def test_post_deposit_binary_upload_fail_if_content_length_missing(
authenticated_client, deposit_collection, sample_archive, tmp_path):
"""The Content-Length header is mandatory"""
tmp_path = str(tmp_path)
url = reverse(COL_IRI, args=[deposit_collection.name])
archive = create_arborescence_archive(tmp_path,
"archive2",
"file2",
b"some content in file",
up_to_size=500)
external_id = "some-external-id"
# when
response = post_archive(
authenticated_client,
url,
archive,
CONTENT_LENGTH=None,
HTTP_SLUG=external_id,
HTTP_IN_PROGRESS="false",
)
# then
assert response.status_code == status.HTTP_400_BAD_REQUEST
assert b"the CONTENT_LENGTH header must be sent." in response.content
with pytest.raises(Deposit.DoesNotExist):
Deposit.objects.get(external_id=external_id)
def test_post_deposit_binary_upload_fail_if_upload_size_limit_exceeded(
authenticated_client, deposit_collection, sample_archive, tmp_path):
"""Binary upload must not exceed the limit set up..."""
tmp_path = str(tmp_path)
url = reverse(COL_IRI, args=[deposit_collection.name])
archive = create_arborescence_archive(tmp_path,
"archive2",
"file2",
b"some content in file",
up_to_size=5000)
external_id = "some-external-id"
# when
response = post_archive(
authenticated_client,
url,
archive,
HTTP_SLUG=external_id,
HTTP_IN_PROGRESS="false",
)
# then
assert response.status_code == status.HTTP_413_REQUEST_ENTITY_TOO_LARGE
assert b"Upload size limit exceeded" in response.content
with pytest.raises(Deposit.DoesNotExist):
Deposit.objects.get(external_id=external_id)
def sample_archive(tmp_path):
"""Returns a sample archive"""
tmp_path = str(tmp_path) # pytest version limitation in previous version
archive = create_arborescence_archive(
tmp_path, "archive1", "file1", b"some content in file"
)
return archive
def test_replace_archive_to_deposit_is_possible(
tmp_path,
partial_deposit,
deposit_collection,
authenticated_client,
sample_archive,
atom_dataset,
):
"""Replace all archive with another one should return a 204 response"""
tmp_path = str(tmp_path)
# given
deposit = partial_deposit
requests = DepositRequest.objects.filter(deposit=deposit, type="archive")
assert len(list(requests)) == 1
check_archive(sample_archive["name"], requests[0].archive.name)
# we have no metadata for that deposit
requests = list(
DepositRequest.objects.filter(deposit=deposit, type="metadata"))
assert len(requests) == 0
response = post_atom(
authenticated_client,
reverse(SE_IRI, args=[deposit_collection.name, deposit.id]),
data=atom_dataset["entry-data1"],
HTTP_SLUG=deposit.external_id,
HTTP_IN_PROGRESS=True,
)
requests = list(
DepositRequest.objects.filter(deposit=deposit, type="metadata"))
assert len(requests) == 1
update_uri = reverse(EM_IRI, args=[deposit_collection.name, deposit.id])
external_id = "some-external-id-1"
archive2 = create_arborescence_archive(tmp_path, "archive2", "file2",
b"some other content in file")
response = put_archive(
authenticated_client,
update_uri,
archive2,
HTTP_SLUG=external_id,
HTTP_IN_PROGRESS="false",
)
assert response.status_code == status.HTTP_204_NO_CONTENT
requests = DepositRequest.objects.filter(deposit=deposit, type="archive")
assert len(list(requests)) == 1
check_archive(archive2["name"], requests[0].archive.name)
# check we did not touch the other parts
requests = list(
DepositRequest.objects.filter(deposit=deposit, type="metadata"))
assert len(requests) == 1
def test_add_archive_to_deposit_is_possible(
tmp_path,
authenticated_client,
deposit_collection,
partial_deposit_with_metadata,
sample_archive,
):
"""Add another archive to a deposit return a 201 response"""
tmp_path = str(tmp_path)
deposit = partial_deposit_with_metadata
requests = DepositRequest.objects.filter(deposit=deposit, type="archive")
assert len(requests) == 1
check_archive(sample_archive["name"], requests[0].archive.name)
requests_meta0 = DepositRequest.objects.filter(deposit=deposit,
type="metadata")
assert len(requests_meta0) == 1
update_uri = reverse(EM_IRI, args=[deposit_collection.name, deposit.id])
external_id = "some-external-id-1"
archive2 = create_arborescence_archive(tmp_path, "archive2", "file2",
b"some other content in file")
response = post_archive(
authenticated_client,
update_uri,
archive2,
HTTP_SLUG=external_id,
HTTP_IN_PROGRESS="false",
)
assert response.status_code == status.HTTP_201_CREATED
requests = DepositRequest.objects.filter(deposit=deposit,
type="archive").order_by("id")
assert len(requests) == 2
# first archive still exists
check_archive(sample_archive["name"], requests[0].archive.name)
# a new one was added
check_archive(archive2["name"], requests[1].archive.name)
# check we did not touch the other parts
requests_meta1 = DepositRequest.objects.filter(deposit=deposit,
type="metadata")
assert len(requests_meta1) == 1
assert set(requests_meta0) == set(requests_meta1)
def test_access_to_existing_deposit_with_multiple_archives(
tmp_path, authenticated_client, deposit_collection, partial_deposit,
sample_archive):
"""Access to deposit should stream a 200 response with its raw contents"""
deposit = partial_deposit
archive2 = create_arborescence_archive(tmp_path, "archive2", "file2",
b"some other content in file")
# Add a second archive to deposit
update_uri = reverse(EM_IRI, args=[deposit_collection.name, deposit.id])
response = authenticated_client.post(
update_uri,
content_type="application/zip", # as zip
data=archive2["data"],
# + headers
CONTENT_LENGTH=archive2["length"],
HTTP_SLUG=deposit.external_id,
HTTP_CONTENT_MD5=archive2["md5sum"],
HTTP_PACKAGING="http://purl.org/net/sword/package/SimpleZip",
HTTP_IN_PROGRESS="false",
HTTP_CONTENT_DISPOSITION="attachment; filename=%s" %
(archive2["name"], ),
)
assert response.status_code == status.HTTP_201_CREATED
for i, url in enumerate(
private_get_raw_url_endpoints(deposit_collection, deposit)):
response = authenticated_client.get(url)
assert response.status_code == status.HTTP_200_OK
assert response["content-type"] == "application/tar"
# write the response stream in a temporary archive
archive_path = join(tmp_path, f"archive_{i}.tar")
with open(archive_path, "wb") as f:
for chunk in response.streaming_content:
f.write(chunk)
# to check its properties are correct
tfile = tarfile.open(archive_path)
assert set(tfile.getnames()) == {".", "./file1", "./file2"}
assert tfile.extractfile("./file1").read() == b"some content in file"
assert tfile.extractfile(
"./file2").read() == b"some other content in file"
def create_deposit_archive_with_archive(
root_path, archive_extension, client, collection_name
):
# we create the holding archive to a given extension
archive = create_arborescence_archive(
root_path,
"archive1",
"file1",
b"some content in file",
extension=archive_extension,
)
# now we create an archive holding the first created archive
invalid_archive = create_archive_with_archive(root_path, "invalid.tgz", archive)
# we deposit it
response = client.post(
reverse(COL_IRI, args=[collection_name]),
content_type="application/x-tar",
data=invalid_archive["data"],
CONTENT_LENGTH=invalid_archive["length"],
HTTP_MD5SUM=invalid_archive["md5sum"],
HTTP_SLUG="external-id",
HTTP_IN_PROGRESS=False,
HTTP_CONTENT_DISPOSITION="attachment; filename=%s" % (invalid_archive["name"],),
)
# then
assert response.status_code == status.HTTP_201_CREATED
response_content = parse_xml(response.content)
deposit_status = response_content.findtext(
"swh:deposit_status", namespaces=NAMESPACES
)
assert deposit_status == DEPOSIT_STATUS_DEPOSITED
deposit_id = int(response_content.findtext("swh:deposit_id", namespaces=NAMESPACES))
deposit = Deposit.objects.get(pk=deposit_id)
assert DEPOSIT_STATUS_DEPOSITED == deposit.status
return deposit
def test_post_deposit_binary_and_post_to_add_another_archive(
authenticated_client, deposit_collection, sample_archive, tmp_path):
"""Updating a deposit should return a 201 with receipt"""
tmp_path = str(tmp_path)
url = reverse(COL_IRI, args=[deposit_collection.name])
external_id = "some-external-id-1"
# when
response = post_archive(
authenticated_client,
url,
sample_archive,
HTTP_SLUG=external_id,
HTTP_IN_PROGRESS="true",
)
# then
assert response.status_code == status.HTTP_201_CREATED
response_content = parse_xml(response.content)
deposit_id = response_content.findtext("swh:deposit_id",
namespaces=NAMESPACES)
deposit = Deposit.objects.get(pk=deposit_id)
assert deposit.status == "partial"
assert deposit.external_id == external_id
assert deposit.collection == deposit_collection
assert deposit.swhid is None
deposit_request = DepositRequest.objects.get(deposit=deposit)
assert deposit_request.deposit == deposit
assert deposit_request.type == "archive"
check_archive(sample_archive["name"], deposit_request.archive.name)
# 2nd archive to upload
archive2 = create_arborescence_archive(tmp_path, "archive2", "file2",
b"some other content in file")
# uri to update the content
update_uri = reverse(EM_IRI, args=[deposit_collection.name, deposit_id])
# adding another archive for the deposit and finalizing it
response = post_archive(
authenticated_client,
update_uri,
archive2,
HTTP_SLUG=external_id,
)
assert response.status_code == status.HTTP_201_CREATED
response_content = parse_xml(response.content)
deposit = Deposit.objects.get(pk=deposit_id)
assert deposit.status == DEPOSIT_STATUS_DEPOSITED
assert deposit.external_id == external_id
assert deposit.collection == deposit_collection
assert deposit.swhid is None
deposit_requests = list(
DepositRequest.objects.filter(deposit=deposit).order_by("id"))
# 2 deposit requests for the same deposit
assert len(deposit_requests) == 2
assert deposit_requests[0].deposit == deposit
assert deposit_requests[0].type == "archive"
check_archive(sample_archive["name"], deposit_requests[0].archive.name)
assert deposit_requests[1].deposit == deposit
assert deposit_requests[1].type == "archive"
check_archive(archive2["name"], deposit_requests[1].archive.name)
# only 1 deposit in db
deposits = Deposit.objects.all()
assert len(deposits) == 1
def test_post_deposit_then_update_refused(authenticated_client,
deposit_collection, sample_archive,
atom_dataset, tmp_path):
"""Updating a deposit with status 'ready' should return a 400"""
tmp_path = str(tmp_path)
url = reverse(COL_IRI, args=[deposit_collection.name])
external_id = "some-external-id-1"
# when
response = post_archive(
authenticated_client,
url,
sample_archive,
HTTP_SLUG=external_id,
HTTP_IN_PROGRESS="false",
)
# then
assert response.status_code == status.HTTP_201_CREATED
response_content = parse_xml(response.content)
deposit_id = response_content.findtext("swh:deposit_id",
namespaces=NAMESPACES)
deposit = Deposit.objects.get(pk=deposit_id)
assert deposit.status == DEPOSIT_STATUS_DEPOSITED
assert deposit.external_id == external_id
assert deposit.collection == deposit_collection
assert deposit.swhid is None
deposit_request = DepositRequest.objects.get(deposit=deposit)
assert deposit_request.deposit == deposit
check_archive(sample_archive["name"], deposit_request.archive.name)
# updating/adding is forbidden
# uri to update the content
edit_iri = reverse("edit_iri", args=[deposit_collection.name, deposit_id])
se_iri = reverse("se_iri", args=[deposit_collection.name, deposit_id])
em_iri = reverse("em_iri", args=[deposit_collection.name, deposit_id])
# Testing all update/add endpoint should fail
# since the status is ready
archive2 = create_arborescence_archive(tmp_path, "archive2", "file2",
b"some content in file 2")
# replacing file is no longer possible since the deposit's
# status is ready
r = put_archive(
authenticated_client,
em_iri,
archive2,
HTTP_SLUG=external_id,
HTTP_IN_PROGRESS="false",
)
assert r.status_code == status.HTTP_400_BAD_REQUEST
assert (ET.fromstring(r.content).findtext("atom:summary",
namespaces=NAMESPACES) ==
"You can only act on deposit with status 'partial'")
# adding file is no longer possible since the deposit's status
# is ready
r = post_archive(
authenticated_client,
em_iri,
archive2,
HTTP_SLUG=external_id,
HTTP_IN_PROGRESS="false",
)
assert r.status_code == status.HTTP_400_BAD_REQUEST
assert (ET.fromstring(r.content).findtext("atom:summary",
namespaces=NAMESPACES) ==
"You can only act on deposit with status 'partial'")
# replacing metadata is no longer possible since the deposit's
# status is ready
r = put_atom(
authenticated_client,
edit_iri,
data=atom_dataset["entry-data-deposit-binary"],
CONTENT_LENGTH=len(atom_dataset["entry-data-deposit-binary"]),
HTTP_SLUG=external_id,
)
assert r.status_code == status.HTTP_400_BAD_REQUEST
assert (ET.fromstring(r.content).findtext("atom:summary",
namespaces=NAMESPACES) ==
"You can only act on deposit with status 'partial'")
# adding new metadata is no longer possible since the
# deposit's status is ready
r = post_atom(
authenticated_client,
se_iri,
data=atom_dataset["entry-data-deposit-binary"],
CONTENT_LENGTH=len(atom_dataset["entry-data-deposit-binary"]),
HTTP_SLUG=external_id,
)
assert r.status_code == status.HTTP_400_BAD_REQUEST
assert (ET.fromstring(r.content).findtext("atom:summary",
namespaces=NAMESPACES) ==
"You can only act on deposit with status 'partial'")
archive_content = b"some content representing archive"
archive = InMemoryUploadedFile(
BytesIO(archive_content),
field_name="archive0",
name="archive0",
content_type="application/zip",
size=len(archive_content),
charset=None,
)
atom_entry = InMemoryUploadedFile(
BytesIO(atom_dataset["entry-data-deposit-binary"].encode("utf-8")),
field_name="atom0",
name="atom0",
content_type='application/atom+xml; charset="utf-8"',
size=len(atom_dataset["entry-data-deposit-binary"]),
charset="utf-8",
)
# replacing multipart metadata is no longer possible since the
# deposit's status is ready
r = authenticated_client.put(
edit_iri,
format="multipart",
data={
"archive": archive,
"atom_entry": atom_entry,
},
)
assert r.status_code == status.HTTP_400_BAD_REQUEST
assert (ET.fromstring(r.content).findtext("atom:summary",
namespaces=NAMESPACES) ==
"You can only act on deposit with status 'partial'")
# adding new metadata is no longer possible since the
# deposit's status is ready
r = authenticated_client.post(
se_iri,
format="multipart",
data={
"archive": archive,
"atom_entry": atom_entry,
},
)
assert r.status_code == status.HTTP_400_BAD_REQUEST
assert (ET.fromstring(r.content).findtext("atom:summary",
namespaces=NAMESPACES) ==
"You can only act on deposit with status 'partial'")