def add_file_to_filesec(workspace, path, filegrp):
"""Add file element to fileGrp element given as parameter.
:param workspace: Workspace directorye from which administrative MD
files and amd reference files searched.
:param path: url encoded path of the file
:param lxml.etree.Element filegrp: fileGrp element
:param str returns: id of file added to fileGrp
:returns: unique identifier of file element
"""
fileid = '_{}'.format(uuid4())
# Create list of IDs of amdID elements
amdids = get_md_references(workspace, path=path)
# Create XML element and add it to fileGrp
file_el = mets.file_elem(fileid,
admid_elements=set(amdids),
loctype='URL',
xlink_href='file://%s' %
encode_path(path, safe='/'),
xlink_type='simple',
groupid=None)
streams = get_objectlist(workspace, path)
if streams:
for stream in streams:
stream_ids = get_md_references(workspace, path=path, stream=stream)
stream_el = mets.stream(admid_elements=stream_ids)
file_el.append(stream_el)
filegrp.append(file_el)
return fileid
def get_reference_lists(**attributes):
"""
Fill the attributes with various lists.
:attributes: The following keys:
workspace: Workspace path
object_refs: XML tree of digital objects.
Will be created if missing.
filelist: ID list of objects. Will be created if missing.
all_amd_refs: All administrative metadata references.
Will be created if missing.
all_dmd_refs: All descriptive metadata references.
Will be created if missing.
:returns: Attributes filled with the lists listed above.
"""
attributes["object_refs"] = attributes.get(
"object_refs",
read_md_references(attributes["workspace"],
"import-object-md-references.jsonl"))
attributes["filelist"] = attributes.get(
"filelist", get_objectlist(attributes["object_refs"]))
attributes["all_amd_refs"] = attributes.get(
"all_amd_refs", read_all_amd_references(attributes["workspace"]))
attributes["all_dmd_refs"] = attributes.get(
"all_dmd_refs",
read_md_references(attributes["workspace"],
"import-description-md-references.jsonl"))
return attributes
def add_file_to_filesec(all_amd_refs, object_refs, path, filegrp):
"""Add file element to fileGrp element given as parameter.
:all_amd_refs: XML element tree of administrative metadata references
:object_refs: XML tree of object references
:path: url encoded path of the file
:filegrp: fileGrp element
:returns: unique identifier of file element
"""
fileid = '_{}'.format(uuid4())
# Create list of IDs of amdID elements
amdids = get_md_references(all_amd_refs, path=path)
# Create XML element and add it to fileGrp
file_el = mets.file_elem(fileid,
admid_elements=set(amdids),
loctype='URL',
xlink_href='file://%s' %
encode_path(path, safe='/'),
xlink_type='simple',
groupid=None)
streams = get_objectlist(object_refs, path)
if streams:
for stream in streams:
stream_ids = get_md_references(all_amd_refs,
path=path,
stream=stream)
stream_el = mets.stream(admid_elements=stream_ids)
file_el.append(stream_el)
filegrp.append(file_el)
return fileid
def copy_objects(workspace, data_dir):
"""Copy digital objects to workspace
"""
files = get_objectlist(workspace)
for source in files:
target = os.path.join(workspace, source)
if not os.path.exists(os.path.dirname(target)):
os.makedirs(os.path.dirname(target))
copyfile(os.path.join(data_dir, source), target)
def copy_objects(workspace, data_dir):
"""
Copy digital objects to workspace.
:workspace: Workspace path
:data_dir: Path to digital objects
"""
files = get_objectlist(read_md_references(
workspace, "import-object-md-references.jsonl"
))
for source in files:
target = os.path.join(workspace, source)
if not os.path.exists(os.path.dirname(target)):
os.makedirs(os.path.dirname(target))
copyfile(os.path.join(data_dir, source), target)
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))