def main(target, xml_file, date, event_uuid, file_uuid, file_grpuse):
"""
Note: xml_file, date and event_uuid are not being used.
"""
if file_grpuse in ("DSPACEMETS", "maildirFile"):
logger.error("File's fileGrpUse in exclusion list, skipping")
return 0
if not FPCommandOutput.objects.filter(file=file_uuid).exists():
logger.error("Warning: Fits has already run on this file. Not running again.")
return 0
_, temp_file = tempfile.mkstemp()
args = ["fits.sh", "-i", target, "-o", temp_file]
try:
logger.info("Executing %s", args)
retcode, stdout, stderr = executeOrRun(
"command", args, printing=False, capture_output=True
)
if retcode != 0:
logger.error(
"fits.sh exited with status code %s, %s, %s", retcode, stdout, stderr
)
return retcode
try:
tree = etree.parse(temp_file)
except:
logger.exception("Failed to read Fits's XML.")
return 2
fits = tree.getroot()
fits = exclude_jhove_properties(fits)
# NOTE: This is hardcoded for now because FPCommandOutput references FPRule for future development,
# when characterization will become user-configurable and be decoupled from FITS specifically.
# Thus a stub rule must exist for FITS; this will be replaced with a real rule in the future.
logger.info("Storing output of file characterization...")
insertIntoFPCommandOutput(
file_uuid,
etree.tostring(fits, pretty_print=False),
"3a19de70-0e42-4145-976b-3a248d43b462",
)
except (OSError, ValueError):
logger.exception("Execution failed")
return 1
finally:
# We are responsible for removing the temporary file and we do it here
# to ensure that it's going to happen whatever occurs inside our try
# block.
os.remove(temp_file)
return 0
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
print output[0]
if output[1] != "":
print >> sys.stderr, output[1]
#it executes check for errors
if retcode != 0:
print >> sys.stderr, "error code:" + retcode.__str__()
print output[1] # sError
quit(retcode)
try:
tree = etree.parse(tempFile)
except:
os.remove(tempFile)
print >> sys.stderr, "Failed to read Fits's xml."
exit(2)
fits = tree.getroot()
os.remove(tempFile)
if excludeJhoveProperties:
fits = excludeJhoveProperties(fits)
# NOTE: This is hardcoded for now because FPCommandOutput references FPRule for future development,
# when characterization will become user-configurable and be decoupled from FITS specifically.
# Thus a stub rule must exist for FITS; this will be replaced with a real rule in the future.
insertIntoFPCommandOutput(fileUUID,
etree.tostring(fits, pretty_print=False),
'3a19de70-0e42-4145-976b-3a248d43b462')
except OSError as ose:
print >> sys.stderr, "Execution failed:", ose
exit(1)
exit(exitCode)
