def main(args_): ''' Launch all the functions for creating an IFI SIP. ''' args = parse_args(args_) start = datetime.datetime.now() inputs = args.i if args.d: try: import clairmeta clairmeta_version = clairmeta.__version__ except ImportError: print( 'Exiting as Clairmeta is not installed. If there is a case for not using clairmeta, please let me know and i can make a workaround' ) sys.exit() print(args) user = ififuncs.determine_user(args) object_entry = get_object_entry(args) sip_path = make_folder_path(os.path.join(args.o), args, object_entry) uuid, uuid_event = determine_uuid(args, sip_path) new_log_textfile = os.path.join(sip_path, 'logs' + '/' + uuid + '_sip_log.log') if args.d: content_title = create_content_title_text(sip_path, args) ififuncs.generate_log(new_log_textfile, 'EVENT = sipcreator.py started') ififuncs.generate_log( new_log_textfile, 'eventDetail=sipcreator.py %s' % ififuncs.get_script_version('sipcreator.py')) ififuncs.generate_log(new_log_textfile, 'Command line arguments: %s' % args) ififuncs.generate_log(new_log_textfile, 'EVENT = agentName=%s' % user) ififuncs.generate_log(new_log_textfile, uuid_event) if not args.sc: ififuncs.generate_log( new_log_textfile, 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=object entry, value=%s' % object_entry) metadata_dir = os.path.join(sip_path, 'metadata') supplemental_dir = os.path.join(metadata_dir, 'supplemental') logs_dir = os.path.join(sip_path, 'logs') if args.accession: accession_number = ififuncs.get_accession_number() reference_number = ififuncs.get_reference_number() parent = ififuncs.ask_question( 'What is the parent record? eg MV 1234. Enter n/a if this is a born digital acquisition with no parent.' ) donor = ififuncs.ask_question( 'Who is the source of acquisition, as appears on the donor agreement? This will not affect Reproductions.' ) depositor_reference = ififuncs.ask_question( 'What is the donor/depositor number? This will not affect Reproductions.' ) acquisition_type = ififuncs.get_acquisition_type('') donation_date = ififuncs.ask_question( 'When was the donation date in DD/MM/YYYY format? Eg. 31/12/1999 - Unfortunately this is NOT using ISO 8601.' ) if args.zip: inputxml, inputtracexml, dfxml = ififuncs.generate_mediainfo_xmls( inputs[0], args.o, uuid, new_log_textfile) if args.manifest: shutil.copy( args.manifest, args.manifest.replace('_manifest.md5', '_manifest-md5.txt')) source_manifest = args.manifest.replace('_manifest.md5', '_manifest-md5.txt') else: source_manifest = os.path.join( args.o, os.path.basename(args.i[0]) + '_manifest-md5.txt') ififuncs.generate_log( new_log_textfile, 'EVENT = message digest calculation, status=started, eventType=messageDigestCalculation, agentName=hashlib, eventDetail=MD5 checksum of source files within ZIP' ) ififuncs.hashlib_manifest(args.i[0], source_manifest, os.path.dirname(args.i[0])) ififuncs.generate_log( new_log_textfile, 'EVENT = message digest calculation, status=finished, eventType=messageDigestCalculation, agentName=hashlib, eventDetail=MD5 checksum of source files within ZIP' ) ififuncs.generate_log( new_log_textfile, 'EVENT = packing, status=started, eventType=packing, agentName=makezip.py, eventDetail=Source object to be packed=%s' % inputs[0]) makezip_judgement, zip_file = makezip.main([ '-i', inputs[0], '-o', os.path.join(sip_path, 'objects'), '-basename', uuid + '.zip' ]) ififuncs.generate_log( new_log_textfile, 'EVENT = packing, status=finished, eventType=packing, agentName=makezip.py, eventDetail=Source object packed into=%s' % zip_file) if makezip_judgement is None: judgement = 'lossless' else: judgement = makezip_judgement ififuncs.generate_log( new_log_textfile, 'EVENT = losslessness verification, status=finished, eventType=messageDigestCalculation, agentName=makezip.py, eventDetail=embedded crc32 checksum validation, eventOutcome=%s' % judgement) ififuncs.generate_log( new_log_textfile, 'EVENT = losslessness verification, status=finished, eventType=messageDigestCalculation, agentName=makezip.py, eventDetail=embedded crc32 checksum validation, eventOutcome=%s' % judgement) else: log_names = move_files(inputs, sip_path, args, user) ififuncs.get_technical_metadata(sip_path, new_log_textfile) ififuncs.hashlib_manifest(metadata_dir, metadata_dir + '/metadata_manifest.md5', metadata_dir) if args.sc: normalise_objects_manifest(sip_path) new_manifest_textfile = consolidate_manifests(sip_path, 'objects', new_log_textfile) if args.zip: ififuncs.generate_log( new_log_textfile, 'EVENT = Message Digest Calculation, status=started, eventType=message digest calculation, eventDetail=%s module=hashlib' % zip_file) ififuncs.manifest_update(new_manifest_textfile, zip_file) ififuncs.generate_log( new_log_textfile, 'EVENT = Message Digest Calculation, status=finished, eventType=message digest calculation, eventDetail=%s module=hashlib' % zip_file) consolidate_manifests(sip_path, 'metadata', new_log_textfile) ififuncs.hashlib_append(logs_dir, new_manifest_textfile, os.path.dirname(os.path.dirname(logs_dir))) if args.supplement: os.makedirs(supplemental_dir) supplement_cmd = [ '-i', args.supplement, '-user', user, '-new_folder', supplemental_dir, os.path.dirname(sip_path), '-copy' ] package_update.main(supplement_cmd) if args.zip: os.makedirs(supplemental_dir) supplement_cmd = [ '-i', [inputxml, inputtracexml, dfxml, source_manifest], '-user', user, '-new_folder', supplemental_dir, os.path.dirname(sip_path), '-copy' ] package_update.main(supplement_cmd) if args.sc: print('Generating Digital Forensics XML') dfxml = accession.make_dfxml(args, sip_path, uuid) ififuncs.generate_log( new_log_textfile, 'EVENT = Metadata extraction - eventDetail=File system metadata extraction using Digital Forensics XML, eventOutcome=%s, agentName=makedfxml' % (dfxml)) ififuncs.manifest_update(new_manifest_textfile, dfxml) sha512_log = manifest.main([sip_path, '-sha512', '-s']) sha512_manifest = os.path.join(os.path.dirname(sip_path), uuid + '_manifest-sha512.txt') ififuncs.merge_logs_append(sha512_log, new_log_textfile, new_manifest_textfile) ififuncs.checksum_replace(sha512_manifest, new_log_textfile, 'sha512') os.remove(sha512_log) ififuncs.sort_manifest(new_manifest_textfile) if not args.quiet: if 'log_names' in locals(): log_report(log_names) finish = datetime.datetime.now() print('\n- %s ran this script at %s and it finished at %s' % (user, start, finish)) if args.d: process_dcp(sip_path, content_title, args, new_manifest_textfile, new_log_textfile, metadata_dir, clairmeta_version) if args.accession: register = accession.make_register() filmographic_dict = ififuncs.extract_metadata(args.csv)[0] for filmographic_record in filmographic_dict: if filmographic_record['Reference Number'].lower( ) == reference_number.lower(): if filmographic_record['Title'] == '': title = filmographic_record[ 'TitleSeries'] + '; ' + filmographic_record['EpisodeNo'] else: title = filmographic_record['Title'] oe_register = make_oe_register() ififuncs.append_csv( oe_register, (object_entry.upper()[:2] + '-' + object_entry[2:], donation_date, '1', '', title, donor, acquisition_type[1], accession_number, 'Representation of %s|Reproduction of %s' % (reference_number, parent), '')) accession_cmd = [ os.path.dirname(sip_path), '-user', user, '-f', '-number', accession_number, '-reference', reference_number, '-register', register, '-csv', args.csv, '-pbcore' ] if not parent.lower() == 'n/a': accession_cmd.extend(['-parent', parent]) accession_cmd.extend(['-donor', donor]) accession_cmd.extend(['-depositor_reference', depositor_reference]) accession_cmd.extend(['-acquisition_type', acquisition_type[2]]) accession_cmd.extend(['-donation_date', donation_date]) print(accession_cmd) accession.main(accession_cmd) return new_log_textfile, new_manifest_textfile
def main(args_): ''' Launch all the functions for creating an IFI SIP. ''' args = parse_args(args_) start = datetime.datetime.now() inputs = args.i print args if args.user: user = args.user else: user = ififuncs.get_user() if args.oe: if args.oe[:2] != 'oe': print 'First two characters must be \'oe\' and last four characters must be four digits' object_entry = ififuncs.get_object_entry() elif len(args.oe[2:]) != 4: print 'First two characters must be \'oe\' and last four characters must be four digits' object_entry = ififuncs.get_object_entry() elif not args.oe[2:].isdigit(): object_entry = ififuncs.get_object_entry() print 'First two characters must be \'oe\' and last four characters must be four digits' else: object_entry = args.oe else: object_entry = ififuncs.get_object_entry() sip_path = make_folder_path(os.path.join(args.o), args, object_entry) if args.u: if ififuncs.validate_uuid4(args.u) is None: uuid = args.u uuid_event = ( 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=UUID, value=%s, module=uuid.uuid4' ) % uuid else: print 'exiting due to invalid UUID' uuid_event = ( 'EVENT = exiting due to invalid UUID supplied on the commmand line: %s' % uuid ) uuid = False else: uuid = os.path.basename(sip_path) uuid_event = ( 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=UUID, value=%s, module=uuid.uuid4' ) % uuid new_log_textfile = os.path.join(sip_path, 'logs' + '/' + uuid + '_sip_log.log') ififuncs.generate_log( new_log_textfile, 'EVENT = sipcreator.py started' ) ififuncs.generate_log( new_log_textfile, 'eventDetail=sipcreator.py %s' % ififuncs.get_script_version('sipcreator.py') ) ififuncs.generate_log( new_log_textfile, 'Command line arguments: %s' % args ) ififuncs.generate_log( new_log_textfile, 'EVENT = agentName=%s' % user ) ififuncs.generate_log( new_log_textfile, uuid_event ) if args.u is False: sys.exit() ififuncs.generate_log( new_log_textfile, 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=object entry, value=%s' % object_entry ) metadata_dir = os.path.join(sip_path, 'metadata') logs_dir = os.path.join(sip_path, 'logs') log_names = move_files(inputs, sip_path, args) get_metadata(sip_path, new_log_textfile) ififuncs.hashlib_manifest( metadata_dir, metadata_dir + '/metadata_manifest.md5', metadata_dir ) new_manifest_textfile = consolidate_manifests(sip_path, 'objects', new_log_textfile) consolidate_manifests(sip_path, 'metadata', new_log_textfile) ififuncs.hashlib_append( logs_dir, new_manifest_textfile, os.path.dirname(os.path.dirname(logs_dir)) ) ififuncs.sort_manifest(new_manifest_textfile) if not args.quiet: log_report(log_names) finish = datetime.datetime.now() print '\n', user, 'ran this script at %s and it finished at %s' % (start, finish) if args.d: content_title = create_content_title_text(args, sip_path) ififuncs.manifest_replace( new_manifest_textfile, os.path.join('objects', os.path.basename(args.i[0])).replace("\\", "/"), os.path.join('objects', content_title).replace("\\", "/") ) return new_log_textfile, new_manifest_textfile
def main(args_): ''' Launch all the functions for creating an IFI SIP. ''' args = parse_args(args_) start = datetime.datetime.now() inputs = args.i if args.d: try: import clairmeta clairmeta_version = clairmeta.__version__ except ImportError: print( 'Exiting as Clairmeta is not installed. If there is a case for not using clairmeta, please let me know and i can make a workaround' ) sys.exit() print(args) user = ififuncs.determine_user(args) object_entry = get_object_entry(args) sip_path = make_folder_path(os.path.join(args.o), args, object_entry) uuid, uuid_event = determine_uuid(args, sip_path) new_log_textfile = os.path.join(sip_path, 'logs' + '/' + uuid + '_sip_log.log') if args.d: content_title = create_content_title_text(sip_path, args) ififuncs.generate_log(new_log_textfile, 'EVENT = sipcreator.py started') ififuncs.generate_log( new_log_textfile, 'eventDetail=sipcreator.py %s' % ififuncs.get_script_version('sipcreator.py')) ififuncs.generate_log(new_log_textfile, 'Command line arguments: %s' % args) ififuncs.generate_log(new_log_textfile, 'EVENT = agentName=%s' % user) ififuncs.generate_log(new_log_textfile, uuid_event) if not args.sc: ififuncs.generate_log( new_log_textfile, 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=object entry, value=%s' % object_entry) metadata_dir = os.path.join(sip_path, 'metadata') supplemental_dir = os.path.join(metadata_dir, 'supplemental') logs_dir = os.path.join(sip_path, 'logs') log_names = move_files(inputs, sip_path, args) ififuncs.get_technical_metadata(sip_path, new_log_textfile) ififuncs.hashlib_manifest(metadata_dir, metadata_dir + '/metadata_manifest.md5', metadata_dir) if args.sc: normalise_objects_manifest(sip_path) new_manifest_textfile = consolidate_manifests(sip_path, 'objects', new_log_textfile) consolidate_manifests(sip_path, 'metadata', new_log_textfile) ififuncs.hashlib_append(logs_dir, new_manifest_textfile, os.path.dirname(os.path.dirname(logs_dir))) if args.supplement: os.makedirs(supplemental_dir) supplement_cmd = [ '-i', args.supplement, '-user', user, '-new_folder', supplemental_dir, os.path.dirname(sip_path), '-copy' ] package_update.main(supplement_cmd) if args.sc: print('Generating Digital Forensics XML') dfxml = accession.make_dfxml(args, sip_path, uuid) ififuncs.generate_log( new_log_textfile, 'EVENT = Metadata extraction - eventDetail=File system metadata extraction using Digital Forensics XML, eventOutcome=%s, agentName=makedfxml' % (dfxml)) ififuncs.manifest_update(new_manifest_textfile, dfxml) sha512_log = manifest.main([sip_path, '-sha512', '-s']) sha512_manifest = os.path.join(os.path.dirname(sip_path), uuid + '_manifest-sha512.txt') ififuncs.merge_logs_append(sha512_log, new_log_textfile, new_manifest_textfile) ififuncs.checksum_replace(sha512_manifest, new_log_textfile, 'sha512') os.remove(sha512_log) ififuncs.sort_manifest(new_manifest_textfile) if not args.quiet: log_report(log_names) finish = datetime.datetime.now() print(('\n', user, 'ran this script at %s and it finished at %s' % (start, finish))) if args.d: process_dcp(sip_path, content_title, args, new_manifest_textfile, new_log_textfile, metadata_dir, clairmeta_version) return new_log_textfile, new_manifest_textfile
def main(args_): ''' Launch all the functions for creating an IFI SIP. ''' args = parse_args(args_) start = datetime.datetime.now() inputs = args.i print args if args.user: user = args.user else: user = ififuncs.get_user() if args.oe: if args.oe[:2] != 'oe': print 'First two characters must be \'oe\' and last four characters must be four digits' object_entry = ififuncs.get_object_entry() elif len(args.oe[2:]) not in range(4, 6): print 'First two characters must be \'oe\' and last four characters must be four digits' object_entry = ififuncs.get_object_entry() elif not args.oe[2:].isdigit(): object_entry = ififuncs.get_object_entry() print 'First two characters must be \'oe\' and last four characters must be four digits' else: object_entry = args.oe else: object_entry = ififuncs.get_object_entry() sip_path = make_folder_path(os.path.join(args.o), args, object_entry) if args.u: if ififuncs.validate_uuid4(args.u) is None: uuid = args.u uuid_event = ( 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=UUID, value=%s, module=uuid.uuid4' ) % uuid else: print 'exiting due to invalid UUID' uuid_event = ( 'EVENT = exiting due to invalid UUID supplied on the commmand line: %s' % uuid) uuid = False else: uuid = os.path.basename(sip_path) uuid_event = ( 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=UUID, value=%s, module=uuid.uuid4') % uuid new_log_textfile = os.path.join(sip_path, 'logs' + '/' + uuid + '_sip_log.log') ififuncs.generate_log(new_log_textfile, 'EVENT = sipcreator.py started') ififuncs.generate_log( new_log_textfile, 'eventDetail=sipcreator.py %s' % ififuncs.get_script_version('sipcreator.py')) ififuncs.generate_log(new_log_textfile, 'Command line arguments: %s' % args) ififuncs.generate_log(new_log_textfile, 'EVENT = agentName=%s' % user) ififuncs.generate_log(new_log_textfile, uuid_event) if args.u is False: sys.exit() ififuncs.generate_log( new_log_textfile, 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=object entry, value=%s' % object_entry) metadata_dir = os.path.join(sip_path, 'metadata') logs_dir = os.path.join(sip_path, 'logs') log_names = move_files(inputs, sip_path, args) get_metadata(sip_path, new_log_textfile) ififuncs.hashlib_manifest(metadata_dir, metadata_dir + '/metadata_manifest.md5', metadata_dir) new_manifest_textfile = consolidate_manifests(sip_path, 'objects', new_log_textfile) consolidate_manifests(sip_path, 'metadata', new_log_textfile) ififuncs.hashlib_append(logs_dir, new_manifest_textfile, os.path.dirname(os.path.dirname(logs_dir))) ififuncs.sort_manifest(new_manifest_textfile) if not args.quiet: log_report(log_names) finish = datetime.datetime.now() print '\n', user, 'ran this script at %s and it finished at %s' % (start, finish) if args.d: content_title = create_content_title_text(args, sip_path) ififuncs.manifest_replace( new_manifest_textfile, os.path.join('objects', os.path.basename(args.i[0])).replace("\\", "/"), os.path.join('objects', content_title).replace("\\", "/")) return new_log_textfile, new_manifest_textfile
def main(args_): ''' Launch all the functions for creating an IFI SIP. ''' args = parse_args(args_) start = datetime.datetime.now() inputs = args.i if args.d: try: import clairmeta except ImportError: print( 'Exiting as Clairmeta is not installed. If there is a case for not using clairmeta, please let me know and i can make a workaround' ) sys.exit() print args if args.user: user = args.user else: user = ififuncs.get_user() if not args.sc: if args.oe: if args.oe[:2] != 'oe': print 'First two characters must be \'oe\' and last four characters must be four digits' object_entry = ififuncs.get_object_entry() elif len(args.oe[2:]) not in range(4, 6): print 'First two characters must be \'oe\' and last four characters must be four digits' object_entry = ififuncs.get_object_entry() elif not args.oe[2:].isdigit(): object_entry = ififuncs.get_object_entry() print 'First two characters must be \'oe\' and last four characters must be four digits' else: object_entry = args.oe else: object_entry = ififuncs.get_object_entry() else: object_entry = 'not_applicable' sip_path = make_folder_path(os.path.join(args.o), args, object_entry) if args.u: if ififuncs.validate_uuid4(args.u) is None: uuid = args.u uuid_event = ( 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=UUID, value=%s, module=uuid.uuid4' ) % uuid else: print 'exiting due to invalid UUID' uuid_event = ( 'EVENT = exiting due to invalid UUID supplied on the commmand line: %s' % uuid) uuid = False else: uuid = os.path.basename(sip_path) uuid_event = ( 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=UUID, value=%s, module=uuid.uuid4') % uuid new_log_textfile = os.path.join(sip_path, 'logs' + '/' + uuid + '_sip_log.log') ififuncs.generate_log(new_log_textfile, 'EVENT = sipcreator.py started') ififuncs.generate_log( new_log_textfile, 'eventDetail=sipcreator.py %s' % ififuncs.get_script_version('sipcreator.py')) ififuncs.generate_log(new_log_textfile, 'Command line arguments: %s' % args) ififuncs.generate_log(new_log_textfile, 'EVENT = agentName=%s' % user) ififuncs.generate_log(new_log_textfile, uuid_event) if args.u is False: sys.exit() if not args.sc: ififuncs.generate_log( new_log_textfile, 'EVENT = eventType=Identifier assignement,' ' eventIdentifierType=object entry, value=%s' % object_entry) metadata_dir = os.path.join(sip_path, 'metadata') supplemental_dir = os.path.join(metadata_dir, 'supplemental') logs_dir = os.path.join(sip_path, 'logs') log_names = move_files(inputs, sip_path, args) get_metadata(sip_path, new_log_textfile) ififuncs.hashlib_manifest(metadata_dir, metadata_dir + '/metadata_manifest.md5', metadata_dir) if args.sc: normalise_objects_manifest(sip_path) new_manifest_textfile = consolidate_manifests(sip_path, 'objects', new_log_textfile) consolidate_manifests(sip_path, 'metadata', new_log_textfile) ififuncs.hashlib_append(logs_dir, new_manifest_textfile, os.path.dirname(os.path.dirname(logs_dir))) if args.supplement: os.makedirs(supplemental_dir) supplement_cmd = [ '-i', args.supplement, '-user', user, '-new_folder', supplemental_dir, os.path.dirname(sip_path), '-copy' ] package_update.main(supplement_cmd) if args.sc: print('Generating Digital Forensics XML') dfxml = accession.make_dfxml(args, sip_path, uuid) ififuncs.generate_log( new_log_textfile, 'EVENT = Metadata extraction - eventDetail=File system metadata extraction using Digital Forensics XML, eventOutcome=%s, agentName=makedfxml' % (dfxml)) ififuncs.manifest_update(new_manifest_textfile, dfxml) sha512_log = manifest.main([sip_path, '-sha512', '-s']) sha512_manifest = os.path.join(os.path.dirname(sip_path), uuid + '_manifest-sha512.txt') ififuncs.merge_logs_append(sha512_log, new_log_textfile, new_manifest_textfile) ififuncs.checksum_replace(sha512_manifest, new_log_textfile, 'sha512') os.remove(sha512_log) ififuncs.sort_manifest(new_manifest_textfile) if not args.quiet: log_report(log_names) finish = datetime.datetime.now() print '\n', user, 'ran this script at %s and it finished at %s' % (start, finish) if args.d: content_title = create_content_title_text(sip_path) new_dcp_path = os.path.join('objects', content_title).replace("\\", "/") absolute_dcp_path = os.path.join(sip_path, new_dcp_path) ififuncs.manifest_replace( new_manifest_textfile, os.path.join('objects', os.path.basename(args.i[0])).replace("\\", "/"), new_dcp_path) ''' a = subprocess.check_output(['python', '-m', 'clairmeta.cli', 'check', '-type', 'dcp', absolute_dcp_path], stderr=subprocess.STDOUT) b = subprocess.check_output(['python', '-m', 'clairmeta.cli', 'probe', '-type', 'dcp', '-format', 'xml', absolute_dcp_path], stderr=subprocess.STDOUT) ''' dcp = DCP(absolute_dcp_path) clairmeta_version = clairmeta.__version__ dcp_dict = dcp.parse() # json_str = json.dumps(dcp_dict , sort_keys=True, indent=2, separators=(',', ': ')) xml_str = dicttoxml.dicttoxml(dcp_dict, custom_root='ClairmetaProbe', ids=False, attr_type=False) xml_pretty = prettyprint_xml(xml_str) status, report = dcp.check() ififuncs.generate_log( new_log_textfile, 'EVENT = eventType=validation, eventOutcome=%s, eventDetail=%s, agentName=Clairmeta version %s' % (status, report, clairmeta_version)) clairmeta_xml = os.path.join(metadata_dir, '%s_clairmeta.xml' % content_title) ififuncs.generate_log( new_log_textfile, 'EVENT = Metadata extraction - eventDetail=Clairmeta DCP metadata extraction, eventOutcome=%s, agentName=Clairmeta version %s' % (clairmeta_xml, clairmeta_version)) with open(clairmeta_xml, 'w') as fo: fo.write(xml_pretty) ififuncs.checksum_replace(new_manifest_textfile, new_log_textfile, 'md5') ififuncs.manifest_update(new_manifest_textfile, clairmeta_xml) print status print report print '\n', user, 'ran this script at %s and it finished at %s' % ( start, finish) return new_log_textfile, new_manifest_textfile