def test_ambiguous_mets_file_is_error(tmpdir):
for i in range(3):
touch(tmpdir / "metadata" / ("METS.%d.xml" % i))
with pytest.raises(
Exception, match="Multiple METS files found in %s/metadata" % tmpdir
):
find_mets_file(str(tmpdir))
def parse_reingest_mets(job, transfer_uuid, transfer_path):
# Parse METS to extract information needed by later microservices
mets_path = find_mets_file(transfer_path)
try:
root = etree.parse(mets_path)
except Exception:
job.pyprint("Error parsing reingest METS", mets_path, " - skipping")
logger.info("Error parsing reingest mets %s - skipping",
mets_path,
exc_info=True)
return
# Get SIP UUID from METS name
sip_uuid = os.path.basename(mets_path).replace("METS.",
"").replace(".xml", "")
# Note: Because DublinCore and PREMIS rights are not database-level foreign keys, this works even though the SIP may not exist yet
parse_mets_to_db.parse_dc(job, sip_uuid, root)
parse_mets_to_db.parse_rights(job, sip_uuid, root)
def test_no_mets_file_is_error(tmpdir):
with pytest.raises(Exception, match="No METS file found in %s/metadata" % tmpdir):
find_mets_file(str(tmpdir))
def test_finds_matching_mets_file(tmpdir):
touch(tmpdir / "metadata" / "METS.1.xml")
assert find_mets_file(str(tmpdir)) == str(tmpdir / "metadata" / "METS.1.xml")