def create_structmap(workspace, filesec, filelist, type_attr=None,
root_type=None):
"""Creates METS document element tree that contains structural map.
:param workspace: directory from which some files are searhed
:param filesec: fileSec element
:param filelist: Sorted list of digital objects (file paths)
:param type_attr: TYPE attribute of structMap element
:param root_type: TYPE attribute of root div element
:returns: structural map element
"""
amdids = get_md_references(workspace, directory='.')
dmdids = get_md_references(workspace, directory='.', ref_type='dmd')
if type_attr == 'Directory-physical':
container_div = mets.div(type_attr='directory', label='.',
dmdid=dmdids, admid=amdids)
else:
root_type = root_type if root_type else 'directory'
container_div = mets.div(type_attr=root_type, dmdid=dmdids,
admid=amdids)
structmap = mets.structmap(type_attr=type_attr)
structmap.append(container_div)
divs = div_structure(filelist)
create_div(workspace, divs, container_div, filesec,
filelist, type_attr=type_attr)
mets_element = mets.mets(child_elements=[structmap])
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 create_ead3_structmap(filegrp, attributes):
"""Create structmap based on ead3 descriptive metadata structure.
:filegrp: fileGrp element
:attributes: The following keys:
all_amd_refs: XML element tree of administrative metadata
references
all_dmd_refs: XML element tree of descriptive metadata
references
filelist: Sorted list of digital objects (file paths)
dmdsec_loc: EAD3 descriptive metadata file
structmap_type: TYPE attribute of structMap element
workspace: Workspace path
"""
structmap = mets.structmap(type_attr=attributes["structmap_type"])
container_div = mets.div(type_attr='logical')
root = ET.parse(attributes["dmdsec_loc"]).getroot()
try:
label = root.xpath(("//ead3:archdesc/@otherlevel | "
"//ead3:archdesc/@level"),
namespaces=NAMESPACES)[0]
except IndexError:
label = 'archdesc'
amdids = get_md_references(attributes["all_amd_refs"], directory='.')
dmdids = get_md_references(attributes["all_dmd_refs"], directory='.')
div_ead = mets.div(type_attr='archdesc',
label=label,
dmdid=dmdids,
admid=amdids)
if root.xpath("//ead3:archdesc/ead3:dsc", namespaces=NAMESPACES):
for elem in root.xpath("//ead3:dsc/*", namespaces=NAMESPACES):
if ET.QName(elem.tag).localname in ALLOWED_C_SUBS:
ead3_c_div(elem, div_ead, filegrp, attributes)
container_div.append(div_ead)
structmap.append(container_div)
mets_element = mets.mets(child_elements=[structmap])
ET.cleanup_namespaces(mets_element)
return ET.ElementTree(mets_element)
def create_ead3_structmap(descfile, workspace, filegrp, filelist, type_attr):
"""Create structmap based on ead3 descriptive metadata structure.
:desc_file: EAD3 descriptive metadata file
:workspace: Workspace path
:structmap: Structmap element
:filegrp: fileGrp element
:filelist: Sorted list of digital objects (file paths)
:type_attr: TYPE attribute of structMap element
"""
structmap = mets.structmap(type_attr=type_attr)
container_div = mets.div(type_attr='logical')
root = ET.parse(descfile).getroot()
try:
label = root.xpath(("//ead3:archdesc/@otherlevel | "
"//ead3:archdesc/@level"),
namespaces=NAMESPACES)[0]
except IndexError:
label = 'archdesc'
amdids = get_md_references(workspace, directory='.')
dmdids = get_md_references(workspace, directory='.', ref_type='dmd')
div_ead = mets.div(type_attr='archdesc',
label=label,
dmdid=dmdids,
admid=amdids)
if len(root.xpath("//ead3:archdesc/ead3:dsc", namespaces=NAMESPACES)) > 0:
for elem in root.xpath("//ead3:dsc/*", namespaces=NAMESPACES):
if ET.QName(elem.tag).localname in ALLOWED_C_SUBS:
ead3_c_div(elem, div_ead, filegrp, workspace, filelist)
container_div.append(div_ead)
structmap.append(container_div)
mets_element = mets.mets(child_elements=[structmap])
ET.cleanup_namespaces(mets_element)
return ET.ElementTree(mets_element)
def create_structmap(filesec, **attributes):
"""
Creates METS document element tree that contains structural map.
:filesec: fileSec element
:attributes: The following keys:
all_amd_refs: XML element tree of administrative metadata
references
all_dmd_refs: XML element tree of descriptive metadata
references
filelist: Sorted list of digital objects (file paths)
structmap_type: TYPE attribute of structMap element
root_type: TYPE attribute of root div element
file_ids: Dict with file paths and identifiers
workspace: Workspace path
:returns: structural map element
"""
attributes = get_reference_lists(**_attribute_values(attributes))
amdids = get_md_references(attributes["all_amd_refs"], directory='.')
dmdids = get_md_references(attributes["all_dmd_refs"], directory='.')
if attributes["structmap_type"] == 'Directory-physical':
container_div = mets.div(type_attr='directory',
label='.',
dmdid=dmdids,
admid=amdids)
else:
container_div = mets.div(type_attr=attributes["root_type"],
dmdid=dmdids,
admid=amdids)
structmap = mets.structmap(type_attr=attributes["structmap_type"])
structmap.append(container_div)
divs = div_structure(attributes["filelist"])
create_div(divs, container_div, filesec, attributes)
mets_element = mets.mets(child_elements=[structmap])
ET.cleanup_namespaces(mets_element)
return ET.ElementTree(mets_element)
def run(self):
"""Create a METS document that contains logical structural map.
Logical structural map is based on dataset metadata retrieved
from Metax.
:returns: ``None``
"""
# Read the generated physical structmap from file
physical_structmap = ET.parse(
os.path.join(self.sip_creation_path, 'structmap.xml'))
# Get dmdsec id from physical_structmap
dmdsec_id = physical_structmap.getroot()[0][0].attrib['DMDID']
# Get provenance id's
provenance_ids = self.get_provenance_ids()
# Init logical structmap
logical_structmap = mets.structmap(type_attr='Fairdata-logical')
mets_structmap = mets.mets(child_elements=[logical_structmap])
# Create logical structmap
categories = self.find_file_categories()
wrapper_div = mets.div(type_attr='logical',
dmdid=[dmdsec_id],
admid=provenance_ids)
for category in categories:
div = mets.div(type_attr=category)
for filename in categories.get(category):
fileid = self.get_fileid(encode_path(filename, safe='/'))
div.append(mets.fptr(fileid))
wrapper_div.append(div)
logical_structmap.append(wrapper_div)
with self.output().open('wb') as output:
output.write(h.serialize(mets_structmap))