def main(args_):
# if multiple file are present, this script will treat them as a single
# instantiation/representation and get aggregate metadata about the whole
# package. For now, this will be a clumsy implementation - the first file
# will provide most metadata. Things like duration/bitrate/filesize
# will be calculated as a whole.
# Although another way would be that every call is looped, and if
# this could catch files that should not be in the package, eg. a 4:2:2
# file in a 4:2:0 package..
# yup - do it that way!
args = parse_args(args_)
all_files = ififuncs.recursive_file_list(args.input)
silence = True
if args.user:
user = args.user
else:
user = ififuncs.get_user()
for dirs in os.listdir(args.input):
if ififuncs.validate_uuid4(dirs) is None:
instantiationIdentif = dirs
Accession_Number = get_accession_number(args.input)
Reference_Number = get_reference_number(args.input)
if args.p:
for root, _, filenames in os.walk(args.input):
if os.path.basename(root) == 'metadata':
metadata_dir = root
elif os.path.basename(root) == 'logs':
logs_dir = root
csv_filename = os.path.join(metadata_dir, Accession_Number + '.csv')
sipcreator_log = os.path.join(logs_dir,
instantiationIdentif + '_sip_log.log')
ififuncs.generate_log(sipcreator_log, 'EVENT = makepbcore.py started')
ififuncs.generate_log(
sipcreator_log, 'eventDetail=makepbcore.py %s' %
ififuncs.get_script_version('makepbcore.py'))
ififuncs.generate_log(sipcreator_log,
'Command line arguments: %s' % args)
ififuncs.generate_log(sipcreator_log, 'EVENT = agentName=%s' % user)
else:
csv_filename = 'blaa.csv'
print ' - Metadata will be stored in %s' % csv_filename
for filenames in os.listdir(args.input):
if '_manifest.md5' in filenames:
md5_manifest = os.path.join(args.input, filenames)
elif 'manifest-sha512.txt' in filenames:
sha512_manifest = os.path.join(args.input, filenames)
make_csv(csv_filename)
ms = 0
FrameCount = 0
instantFileSize = 0
instantFileSize_gigs = 0
for source in all_files:
metadata = subprocess.check_output(
['mediainfo', '--Output=PBCore2', source])
root = etree.fromstring(metadata)
print(' - Analysing %s') % source
pbcore_namespace = root.xpath('namespace-uri(.)')
track_type = root.xpath('//ns:essenceTrackType',
namespaces={'ns': pbcore_namespace})
if len(track_type) > 0:
for track in track_type:
if track.text == 'Video':
essenceTrackEncodvid = get_metadata(
"ns:essenceTrackEncoding", track.getparent(),
pbcore_namespace)
vcodec_attributes = get_attributes(track.getparent(),
pbcore_namespace)
elif track.text == 'Audio':
silence = False
essenceTrackEncod_au = get_metadata(
"ns:essenceTrackEncoding", track.getparent(),
pbcore_namespace)
acodec_attributes = get_attributes(track.getparent(),
pbcore_namespace)
ScanType = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='ScanType']", root,
pbcore_namespace)
matrix_coefficients = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='matrix_coefficients']",
root, pbcore_namespace)
transfer_characteris = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='transfer_characteristics']",
root, pbcore_namespace)
colour_primaries = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='color_primaries']",
root, pbcore_namespace)
FrameCount += int(
get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='FrameCount']",
root, pbcore_namespace))
instantFileSize += int(
get_metadata("//ns:instantiationFileSize", root, pbcore_namespace))
instantDataRate = round(
float(
ififuncs.get_mediainfo('OverallBitRate',
'--inform=General;%OverallBitRate%',
source)) / 1000 / 1000, 2)
ms += ififuncs.get_milliseconds(source)
ColorSpace = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='ColorSpace']", root,
pbcore_namespace)
ChromaSubsampling = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='ChromaSubsampling']",
root, pbcore_namespace)
instantMediaty = get_metadata("//ns:instantiationMediaType", root,
pbcore_namespace)
essenceFrameSize = get_metadata("//ns:essenceTrackFrameSize", root,
pbcore_namespace)
essenceAspectRatio = get_metadata("//ns:essenceTrackAspectRatio", root,
pbcore_namespace)
PixelAspectRatio = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='PixelAspectRatio']",
root, pbcore_namespace)
instantiationStandar = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Format']", root,
pbcore_namespace)
essenceFrameRate = get_metadata("//ns:essenceTrackFrameRate", root,
pbcore_namespace)
essenceTrackSampling = get_metadata("//ns:essenceTrackSamplingRate",
root, pbcore_namespace)
Interlacement = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Interlacement']",
root, pbcore_namespace)
Compression_Mode = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Compression_Mode']",
root, pbcore_namespace)
instantiationDate_modified = get_metadata(
"//ns:instantiationDate[@dateType='file modification']", root,
pbcore_namespace)
pix_fmt = ififuncs.get_ffmpeg_fmt(source, 'video')
audio_fmt = ififuncs.get_ffmpeg_fmt(source, 'audio')
if not silence:
audio_codecid = acodec_attributes['ref']
essenceBitDepth_au = ififuncs.get_mediainfo(
'duration', '--inform=Audio;%BitDepth%', source)
else:
audio_codecid = 'n/a'
essenceBitDepth_au = 'n/a'
essenceTrackEncod_au = 'n/a'
video_codecid = vcodec_attributes['ref']
try:
video_codec_version = vcodec_attributes['version']
except KeyError:
video_codec_version = 'n/a'
try:
video_codec_profile = vcodec_attributes['annotation'][8:]
except KeyError:
video_codec_profile = 'n/a'
tc = ififuncs.convert_millis(ms)
instantiationDuratio = ififuncs.convert_timecode(25, tc)
Donor = ''
Edited_By = user
Date_Created = ''
Date_Last_Modified = ''
Film_Or_Tape = 'Digital File'
Date_Of_Donation = ''
Habitat = ''
Type_Of_Deposit = ''
Depositor_Reference = ''
Master_Viewing = 'Preservation Master'
Language_Version = ''
Condition_Rating = ''
Companion_Elements = ''
EditedNew = user
FIO = 'In'
CollectionTitle = ''
Created_By = user
instantiationDimensi = ''
instantiationLocatio = ''
instantTimeStart = ''
instantFileSize_gigs = round(
float(instantFileSize) / 1024 / 1024 / 1024, 3)
instantColors = ''
instantLanguage = ''
instantAltMo = 'n/a'
essenceBitDepth_vid = ififuncs.get_mediainfo('duration',
'--inform=Video;%BitDepth%',
source)
instantiationChanCon = ''
ififuncs.append_csv(csv_filename, [
Reference_Number, Donor, Edited_By, Date_Created, Date_Last_Modified,
Film_Or_Tape, Date_Of_Donation, Accession_Number, Habitat,
Type_Of_Deposit, Depositor_Reference, Master_Viewing, Language_Version,
Condition_Rating, Companion_Elements, EditedNew, FIO, CollectionTitle,
Created_By, instantiationIdentif, instantiationDate_modified,
instantiationDimensi, instantiationStandar, instantiationLocatio,
instantMediaty, instantFileSize, instantFileSize_gigs,
instantTimeStart, instantDataRate, instantColors, instantLanguage,
instantAltMo, essenceTrackEncodvid, essenceFrameRate,
essenceTrackSampling, essenceBitDepth_vid, essenceFrameSize,
essenceAspectRatio, essenceTrackEncod_au, essenceBitDepth_au,
instantiationDuratio, instantiationChanCon, PixelAspectRatio,
FrameCount, ColorSpace, ChromaSubsampling, ScanType, Interlacement,
Compression_Mode, colour_primaries, transfer_characteris,
matrix_coefficients, pix_fmt, audio_fmt, audio_codecid, video_codecid,
video_codec_version, video_codec_profile
])
if args.p:
ififuncs.generate_log(
sipcreator_log,
'EVENT = Metadata extraction - eventDetail=Technical record creation using PBCore, eventOutcome=%s, agentName=makepbcore'
% (csv_filename))
ififuncs.generate_log(sipcreator_log, 'EVENT = makepbcore.py finished')
ififuncs.checksum_replace(md5_manifest, sipcreator_log, 'md5')
ififuncs.checksum_replace(sha512_manifest, sipcreator_log, 'sha512')
ififuncs.manifest_update(md5_manifest, csv_filename)
print ' - Updating %s with %s' % (md5_manifest, csv_filename)
ififuncs.sha512_update(sha512_manifest, csv_filename)
print ' - Updating %s with %s' % (sha512_manifest, csv_filename)
def main(args_):
'''
Launches the various functions that will accession a package
'''
args = parse_args(args_)
source = args.input
uuid_directory = ififuncs.check_for_sip([source])
if uuid_directory is not None:
oe_path = os.path.dirname(uuid_directory)
oe_number = os.path.basename(oe_path)
if args.user:
user = args.user
else:
user = ififuncs.get_user()
if args.number:
if args.number[:3] != 'aaa':
print 'First three characters must be \'aaa\' and last four characters must be four digits'
accession_number = ififuncs.get_accession_number()
elif len(args.number[3:]) != 4:
accession_number = ififuncs.get_accession_number()
print 'First three characters must be \'aaa\' and last four characters must be four digits'
elif not args.number[3:].isdigit():
accession_number = ififuncs.get_accession_number()
print 'First three characters must be \'aaa\' and last four characters must be four digits'
else:
accession_number = args.number
else:
accession_number = ififuncs.get_accession_number()
if args.pbcore:
if args.reference:
Reference_Number = args.reference.upper()
else:
Reference_Number = ififuncs.get_reference_number()
if args.acquisition_type:
acquisition_type = ififuncs.get_acquisition_type(
args.acquisition_type)
print acquisition_type
accession_path = os.path.join(os.path.dirname(oe_path),
accession_number)
uuid = os.path.basename(uuid_directory)
new_uuid_path = os.path.join(accession_path, uuid)
logs_dir = os.path.join(new_uuid_path, 'logs')
sipcreator_log = os.path.join(logs_dir, uuid) + '_sip_log.log'
if args.force:
proceed = 'Y'
else:
proceed = ififuncs.ask_yes_no('Do you want to rename %s with %s' %
(oe_number, accession_number))
if proceed == 'Y':
os.rename(oe_path, accession_path)
if args.register:
register = args.register
else:
register = make_register()
ififuncs.append_csv(register,
(oe_number.upper()[:2] + '-' + oe_number[2:],
accession_number, '', '', '', '', '', ''))
ififuncs.generate_log(sipcreator_log, 'EVENT = accession.py started')
ififuncs.generate_log(
sipcreator_log, 'eventDetail=accession.py %s' %
ififuncs.get_script_version('accession.py'))
ififuncs.generate_log(sipcreator_log,
'Command line arguments: %s' % args)
ififuncs.generate_log(sipcreator_log, 'EVENT = agentName=%s' % user)
ififuncs.generate_log(
sipcreator_log, 'EVENT = eventType=Identifier assignment,'
' eventIdentifierType=accession number, value=%s' %
accession_number)
ififuncs.generate_log(
sipcreator_log, 'EVENT = eventType=accession,'
' eventIdentifierType=accession number, value=%s' %
accession_number)
sip_manifest = os.path.join(accession_path, uuid) + '_manifest.md5'
sha512_log = manifest.main([new_uuid_path, '-sha512', '-s'])
sha512_manifest = os.path.join(os.path.dirname(new_uuid_path),
uuid + '_manifest-sha512.txt')
ififuncs.merge_logs_append(sha512_log, sipcreator_log, sip_manifest)
os.remove(sha512_log)
print('Generating Digital Forensics XML')
dfxml = make_dfxml(args, new_uuid_path, uuid)
ififuncs.generate_log(
sipcreator_log,
'EVENT = Metadata extraction - eventDetail=File system metadata extraction using Digital Forensics XML, eventOutcome=%s, agentName=makedfxml'
% (dfxml))
# this is inefficient. The script should not have to ask for reference
# number twice if someone wants to insert the filmographic but do not
# want to make the pbcore csv, perhaps because the latter already exists.
if args.csv:
metadata_dir = os.path.join(new_uuid_path, 'metadata')
package_filmographic = os.path.join(
metadata_dir, Reference_Number + '_filmographic.csv')
insert_filmographic(args.csv, Reference_Number,
package_filmographic)
ififuncs.generate_log(
sipcreator_log,
'EVENT = Metadata extraction - eventDetail=Filmographic descriptive metadata added to metadata folder, eventOutcome=%s, agentName=accession.py'
% (package_filmographic))
ififuncs.manifest_update(sip_manifest, package_filmographic)
ififuncs.sha512_update(sha512_manifest, package_filmographic)
print('Filmographic descriptive metadata added to metadata folder')
ififuncs.generate_log(sipcreator_log, 'EVENT = accession.py finished')
ififuncs.checksum_replace(sip_manifest, sipcreator_log, 'md5')
ififuncs.checksum_replace(sha512_manifest, sipcreator_log, 'sha512')
ififuncs.manifest_update(sip_manifest, dfxml)
ififuncs.sha512_update(sha512_manifest, dfxml)
if args.pbcore:
makepbcore_cmd = [
accession_path, '-p', '-user', user, '-reference',
Reference_Number
]
if args.parent:
makepbcore_cmd.extend(['-parent', args.parent])
if args.acquisition_type:
makepbcore_cmd.extend(
['-acquisition_type', args.acquisition_type])
if args.donor:
makepbcore_cmd.extend(['-donor', args.donor])
if args.donor:
makepbcore_cmd.extend(
['-depositor_reference', args.depositor_reference])
if args.donation_date:
makepbcore_cmd.extend(['-donation_date', args.donation_date])
makepbcore.main(makepbcore_cmd)
else:
print 'not a valid package. The input should include a package that has been through Object Entry'
def main(args_):
'''
Launches the various functions that will accession a package
'''
args = parse_args(args_)
input = args.input
uuid_directory = ififuncs.check_for_sip([input])
if uuid_directory is not None:
oe_path = os.path.dirname(uuid_directory)
oe_number = os.path.basename(oe_path)
if args.user:
user = args.user
else:
user = ififuncs.get_user()
if args.number:
if args.number[:3] != 'aaa':
print 'First three characters must be \'aaa\' and last four characters must be four digits'
accession_number = ififuncs.get_accession_number()
elif len(args.number[3:]) != 4:
accession_number = ififuncs.get_accession_number()
print 'First three characters must be \'aaa\' and last four characters must be four digits'
elif not args.number[3:].isdigit():
accession_number = ififuncs.get_accession_number()
print 'First three characters must be \'aaa\' and last four characters must be four digits'
else:
accession_number = args.number
else:
accession_number = ififuncs.get_accession_number()
accession_path = os.path.join(
os.path.dirname(oe_path), accession_number
)
uuid = os.path.basename(uuid_directory)
new_uuid_path = os.path.join(accession_path, uuid)
logs_dir = os.path.join(new_uuid_path, 'logs')
sipcreator_log = os.path.join(logs_dir, uuid) + '_sip_log.log'
if args.force:
proceed = 'Y'
else:
proceed = ififuncs.ask_yes_no(
'Do you want to rename %s with %s' % (oe_number, accession_number)
)
if proceed == 'Y':
os.rename(oe_path, accession_path)
register = make_register()
ififuncs.append_csv(register, (oe_number.upper()[:2] + '-' + oe_number[2:6], accession_number, '','','','', ''))
ififuncs.generate_log(
sipcreator_log,
'EVENT = accession.py started'
)
ififuncs.generate_log(
sipcreator_log,
'eventDetail=accession.py %s' % ififuncs.get_script_version('accession.py')
)
ififuncs.generate_log(
sipcreator_log,
'Command line arguments: %s' % args
)
ififuncs.generate_log(
sipcreator_log,
'EVENT = agentName=%s' % user
)
ififuncs.generate_log(
sipcreator_log,
'EVENT = eventType=Identifier assignment,'
' eventIdentifierType=accession number, value=%s'
% accession_number
)
ififuncs.generate_log(
sipcreator_log,
'EVENT = eventType=accession,'
' eventIdentifierType=accession number, value=%s'
% accession_number
)
sip_manifest = os.path.join(
accession_path, uuid
) + '_manifest.md5'
sha512_log = manifest.main([new_uuid_path, '-sha512', '-s'])
sha512_manifest = os.path.join(
os.path.dirname(new_uuid_path), uuid + '_manifest-sha512.txt'
)
ififuncs.merge_logs_append(sha512_log, sipcreator_log, sip_manifest)
os.remove(sha512_log)
dfxml = make_dfxml(args, new_uuid_path, uuid)
ififuncs.generate_log(
sipcreator_log,
'EVENT = Metadata extraction - eventDetail=File system metadata extraction using Digital Forensics XML, eventOutcome=%s, agentName=makedfxml' % (dfxml)
)
ififuncs.generate_log(
sipcreator_log,
'EVENT = accession.py finished'
)
ififuncs.checksum_replace(sip_manifest, sipcreator_log, 'md5')
ififuncs.checksum_replace(sha512_manifest, sipcreator_log, 'sha512')
ififuncs.manifest_update(sip_manifest, dfxml)
ififuncs.sha512_update(sha512_manifest, dfxml)
if args.pbcore:
makepbcore.main([accession_path, '-p', '-user', user])
else:
print 'not a valid package. The input should include a package that has been through Object Entry'
def main(args_):
# if multiple file are present, this script will treat them as a single
# instantiation/representation and get aggregate metadata about the whole
# package. For now, this will be a clumsy implementation - the first file
# will provide most metadata. Things like duration/bitrate/filesize
# will be calculated as a whole.
# Although another way would be that every call is looped, and if
# this could catch files that should not be in the package, eg. a 4:2:2
# file in a 4:2:0 package..
# yup - do it that way!
args = parse_args(args_)
all_files = ififuncs.recursive_file_list(args.input)
silence = True
if args.user:
user = args.user
else:
user = ififuncs.get_user()
acquisition_type = ''
if args.acquisition_type:
acquisition_type = ififuncs.get_acquisition_type(
args.acquisition_type)[0]
for dirs in os.listdir(args.input):
if ififuncs.validate_uuid4(dirs) is None:
instantiationIdentif = dirs
Accession_Number = get_accession_number(args.input)
if args.reference:
Reference_Number = args.reference.upper()
else:
Reference_Number = get_reference_number(args.input)
if args.p:
for root, _, filenames in os.walk(args.input):
if os.path.basename(root) == 'metadata':
metadata_dir = root
elif os.path.basename(root) == 'logs':
logs_dir = root
csv_filename = os.path.join(metadata_dir,
Accession_Number + '_pbcore.csv')
sipcreator_log = os.path.join(logs_dir,
instantiationIdentif + '_sip_log.log')
ififuncs.generate_log(sipcreator_log, 'EVENT = makepbcore.py started')
ififuncs.generate_log(
sipcreator_log, 'eventDetail=makepbcore.py %s' %
ififuncs.get_script_version('makepbcore.py'))
ififuncs.generate_log(sipcreator_log,
'Command line arguments: %s' % args)
ififuncs.generate_log(sipcreator_log, 'EVENT = agentName=%s' % user)
else:
csv_filename = 'blaa.csv'
print ' - Metadata will be stored in %s' % csv_filename
for filenames in os.listdir(args.input):
if '_manifest.md5' in filenames:
md5_manifest = os.path.join(args.input, filenames)
elif 'manifest-sha512.txt' in filenames:
sha512_manifest = os.path.join(args.input, filenames)
make_csv(csv_filename)
ms = 0
FrameCount = 0
instantFileSize_byte = 0
instantFileSize_gigs = 0
scan_types = []
matrix_list = []
transfer_list = []
colour_primaries_list = []
color_spaces = []
chroma = []
frame_sizes = []
par_list = []
container_list = []
fps_list = []
sample_rate_list = []
track_count_list = []
interlace_list = []
compression_list = []
pix_fmt_list = []
audio_fmt_list = []
audio_codecid_list = []
audio_codec_list = []
au_bitdepth_list = []
video_codecid_list = []
video_codec_version_list = []
video_codec_profile_list = []
timecode_list = []
channels_list = []
for source in all_files:
metadata = subprocess.check_output(
['mediainfo', '--Output=PBCore2', source])
root = etree.fromstring(metadata)
print(' - Analysing %s') % source
pbcore_namespace = root.xpath('namespace-uri(.)')
track_type = root.xpath('//ns:essenceTrackType',
namespaces={'ns': pbcore_namespace})
if len(track_type) > 0:
for track in track_type:
if track.text == 'Video':
essenceTrackEncodvid = get_metadata(
"ns:essenceTrackEncoding", track.getparent(),
pbcore_namespace)
vcodec_attributes = get_attributes(track.getparent(),
pbcore_namespace)
elif track.text == 'Audio':
silence = False
essenceTrackEncod_au = get_metadata(
"//ns:essenceTrackEncoding", track.getparent(),
pbcore_namespace)
audio_codec_list.append(essenceTrackEncod_au)
acodec_attributes = get_attributes(track.getparent(),
pbcore_namespace)
try:
audio_codecid = acodec_attributes['ref']
except KeyError:
audio_codecid = 'n/a'
essenceTrackSampling = ififuncs.get_mediainfo(
'samplerate', '--inform=Audio;%SamplingRate_String%',
source)
sample_rate_list.append(essenceTrackSampling)
essenceBitDepth_au = get_metadata(
"//ns:essenceTrackBitDepth", root, pbcore_namespace)
audio_codecid_list.append(audio_codecid)
au_bitdepth_list.append(essenceBitDepth_au)
channels = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='Channel(s)']",
track.getparent(), pbcore_namespace)
channels_list.append(channels)
ScanType = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='ScanType']", root,
pbcore_namespace)
scan_types.append(ScanType)
matrix_coefficients = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='matrix_coefficients']",
root, pbcore_namespace)
timecode_source, starting_timecode = get_timecode(
pbcore_namespace, root, source)
timecode_list.append(starting_timecode)
matrix_list.append(matrix_coefficients)
transfer_characteris = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='transfer_characteristics']",
root, pbcore_namespace)
transfer_list.append(transfer_characteris)
colour_primaries = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='color_primaries']",
root, pbcore_namespace)
colour_primaries_list.append(colour_primaries)
FrameCount += int(
get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='FrameCount']",
root, pbcore_namespace))
instantFileSize_byte += int(
get_metadata("//ns:instantiationFileSize", root, pbcore_namespace))
instantDataRate = round(
float(
ififuncs.get_mediainfo('OverallBitRate',
'--inform=General;%OverallBitRate%',
source)) / 1000 / 1000, 2)
instantTracks = ififuncs.get_number_of_tracks(source)
track_count_list.append(instantTracks)
ms += ififuncs.get_milliseconds(source)
ColorSpace = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='ColorSpace']", root,
pbcore_namespace)
color_spaces.append(ColorSpace)
ChromaSubsampling = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='ChromaSubsampling']",
root, pbcore_namespace)
chroma.append(ChromaSubsampling)
instantMediaty = get_metadata("//ns:instantiationMediaType", root,
pbcore_namespace)
essenceFrameSize = get_metadata("//ns:essenceTrackFrameSize", root,
pbcore_namespace)
frame_sizes.append(essenceFrameSize)
PixelAspectRatio = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='PixelAspectRatio']",
root, pbcore_namespace)
par_list.append(PixelAspectRatio)
instantiationStandar = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Format']", root,
pbcore_namespace)
container_list.append(instantiationStandar)
essenceFrameRate = get_metadata("//ns:essenceTrackFrameRate", root,
pbcore_namespace)
fps_list.append(essenceFrameRate)
essenceAspectRatio = ififuncs.get_mediainfo(
'DAR', '--inform=Video;%DisplayAspectRatio_String%', source)
Interlacement = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Interlacement']",
root, pbcore_namespace)
interlace_list.append(Interlacement)
Compression_Mode = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Compression_Mode']",
root, pbcore_namespace)
colour_range = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='colour_range']",
root, pbcore_namespace)
format_version = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Format_Version']",
root, pbcore_namespace)
app_company_name = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Encoded_Application_CompanyName']",
root, pbcore_namespace)
app_name = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Encoded_Application_Name']",
root, pbcore_namespace)
app_version = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Encoded_Application_Version']",
root, pbcore_namespace)
library_name = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Encoded_Library_Name']",
root, pbcore_namespace)
library_version = get_metadata(
"//ns:instantiationAnnotation[@annotationType='Encoded_Library_Version']",
root, pbcore_namespace)
compression_list.append(Compression_Mode)
instantiationDate_mo = get_metadata(
"//ns:instantiationDate[@dateType='file modification']", root,
pbcore_namespace)
instantDate_other = 'n/a'
instantDate_type = 'n/a'
pix_fmt = ififuncs.get_ffmpeg_fmt(source, 'video')
pix_fmt_list.append(pix_fmt)
audio_fmt = ififuncs.get_ffmpeg_fmt(source, 'audio')
audio_fmt_list.append(audio_fmt)
if silence:
audio_codecid = 'n/a'
essenceBitDepth_au = 'n/a'
essenceTrackEncod_au = 'n/a'
essenceTrackSampling = 'n/a'
video_codecid = vcodec_attributes['ref']
video_codecid_list.append(video_codecid)
try:
video_codec_version = vcodec_attributes['version']
except KeyError:
video_codec_version = 'n/a'
try:
video_codec_profile = vcodec_attributes['annotation'][8:]
except KeyError:
video_codec_profile = 'n/a'
video_codec_version_list.append(video_codec_version)
video_codec_profile_list.append(video_codec_profile)
metadata_error = ''
metadata_list = [
scan_types, matrix_list, transfer_list, colour_primaries_list,
color_spaces, chroma, frame_sizes, par_list, container_list, fps_list,
sample_rate_list, track_count_list, interlace_list, compression_list,
pix_fmt_list, audio_fmt_list, audio_codecid_list, audio_codec_list,
au_bitdepth_list, video_codecid_list, video_codec_version_list,
video_codec_profile_list, channels_list, timecode_list
]
for i in metadata_list:
if len(set(i)) > 1:
metadata_error += 'WARNING - Your metadata values are not the same for all files: %s\n' % set(
i)
print metadata_error
if args.p:
ififuncs.generate_log(
sipcreator_log,
'EVENT = Metadata mismatch - Your metadata values are not the same for all files: %s'
% set(i))
tc = ififuncs.convert_millis(ms)
instantiationDuratio = ififuncs.convert_timecode(25, tc)
if args.donor:
Donor = args.donor
else:
Donor = ''
Edited_By = user
Date_Created = ''
Date_Last_Modified = ''
Film_Or_Tape = 'Digital AV Object'
Date_Of_Donation = ''
reproduction_creator = ''
if args.acquisition_type:
if acquisition_type == 'Reproduction':
Date_Of_Donation = instantiationDate_mo.split('T')[0]
# if a reproduction, then there's no Donor/transfer of title.
Donor = 'n/a'
if ififuncs.find_concat_user(
sipcreator_log) == 'Aoife Fitzmaurice':
reproduction_creator = 'Aoife Fitzmaurice (2016)'
elif ififuncs.find_concat_user(
sipcreator_log) == 'Kieran O\'Leary':
reproduction_creator = 'Kieran O\'Leary (2013)'
Habitat = ''
backup_habitat = ''
Type_Of_Deposit = acquisition_type
if args.depositor_reference:
Depositor_Reference = args.depositor_reference
else:
Depositor_Reference = ''
Master_Viewing = 'Preservation Object'
Language_Version = ''
Condition_Rating = ''
Companion_Elements = ''
TTape_Origin = args.parent
EditedNew = user
FIO = 'In'
CollectionTitle = ''
Created_By = user
instantTimeStart = 'n/a'
instantFileSize_gigs = round(
float(instantFileSize_byte) / 1024 / 1024 / 1024, 3)
instantColors = 'n/a'
instantLanguage = 'n/a'
instantAltMo = 'n/a'
essenceBitDepth_vid = ififuncs.get_mediainfo('duration',
'--inform=Video;%BitDepth%',
source)
instantiationChanCon = 'n/a'
colour_range = colour_range
format_version = format_version
TimeCode_FirstFrame = process_mixed_values(timecode_list)
TimeCode_Source = timecode_source
reproduction_reason = ''
dig_object_descrip = ififuncs.get_digital_object_descriptor(args.input)
ififuncs.append_csv(csv_filename, [
Reference_Number,
Donor,
Edited_By,
Date_Created,
Date_Last_Modified,
Film_Or_Tape,
Date_Of_Donation,
Accession_Number,
Habitat,
backup_habitat,
TTape_Origin,
Type_Of_Deposit,
Depositor_Reference,
Master_Viewing,
Language_Version,
Condition_Rating,
Companion_Elements,
EditedNew,
FIO,
CollectionTitle,
Created_By,
instantiationIdentif,
instantDate_other,
instantDate_type,
instantiationDate_mo,
instantiationStandar,
instantMediaty,
instantFileSize_byte,
instantFileSize_gigs,
instantTimeStart,
instantDataRate,
instantTracks,
instantColors,
instantLanguage,
instantAltMo,
essenceTrackEncodvid,
essenceFrameRate,
essenceTrackSampling,
essenceBitDepth_vid,
essenceFrameSize,
essenceAspectRatio,
essenceTrackEncod_au,
essenceBitDepth_au,
instantiationDuratio,
instantiationChanCon,
PixelAspectRatio,
FrameCount,
ColorSpace,
ChromaSubsampling,
ScanType,
Interlacement,
Compression_Mode,
colour_primaries,
transfer_characteris,
matrix_coefficients,
pix_fmt,
audio_fmt,
audio_codecid,
video_codecid,
video_codec_version,
video_codec_profile,
channels,
colour_range,
format_version,
TimeCode_FirstFrame,
TimeCode_Source,
app_company_name,
app_name,
app_version,
library_name,
library_version,
reproduction_creator,
reproduction_reason,
dig_object_descrip,
])
if args.p:
ififuncs.generate_log(
sipcreator_log,
'EVENT = Metadata extraction - eventDetail=Technical record creation using PBCore, eventOutcome=%s, agentName=makepbcore'
% (csv_filename))
ififuncs.generate_log(sipcreator_log, 'EVENT = makepbcore.py finished')
ififuncs.checksum_replace(md5_manifest, sipcreator_log, 'md5')
ififuncs.checksum_replace(sha512_manifest, sipcreator_log, 'sha512')
ififuncs.manifest_update(md5_manifest, csv_filename)
print ' - Updating %s with %s' % (md5_manifest, csv_filename)
ififuncs.sha512_update(sha512_manifest, csv_filename)
print ' - Updating %s with %s' % (sha512_manifest, csv_filename)
print metadata_error
def main(args_):
# if multiple file are present, this script will treat them as a single
# instantiation/representation and get aggregate metadata about the whole
# package. For now, this will be a clumsy implementation - the first file
# will provide most metadata. Things like duration/bitrate/filesize
# will be calculated as a whole.
# Although another way would be that every call is looped, and if
# this could catch files that should not be in the package, eg. a 4:2:2
# file in a 4:2:0 package..
# yup - do it that way!
args = parse_args(args_)
all_files = ififuncs.recursive_file_list(args.input)
silence = True
audio_only = True
if args.user:
user = args.user
else:
user = ififuncs.get_user()
acquisition_type = ''
if args.acquisition_type:
acquisition_type = ififuncs.get_acquisition_type(
args.acquisition_type)[0]
instantiationIdentif = ''
for dirs in os.listdir(args.input):
if ififuncs.validate_uuid4(dirs) is None:
instantiationIdentif = dirs
Accession_Number = get_accession_number(args.input)
if args.reference:
Reference_Number = args.reference.upper()
else:
Reference_Number = get_reference_number(args.input)
if args.p:
for root, _, filenames in os.walk(args.input):
if os.path.basename(root) == 'metadata':
metadata_dir = root
elif os.path.basename(root) == 'logs':
logs_dir = root
csv_filename = os.path.join(
metadata_dir,
Accession_Number + '_%s_pbcore.csv' % Reference_Number)
sipcreator_log = os.path.join(logs_dir,
instantiationIdentif + '_sip_log.log')
ififuncs.generate_log(sipcreator_log, 'EVENT = makepbcore.py started')
ififuncs.generate_log(
sipcreator_log, 'eventDetail=makepbcore.py %s' %
ififuncs.get_script_version('makepbcore.py'))
ififuncs.generate_log(sipcreator_log,
'Command line arguments: %s' % args)
ififuncs.generate_log(sipcreator_log, 'EVENT = agentName=%s' % user)
else:
csv_filename = 'blaa.csv'
print((' - Metadata will be stored in %s' % csv_filename))
for filenames in os.listdir(args.input):
if '_manifest.md5' in filenames:
md5_manifest = os.path.join(args.input, filenames)
elif 'manifest-sha512.txt' in filenames:
sha512_manifest = os.path.join(args.input, filenames)
make_csv(csv_filename)
ms = 0
FrameCount = 0
instantFileSize_byte = 0
instantFileSize_gigs = 0
scan_types = []
matrix_list = []
transfer_list = []
colour_primaries_list = []
color_spaces = []
chroma = []
frame_sizes = []
par_list = []
container_list = []
fps_list = []
sample_rate_list = []
track_count_list = []
interlace_list = []
compression_list = []
pix_fmt_list = []
audio_fmt_list = []
audio_codecid_list = []
audio_codec_list = []
au_bitdepth_list = []
video_codecid_list = []
video_codec_version_list = []
video_codec_profile_list = []
timecode_list = []
channels_list = []
stl = False
subtitle_check = ififuncs.get_digital_object_descriptor(args.input)
if 'STL' in subtitle_check:
stl = True
for source in all_files:
metadata = subprocess.check_output(
['mediainfo', '--Output=PBCore2', source])
new_metadata = subprocess.check_output(
['mediainfo', '--Output=XML', source])
try:
root = etree.fromstring(metadata)
new_root = etree.fromstring(new_metadata)
except lxml.etree.XMLSyntaxError:
print('Windows encoding detected - transforming into utf-8')
root = etree.fromstring(metadata.decode('cp1252').encode('utf-8'))
new_root = etree.fromstring(
new_metadata.decode('cp1252').encode('utf-8'))
print(((' - Analysing %s') % source))
pbcore_namespace = root.xpath('namespace-uri(.)')
mediainfo_namespace = new_root.xpath('namespace-uri(.)')
track_type = root.xpath('//ns:essenceTrackType',
namespaces={'ns': pbcore_namespace})
new_track_type = new_root.xpath('//ns:track',
namespaces={'ns': mediainfo_namespace})
if len(new_track_type) > 0:
for track in new_track_type:
if track.attrib['type'] == 'Video':
audio_only = False
essenceTrackEncodvid = ififuncs.get_metadata(
"ns:Format", track, mediainfo_namespace)
#vcodec_attributes = get_attributes(track.getparent(), pbcore_namespace)
#vcodec_attributes = 'TODO'
video_codecid = ififuncs.get_metadata(
"ns:CodecID", track, mediainfo_namespace)
video_codec_version = ififuncs.get_metadata(
"ns:Format_Version", track, mediainfo_namespace)
video_codec_profile = ififuncs.get_metadata(
"ns:Format_Profile", track, mediainfo_namespace)
video_codec_version_list.append(video_codec_version)
video_codec_profile_list.append(video_codec_profile)
elif track.attrib['type'] == 'Audio':
silence = False
essenceTrackEncod_au = ififuncs.get_metadata(
"ns:Format", track, mediainfo_namespace)
audio_codec_list.append(essenceTrackEncod_au)
#acodec_attributes = get_attributes(track.getparent(), pbcore_namespace)
audio_codecid = ififuncs.get_metadata(
"ns:CodecID", track, mediainfo_namespace)
essenceTrackSampling = ififuncs.get_mediainfo(
'samplerate', '--inform=Audio;%SamplingRate_String%',
source)
sample_rate_list.append(essenceTrackSampling)
essenceBitDepth_au = ififuncs.get_metadata(
"ns:BitDepth", track, mediainfo_namespace)
audio_codecid_list.append(audio_codecid)
au_bitdepth_list.append(essenceBitDepth_au)
channels = ififuncs.get_metadata("//ns:Channels", track,
mediainfo_namespace)
channels_list.append(channels)
if audio_only:
essenceTrackEncodvid = 'n/a'
video_codecid = 'n/a'
video_codec_version = 'n/a'
video_codec_profile = 'n/a'
ScanType = ififuncs.get_metadata("//ns:ScanType", new_root,
mediainfo_namespace)
scan_types.append(ScanType)
matrix_coefficients = ififuncs.get_metadata("//ns:matrix_coefficients",
new_root,
mediainfo_namespace)
timecode_source, starting_timecode = get_timecode(
pbcore_namespace, root, source)
timecode_list.append(starting_timecode)
matrix_list.append(matrix_coefficients)
transfer_characteris = ififuncs.get_metadata(
"//ns:transfer_characteristics", new_root, mediainfo_namespace)
transfer_list.append(transfer_characteris)
colour_primaries = ififuncs.get_metadata("//ns:colour_primaries",
new_root, mediainfo_namespace)
colour_primaries_list.append(colour_primaries)
try:
if audio_only:
FrameCount = 'n/a'
else:
# increment if multiple objects are present
try:
FrameCount += int(
ififuncs.get_metadata("//ns:FrameCount", new_root,
mediainfo_namespace))
except ValueError:
# don't increment if multiple values are returned as str
FrameCount = ififuncs.get_metadata("//ns:FrameCount",
new_root,
mediainfo_namespace)
except TypeError:
# workaround for silent pic in DCP
FrameCount = 'n/a'
instantFileSize_byte += int(
ififuncs.get_metadata("//ns:FileSize", new_root,
mediainfo_namespace))
instantDataRate = round(
float(
ififuncs.get_mediainfo('OverallBitRate',
'--inform=General;%OverallBitRate%',
source)) / 1000 / 1000, 2)
instantTracks = ififuncs.get_number_of_tracks(source)
track_count_list.append(instantTracks)
if stl is True:
track_count_list.append('STL sidecar')
ms += ififuncs.get_milliseconds(source)
ColorSpace = ififuncs.get_metadata("//ns:ColorSpace", new_root,
mediainfo_namespace)
color_spaces.append(ColorSpace)
ChromaSubsampling = get_metadata("//ns:ChromaSubsampling", new_root,
mediainfo_namespace)
chroma.append(ChromaSubsampling)
instantMediaty = get_metadata("//ns:instantiationMediaType", root,
pbcore_namespace)
if audio_only:
essenceFrameSize = 'n/a'
else:
essenceFrameSize = get_metadata("//ns:essenceTrackFrameSize", root,
pbcore_namespace)
frame_sizes.append(essenceFrameSize)
PixelAspectRatio = ififuncs.get_metadata("//ns:PixelAspectRatio",
new_root, mediainfo_namespace)
par_list.append(PixelAspectRatio)
general_root = new_root.xpath("//ns:track[@type='General']",
namespaces={'ns':
mediainfo_namespace})[0]
instantiationStandar = ififuncs.get_metadata("ns:Format", general_root,
mediainfo_namespace)
container_list.append(instantiationStandar)
essenceFrameRate = ififuncs.get_metadata("//ns:FrameRate", new_root,
mediainfo_namespace)
fps_list.append(essenceFrameRate)
essenceAspectRatio = ififuncs.get_mediainfo(
'DAR', '--inform=Video;%DisplayAspectRatio_String%', source)
Interlacement = ififuncs.get_metadata("//ns:ScanOrder", new_root,
mediainfo_namespace)
# FFV1/MKV seems to have this scanorder metadata here rather than Interlacement
# FFV1/MKV is the only example I've seen so far that behaves like this :|
# It could be that Interlacement is set at a codec level for FFV1, but others are
# declared at the container level..
if Interlacement == 'n/a':
Interlacement = get_metadata(
"//ns:essenceTrackAnnotation[@annotationType='ScanOrder']",
root, pbcore_namespace)
interlace_list.append(Interlacement)
Compression_Mode = ififuncs.get_metadata("//ns:Compression_Mode",
new_root, mediainfo_namespace)
colour_range = ififuncs.get_metadata("//ns:colour_range", new_root,
mediainfo_namespace)
# this needs to be clarified as it exists in general and codec
format_version = ififuncs.get_metadata("ns:Format_Version",
general_root,
mediainfo_namespace)
app_company_name = ififuncs.get_metadata(
"//ns:Encoded_Application_CompanyName", new_root,
mediainfo_namespace)
app_name = ififuncs.get_metadata("//ns:Encoded_Application_Name",
new_root, mediainfo_namespace)
app_version = ififuncs.get_metadata("//ns:Encoded_Application_Version",
new_root, mediainfo_namespace)
library_name = ififuncs.get_metadata("//ns:Encoded_Library_Name",
new_root, mediainfo_namespace)
if library_name == 'n/a':
library_name = ififuncs.get_metadata("//ns:Encoded_Library",
general_root,
mediainfo_namespace)
library_version = ififuncs.get_metadata("//ns:Encoded_Library_Version",
new_root, mediainfo_namespace)
compression_list.append(Compression_Mode)
instantiationDate_mo = get_metadata(
"//ns:instantiationDate[@dateType='file modification']", root,
pbcore_namespace)
instantDate_other = 'n/a'
instantDate_type = 'n/a'
pix_fmt = ififuncs.get_ffmpeg_fmt(source, 'video')
pix_fmt_list.append(pix_fmt)
audio_fmt = ififuncs.get_ffmpeg_fmt(source, 'audio')
audio_fmt_list.append(audio_fmt)
essenceBitDepth_vid = ififuncs.get_mediainfo(
'duration', '--inform=Video;%BitDepth%', source)
if silence:
audio_codecid = 'n/a'
essenceBitDepth_au = 'n/a'
essenceTrackEncod_au = 'n/a'
essenceTrackSampling = 'n/a'
channels = 'n/a'
'''
video_codecid = vcodec_attributes['ref']
video_codecid_list.append(video_codecid)
try:
video_codec_version = vcodec_attributes['version']
except KeyError:
video_codec_version = 'n/a'
try:
video_codec_profile = vcodec_attributes['annotation'][8:]
except KeyError:
video_codec_profile = 'n/a'
'''
metadata_error = ''
metadata_list = [
scan_types, matrix_list, transfer_list, colour_primaries_list,
color_spaces, chroma, frame_sizes, par_list, container_list, fps_list,
sample_rate_list, track_count_list, interlace_list, compression_list,
pix_fmt_list, audio_fmt_list, audio_codecid_list, audio_codec_list,
au_bitdepth_list, video_codecid_list, video_codec_version_list,
video_codec_profile_list, channels_list, timecode_list
]
for i in metadata_list:
if len(set(i)) > 1:
metadata_error += 'WARNING - Your metadata values are not the same for all files - but this could be a false positive if dealing with atomised audio and video as with DCP: %s\n' % set(
i)
if args.p:
ififuncs.generate_log(
sipcreator_log,
'EVENT = Metadata mismatch - Your metadata values are not the same for all files - but this could be a false positive if dealing with atomised audio and video as with DCP: %s'
% set(i))
print(metadata_error)
tc = ififuncs.convert_millis(ms)
instantiationDuratio = ififuncs.convert_timecode(25, tc)
if args.donor:
Donor = args.donor
else:
Donor = ''
Edited_By = user
Date_Created = ''
Date_Last_Modified = ''
Film_Or_Tape = 'Digital AV Object'
Date_Of_Donation = ''
if args.reproduction_creator:
reproduction_creator = args.reproduction_creator
else:
reproduction_creator = ''
if args.acquisition_type:
if acquisition_type == 'Reproduction':
Date_Of_Donation = instantiationDate_mo.split('T')[0]
# if a reproduction, then there's no Donor/transfer of title.
Donor = 'n/a'
else:
Date_Of_Donation = args.donation_date
Habitat = ''
backup_habitat = ''
Type_Of_Deposit = acquisition_type
if args.depositor_reference:
Depositor_Reference = args.depositor_reference
else:
Depositor_Reference = ''
Master_Viewing = 'Preservation Object'
Language_Version = ''
Condition_Rating = ''
Companion_Elements = ''
TTape_Origin = args.parent
EditedNew = user
FIO = 'In'
CollectionTitle = ''
Created_By = user
instantTimeStart = 'n/a'
instantFileSize_gigs = round(
float(instantFileSize_byte) / 1024 / 1024 / 1024, 3)
instantColors = 'n/a'
instantLanguage = 'n/a'
instantAltMo = 'n/a'
instantiationChanCon = 'n/a'
'''
no idea why these are here
colour_range = colour_range
format_version = format_version
'''
TimeCode_FirstFrame = process_mixed_values(timecode_list)
pix_fmt = process_mixed_values(pix_fmt_list)
audio_fmt = process_mixed_values(audio_fmt_list)
instantTracks = process_mixed_values(track_count_list)
TimeCode_Source = timecode_source
reproduction_reason = ''
dig_object_descrip = ififuncs.get_digital_object_descriptor(args.input)
if 'STL' in dig_object_descrip:
dig_object_descrip = 'AS-11 package'
dcp_check = ififuncs.find_cpl(args.input)
if dcp_check is not None:
essenceFrameSize, ChromaSubsampling, ColorSpace, FrameCount, essenceAspectRatio, instantiationDuratio, PixelAspectRatio, ScanType, dig_object_descrip, instantTracks, instantDataRate, essenceBitDepth_vid, instantMediaty = check_dcp(
dcp_check)
ififuncs.append_csv(csv_filename, [
Reference_Number,
Donor,
Edited_By,
Date_Created,
Date_Last_Modified,
Film_Or_Tape,
Date_Of_Donation,
Accession_Number,
Habitat,
backup_habitat,
TTape_Origin,
Type_Of_Deposit,
Depositor_Reference,
Master_Viewing,
Language_Version,
Condition_Rating,
Companion_Elements,
EditedNew,
FIO,
CollectionTitle,
Created_By,
instantiationIdentif,
instantDate_other,
instantDate_type,
instantiationDate_mo,
instantiationStandar,
instantMediaty,
instantFileSize_byte,
instantFileSize_gigs,
instantTimeStart,
instantDataRate,
instantTracks,
instantColors,
instantLanguage,
instantAltMo,
essenceTrackEncodvid,
essenceFrameRate,
essenceTrackSampling,
essenceBitDepth_vid,
essenceFrameSize,
essenceAspectRatio,
essenceTrackEncod_au,
essenceBitDepth_au,
instantiationDuratio,
instantiationChanCon,
PixelAspectRatio,
FrameCount,
ColorSpace,
ChromaSubsampling,
ScanType,
Interlacement,
Compression_Mode,
colour_primaries,
transfer_characteris,
matrix_coefficients,
pix_fmt,
audio_fmt,
audio_codecid,
video_codecid,
video_codec_version,
video_codec_profile,
channels,
colour_range,
format_version,
TimeCode_FirstFrame,
TimeCode_Source,
app_company_name,
app_name,
app_version,
library_name,
library_version,
reproduction_creator,
reproduction_reason,
dig_object_descrip,
])
if args.p:
ififuncs.generate_log(
sipcreator_log,
'EVENT = Metadata extraction - eventDetail=Technical record creation using PBCore, eventOutcome=%s, agentName=makepbcore'
% (csv_filename))
ififuncs.generate_log(sipcreator_log, 'EVENT = makepbcore.py finished')
ififuncs.checksum_replace(md5_manifest, sipcreator_log, 'md5')
ififuncs.checksum_replace(sha512_manifest, sipcreator_log, 'sha512')
ififuncs.manifest_update(md5_manifest, csv_filename)
print((' - Updating %s with %s' % (md5_manifest, csv_filename)))
ififuncs.sha512_update(sha512_manifest, csv_filename)
print((' - Updating %s with %s' % (sha512_manifest, csv_filename)))
print(metadata_error)