def test_collect_mdsec_elements(self):
f1 = metsrw.FSEntry('file1.txt', file_uuid=str(uuid.uuid4()))
f1.amdsecs.append(metsrw.AMDSec())
f1.dmdsecs.append(metsrw.SubSection('dmdSec', None))
f2 = metsrw.FSEntry('file2.txt', file_uuid=str(uuid.uuid4()))
f2.dmdsecs.append(metsrw.SubSection('dmdSec', None))
mw = metsrw.METSDocument()
elements = mw._collect_mdsec_elements([f1, f2])
# Check ordering - dmdSec before amdSec
assert isinstance(elements, list)
assert len(elements) == 3
assert isinstance(elements[0], metsrw.SubSection)
assert elements[0].subsection == 'dmdSec'
assert isinstance(elements[1], metsrw.SubSection)
assert elements[1].subsection == 'dmdSec'
assert isinstance(elements[2], metsrw.AMDSec)
def add_new_files(mets, sip_uuid, sip_dir):
"""
Add new files to structMap, fileSec.
This supports adding new metadata or preservation files.
If a new file is a metadata.csv, parse it to create dmdSecs.
"""
# Find new files
# How tell new file from old with same name? Check hash?
# QUESTION should the metadata.csv be parsed and only updated if different even if one already existed?
new_files = []
old_mets_rel_path = _get_old_mets_rel_path(sip_uuid)
metadata_csv = None
objects_dir = os.path.join(sip_dir, 'objects')
for dirpath, _, filenames in os.walk(objects_dir):
for filename in filenames:
# Find in METS
current_loc = os.path.join(dirpath, filename).replace(
sip_dir, '%SIPDirectory%', 1)
rel_path = current_loc.replace('%SIPDirectory%', '', 1)
print('Looking for', rel_path, 'in METS')
fsentry = mets.get_file(path=rel_path)
if fsentry is None:
# If not in METS (and is not old METS), get File object and
# store for later
if rel_path != old_mets_rel_path:
print(rel_path, 'not found in METS, must be new file')
f = models.File.objects.get(currentlocation=current_loc,
sip_id=sip_uuid)
new_files.append(f)
if rel_path == 'objects/metadata/metadata.csv':
metadata_csv = f
else:
print(rel_path, 'found in METS, no further work needed')
if not new_files:
return mets
# Set global counters so getAMDSec will work
createmets2.globalAmdSecCounter = int(
mets.tree.xpath('count(mets:amdSec)', namespaces=ns.NSMAP))
createmets2.globalTechMDCounter = int(
mets.tree.xpath('count(mets:amdSec/mets:techMD)', namespaces=ns.NSMAP))
createmets2.globalDigiprovMDCounter = int(
mets.tree.xpath('count(mets:amdSec/mets:digiprovMD)',
namespaces=ns.NSMAP))
objects_fsentry = mets.get_file(label='objects', type='Directory')
for f in new_files:
# Create amdSecs
print('Adding amdSec for', f.currentlocation, '(', f.uuid, ')')
amdsec, amdid = createmets2.getAMDSec(
fileUUID=f.uuid,
filePath=None, # Only needed if use=original
use=f.filegrpuse,
type=None, # Not used
sip_uuid=sip_uuid,
transferUUID=None, # Only needed if use=original
itemdirectoryPath=None, # Only needed if use=original
typeOfTransfer=None, # Only needed if use=original
baseDirectoryPath=sip_dir,
)
print(f.uuid, 'has amdSec with ID', amdid)
# Create parent directories if needed
dirs = os.path.dirname(
f.currentlocation.replace('%SIPDirectory%objects/', '',
1)).split('/')
parent_fsentry = objects_fsentry
for dirname in (d for d in dirs if d):
child = mets.get_file(type='Directory', label=dirname)
if child is None:
child = metsrw.FSEntry(
path=None,
type='Directory',
label=dirname,
)
parent_fsentry.add_child(child)
parent_fsentry = child
derived_from = None
if f.original_file_set.exists():
original_f = f.original_file_set.get().source_file
derived_from = mets.get_file(file_uuid=original_f.uuid)
entry = metsrw.FSEntry(
path=f.currentlocation.replace('%SIPDirectory%', '', 1),
use=f.filegrpuse,
type='Item',
file_uuid=f.uuid,
derived_from=derived_from,
)
metsrw_amdsec = metsrw.AMDSec(tree=amdsec, section_id=amdid)
entry.amdsecs.append(metsrw_amdsec)
parent_fsentry.add_child(entry)
# Parse metadata.csv and add dmdSecs
if metadata_csv:
mets = update_metadata_csv(mets, metadata_csv, sip_uuid, sip_dir)
return mets
def test_identifier(self):
# should be in the format 'amdSec_1'
amdsec = metsrw.AMDSec()
assert amdsec.id_string()
def test_tree_overwrites_serialize(self):
elem = etree.Element('temp')
amdsec = metsrw.AMDSec(tree=elem, section_id='id1')
assert amdsec.serialize() == elem
def test_tree_no_id(self):
with pytest.raises(ValueError) as excinfo:
metsrw.AMDSec(tree=etree.Element('amdSec'))
assert 'section_id' in str(excinfo.value)
def add_new_files(job, mets, sip_uuid, sip_dir):
"""
Add new files to structMap, fileSec.
This supports adding new metadata or preservation files.
If a new file is a metadata.csv, parse it to create dmdSecs.
"""
# Find new files
# How tell new file from old with same name? Check hash?
# QUESTION should the metadata.csv be parsed and only updated if different
# even if one already existed?
new_files = []
old_mets_rel_path = _get_old_mets_rel_path(sip_uuid)
metadata_csv = None
objects_dir = os.path.join(sip_dir, "objects")
for dirpath, _, filenames in os.walk(objects_dir):
for filename in filenames:
# Find in METS
current_loc = os.path.join(dirpath, filename).replace(
sip_dir, "%SIPDirectory%", 1
)
rel_path = current_loc.replace("%SIPDirectory%", "", 1)
job.pyprint("Looking for", rel_path, "in METS")
fsentry = mets.get_file(path=rel_path)
if fsentry is None:
# If not in METS (and is not old METS), get File object and
# store for later
if rel_path != old_mets_rel_path:
job.pyprint(rel_path, "not found in METS, must be new file")
f = models.File.objects.get(
currentlocation=current_loc, sip_id=sip_uuid
)
new_files.append(f)
if rel_path == "objects/metadata/metadata.csv":
metadata_csv = f
else:
job.pyprint(rel_path, "found in METS, no further work needed")
if not new_files:
return mets
# Set global counters so getAMDSec will work
state = createmets2.MetsState(
globalAmdSecCounter=metsrw.AMDSec.get_current_id_count(),
globalTechMDCounter=metsrw.SubSection.get_current_id_count("techMD"),
globalDigiprovMDCounter=metsrw.SubSection.get_current_id_count("digiprovMD"),
)
objects_fsentry = mets.get_file(label="objects", type="Directory")
for f in new_files:
# Create amdSecs
job.pyprint("Adding amdSec for", f.currentlocation, "(", f.uuid, ")")
amdsec, amdid = createmets2.getAMDSec(
job,
fileUUID=f.uuid,
filePath=None, # Only needed if use=original
use=f.filegrpuse,
sip_uuid=sip_uuid,
transferUUID=None, # Only needed if use=original
itemdirectoryPath=None, # Only needed if use=original
typeOfTransfer=None, # Only needed if use=original
baseDirectoryPath=sip_dir,
state=state,
)
job.pyprint(f.uuid, "has amdSec with ID", amdid)
# Create parent directories if needed
dirs = os.path.dirname(
f.currentlocation.replace("%SIPDirectory%objects/", "", 1)
).split("/")
parent_fsentry = objects_fsentry
for dirname in (d for d in dirs if d):
child = mets.get_file(type="Directory", label=dirname)
if child is None:
child = metsrw.FSEntry(path=None, type="Directory", label=dirname)
parent_fsentry.add_child(child)
parent_fsentry = child
derived_from = None
if f.original_file_set.exists():
original_f = f.original_file_set.get().source_file
derived_from = mets.get_file(file_uuid=original_f.uuid)
entry = metsrw.FSEntry(
path=f.currentlocation.replace("%SIPDirectory%", "", 1),
use=f.filegrpuse,
type="Item",
file_uuid=f.uuid,
derived_from=derived_from,
)
metsrw_amdsec = metsrw.AMDSec(tree=amdsec, section_id=amdid)
entry.amdsecs.append(metsrw_amdsec)
parent_fsentry.add_child(entry)
# Parse metadata.csv and add dmdSecs
if metadata_csv:
mets = update_metadata_csv(job, mets, metadata_csv, sip_uuid, sip_dir, state)
return mets
