def test_put(client, db, bucket, permissions, multipart, multipart_url,
get_md5, get_json):
"""Test part upload."""
cases = [
(None, 404),
('auth', 404),
('objects', 404), # TODO - use 403 instead
('bucket', 200),
('location', 200),
]
for user, expected in cases:
login_user(client, permissions[user])
data = b'a' * multipart.chunk_size
res = client.put(
multipart_url + '&partNumber={0}'.format(1),
input_stream=BytesIO(data),
)
assert res.status_code == expected
if res.status_code == 200:
assert res.get_etag()[0] == get_md5(data)
# Assert content
with open(multipart.file.uri, 'rb') as fp:
fp.seek(multipart.chunk_size)
content = fp.read(multipart.chunk_size)
assert content == data
assert Part.count(multipart) == 1
assert Part.get_or_none(multipart, 1).checksum == get_md5(data)
def test_post_by_reference_segmented(api, users, location, task_delay):
with api.test_request_context(), api.test_client() as client:
# Assemble a segmented upload from parts, and complete it
segmented_upload_record: SegmentedUploadRecord = SegmentedUploadRecord.create(
{}
)
multipart_object = MultipartObject.create(
bucket=segmented_upload_record.bucket,
key="some-key",
size=15,
chunk_size=10,
)
Part.create(multipart_object, 0, stream=io.BytesIO(b"abcdefghij"))
Part.create(multipart_object, 1, stream=io.BytesIO(b"klmno"))
multipart_object.complete()
login(client)
ttl = (
datetime.datetime.now(tz=datetime.timezone.utc)
+ datetime.timedelta(0, 3600)
).isoformat()
response = client.post(
"/sword/service-document",
data=json.dumps(
{
"@context": JSON_LD_CONTEXT,
"@type": "ByReference",
"byReferenceFiles": [
{
"@id": f"http://localhost/sword/staging/{segmented_upload_record.id}",
"contentDisposition": "attachment; filename=some-resource.json",
"contentType": "application/json",
"dereference": True,
"ttl": ttl,
}
],
}
),
headers={
"Content-Disposition": "attachment; by-reference=true",
"Content-Type": "application/ld+json",
},
)
assert response.status_code == HTTPStatus.CREATED
object_version = ObjectVersion.query.one()
tags = TagManager(object_version)
assert tags == {
ObjectTagKey.Packaging: "http://purl.org/net/sword/3.0/package/Binary",
ObjectTagKey.ByReferenceTemporaryID: str(segmented_upload_record.id),
ObjectTagKey.FileState: FileState.Pending,
ObjectTagKey.ByReferenceDereference: "true",
ObjectTagKey.ByReferenceNotDeleted: "true",
ObjectTagKey.OriginalDeposit: "true",
ObjectTagKey.ByReferenceTTL: ttl,
}
def test_put(client, db, bucket, permissions, multipart, multipart_url,
get_md5, get_json):
"""Test part upload."""
cases = [
(None, 404),
('auth', 404),
('objects', 404), # TODO - use 403 instead
('bucket', 200),
('location', 200),
]
for user, expected in cases:
login_user(client, permissions[user])
data = b'a' * multipart.chunk_size
res = client.put(
multipart_url + '&partNumber={0}'.format(1),
input_stream=BytesIO(data),
)
assert res.status_code == expected
if res.status_code == 200:
assert res.get_etag()[0] == get_md5(data)
# Assert content
with open(multipart.file.uri, 'rb') as fp:
fp.seek(multipart.chunk_size)
content = fp.read(multipart.chunk_size)
assert content == data
assert Part.count(multipart) == 1
assert Part.get_or_none(multipart, 1).checksum == get_md5(data)
def test_by_reference_sets_tag(api, users, location, task_delay):
with api.test_request_context():
# Assemble a segmented upload from parts, and complete it
segmented_upload_record: SegmentedUploadRecord = SegmentedUploadRecord.create(
{}
)
multipart_object = MultipartObject.create(
bucket=segmented_upload_record.bucket,
key="some-key",
size=15,
chunk_size=10,
)
Part.create(multipart_object, 0, stream=io.BytesIO(b"abcdefghij"))
Part.create(multipart_object, 1, stream=io.BytesIO(b"klmno"))
multipart_object.complete()
record: SWORDDeposit = SWORDDeposit.create({})
record.set_by_reference_files(
[
ByReferenceFileDefinition(
temporary_id=segmented_upload_record.id,
content_disposition="attachment; filename=something.txt",
content_type="text/plain",
packaging=PackagingFormat.Binary,
dereference=True,
),
],
lambda *args: True,
"http://localhost/",
)
object_version = ObjectVersion.query.one()
tags = TagManager(object_version)
assert tags == {
ObjectTagKey.OriginalDeposit: "true",
ObjectTagKey.ByReferenceTemporaryID: str(segmented_upload_record.id),
ObjectTagKey.Packaging: "http://purl.org/net/sword/3.0/package/Binary",
ObjectTagKey.FileState: FileState.Pending,
ObjectTagKey.ByReferenceDereference: "true",
ObjectTagKey.ByReferenceNotDeleted: "true",
}
tasks.dereference_object(record.id, object_version.version_id)
assert object_version.file.storage().open().read() == b"abcdefghijklmno"
def test_multipart_full(app, db, bucket):
"""Test full multipart object."""
app.config.update(
dict(
FILES_REST_MULTIPART_CHUNKSIZE_MIN=5 * 1024 * 1024,
FILES_REST_MULTIPART_CHUNKSIZE_MAX=5 * 1024 * 1024 * 1024,
))
# Initial parameters
chunks = 20
chunk_size = 5 * 1024 * 1024 # 5 MiB
last_chunk = 1024 * 1024 # 1 MiB
size = (chunks - 1) * chunk_size + last_chunk
# Initiate
mp = MultipartObject.create(bucket,
'testfile',
size=size,
chunk_size=chunk_size)
db.session.commit()
# Create parts
for i in range(chunks):
part_size = chunk_size if i < chunks - 1 else last_chunk
Part.create(mp, i, stream=make_stream(part_size))
db.session.commit()
# Complete
mp.complete()
db.session.commit()
# Merge parts.
pre_size = mp.bucket.size
mp.merge_parts()
db.session.commit()
# Test size update
bucket = Bucket.get(bucket.id)
assert bucket.size == pre_size
app.config.update(
dict(
FILES_REST_MULTIPART_CHUNKSIZE_MIN=2,
FILES_REST_MULTIPART_CHUNKSIZE_MAX=20,
))
def test_multipart_full(app, db, bucket):
"""Test full multipart object."""
app.config.update(dict(
FILES_REST_MULTIPART_CHUNKSIZE_MIN=5 * 1024 * 1024,
FILES_REST_MULTIPART_CHUNKSIZE_MAX=5 * 1024 * 1024 * 1024,
))
# Initial parameters
chunks = 20
chunk_size = 5 * 1024 * 1024 # 5 MiB
last_chunk = 1024 * 1024 # 1 MiB
size = (chunks - 1) * chunk_size + last_chunk
# Initiate
mp = MultipartObject.create(
bucket, 'testfile', size=size, chunk_size=chunk_size)
db.session.commit()
# Create parts
for i in range(chunks):
part_size = chunk_size if i < chunks - 1 else last_chunk
Part.create(mp, i, stream=make_stream(part_size))
db.session.commit()
# Complete
mp.complete()
db.session.commit()
# Merge parts.
pre_size = mp.bucket.size
mp.merge_parts()
db.session.commit()
# Test size update
bucket = Bucket.get(bucket.id)
assert bucket.size == pre_size
app.config.update(dict(
FILES_REST_MULTIPART_CHUNKSIZE_MIN=2,
FILES_REST_MULTIPART_CHUNKSIZE_MAX=20,
))
def test_get(client, db, bucket, permissions, multipart, multipart_url,
get_json):
"""Test get parts."""
Part.create(multipart, 0)
Part.create(multipart, 1)
Part.create(multipart, 3)
db.session.commit()
cases = [
(None, 404),
('auth', 404),
('objects', 404),
('bucket', 200),
('location', 200),
]
for user, expected in cases:
login_user(client, permissions[user])
res = client.get(multipart_url)
assert res.status_code == expected
if res.status_code == 200:
data = get_json(res)
assert len(data['parts']) == 3
def test_get(client, db, bucket, permissions, multipart, multipart_url,
get_json):
"""Test get parts."""
Part.create(multipart, 0)
Part.create(multipart, 1)
Part.create(multipart, 3)
db.session.commit()
cases = [
(None, 404),
('auth', 404),
('objects', 404),
('bucket', 200),
('location', 200),
]
for user, expected in cases:
login_user(client, permissions[user])
res = client.get(multipart_url)
assert res.status_code == expected
if res.status_code == 200:
data = get_json(res)
assert len(data['parts']) == 3
def parts(db, multipart):
"""All parts for a multipart object."""
items = []
for i in range(multipart.last_part_number + 1):
chunk_size = multipart.chunk_size \
if not i == multipart.last_part_number \
else multipart.last_part_size
p = Part.create(multipart,
i,
stream=BytesIO(u'{0}'.format(i).encode('ascii') *
chunk_size))
items.append(p)
db.session.commit()
return items
def parts(db, multipart):
"""All parts for a multipart object."""
items = []
for i in range(multipart.last_part_number + 1):
chunk_size = multipart.chunk_size \
if not i == multipart.last_part_number \
else multipart.last_part_size
p = Part.create(
multipart,
i,
stream=BytesIO(u'{0}'.format(i).encode('ascii') * chunk_size)
)
items.append(p)
db.session.commit()
return items
def test_part_creation(app, db, bucket, get_sha256):
"""Test part creation."""
assert bucket.size == 0
mp = MultipartObject.create(bucket, 'test.txt', 5, 2)
db.session.commit()
assert bucket.size == 5
Part.create(mp, 2, stream=BytesIO(b'p'))
Part.create(mp, 0, stream=BytesIO(b'p1'))
Part.create(mp, 1, stream=BytesIO(b'p2'))
db.session.commit()
assert bucket.size == 5
mp.complete()
db.session.commit()
assert bucket.size == 5
# Assert checksum of part.
m = hashlib.sha256()
m.update(b'p2')
assert "sha256:{0}".format(m.hexdigest()) == \
Part.get_or_none(mp, 1).checksum
obj = mp.merge_parts()
db.session.commit()
assert bucket.size == 5
assert MultipartObject.query.count() == 0
assert Part.query.count() == 0
assert obj.file.size == 5
assert obj.file.checksum == get_sha256(b'p1p2p')
assert obj.file.storage().open().read() == b'p1p2p'
assert obj.file.writable is False
assert obj.file.readable is True
assert obj.version_id == ObjectVersion.get(bucket, 'test.txt').version_id
def test_part_creation(app, db, bucket, get_md5):
"""Test part creation."""
assert bucket.size == 0
mp = MultipartObject.create(bucket, 'test.txt', 5, 2)
db.session.commit()
assert bucket.size == 5
Part.create(mp, 2, stream=BytesIO(b'p'))
Part.create(mp, 0, stream=BytesIO(b'p1'))
Part.create(mp, 1, stream=BytesIO(b'p2'))
db.session.commit()
assert bucket.size == 5
mp.complete()
db.session.commit()
assert bucket.size == 5
# Assert checksum of part.
m = hashlib.md5()
m.update(b'p2')
assert "md5:{0}".format(m.hexdigest()) == Part.get_or_none(mp, 1).checksum
obj = mp.merge_parts()
db.session.commit()
assert bucket.size == 5
assert MultipartObject.query.count() == 0
assert Part.query.count() == 0
assert obj.file.size == 5
assert obj.file.checksum == get_md5(b'p1p2p')
assert obj.file.storage().open().read() == b'p1p2p'
assert obj.file.writable is False
assert obj.file.readable is True
assert obj.version_id == ObjectVersion.get(bucket, 'test.txt').version_id