def add_file_metadata(manifest,
source_url=None,
local_path=None,
media_type=None,
retrieved_on=None,
retrieved_by=None,
created_on=None,
created_by=None,
authored_on=None,
authored_by=None,
conforms_to=None,
bundled_as=None,
update_existing=False):
check_input(manifest)
if not source_url and not local_path:
raise ValueError(
"Error while adding file metadata to RO manifest. "
"At least one of the parameters \"source_url\" or \"local_path\" must be specified"
)
path = source_url
if local_path:
path = local_path
elif bundled_as:
filename = bundled_as.get("filename")
path = filename
if not conforms_to:
file_ext = os.path.splitext(path)[1][1:]
file_ext = file_ext.lstrip(".") if file_ext else None
conforms_to = FILETYPE_ONTOLOGY_MAP.get(file_ext, None)
if not media_type:
media_type = guess_mime_type(path)
uri = source_url = escape_uri(source_url)
retrieved_from = None
if local_path:
uri = escape_uri(ensure_payload_path_prefix(local_path))
if source_url:
retrieved_from = dict(retrievedFrom=source_url)
add_provenance(add_aggregate(manifest,
uri=uri,
mediatype=media_type,
conforms_to=conforms_to,
bundled_as=bundled_as,
update_existing=update_existing),
retrieved_from=retrieved_from,
retrieved_on=retrieved_on,
retrieved_by=retrieved_by,
created_on=created_on,
created_by=created_by,
authored_on=authored_on,
authored_by=authored_by)
def _make_fetch_file(path, remote_entries):
fetch_file_path = os.path.join(path, "fetch.txt")
if not remote_entries:
if os.path.isfile(fetch_file_path):
os.remove(fetch_file_path)
return
LOGGER.info('Writing fetch.txt')
with open_text_file(fetch_file_path, 'w') as fetch_file:
for filename in sorted(remote_entries.keys()):
fetch_file.write("%s\t%s\t%s\n" %
(escape_uri(remote_entries[filename]['url']),
remote_entries[filename]['length'],
escape_uri(_denormalize_filename(filename))))