def main(job, sip_uuid, dc_path):
# If there's no metadata, that's not an error, and just keep going
if not os.path.exists(dc_path):
job.pyprint("DC metadata not found; exiting", "(at", dc_path + ")")
return 0
job.pyprint("Loading DC metadata from", dc_path)
with open(dc_path) as json_data:
data = json.load(json_data)
dc = models.DublinCore(metadataappliestoidentifier=sip_uuid,
metadataappliestotype_id=INGEST_METADATA_TYPE)
for key, value in data.items():
try:
setattr(dc, key, value)
except AttributeError:
job.pyprint("Invalid DC attribute:", key, file=sys.stderr)
dc.save()
# ``dc.json`` was copied to ingest so the code above could read it, but we
# don't need it anymore so we're removing it.
try:
job.pyprint('Removing "dc.json":', dc_path)
os.remove(dc_path)
except Exception as err:
job.pyprint('Unable to remove "dc.json":', err)
return 0
def parse_dc(sip_uuid, root):
"""
Parse SIP-level DublinCore metadata into the DublinCore table.
Deletes existing entries associated with this SIP.
:param str sip_uuid: UUID of the SIP to parse the metadata for.
:param root: root Element of the METS file.
:return: DublinCore DB object, or None
"""
# Delete existing DC
models.DublinCore.objects.filter(metadataappliestoidentifier=sip_uuid, metadataappliestotype_id=MD_TYPE_SIP_ID).delete()
# Parse DC
dmds = root.xpath('mets:dmdSec/mets:mdWrap[@MDTYPE="DC"]/parent::*', namespaces=ns.NSMAP)
dc_model = None
# Find which DC to parse into DB
if len(dmds) > 0:
DC_TERMS_MATCHING = {
'title': 'title',
'creator': 'creator',
'subject': 'subject',
'description': 'description',
'publisher': 'publisher',
'contributor': 'contributor',
'date': 'date',
'type': 'type',
'format': 'format',
'identifier': 'identifier',
'source': 'source',
'relation': 'relation',
'language': 'language',
'coverage': 'coverage',
'rights': 'rights',
'isPartOf': 'is_part_of',
}
# Want most recently updated
dmds = sorted(dmds, key=lambda e: e.get('CREATED'))
# Only want SIP DC, not file DC
div = root.find('mets:structMap/mets:div/mets:div[@TYPE="Directory"][@LABEL="objects"]', namespaces=ns.NSMAP)
dmdids = div.get('DMDID')
# No SIP DC
if dmdids is None:
return
dmdids = dmdids.split()
for dmd in dmds[::-1]: # Reversed
if dmd.get('ID') in dmdids:
dc_xml = dmd.find('mets:mdWrap/mets:xmlData/dcterms:dublincore', namespaces=ns.NSMAP)
break
dc_model = models.DublinCore(
metadataappliestoidentifier=sip_uuid,
metadataappliestotype_id=MD_TYPE_SIP_ID,
status=models.METADATA_STATUS_REINGEST,
)
print('Dublin Core:')
for elem in dc_xml:
tag = elem.tag.replace(ns.dctermsBNS, '', 1).replace(ns.dcBNS, '', 1)
print(tag, elem.text)
setattr(dc_model, DC_TERMS_MATCHING[tag], elem.text)
dc_model.save()
return dc_model
def transfer_metadata_edit(request, uuid, id=None):
if id:
dc = models.DublinCore.objects.get(pk=id)
else:
try:
dc = models.DublinCore.objects.get(
metadataappliestotype=transfer_metadata_type_id(),
metadataappliestoidentifier__exact=uuid,
)
return redirect("transfer:transfer_metadata_edit", uuid, dc.id)
except models.DublinCore.DoesNotExist:
dc = models.DublinCore(
metadataappliestotype=transfer_metadata_type_id(),
metadataappliestoidentifier=uuid,
)
if request.method == "POST":
form = DublinCoreMetadataForm(request.POST, instance=dc)
if form.is_valid():
dc = form.save()
return redirect("transfer:transfer_metadata_list", uuid)
else:
form = DublinCoreMetadataForm(instance=dc)
jobs = models.Job.objects.filter(sipuuid=uuid)
name = jobs.get_directory_name()
return render(request, "transfer/metadata_edit.html", locals())
def main(sip_uuid, dc_path):
# If there's no metadata, that's not an error, and just keep going
if not os.path.exists(dc_path):
print("DC metadata not found; exiting", "(at", dc_path + ")")
return 0
print("Loading DC metadata from", dc_path)
with open(dc_path) as json_data:
data = json.load(json_data)
dc = models.DublinCore(metadataappliestoidentifier=sip_uuid,
metadataappliestotype_id=INGEST_METADATA_TYPE)
for key, value in data.items():
try:
setattr(dc, key, value)
except AttributeError:
print("Invalid DC attribute:", key, file=sys.stderr)
dc.save()
return 0
def transfer_metadata_edit(request, uuid, id=None):
if id:
dc = models.DublinCore.objects.get(pk=id)
else:
try:
dc = models.DublinCore.objects.get(
metadataappliestotype=transfer_metadata_type_id(),
metadataappliestoidentifier__exact=uuid)
return redirect('components.transfer.views.transfer_metadata_edit',
uuid, dc.id)
except models.DublinCore.DoesNotExist:
dc = models.DublinCore(
metadataappliestotype=transfer_metadata_type_id(),
metadataappliestoidentifier=uuid)
fields = [
'title', 'creator', 'subject', 'description', 'publisher',
'contributor', 'date', 'type', 'format', 'identifier', 'source',
'relation', 'language', 'coverage', 'rights'
]
if request.method == 'POST':
form = DublinCoreMetadataForm(request.POST)
if form.is_valid():
for item in fields:
if item not in form.cleaned_data:
continue
setattr(dc, item, form.cleaned_data[item])
dc.save()
return redirect('components.transfer.views.transfer_metadata_list',
uuid)
else:
initial = {}
for item in fields:
initial[item] = getattr(dc, item)
form = DublinCoreMetadataForm(initial=initial)
jobs = models.Job.objects.filter(sipuuid=uuid, subjobof='')
name = utils.get_directory_name_from_job(jobs)
return render(request, 'transfer/metadata_edit.html', locals())
def ingest_metadata_edit(request, uuid, id=None):
if id:
dc = models.DublinCore.objects.get(pk=id)
else:
# Right now we only support linking metadata to the Ingest
try:
dc = models.DublinCore.objects.get_sip_metadata(uuid)
return HttpResponseRedirect(
reverse('components.ingest.views.ingest_metadata_edit',
args=[uuid, dc.id]))
except ObjectDoesNotExist:
dc = models.DublinCore(
metadataappliestotype=ingest_sip_metadata_type_id(),
metadataappliestoidentifier=uuid)
fields = [
'title', 'creator', 'subject', 'description', 'publisher',
'contributor', 'date', 'type', 'format', 'identifier', 'source',
'relation', 'language', 'coverage', 'rights'
]
if request.method == 'POST':
form = DublinCoreMetadataForm(request.POST)
if form.is_valid():
for item in fields:
setattr(dc, item, form.cleaned_data[item])
dc.save()
return HttpResponseRedirect(
reverse('components.ingest.views.ingest_metadata_list',
args=[uuid]))
else:
initial = {}
for item in fields:
initial[item] = getattr(dc, item)
form = DublinCoreMetadataForm(initial=initial)
jobs = models.Job.objects.filter(sipuuid=uuid, subjobof='')
name = utils.get_directory_name(jobs[0])
return render(request, 'ingest/metadata_edit.html', locals())
def ingest_metadata_edit(request, uuid, id=None):
if id:
# If we have the ID of the DC object, use that - Edit
dc = models.DublinCore.objects.get(pk=id)
else:
# Otherwise look for a SIP with the provided UUID, creating a new one
# if needed. Not using get_or_create because that save the empty
# object, even if the form is not submitted.
sip_type_id = ingest_sip_metadata_type_id()
try:
dc = models.DublinCore.objects.get(
metadataappliestotype=sip_type_id,
metadataappliestoidentifier=uuid)
id = dc.id
except models.DublinCore.DoesNotExist:
dc = models.DublinCore(
metadataappliestotype=sip_type_id,
metadataappliestoidentifier=uuid)
# If the SIP is an AIC, use the AIC metadata form
if 'AIC' in models.SIP.objects.get(uuid=uuid).sip_type:
form = ingest_forms.AICDublinCoreMetadataForm(request.POST or None,
instance=dc)
dc_type = "Archival Information Collection"
else:
form = ingest_forms.DublinCoreMetadataForm(request.POST or None,
instance=dc)
dc_type = "Archival Information Package"
if form.is_valid():
dc = form.save()
dc.type = dc_type
dc.save()
return redirect('components.ingest.views.ingest_metadata_list', uuid)
jobs = models.Job.objects.filter(sipuuid=uuid, subjobof='')
name = utils.get_directory_name_from_job(jobs[0])
return render(request, 'ingest/metadata_edit.html', locals())
def aic_metadata_add(request, uuid):
sip_type_id = ingest_sip_metadata_type_id()
try:
dc = models.DublinCore.objects.get(metadataappliestotype=sip_type_id,
metadataappliestoidentifier=uuid)
id = dc.id
except models.DublinCore.DoesNotExist:
dc = models.DublinCore(metadataappliestotype=sip_type_id,
metadataappliestoidentifier=uuid)
form = ingest_forms.AICDublinCoreMetadataForm(request.POST or None,
instance=dc)
if form.is_valid():
# Save the metadata
dc = form.save()
dc.type = "Archival Information Collection"
dc.save()
# Start the MicroServiceChainLink for the AIC
shared_dir = django_settings.SHARED_DIRECTORY
source = os.path.join(shared_dir, 'tmp', uuid)
watched_dir = django_settings.WATCH_DIRECTORY
name = dc.title if dc.title else dc.identifier
name = slugify(name).replace('-', '_')
dir_name = '{name}-{uuid}'.format(name=name, uuid=uuid)
destination = os.path.join(watched_dir, 'system', 'createAIC',
dir_name)
destination_db = destination.replace(shared_dir, '%sharedPath%') + '/'
models.SIP.objects.filter(uuid=uuid).update(currentpath=destination_db)
shutil.move(source, destination)
return redirect('ingest_index')
name = dc.title or "New AIC"
aic = True
return render(request, 'ingest/metadata_edit.html', locals())
def parse_dc(job, sip_uuid, root):
"""
Parse SIP-level DublinCore metadata into the DublinCore table.
Deletes existing entries associated with this SIP.
:param str sip_uuid: UUID of the SIP to parse the metadata for.
:param root: root Element of the METS file.
:return: DublinCore DB object, or None
"""
# Delete existing DC
models.DublinCore.objects.filter(
metadataappliestoidentifier=sip_uuid, metadataappliestotype_id=MD_TYPE_SIP_ID
).delete()
# Parse DC
dmds = root.xpath(
'mets:dmdSec/mets:mdWrap[@MDTYPE="DC"]/parent::*', namespaces=ns.NSMAP
)
dc_model = None
# Find which DC to parse into DB
if len(dmds) > 0:
DC_TERMS_MATCHING = {
"title": "title",
"creator": "creator",
"subject": "subject",
"description": "description",
"publisher": "publisher",
"contributor": "contributor",
"date": "date",
"type": "type",
"format": "format",
"identifier": "identifier",
"source": "source",
"relation": "relation",
"language": "language",
"coverage": "coverage",
"rights": "rights",
"isPartOf": "is_part_of",
}
# Want most recently updated
dmds = sorted(dmds, key=lambda e: e.get("CREATED"))
# Only want SIP DC, not file DC
div = root.find(
'mets:structMap/mets:div/mets:div[@TYPE="Directory"][@LABEL="objects"]',
namespaces=ns.NSMAP,
)
dmdids = div.get("DMDID")
# No SIP DC
if dmdids is None:
return
dmdids = dmdids.split()
for dmd in dmds[::-1]: # Reversed
if dmd.get("ID") in dmdids:
dc_xml = dmd.find(
"mets:mdWrap/mets:xmlData/dcterms:dublincore", namespaces=ns.NSMAP
)
break
dc_model = models.DublinCore(
metadataappliestoidentifier=sip_uuid,
metadataappliestotype_id=MD_TYPE_SIP_ID,
status=models.METADATA_STATUS_REINGEST,
)
job.pyprint("Dublin Core:")
for elem in dc_xml:
tag = elem.tag.replace(ns.dctermsBNS, "", 1).replace(ns.dcBNS, "", 1)
job.pyprint(tag, elem.text)
if elem.text is not None:
setattr(dc_model, DC_TERMS_MATCHING[tag], elem.text)
dc_model.save()
return dc_model
