def compile_structmap(**kwargs):
"""Generate METS file section and structural map based on
created/imported administrative metada and descriptive metadata.
:kwargs: Given arguments:
workspace: Workspace directory
structmap_type: Type of structmap
root_type: Type of root div
dmdsec_loc: Location of structured descriptive metadata
file_ids: Dict to be populated with file paths and IDs
stdout: True to print output to stdout
"""
attributes = _attribute_values(kwargs)
# Create an event documenting the structmap creation
_create_event(workspace=attributes["workspace"],
structmap_type=attributes["structmap_type"],
root_type=attributes["root_type"])
# Get reference list only after the structmap creation event
attributes = get_reference_lists(**attributes)
if attributes["structmap_type"] == 'EAD3-logical':
# If structured descriptive metadata for structMap divs is used, also
# the fileSec element (apparently?) is different. The
# create_ead3_structmap function populates the fileGrp element.
filegrp = mets.filegrp()
filesec_element = mets.filesec(child_elements=[filegrp])
filesec = mets.mets(child_elements=[filesec_element])
structmap = create_ead3_structmap(filegrp, attributes)
else:
(filesec, file_ids) = create_filesec(**attributes)
# Add file path and ID dict to attributes
attributes['file_ids'] = file_ids
structmap = create_structmap(filesec.getroot(), **attributes)
if attributes["stdout"]:
print(xml_utils.serialize(filesec).decode("utf-8"))
print(xml_utils.serialize(structmap).decode("utf-8"))
output_sm_file = os.path.join(attributes["workspace"], 'structmap.xml')
output_fs_file = os.path.join(attributes["workspace"], 'filesec.xml')
if not os.path.exists(os.path.dirname(output_sm_file)):
os.makedirs(os.path.dirname(output_sm_file))
if not os.path.exists(os.path.dirname(output_fs_file)):
os.makedirs(os.path.dirname(output_fs_file))
with open(output_sm_file, 'wb+') as outfile:
outfile.write(xml_utils.serialize(structmap))
with open(output_fs_file, 'wb+') as outfile:
outfile.write(xml_utils.serialize(filesec))
print("compile_structmap created files: %s %s" %
(output_sm_file, output_fs_file))
def create_filesec(workspace, filelist):
"""Creates METS document element tree that contains fileSec element.
"""
filegrp = mets.filegrp()
filesec = mets.filesec(child_elements=[filegrp])
create_filegrp(workspace, filegrp, filelist)
mets_element = mets.mets(child_elements=[filesec])
ET.cleanup_namespaces(mets_element)
return ET.ElementTree(mets_element)
def main(arguments=None):
"""The main method for compile_sturctmap"""
args = parse_arguments(arguments)
structmap = mets.structmap(type_attr=args.type_attr)
mets_structmap = mets.mets(child_elements=[structmap])
filegrp = mets.filegrp()
filesec = mets.filesec(child_elements=[filegrp])
mets_filesec = mets.mets(child_elements=[filesec])
_, dmdsec_id = ids_for_files(args.workspace,
None,
'dmdsec.xml',
dash_count=0)
if args.dmdsec_struct == 'ead3':
container_div = mets.div(type_attr='logical')
structmap.append(container_div)
create_ead3_structmap(args.dmdsec_loc, args.workspace, container_div,
filegrp, dmdsec_id)
else:
amdids = get_links_event_agent(args.workspace, None)
container_div = mets.div(type_attr='directory',
dmdid=dmdsec_id,
admid=amdids)
structmap.append(container_div)
divs = div_structure(args.workspace)
create_structmap(args.workspace, divs, container_div, filegrp)
if args.stdout:
print h.serialize(mets_filesec)
print h.serialize(mets_structmap)
output_sm_file = os.path.join(args.workspace, 'structmap.xml')
output_fs_file = os.path.join(args.workspace, 'filesec.xml')
if not os.path.exists(os.path.dirname(output_sm_file)):
os.makedirs(os.path.dirname(output_sm_file))
if not os.path.exists(os.path.dirname(output_fs_file)):
os.makedirs(os.path.dirname(output_fs_file))
with open(output_sm_file, 'w+') as outfile:
outfile.write(h.serialize(mets_structmap))
with open(output_fs_file, 'w+') as outfile:
outfile.write(h.serialize(mets_filesec))
print "compile_structmap created files: %s %s" % (output_sm_file,
output_fs_file)
return 0
def compile_structmap(workspace="./workspace/",
structmap_type=None,
root_type=None,
dmdsec_loc=None,
stdout=False):
"""Generate METS file section and structural map based on
created/imported administrative metada and descriptive metadata.
"""
filelist = get_objectlist(workspace)
if structmap_type == 'EAD3-logical':
# If structured descriptive metadata for structMap divs is used, also
# the fileSec element (apparently?) is different. The
# create_ead3_structmap function populates the fileGrp element.
filegrp = mets.filegrp()
filesec_element = mets.filesec(child_elements=[filegrp])
filesec = mets.mets(child_elements=[filesec_element])
structmap = create_ead3_structmap(dmdsec_loc, workspace, filegrp,
filelist, structmap_type)
else:
filesec = create_filesec(workspace, filelist)
structmap = create_structmap(workspace, filesec.getroot(), filelist,
structmap_type, root_type)
if stdout:
print(xml_utils.serialize(filesec).decode("utf-8"))
print(xml_utils.serialize(structmap).decode("utf-8"))
output_sm_file = os.path.join(workspace, 'structmap.xml')
output_fs_file = os.path.join(workspace, 'filesec.xml')
if not os.path.exists(os.path.dirname(output_sm_file)):
os.makedirs(os.path.dirname(output_sm_file))
if not os.path.exists(os.path.dirname(output_fs_file)):
os.makedirs(os.path.dirname(output_fs_file))
with open(output_sm_file, 'wb+') as outfile:
outfile.write(xml_utils.serialize(structmap))
with open(output_fs_file, 'wb+') as outfile:
outfile.write(xml_utils.serialize(filesec))
print("compile_structmap created files: %s %s" %
(output_sm_file, output_fs_file))
def test_get_fileid():
"""Test get_fileid function. Create a fileGrp element with few files and
test that the function finds correct file IDs.
"""
# Create fileGrp element that contains three file elements with different
# identifiers and paths
files = [
mets.file_elem(file_id='identifier%s' % num,
admid_elements=['foo', 'bar'],
loctype='foo',
xlink_href='file://path/to/file+name%s' % num,
xlink_type='foo') for num in range(3)
]
filegrp = mets.filegrp(child_elements=files)
assert compile_structmap.get_fileid(filegrp, 'path/to/file name1') \
== 'identifier1'
def test_add_fptrs_div_ead(testpath, run_cli, hrefs, length, child_elem,
order):
"""Tests the add_fptrs_div_ead function by asserting that the c_div
element has been modified with fptrs and divs correctly according to
the test cases.
"""
create_test_data(testpath, run_cli, order=order)
div_elem = '<mets:div xmlns:mets="http://www.loc.gov/METS/"></mets:div>'
xml = ET.fromstring(div_elem)
attrs = {}
attrs["all_amd_refs"] = read_md_references(
testpath, "import-object-md-references.jsonl")
attrs["object_refs"] = attrs["all_amd_refs"]
attrs["workspace"] = testpath
attrs["filelist"] = [
'tests/data/structured/Publication files/publication.txt',
'tests/data/structured/Software files/koodi.java'
]
filegrp = filegrp = mets.filegrp()
c_div = compile_structmap.add_fptrs_div_ead(xml, hrefs, filegrp, attrs)
# Child elements are either new divs or fptrs
assert c_div.xpath(
'./*')[0].tag == '{http://www.loc.gov/METS/}%s' % child_elem
# Number of child elements should equal the number of valid hrefs
assert len(c_div.xpath('./*')) == length
# Number of fptr elements should equal the number of valid hrefs
assert len(c_div.findall('.//{http://www.loc.gov/METS/}fptr')) == length
# If file properties exist, it is written to the divs
if order and length == 1:
assert 'ORDER' in c_div.attrib
elif order:
assert 'ORDER' in c_div.xpath('./*')[0].attrib
assert c_div.xpath('./*')[0].get('TYPE') == 'dao'
else:
assert 'ORDER' not in c_div.attrib
def create_filesec(**attributes):
"""
Creates METS document element tree that contains fileSec element.
:attributes: Attribute values as a dict
all_amd_refs: XML element tree of administrative metadata
references
filelist: Sorted list of digital objects (file paths)
:returns: A tuple of METS XML Element tree including file section
element and a dict of file paths and identifiers
"""
attributes = get_reference_lists(**_attribute_values(attributes))
filegrp = mets.filegrp()
filesec = mets.filesec(child_elements=[filegrp])
file_ids = {}
for path in attributes["filelist"]:
fileid = add_file_to_filesec(attributes["all_amd_refs"],
attributes["object_refs"], path, filegrp)
file_ids[path] = fileid
mets_element = mets.mets(child_elements=[filesec])
ET.cleanup_namespaces(mets_element)
return (ET.ElementTree(mets_element), file_ids)