INVALID_PICTURE_FRAME_RATE_FOR_2K:
1882                 return String::compose("The frame rate %1 of picture asset %2 is not allowed for 2K DCPs.", note.note().get(), note.file()->filename());
1883         case VerificationNote::Code::INVALID_PICTURE_FRAME_RATE_FOR_4K:
1884                 return String::compose("The frame rate %1 of picture asset %2 is not allowed for 4K DCPs.", note.note().get(), note.file()->filename());
1885         case VerificationNote::Code::INVALID_PICTURE_ASSET_RESOLUTION_FOR_3D:
1886                 return "3D 4K DCPs are not allowed.";
1887         case VerificationNote::Code::INVALID_CLOSED_CAPTION_XML_SIZE_IN_BYTES:
1888                 return String::compose("The size %1 of the closed caption asset %2 is larger than the 256KB maximum.", note.note().get(), note.file()->filename());
1889         case VerificationNote::Code::INVALID_TIMED_TEXT_SIZE_IN_BYTES:
1890                 return String::compose("The size %1 of the timed text asset %2 is larger than the 115MB maximum.", note.note().get(), note.file()->filename());
1891         case VerificationNote::Code::INVALID_TIMED_TEXT_FONT_SIZE_IN_BYTES:
1892                 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());
1893         case VerificationNote::Code::MISSING_SUBTITLE_LANGUAGE:
1894                 return String::compose("The XML for the SMPTE subtitle asset %1 has no <Language> tag.", note.file()->filename());
1895         case VerificationNote::Code::MISMATCHED_SUBTITLE_LANGUAGES:
1896                 return "Some subtitle assets have different <Language> tags than others";
1897         case VerificationNote::Code::MISSING_SUBTITLE_START_TIME:
1898                 return String::compose("The XML for the SMPTE subtitle asset %1 has no <StartTime> tag.", note.file()->filename());
1899         case VerificationNote::Code::INVALID_SUBTITLE_START_TIME:
1900                 return String::compose("The XML for a SMPTE subtitle asset %1 has a non-zero <StartTime> tag.", note.file()->filename());
1901         case VerificationNote::Code::INVALID_SUBTITLE_FIRST_TEXT_TIME:
1902                 return "The first subtitle or closed caption is less than 4 seconds from the start of the DCP.";
1903         case VerificationNote::Code::INVALID_SUBTITLE_DURATION:
1904                 return "At least one subtitle lasts less than 15 frames.";
1905         case VerificationNote::Code::INVALID_SUBTITLE_SPACING:
1906                 return "At least one pair of subtitles is separated by less than 2 frames.";
1907         case VerificationNote::Code::SUBTITLE_OVERLAPS_REEL_BOUNDARY:
1908                 return "At least one subtitle extends outside of its reel.";
1909         case VerificationNote::Code::INVALID_SUBTITLE_LINE_COUNT:
1910                 return "There are more than 3 subtitle lines in at least one place in the DCP.";
1911         case VerificationNote::Code::NEARLY_INVALID_SUBTITLE_LINE_LENGTH:
1912                 return "There are more than 52 characters in at least one subtitle line.";
1913         case VerificationNote::Code::INVALID_SUBTITLE_LINE_LENGTH:
1914                 return "There are more than 79 characters in at least one subtitle line.";
1915         case VerificationNote::Code::INVALID_CLOSED_CAPTION_LINE_COUNT:
1916                 return "There are more than 3 closed caption lines in at least one place.";
1917         case VerificationNote::Code::INVALID_CLOSED_CAPTION_LINE_LENGTH:
1918                 return "There are more than 32 characters in at least one closed caption line.";
1919         case VerificationNote::Code::INVALID_SOUND_FRAME_RATE:
1920                 return String::compose("The sound asset %1 has a sampling rate of %2", note.file()->filename(), note.note().get());
1921         case VerificationNote::Code::MISSING_CPL_ANNOTATION_TEXT:
1922                 return String::compose("The CPL %1 has no <AnnotationText> tag.", note.note().get());
1923         case VerificationNote::Code::MISMATCHED_CPL_ANNOTATION_TEXT:
1924                 return String::compose("The CPL %1 has an <AnnotationText> which differs from its <ContentTitleText>.", note.note().get());
1925         case VerificationNote::Code::MISMATCHED_ASSET_DURATION:
1926                 return "All assets in a reel do not have the same duration.";
1927         case VerificationNote::Code::MISSING_MAIN_SUBTITLE_FROM_SOME_REELS:
1928                 return "At least one reel contains a subtitle asset, but some reel(s) do not.";
1929         case VerificationNote::Code::MISMATCHED_CLOSED_CAPTION_ASSET_COUNTS:
1930                 return "At least one reel has closed captions, but reels have different numbers of closed caption assets.";
1931         case VerificationNote::Code::MISSING_SUBTITLE_ENTRY_POINT:
1932                 return String::compose("The subtitle asset %1 has no <EntryPoint> tag.", note.note().get());
1933         case VerificationNote::Code::INCORRECT_SUBTITLE_ENTRY_POINT:
1934                 return String::compose("The subtitle asset %1 has an <EntryPoint> other than 0.", note.note().get());
1935         case VerificationNote::Code::MISSING_CLOSED_CAPTION_ENTRY_POINT:
1936                 return String::compose("The closed caption asset %1 has no <EntryPoint> tag.", note.note().get());
1937         case VerificationNote::Code::INCORRECT_CLOSED_CAPTION_ENTRY_POINT:
1938                 return String::compose("The closed caption asset %1 has an <EntryPoint> other than 0.", note.note().get());
1939         case VerificationNote::Code::MISSING_HASH:
1940                 return String::compose("The asset %1 has no <Hash> tag in the CPL.", note.note().get());
1941         case VerificationNote::Code::MISSING_FFEC_IN_FEATURE:
1942                 return "The DCP is marked as a Feature but there is no FFEC (first frame of end credits) marker.";
1943         case VerificationNote::Code::MISSING_FFMC_IN_FEATURE:
1944                 return "The DCP is marked as a Feature but there is no FFMC (first frame of moving credits) marker.";
1945         case VerificationNote::Code::MISSING_FFOC:
1946                 return "There should be a FFOC (first frame of content) marker.";
1947         case VerificationNote::Code::MISSING_LFOC:
1948                 return "There should be a LFOC (last frame of content) marker.";
1949         case VerificationNote::Code::INCORRECT_FFOC:
1950                 return String::compose("The FFOC marker is %1 instead of 1", note.note().get());
1951         case VerificationNote::Code::INCORRECT_LFOC:
1952                 return String::compose("The LFOC marker is %1 instead of 1 less than the duration of the last reel.", note.note().get());
1953         case VerificationNote::Code::MISSING_CPL_METADATA:
1954                 return String::compose("The CPL %1 has no <CompositionMetadataAsset> tag.", note.note().get());
1955         case VerificationNote::Code::MISSING_CPL_METADATA_VERSION_NUMBER:
1956                 return String::compose("The CPL %1 has no <VersionNumber> in its <CompositionMetadataAsset>.", note.note().get());
1957         case VerificationNote::Code::MISSING_EXTENSION_METADATA:
1958                 return String::compose("The CPL %1 has no <ExtensionMetadata> in its <CompositionMetadataAsset>.", note.note().get());
1959         case VerificationNote::Code::INVALID_EXTENSION_METADATA:
1960                 return String::compose("The CPL %1 has a malformed <ExtensionMetadata> (%2).", note.file()->filename(), note.note().get());
1961         case VerificationNote::Code::UNSIGNED_CPL_WITH_ENCRYPTED_CONTENT:
1962                 return String::compose("The CPL %1, which has encrypted content, is not signed.", note.note().get());
1963         case VerificationNote::Code::UNSIGNED_PKL_WITH_ENCRYPTED_CONTENT:
1964                 return String::compose("The PKL %1, which has encrypted content, is not signed.", note.note().get());
1965         case VerificationNote::Code::MISMATCHED_PKL_ANNOTATION_TEXT_WITH_CPL:
1966                 return String::compose("The PKL %1 has only one CPL but its <AnnotationText> does not match the CPL's <ContentTitleText>.", note.note().get());
1967         case VerificationNote::Code::PARTIALLY_ENCRYPTED:
1968                 return "Some assets are encrypted but some are not.";
1969         case VerificationNote::Code::INVALID_JPEG2000_CODESTREAM:
1970                 return String::compose("The JPEG2000 codestream for at least one frame is invalid (%1).", note.note().get());
1971         case VerificationNote::Code::INVALID_JPEG2000_GUARD_BITS_FOR_2K:
1972                 return String::compose("The JPEG2000 codestream uses %1 guard bits in a 2K image instead of 1.", note.note().get());
1973         case VerificationNote::Code::INVALID_JPEG2000_GUARD_BITS_FOR_4K:
1974                 return String::compose("The JPEG2000 codestream uses %1 guard bits in a 4K image instead of 2.", note.note().get());
1975         case VerificationNote::Code::INVALID_JPEG2000_TILE_SIZE:
1976                 return "The JPEG2000 tile size is not the same as the image size.";
1977         case VerificationNote::Code::INVALID_JPEG2000_CODE_BLOCK_WIDTH:
1978                 return String::compose("The JPEG2000 codestream uses a code block width of %1 instead of 32.", note.note().get());
1979         case VerificationNote::Code::INVALID_JPEG2000_CODE_BLOCK_HEIGHT:
1980                 return String::compose("The JPEG2000 codestream uses a code block height of %1 instead of 32.", note.note().get());
1981         case VerificationNote::Code::INCORRECT_JPEG2000_POC_MARKER_COUNT_FOR_2K:
1982                 return String::compose("%1 POC markers found in 2K JPEG2000 codestream instead of 0.", note.note().get());
1983         case VerificationNote::Code::INCORRECT_JPEG2000_POC_MARKER_COUNT_FOR_4K:
1984                 return String::compose("%1 POC markers found in 4K JPEG2000 codestream instead of 1.", note.note().get());
1985         case VerificationNote::Code::INCORRECT_JPEG2000_POC_MARKER:
1986                 return String::compose("Incorrect POC marker content found (%1).", note.note().get());
1987         case VerificationNote::Code::INVALID_JPEG2000_POC_MARKER_LOCATION:
1988                 return "POC marker found outside main header.";
1989         case VerificationNote::Code::INVALID_JPEG2000_TILE_PARTS_FOR_2K:
1990                 return String::compose("The JPEG2000 codestream has %1 tile parts in a 2K image instead of 3.", note.note().get());
1991         case VerificationNote::Code::INVALID_JPEG2000_TILE_PARTS_FOR_4K:
1992                 return String::compose("The JPEG2000 codestream has %1 tile parts in a 4K image instead of 6.", note.note().get());
1993         case VerificationNote::Code::MISSING_JPEG200_TLM_MARKER:
1994                 return "No TLM marker was found in a JPEG2000 codestream.";
1995         case VerificationNote::Code::MISMATCHED_TIMED_TEXT_RESOURCE_ID:
1996                 return "The Resource ID in a timed text MXF did not match the ID of the contained XML.";
1997         case VerificationNote::Code::INCORRECT_TIMED_TEXT_ASSET_ID:
1998                 return "The Asset ID in a timed text MXF is the same as the Resource ID or that of the contained XML.";
1999         case VerificationNote::Code::MISMATCHED_TIMED_TEXT_DURATION:
2000         {
2001                 vector<string> parts;
2002                 boost::split (parts, note.note().get(), boost::is_any_of(" "));
2003                 DCP_ASSERT (parts.size() == 2);
2004                 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]);
2005         }
2006         case VerificationNote::Code::MISSED_CHECK_OF_ENCRYPTED:
2007                 return "Some aspect of this DCP could not be checked because it is encrypted.";
2008         case VerificationNote::Code::EMPTY_TEXT:
2009                 return "There is an empty <Text> node in a subtitle or closed caption.";
2010         case VerificationNote::Code::MISMATCHED_CLOSED_CAPTION_VALIGN:
2011                 return "Some closed <Text> or <Image> nodes have different vertical alignments within a <Subtitle>.";
2012         case VerificationNote::Code::INCORRECT_CLOSED_CAPTION_ORDERING:
2013                 return "Some closed captions are not listed in the order of their vertical position.";
2014         case VerificationNote::Code::UNEXPECTED_ENTRY_POINT:
2015                 return "There is an <EntryPoint> node inside a <MainMarkers>.";
2016         case VerificationNote::Code::UNEXPECTED_DURATION:
2017                 return "There is an <Duration> node inside a <MainMarkers>.";
2018         case VerificationNote::Code::INVALID_CONTENT_KIND:
2019                 return String::compose("<ContentKind> has an invalid value %1.", note.note().get());
2020         case VerificationNote::Code::INVALID_MAIN_PICTURE_ACTIVE_AREA:
2021                 return String::compose("<MainPictureActiveaArea> has an invalid value: %1", note.note().get());
2022         case VerificationNote::Code::DUPLICATE_ASSET_ID_IN_PKL:
2023                 return String::compose("The PKL %1 has more than one asset with the same ID.", note.note().get());
2024         case VerificationNote::Code::DUPLICATE_ASSET_ID_IN_ASSETMAP:
2025                 return String::compose("The ASSETMAP %1 has more than one asset with the same ID.", note.note().get());
2026         case VerificationNote::Code::MISSING_SUBTITLE:
2027                 return String::compose("The subtitle asset %1 has no subtitles.", note.note().get());
2028         case VerificationNote::Code::INVALID_SUBTITLE_ISSUE_DATE:
2029                 return String::compose("<IssueDate> has an invalid value: %1", note.note().get());
2030         case VerificationNote::Code::MISMATCHED_SOUND_CHANNEL_COUNTS:
2031                 return String::compose("The sound assets do not all have the same channel count; the first to differ is %1", note.file()->filename());
2032         case VerificationNote::Code::INVALID_MAIN_SOUND_CONFIGURATION:
2033                 return String::compose("<MainSoundConfiguration> has an invalid value: %1", note.note().get());
2034         case VerificationNote::Code::MISSING_FONT:
2035                 return String::compose("The font file for font ID \"%1\" was not found, or was not referred to in the ASSETMAP.", note.note().get());
2036         case VerificationNote::Code::INVALID_JPEG2000_TILE_PART_SIZE:
2037                 return String::compose(
2038                         "Frame %1 has an image component that is too large (component %2 is %3 bytes in size).",
2039                         note.frame().get(), note.component().get(), note.size().get()
2040                         );
2041         case VerificationNote::Code::INCORRECT_SUBTITLE_NAMESPACE_COUNT:
2042                 return String::compose("The XML in the subtitle asset %1 has more than one namespace declaration.", note.note().get());
2043         }
2044
2045         return "";
2046 }
2047
2048
2049 bool
2050 dcp::operator== (dcp::VerificationNote const& a, dcp::VerificationNote const& b)
2051 {
2052         return a.type() == b.type() && a.code() == b.code() && a.note() == b.note() && a.file() == b.file() && a.line() == b.line();
2053 }
2054
2055
2056 bool
2057 dcp::operator< (dcp::VerificationNote const& a, dcp::VerificationNote const& b)
2058 {
2059         if (a.type() != b.type()) {
2060                 return a.type() < b.type();
2061         }
2062
2063         if (a.code() != b.code()) {
2064                 return a.code() < b.code();
2065         }
2066
2067         if (a.note() != b.note()) {
2068                 return a.note().get_value_or("") < b.note().get_value_or("");
2069         }
2070
2071         if (a.file() != b.file()) {
2072                 return a.file().get_value_or("") < b.file().get_value_or("");
2073         }
2074
2075         return a.line().get_value_or(0) < b.line().get_value_or(0);
2076 }
2077
2078
2079 std::ostream&
2080 dcp::operator<< (std::ostream& s, dcp::VerificationNote const& note)
2081 {
2082         s << note_to_string (note);
2083         if (note.note()) {
2084                 s << " [" << note.note().get() << "]";
2085         }
2086         if (note.file()) {
2087                 s << " [" << note.file().get() << "]";
2088         }
2089         if (note.line()) {
2090                 s << " [" << note.line().get() << "]";
2091         }
2092         return s;
2093 }
2094