def main(job, file_path, file_uuid, sip_uuid, shared_path, file_type):
"""Entry point for policy checker."""
setup_dicts(mcpclient_settings)
policy_checker = PolicyChecker(job, file_path, file_uuid, sip_uuid,
shared_path, file_type)
return policy_checker.check()
def main(task_uuid, file_uuid):
setup_dicts(mcpclient_settings)
succeeded = True
file_ = File.objects.get(uuid=file_uuid)
# Normally we don't transcribe derivatives (access copies, preservation copies);
# however, some useful transcription tools can't handle some formats that
# are common as the primary copies. For example, tesseract can't handle JPEG2000.
# If there are no rules for the primary format passed in, try to look at each
# derivative until a transcribable derivative is found.
#
# Skip derivatives to avoid double-scanning them; only look at them as a fallback.
if file_.filegrpuse != "original":
print('{} is not an original; not transcribing'.format(file_uuid), file=sys.stderr)
return 0
rules = fetch_rules_for(file_)
if not rules:
file_, rules = fetch_rules_for_derivatives(file_)
if not rules:
print('No rules found for file {} and its derivatives; not transcribing'.format(file_uuid), file=sys.stderr)
return 0
else:
if file_.filegrpuse == "original":
noun = "original"
else:
noun = file_.filegrpuse + " derivative"
print('Transcribing {} {}'.format(noun, file_.uuid), file=sys.stderr)
rd = ReplacementDict.frommodel(file_=file_, type_='file')
for rule in rules:
script = rule.command.command
if rule.command.script_type in ('bashScript', 'command'):
script, = rd.replace(script)
args = []
else:
args = rd.to_gnu_options
exitstatus, stdout, stderr = executeOrRun(rule.command.script_type,
script, arguments=args)
if exitstatus != 0:
succeeded = False
output_path = rd.replace(rule.command.output_location)[0]
relative_path = output_path.replace(rd['%SIPDirectory%'], '%SIPDirectory%')
event = insert_transcription_event(exitstatus, file_uuid, rule, relative_path)
if os.path.isfile(output_path):
insert_file_into_database(file_uuid, rd['%SIPUUID%'], event, rule, output_path, relative_path)
return 0 if succeeded else 1
def main(file_path, file_uuid, sip_uuid, shared_path, file_type):
setup_dicts(mcpclient_settings)
policy_checker = PolicyChecker(
file_path, file_uuid, sip_uuid, shared_path, file_type)
return policy_checker.check()
def main(job, file_path, file_uuid, sip_uuid):
setup_dicts(mcpclient_settings)
failed = False
# Check to see whether the file has already been characterized; don't try
# to characterize it a second time if so.
if FPCommandOutput.objects.filter(file_id=file_uuid).count() > 0:
return 0
try:
format = FormatVersion.active.get(
fileformatversion__file_uuid=file_uuid)
except FormatVersion.DoesNotExist:
rules = format = None
if format:
rules = FPRule.active.filter(format=format.uuid,
purpose="characterization")
# Characterization always occurs - if nothing is specified, get one or more
# defaults specified in the FPR.
if not rules:
rules = FPRule.active.filter(purpose="default_characterization")
for rule in rules:
if (rule.command.script_type == "bashScript"
or rule.command.script_type == "command"):
args = []
command_to_execute = replace_string_values(rule.command.command,
file_=file_uuid,
sip=sip_uuid,
type_="file")
else:
rd = ReplacementDict.frommodel(file_=file_uuid,
sip=sip_uuid,
type_="file")
args = rd.to_gnu_options()
command_to_execute = rule.command.command
exitstatus, stdout, stderr = executeOrRun(
rule.command.script_type,
command_to_execute,
arguments=args,
capture_output=True,
)
job.write_output(stdout)
job.write_error(stderr)
if exitstatus != 0:
job.write_error(
"Command {} failed with exit status {}; stderr:".format(
rule.command.description, exitstatus))
failed = True
continue
# fmt/101 is XML - we want to collect and package any XML output, while
# allowing other commands to execute without actually collecting their
# output in the event that they are writing their output to disk.
# FPCommandOutput can have multiple rows for a given file,
# distinguished by the rule that produced it.
if (rule.command.output_format
and rule.command.output_format.pronom_id == "fmt/101"):
try:
etree.fromstring(stdout)
insertIntoFPCommandOutput(file_uuid, stdout, rule.uuid)
job.write_output(
'Saved XML output for command "{}" ({})'.format(
rule.command.description, rule.command.uuid))
except etree.XMLSyntaxError:
failed = True
job.write_error(
'XML output for command "{}" ({}) was not valid XML; not saving to database'
.format(rule.command.description, rule.command.uuid))
else:
job.write_error(
'Tool output for command "{}" ({}) is not XML; not saving to database'
.format(rule.command.description, rule.command.uuid))
if failed:
return 255
else:
return 0
def main(job, file_path, file_uuid, sip_uuid, shared_path, file_type):
setup_dicts(mcpclient_settings)
validator = Validator(job, file_path, file_uuid, sip_uuid, shared_path, file_type)
return validator.validate()
def main(opts):
""" Find and execute normalization commands on input file. """
# TODO fix for maildir working only on attachments
setup_dicts(mcpclient_settings)
# Find the file and it's FormatVersion (file identification)
try:
file_ = File.objects.get(uuid=opts.file_uuid)
except File.DoesNotExist:
print('File with uuid',
opts.file_uuid,
'does not exist in database.',
file=sys.stderr)
return NO_RULE_FOUND
print('File found:', file_.uuid, file_.currentlocation)
# Unless normalization file group use is submissionDocumentation, skip the
# submissionDocumentation directory
if opts.normalize_file_grp_use != "submissionDocumentation" and file_.currentlocation.startswith(
'%SIPDirectory%objects/submissionDocumentation'):
print('File', os.path.basename(opts.file_path),
'in objects/submissionDocumentation, skipping')
return SUCCESS
# Only normalize files where the file's group use and normalize group use match
if file_.filegrpuse != opts.normalize_file_grp_use:
print(os.path.basename(opts.file_path), 'is file group usage',
file_.filegrpuse, 'instead of ', opts.normalize_file_grp_use,
' - skipping')
return SUCCESS
# For re-ingest: clean up old derivations
# If the file already has a Derivation with the same purpose, remove it and mark the derived file as deleted
derivatives = Derivation.objects.filter(
source_file=file_, derived_file__filegrpuse=opts.purpose)
for derivative in derivatives:
print(opts.purpose, 'derivative', derivative.derived_file_id,
'already exists, marking as deleted')
File.objects.filter(uuid=derivative.derived_file_id).update(
filegrpuse='deleted')
# Don't create events for thumbnail files
if opts.purpose != 'thumbnail':
databaseFunctions.insertIntoEvents(
fileUUID=derivative.derived_file_id,
eventType='deletion',
)
derivatives.delete()
# If a file has been manually normalized for this purpose, skip it
manually_normalized_file = check_manual_normalization(opts)
if manually_normalized_file:
print(os.path.basename(opts.file_path),
'was already manually normalized into',
manually_normalized_file.currentlocation)
if 'preservation' in opts.purpose:
# Add derivation link and associated event
insert_derivation_event(
original_uuid=opts.file_uuid,
output_uuid=manually_normalized_file.uuid,
derivation_uuid=str(uuid.uuid4()),
event_detail_output="manual normalization",
outcome_detail_note=None,
)
return SUCCESS
do_fallback = False
format_id = get_object_or_None(FileFormatVersion, file_uuid=opts.file_uuid)
# Look up the normalization command in the FPR
if format_id:
print('File format:', format_id.format_version)
try:
rule = FPRule.active.get(format=format_id.format_version,
purpose=opts.purpose)
except FPRule.DoesNotExist:
do_fallback = True
# Try with default rule if no format_id or rule was found
if format_id is None or do_fallback:
try:
rule = get_default_rule(opts.purpose)
print(os.path.basename(file_.currentlocation),
"not identified or without rule",
"- Falling back to default", opts.purpose, "rule")
except FPRule.DoesNotExist:
print('Not normalizing', os.path.basename(file_.currentlocation),
' - No rule or default rule found to normalize for',
opts.purpose)
return NO_RULE_FOUND
print('Format Policy Rule:', rule)
command = rule.command
print('Format Policy Command', command.description)
replacement_dict = get_replacement_dict(opts)
cl = transcoder.CommandLinker(rule, command, replacement_dict, opts,
once_normalized)
exitstatus = cl.execute()
# If the access/thumbnail normalization command has errored AND a
# derivative was NOT created, then we run the default access/thumbnail
# rule. Note that we DO need to check if the derivative file exists. Even
# when a verification command exists for the normalization command, the
# transcoder.py::Command.execute method will only run the verification
# command if the normalization command returns a 0 exit code.
# Errored thumbnail normalization also needs to result in default thumbnail
# normalization; if not, then a transfer with a single file that failed
# thumbnail normalization will result in a failed SIP at "Prepare DIP: Copy
# thumbnails to DIP directory"
if (exitstatus != 0 and opts.purpose in ('access', 'thumbnail')
and cl.commandObject.output_location
and (not os.path.isfile(cl.commandObject.output_location))):
# Fall back to default rule
try:
fallback_rule = get_default_rule(opts.purpose)
print(opts.purpose,
'normalization failed, falling back to default',
opts.purpose, 'rule')
except FPRule.DoesNotExist:
print('Not retrying normalizing for',
os.path.basename(file_.currentlocation),
' - No default rule found to normalize for', opts.purpose)
fallback_rule = None
# Don't re-run the same command
if fallback_rule and fallback_rule.command != command:
print('Fallback Format Policy Rule:', fallback_rule)
command = fallback_rule.command
print('Fallback Format Policy Command', command.description)
# Use existing replacement dict
cl = transcoder.CommandLinker(fallback_rule, command,
replacement_dict, opts,
once_normalized)
exitstatus = cl.execute()
# Store thumbnails locally for use during AIP searches
# TODO is this still needed, with the storage service?
if 'thumbnail' in opts.purpose:
thumbnail_filepath = cl.commandObject.output_location
thumbnail_storage_dir = os.path.join(
mcpclient_settings.SHARED_DIRECTORY,
'www',
'thumbnails',
opts.sip_uuid,
)
try:
os.makedirs(thumbnail_storage_dir)
except OSError as e:
if e.errno == errno.EEXIST and os.path.isdir(
thumbnail_storage_dir):
pass
else:
raise
thumbnail_basename, thumbnail_extension = os.path.splitext(
thumbnail_filepath)
thumbnail_storage_file = os.path.join(
thumbnail_storage_dir,
opts.file_uuid + thumbnail_extension,
)
shutil.copyfile(thumbnail_filepath, thumbnail_storage_file)
if not exitstatus == 0:
print('Command', command.description, 'failed!', file=sys.stderr)
return RULE_FAILED
else:
print('Successfully normalized ', os.path.basename(opts.file_path),
'for', opts.purpose)
return SUCCESS
