def test_10_Multipart_POST_then_update_on_EM_IRI(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER,
user_name="sword",
user_pass="******",
download_service_document=True)
e = Entry(title="Foo",
id="asidjasidj",
dcterms_appendix="blah blah",
dcterms_title="foo bar")
deposit_receipt = conn.create(
payload="Multipart_POST_then_update_on_EM_IRI",
metadata_entry=e,
mimetype="text/plain",
filename="readme.txt",
packaging='http://purl.org/net/sword/package/Binary',
col_iri=conn.workspaces[0][1][0].href,
in_progress=True)
assert deposit_receipt.edit_media != None
dr = conn.update(
payload="Multipart_POST_then_update_on_EM_IRI -- updated resource",
mimetype="text/plain",
filename="readthis.txt",
packaging="http://purl.org/net/sword/package/Binary",
edit_media_iri=deposit_receipt.edit_media)
assert dr.code == 204 # empty response
def test_20_Multipart_POST_to_se_iri(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER,
user_name="sword",
user_pass="******",
download_service_document=True)
e = Entry(title="Foo",
id="asidjasidj",
dcterms_appendix="blah blah",
dcterms_title="foo bar")
deposit_receipt = conn.create(
payload="Multipart_POST_then_update_on_EM_IRI",
metadata_entry=e,
mimetype="text/plain",
filename="readme.txt",
packaging='http://purl.org/net/sword/package/Binary',
col_iri=conn.sd.workspaces[0][1][0].href,
in_progress=True)
assert deposit_receipt.se_iri != None
e.add_fields(dcterms_identifier="doi://multipart_update_to_SE_IRI")
dr = conn.append(
se_iri=deposit_receipt.se_iri,
payload=
"Multipart_POST_then_appending_file_on_SE_IRI -- updated resource",
mimetype="text/plain",
filename="readthisextrafile.txt",
packaging="http://purl.org/net/sword/package/Binary",
metadata_entry=e)
print dr.code
assert dr.code == 201
def test_15_Metadata_POST_to_sss_w_coliri(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
e = Entry(title="Foo", id="asidjasidj", dcterms_appendix="blah blah", dcterms_title="foo bar")
dr = conn.create(metadata_entry = e,
col_iri = conn.sd.workspaces[0][1][0].href,
in_progress=True)
assert dr.code == 201
def preflight_submission(self):
opener = self.get_opener()
conn = Connection(self.sword2_sd_url,
error_response_raises_exceptions=False,
http_impl=UrlLib2Layer(opener))
logger.debug("Retrieving the service document")
conn.get_service_document()
logger.debug("Retrieved the service document")
self.assertIsNotNone(conn.sd)
self.assertIsNotNone(conn.sd.workspaces)
self.assertNotEqual(len(conn.sd.workspaces), 0)
workspace = conn.sd.workspaces[0][1]
# we require there to be at least one collection
self.assertNotEqual(len(workspace), 0)
col = workspace[0]
testid = "testid_" + str(uuid.uuid4())
logger.debug("col iri = " + str(col.href))
e = Entry(id=testid,
title="test title",
dcterms_abstract="test description")
print str(e)
receipt = conn.create(col_iri=col.href,
metadata_entry=e,
suggested_identifier=testid)
#col.href=http://192.168.2.237/swordv2/silo/test-silo
self.assertIsNotNone(receipt)
self.assertEquals(receipt.code, 201)
return receipt.location
def test_04_advanced_create_resource_with_package(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
on_behalf_of=SSS_OBO,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip',
in_progress=True,
suggested_identifier="abcdefghijklmnop")
assert receipt.code == 201
assert receipt.location != None
# these last two assertions are contingent on if we actually get a
# receipt back from the server (which we might not legitimately get)
assert receipt.dom is None or receipt.parsed == True
assert receipt.dom is None or receipt.valid == True
def test_41_get_service_document_unauthorised(self):
conn = Connection(SSS_URL,
user_name="alsdkfjsdz",
user_pass="******",
http_impl=http)
conn.get_service_document()
assert conn.sd is None
def test_05_basic_create_resource_with_multipart(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="Foo",
id="asidjasidj",
dcterms_abstract="abstract",
dcterms_title="my title")
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
metadata_entry=e,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
assert receipt.code == 201
assert receipt.location != None
# these last two assertions are contingent on if we actually get a
# receipt back from the server (which we might not legitimately get)
assert receipt.dom is None or receipt.parsed == True
assert receipt.dom is None or receipt.valid == True
def __init__(self, owner):
raise NotImplementedError
c = Connection(SD_URI, user_name = owner.username, user_pass=owner.password)
c.get_service_document()
# pick the first collection within the first workspace:
workspace_1_title, workspace_1_collections = c.workspaces[0]
collection = workspace_1_collections[0]
# upload "package.zip" to this collection as a new (binary) resource:
with open("package.zip", "r") as pkg:
receipt = c.create(col_iri = collection.href,
payload = pkg,
mimetype = "application/zip",
filename = "package.zip",
packaging = 'http://purl.org/net/sword/package/Binary',
in_progress = True) # As the deposit isn't yet finished
# Add a metadata record to this newly created resource (or 'container')
from sword2 import Entry
# Entry can be passed keyword parameters to add metadata to the entry (namespace + '_' + tagname)
e = Entry(id="atomid",
title="atom-title",
dcterms_abstract = "Info about the resource....",
)
# to add a new namespace:
e.register_namespace('skos', 'http://www.w3.org/2004/02/skos/core#')
e.add_field("skos_Concept", "...")
# Update the metadata entry to the resource:
updated_receipt = c.update(metadata_entry = e,
dr = receipt, # use the receipt to discover the right URI to use
in_progress = False) # finish the deposit
def preflight_submission(self):
opener = self.get_opener()
conn = Connection(self.sword2_sd_url, error_response_raises_exceptions=False, http_impl=UrlLib2Layer(opener))
logger.debug("Retrieving the service document")
conn.get_service_document()
logger.debug("Retrieved the service document")
self.assertIsNotNone(conn.sd)
self.assertIsNotNone(conn.sd.workspaces)
self.assertNotEqual(len(conn.sd.workspaces),0)
workspace = conn.sd.workspaces[0][1]
# we require there to be at least one collection
self.assertNotEqual(len(workspace),0)
col = workspace[0]
testid = "testid_"+str(uuid.uuid4())
logger.debug("col iri = " + str(col.href))
e = Entry(id=testid, title="test title", dcterms_abstract="test description")
print str(e)
receipt = conn.create(col_iri=col.href, metadata_entry=e, suggested_identifier=testid)
#col.href=http://192.168.2.237/swordv2/silo/test-silo
self.assertIsNotNone(receipt)
self.assertEquals(receipt.code,201)
return receipt.location
def test_08_advanced_create_resource_with_entry(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
on_behalf_of=SSS_OBO)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="An entry only deposit",
id="asidjasidj",
dcterms_abstract="abstract",
dcterms_identifier="http://whatever/")
e.register_namespace("oxds", "http://databank.ox.ac.uk/terms/")
e.add_field("oxds_whatever", "whatever")
receipt = conn.create(col_iri=col.href,
metadata_entry=e,
in_progress=True,
suggested_identifier=str(uuid.uuid4()))
assert receipt.code == 201
assert receipt.location != None
# these last two assertions are contingent on if we actually get a
# receipt back from the server (which we might not legitimately get)
assert receipt.dom is None or receipt.parsed == True
assert receipt.dom is None or receipt.valid == True
def test_24_get_sword_statement(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER,
user_name="sword",
user_pass="******",
download_service_document=True)
e = Entry(title="Foo",
id="asidjasidj",
dcterms_appendix="blah blah",
dcterms_title="foo bar")
deposit_receipt = conn.create(
payload="Multipart_POST_then_update_on_EM_IRI",
metadata_entry=e,
mimetype="text/plain",
filename="readme.txt",
packaging='http://purl.org/net/sword/package/Binary',
col_iri=conn.sd.workspaces[0][1][0].href,
in_progress=True)
ss_iri = None
for item_dict in deposit_receipt.links[
'http://purl.org/net/sword/terms/statement']:
if item_dict.has_key('type') and item_dict.get(
'type', None) == "application/atom+xml;type=feed":
ss_iri = item_dict.get('href')
assert ss_iri != None
ss = conn.get_atom_sword_statement(ss_iri)
assert ss != None
assert ss.entries[0].metadata.get('sword_depositedBy') == 'sword'
def test_12_Metadata_POST_to_sss(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
e = Entry(title="Foo", id="asidjasidj", dcterms_appendix="blah blah", dcterms_title="foo bar")
resp = conn.create(metadata_entry = e,
workspace='Main Site',
collection=conn.sd.workspaces[0][1][0].title,
in_progress=True)
assert resp != None
def complete_submission(self, edit_uri):
opener = self.get_opener()
conn = Connection(self.sword2_sd_url,
error_response_raises_exceptions=False,
http_impl=UrlLib2Layer(opener))
receipt = None
try:
receipt = conn.get_deposit_receipt(edit_uri)
except urllib2.URLError as e:
# The sword2 client does not catch network errors like this one,
# which indicates that the url couldn't be reached at all
# don't do anything about it here - we'll try again in a moment
# and then error out appropriately later
pass
# at this stage we need to ensure that we actually got back a deposit
# receipt
i = 0
while (receipt is None
or receipt.code >= 400) and i < self.retry_limit:
err = None
if receipt is None:
err = "<unable to reach server>"
else:
err = str(receipt.code)
logger.debug(
"Attempt to retrieve Entry Document failed with error " +
str(err) + " ... trying again in " + str(self.retry_delay) +
" seconds")
i += 1
time.sleep(self.retry_delay)
try:
receipt = conn.get_deposit_receipt(edit_uri)
except urllib2.URLError as e:
# The sword2 client does not catch network errors like this one,
# which indicates that the url couldn't be reached at all
# just try again up to the retry_limit
continue
self.assertIsNotNone(receipt)
self.assertEquals(receipt.code, 200)
# if we get to here we can go ahead with the deposit for real
with open(self.zipFileName, "rb") as data:
new_receipt = conn.update(
dr=receipt,
payload=data,
mimetype="application/zip",
filename=self.dataset_identifier + ".zip",
packaging='http://dataflow.ox.ac.uk/package/DataBankBagIt')
self.assertIsNotNone(new_receipt)
self.assertEquals(new_receipt.code, 204)
return
def test_04_init_from_sss_then_get_doc(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******")
assert conn.sd_iri == "http://localhost:%s/sd-uri" % PORT_NUMBER
assert conn.sd == None # Not asked to get sd doc yet
conn.get_service_document()
assert conn.sd != None
assert conn.sd.parsed == True
assert conn.sd.valid == True
assert len(conn.sd.workspaces) == 1
def testServiceDocumentAccess(self):
opener = self.get_opener()
for i in range(10):
conn = Connection(self.sword2_sd_url, error_response_raises_exceptions=False, http_impl=UrlLib2Layer(opener))
conn.get_service_document()
self.assertIsNotNone(conn.sd, "Service document None (loop %d)"%(i))
self.assertIsNotNone(conn.sd.workspaces, "Service document workspace None (loop %d)"%(i))
self.assertNotEqual(len(conn.sd.workspaces),0, "Service document worksoacxe count %d (loop %d)"%(len(conn.sd.workspaces),i))
return
def test_03_init_then_load_from_string_t_history(self):
conn = Connection("http://example.org/service-doc")
assert conn.sd_iri == "http://example.org/service-doc"
assert conn.sd == None
conn.load_service_document(long_service_doc)
# Should have made a two client 'transactions', the init and subsequent XML load
assert len(conn.history) == 2
assert conn.history[0]['type'] == "init"
assert conn.history[1]['type'] == "SD Parse"
def preflight_submission(self, dataset, opener, repository, silo ):
logger.debug("Carrying out pre-flight submission")
# verify that we can get a service document, and that there
# is at least one silo and that we can authenticate
if repository.sword2_sd_url is None:
raise SwordServiceError("No sword2 service-document URL for repository configuration")
# get the service document (for which we must be authenticated)
conn = Connection(repository.sword2_sd_url, error_response_raises_exceptions=False, http_impl=UrlLib2Layer(opener))
conn.get_service_document()
# we require there to be at least one workspace
if conn.sd is None:
raise SwordServiceError("did not successfully retrieve a service document")
if conn.sd.workspaces is None:
raise SwordServiceError("no workspaces defined in service document")
if len(conn.sd.workspaces) == 0:
raise SwordServiceError("no workspaces defined in service document")
workspace = conn.sd.workspaces[0][1]
# we require there to be at least one collection
if len(workspace) == 0:
raise SwordServiceError("no collections defined in workspace")
# FIXME: we don't currently have a mechanism to make decisions about
# which collection to put stuff in, so we just put stuff in the first
# one for the time being
col = workspace[0]
silohref = repository.homepage + "swordv2/silo/" + silo
# assemble the entry ready for deposit, using the basic metadata
# FIXME: is there anything further we need to do about the metadata here?
e = Entry(id=dataset.identifier, title=dataset.title, dcterms_abstract=dataset.description)
# create the item using the metadata-only approach (suppress errors along the way,
# we'll check for them below)
#receipt = conn.create(col_iri=col.href, metadata_entry=e, suggested_identifier=dataset.identifier)
logger.debug( "Deposit is being created" )
receipt = conn.create(col_iri=silohref, metadata_entry=e, suggested_identifier=dataset.identifier)
logger.debug( "Deposit created" )
# check for errors
if receipt.code >= 400:
# this is an error
logger.debug("Received error message from server: " + receipt.to_xml())
if receipt.error_href == "http://databank.ox.ac.uk/errors/DatasetConflict":
raise SwordSlugRejected()
raise SwordDepositError(receipt)
logger.debug("Deposit carried out to: " + receipt.location)
# return receipt.location
return (receipt.alternate,receipt.location)
def test_34_check_metadata_only_state(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="An entry only deposit", id="asidjasidj", dcterms_abstract="abstract", dcterms_identifier="http://whatever/")
receipt = conn.create(col_iri = col.href, metadata_entry = e)
statement = conn.get_ore_sword_statement(receipt.ore_statement_iri)
assert len(statement.states) == 1
assert statement.states[0][0] == "http://databank.ox.ac.uk/state/EmptyContainer"
def test_06_Simple_POST_to_sss(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
resp = conn.create(payload = "Payload is just a load of text",
mimetype = "text/plain",
filename = "readme.txt",
packaging = 'http://purl.org/net/sword/package/Binary',
workspace = 'Main Site',
collection = conn.sd.workspaces[0][1][0].title,
in_progress=True,
metadata_entry=None)
assert resp.code == 201
def test_02_init_then_load_from_string(self):
conn = Connection("http://example.org/service-doc")
assert conn.sd_iri == "http://example.org/service-doc"
assert conn.sd == None
conn.load_service_document(long_service_doc)
assert conn.sd != None
assert len(conn.sd.workspaces) == 2
assert len(conn.workspaces) == 2
assert conn.sd.workspaces[0][0] == "Main Site"
assert conn.sd.workspaces[1][0] == "Sub-site"
assert len(conn.sd.workspaces[1][1]) == 2
def test_02_get_service_document_on_behalf_of(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW, on_behalf_of=SSS_OBO)
conn.get_service_document()
# given that the client is fully functional, testing that the
# service document parses and is valid is sufficient. This, obviously,
# doesn't test the validation routine itself.
assert conn.sd != None
assert conn.sd.parsed == True
assert conn.sd.valid == True
assert len(conn.sd.workspaces) == 1
def test_04_init_from_sss_then_get_doc(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER,
user_name="sword",
user_pass="******")
assert conn.sd_iri == "http://localhost:%s/sd-uri" % PORT_NUMBER
assert conn.sd == None # Not asked to get sd doc yet
conn.get_service_document()
assert conn.sd != None
assert conn.sd.parsed == True
assert conn.sd.valid == True
assert len(conn.sd.workspaces) == 1
def test_01_get_service_document(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW)
conn.get_service_document()
# given that the client is fully functional, testing that the
# service document parses and is valid is sufficient. This, obviously,
# doesn't test the validation routine itself.
assert conn.sd != None
assert conn.sd.parsed == True
assert conn.sd.valid == True
assert len(conn.sd.workspaces) == 1
def test_08_Simple_POST_to_sss_w_coliri(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
e = Entry(title="Foo", id="asidjasidj", dcterms_appendix="blah blah", dcterms_title="foo bar")
resp = conn.create(payload = "Payload is just a load of text",
mimetype = "text/plain",
filename = "readme.txt",
packaging = 'http://purl.org/net/sword/package/Binary',
col_iri = conn.sd.workspaces[0][1][0].href,
in_progress=True,
metadata_entry=None)
assert resp.code == 201
def test_07_Multipart_POST_to_sss(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
e = Entry(title="Foo", id="asidjasidj", dcterms_appendix="blah blah", dcterms_title="foo bar")
resp = conn.create(payload = "Multipart payload here",
metadata_entry = e,
mimetype = "text/plain",
filename = "readme.txt",
packaging = 'http://purl.org/net/sword/package/Binary',
workspace='Main Site',
collection=conn.sd.workspaces[0][1][0].title,
in_progress=True)
assert resp.code == 201
def test_15_Metadata_POST_to_sss_w_coliri(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER,
user_name="sword",
user_pass="******",
download_service_document=True)
e = Entry(title="Foo",
id="asidjasidj",
dcterms_appendix="blah blah",
dcterms_title="foo bar")
dr = conn.create(metadata_entry=e,
col_iri=conn.sd.workspaces[0][1][0].href,
in_progress=True)
assert dr.code == 201
def test_22_Create_deposit_and_delete_deposit(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
e = Entry(title="Foo", id="asidjasidj", dcterms_appendix="blah blah", dcterms_title="foo bar")
deposit_receipt = conn.create(payload = "Multipart_POST_then_update_on_EM_IRI",
metadata_entry = e,
mimetype = "text/plain",
filename = "readme.txt",
packaging = 'http://purl.org/net/sword/package/Binary',
col_iri = conn.sd.workspaces[0][1][0].href,
in_progress=True)
assert deposit_receipt.edit != None
dr = conn.delete(resource_iri = deposit_receipt.edit)
assert dr.code == 204 or dr.code == 200
def complete_submission(self,edit_uri):
opener = self.get_opener()
conn = Connection(self.sword2_sd_url, error_response_raises_exceptions=False, http_impl=UrlLib2Layer(opener))
receipt = None
try:
receipt = conn.get_deposit_receipt(edit_uri)
except urllib2.URLError as e:
# The sword2 client does not catch network errors like this one,
# which indicates that the url couldn't be reached at all
# don't do anything about it here - we'll try again in a moment
# and then error out appropriately later
pass
# at this stage we need to ensure that we actually got back a deposit
# receipt
i = 0
while (receipt is None or receipt.code >= 400) and i < self.retry_limit:
err = None
if receipt is None:
err = "<unable to reach server>"
else:
err = str(receipt.code)
logger.debug("Attempt to retrieve Entry Document failed with error " + str(err) + " ... trying again in " + str(self.retry_delay) + " seconds")
i += 1
time.sleep(self.retry_delay)
try:
receipt = conn.get_deposit_receipt(edit_uri)
except urllib2.URLError as e:
# The sword2 client does not catch network errors like this one,
# which indicates that the url couldn't be reached at all
# just try again up to the retry_limit
continue
self.assertIsNotNone(receipt)
self.assertEquals(receipt.code,200)
# if we get to here we can go ahead with the deposit for real
with open( self.zipFileName, "rb") as data:
new_receipt = conn.update(dr = receipt,
payload=data,
mimetype="application/zip",
filename=self.dataset_identifier + ".zip",
packaging='http://dataflow.ox.ac.uk/package/DataBankBagIt')
self.assertIsNotNone(new_receipt)
self.assertEquals(new_receipt.code,204)
return
def test_31_delete_container(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
on_behalf_of=SSS_OBO,
error_response_raises_exceptions=False,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="Multipart deposit",
id="asidjasidj",
dcterms_abstract="abstract",
dcterms_identifier="http://whatever/")
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
metadata_entry=e,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
# ensure that we have a receipt (the server may not give us one
# by default)
edit_iri = receipt.location
receipt = conn.get_deposit_receipt(edit_iri)
# delete the container
new_receipt = conn.delete_container(dr=receipt)
assert new_receipt.code == 204
assert new_receipt.dom is None
# the next check is that this 404s appropriately now
another_receipt = conn.get_deposit_receipt(edit_iri)
def test_23_Finish_in_progress_deposit(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
e = Entry(title="Foo", id="asidjasidj", dcterms_appendix="blah blah", dcterms_title="foo bar")
deposit_receipt = conn.create(payload = "Multipart_POST_then_update_on_EM_IRI",
metadata_entry = e,
mimetype = "text/plain",
filename = "readme.txt",
packaging = 'http://purl.org/net/sword/package/Binary',
col_iri = conn.sd.workspaces[0][1][0].href,
in_progress=True)
assert deposit_receipt.edit != None
dr = conn.complete_deposit(se_iri = deposit_receipt.se_iri)
print "This will fail until the sss.py SWORD2 server responds properly, rather than with code 201"
assert dr.code == 200
def test_12_Metadata_POST_to_sss(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER,
user_name="sword",
user_pass="******",
download_service_document=True)
e = Entry(title="Foo",
id="asidjasidj",
dcterms_appendix="blah blah",
dcterms_title="foo bar")
resp = conn.create(metadata_entry=e,
workspace='Main Site',
collection=conn.sd.workspaces[0][1][0].title,
in_progress=True)
assert resp != None
def test_07_basic_create_resource_with_entry(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="An entry only deposit", id="asidjasidj", dcterms_abstract="abstract", dcterms_identifier="http://whatever/")
receipt = conn.create(col_iri = col.href,
metadata_entry = e)
assert receipt.code == 201
assert receipt.location != None
# these last two assertions are contingent on if we actually get a
# receipt back from the server (which we might not legitimately get)
assert receipt.dom is None or receipt.parsed == True
assert receipt.dom is None or receipt.valid == True
def test_07_basic_create_resource_with_entry(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="An entry only deposit", id="asidjasidj", dcterms_abstract="abstract", dcterms_identifier="http://whatever/")
receipt = conn.create(col_iri = col.href,
metadata_entry = e)
assert receipt.code == 201
assert receipt.location != None
# these last two assertions are contingent on if we actually get a
# receipt back from the server (which we might not legitimately get)
assert receipt.dom is None or receipt.parsed == True
assert receipt.dom is None or receipt.valid == True
def test_06_Simple_POST_to_sss(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER,
user_name="sword",
user_pass="******",
download_service_document=True)
resp = conn.create(
payload="Payload is just a load of text",
mimetype="text/plain",
filename="readme.txt",
packaging='http://purl.org/net/sword/package/Binary',
workspace='Main Site',
collection=conn.sd.workspaces[0][1][0].title,
in_progress=True,
metadata_entry=None)
assert resp.code == 201
def test_09_basic_retrieve_deposit_receipt(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="An entry only deposit", id="asidjasidj", dcterms_abstract="abstract", dcterms_identifier="http://whatever/")
receipt = conn.create(col_iri = col.href, metadata_entry = e)
# we're going to work with the location
assert receipt.location != None
new_receipt = conn.get_deposit_receipt(receipt.location)
assert new_receipt.code == 200
assert new_receipt.parsed == True
assert new_receipt.valid == True
def test_17_Simple_POST_and_GET(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
col_iri = conn.sd.workspaces[0][1][0].href # pick the first collection
dr = conn.create(payload = "Simple_POST_and_GET",
mimetype = "text/plain",
filename = "readme.txt",
packaging = 'http://purl.org/net/sword/package/Binary',
col_iri = col_iri,
in_progress=True,
metadata_entry=None)
assert dr.code == 201
# Now to GET that resource with no prescribed for packaging
content_object = conn.get_resource(dr.cont_iri)
# Can't guarantee that sss.py won't mangle submissions, so can't validate response at this moment
assert content_object != None
def test_14_Invalid_Packaging_cached_receipt(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True, honour_receipts=True)
col_iri = conn.sd.workspaces[0][1][0].href # pick the first collection
dr = conn.create(payload = "Payload is just a load of text",
mimetype = "text/plain",
filename = "readme.txt",
packaging = 'http://purl.org/net/sword/package/Binary',
col_iri = col_iri,
in_progress=True)
# Now to GET that resource with invalid packaging
try:
content = conn.get_resource(dr.cont_iri, packaging="foofar")
assert 1 == 0 # fail
except PackagingFormatNotAvailable:
# test the 'honour_receipts' flag and cached deposit
pass
def test_18_Metadata_POST_to_se_iri(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
e = Entry(title="Foo", id="asidjasidj", dcterms_appendix="blah blah", dcterms_title="foo bar")
deposit_receipt = conn.create(payload = "Multipart_POST_then_update_on_EM_IRI",
metadata_entry = e,
mimetype = "text/plain",
filename = "readme.txt",
packaging = 'http://purl.org/net/sword/package/Binary',
col_iri = conn.sd.workspaces[0][1][0].href,
in_progress=True)
assert deposit_receipt.se_iri != None
e.add_fields(dcterms_identifier="doi://somerubbish", dcterms_foo="blah blah")
dr = conn.append(se_iri = deposit_receipt.se_iri,
metadata_entry = e,
in_progress=False)
assert dr.code == 200
def testServiceDocumentAccess(self):
opener = self.get_opener()
for i in range(10):
conn = Connection(self.sword2_sd_url,
error_response_raises_exceptions=False,
http_impl=UrlLib2Layer(opener))
conn.get_service_document()
self.assertIsNotNone(conn.sd,
"Service document None (loop %d)" % (i))
self.assertIsNotNone(
conn.sd.workspaces,
"Service document workspace None (loop %d)" % (i))
self.assertNotEqual(
len(conn.sd.workspaces), 0,
"Service document worksoacxe count %d (loop %d)" %
(len(conn.sd.workspaces), i))
return
def _check_dataset(self, dataset_submission):
retry_counter = 0
exception = None
while retry_counter < SwordStatementCheckThread.retry_count:
try:
# logger.info("Checking state of dataset at " + dataset_submission.remote_url)
opener = openers.get_opener(dataset_submission.repository,
dataset_submission.submitting_user)
conn = Connection(error_response_raises_exceptions=False,
http_impl=UrlLib2Layer(opener))
receipt = conn.get_deposit_receipt(
dataset_submission.remote_url)
statement = conn.get_ore_sword_statement(
receipt.ore_statement_iri)
for state_uri, state_desc in statement.states:
logger.info("Dataset has state URI: " + state_uri)
if state_uri in ERROR_STATES:
dataset_submission.status = 'error'
logger.info(
"URI: " + state_uri +
" is an error state ... setting 'error' state on submission record"
)
break
dataset_submission.last_accessed = datetime.datetime.now()
dataset_submission.save()
time.sleep(SwordStatementCheckThread.throttle)
except urllib2.URLError as e:
# if we get an exception, try again up to the limit
logger.info(
"Got error connecting to the server ... retrying " +
str(retry_counter + 1) + " of " +
str(SwordStatementCheckThread.retry_count))
retry_counter += 1
exception = e
time.sleep(SwordStatementCheckThread.retry_delay)
continue
else:
# if we don't get an exception, we're done
return
# if we don't return from the else statement above, it means the retries
# all failed, and we have a problem. Raise the last thrown exception.
raise exception
def test_36_error_bad_request(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW,
error_response_raises_exceptions=False)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(col_iri = col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging = 'http://purl.org/net/sword/package/SimpleZip',
in_progress = "Invalid", # the API seems to allow this!
suggested_identifier = "zyxwvutsrq")
assert receipt.code == 400
assert isinstance(receipt, Error_Document)
assert receipt.error_href == "http://purl.org/net/sword/error/ErrorBadRequest"
def test_19_advanced_replace_metadata(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
on_behalf_of=SSS_OBO,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="An entry only deposit",
id="asidjasidj",
dcterms_abstract="abstract",
dcterms_identifier="http://whatever/")
receipt = conn.create(col_iri=col.href, metadata_entry=e)
# ensure that we have a receipt (the server may not give us one
# by default)
receipt = conn.get_deposit_receipt(receipt.location)
# now do the replace
ne = Entry(title="A metadata update",
id="asidjasidj",
dcterms_abstract="new abstract",
dcterms_identifier="http://elsewhere/")
new_receipt = conn.update(dr=receipt,
metadata_entry=ne,
in_progress=True)
assert new_receipt.code == 204 or new_receipt.code == 200
if new_receipt.code == 204:
assert new_receipt.dom is None
if new_receipt.code == 200:
assert new_receipt.parsed == True
assert new_receipt.valid == True
def test_33_get_ore_statement(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
on_behalf_of=SSS_OBO,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="Multipart deposit",
id="asidjasidj",
dcterms_abstract="abstract",
dcterms_identifier="http://whatever/")
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
metadata_entry=e,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
# ensure that we have a receipt (the server may not give us one
# by default)
edit_iri = receipt.location
receipt = conn.get_deposit_receipt(edit_iri)
assert receipt.ore_statement_iri is not None
# get the statement
statement = conn.get_ore_sword_statement(receipt.ore_statement_iri)
assert isinstance(statement, Ore_Sword_Statement)
def test_17_advanced_replace_file_content(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
on_behalf_of=SSS_OBO,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
# ensure that we have a receipt (the server may not give us one
# by default)
receipt = conn.get_deposit_receipt(receipt.location)
# now do the replace
with open(PACKAGE) as pkg:
new_receipt = conn.update(
dr=receipt,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="update.zip",
packaging='http://purl.org/net/sword/package/SimpleZip',
metadata_relevant=True)
assert new_receipt.code == 204
assert new_receipt.dom is None
def test_28_advanced_add_metadata(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
on_behalf_of=SSS_OBO,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="Multipart deposit",
id="asidjasidj",
dcterms_abstract="abstract",
dcterms_identifier="http://whatever/")
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
metadata_entry=e,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
# ensure that we have a receipt (the server may not give us one
# by default)
receipt = conn.get_deposit_receipt(receipt.location)
ne = Entry(title="Multipart deposit",
id="asidjasidj",
dcterms_identifier="http://another/",
dcterms_creator="Me!",
dcterms_rights="CC0")
new_receipt = conn.append(dr=receipt,
metadata_entry=ne,
in_progress=True)
assert new_receipt.code == 200
def test_26_advanced_add_content_to_resource_package(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
on_behalf_of=SSS_OBO,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
receipt = conn.get_deposit_receipt(receipt.location)
with open(PACKAGE) as pkg:
new_receipt = conn.add_file_to_resource(
receipt.edit_media,
pkg,
"addition.zip",
mimetype=PACKAGE_MIME,
packaging="http://purl.org/net/sword/package/SimpleZip",
metadata_relevant=True)
assert new_receipt.code >= 200 and new_receipt.code < 400
assert new_receipt.location is not None
assert new_receipt.location == receipt.edit_media
def test_14_error_retrieve_content_em_iri(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
error_response_raises_exceptions=False,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
# ensure that we have a receipt (the server may not give us one
# by default)
receipt = conn.get_deposit_receipt(receipt.location)
error = 'http://purl.org/net/sword/package/IJustMadeThisUp'
response = conn.get_resource(content_iri=receipt.edit_media,
packaging=error)
assert response.code == 406
assert isinstance(response, Error_Document)
assert response.error_href == "http://purl.org/net/sword/error/ErrorContent"
def test_13_advanced_retrieve_content_em_iri(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
# ensure that we have a receipt (the server may not give us one
# by default)
receipt = conn.get_deposit_receipt(receipt.location)
packaging = 'http://purl.org/net/sword/package/SimpleZip'
if receipt.packaging is not None and len(receipt.packaging) > 0:
packaging = receipt.packaging[0]
resource = conn.get_resource(content_iri=receipt.edit_media,
packaging=packaging,
on_behalf_of=SSS_OBO)
assert resource.code == 200
assert resource.content is not None
def test_15_retrieve_content_em_iri_as_feed(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
# ensure that we have a receipt (the server may not give us one
# by default)
receipt = conn.get_deposit_receipt(receipt.location)
# we're going to work with the edit_media_feed iri
assert receipt.edit_media_feed is not None
response = conn.get_resource(content_iri=receipt.edit_media_feed)
assert response.code == 200
assert response.content is not None
# the response should be an xml document, so let's see if we can parse
# it. This should give us an exception which will fail the test if not
dom = etree.fromstring(response.content)
def test_22_delete_content(self):
conn = Connection(SSS_URL,
user_name=SSS_UN,
user_pass=SSS_PW,
on_behalf_of=SSS_OBO,
http_impl=http)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
e = Entry(title="Multipart deposit",
id="asidjasidj",
dcterms_abstract="abstract",
dcterms_identifier="http://whatever/")
with open(PACKAGE) as pkg:
receipt = conn.create(
col_iri=col.href,
metadata_entry=e,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging='http://purl.org/net/sword/package/SimpleZip')
# ensure that we have a receipt (the server may not give us one
# by default)
receipt = conn.get_deposit_receipt(receipt.location)
# now delete the content but not the container
new_receipt = conn.delete_content_of_resource(dr=receipt)
assert new_receipt.code == 204
assert new_receipt.dom is None
def test_31_delete_container(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW, on_behalf_of=SSS_OBO,
error_response_raises_exceptions=False)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
#e = Entry(title="Multipart deposit", id="asidjasidj", dcterms_abstract="abstract", dcterms_identifier="http://whatever/")
with open(PACKAGE) as pkg:
receipt = conn.create(col_iri = col.href,
#metadata_entry = e,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging = 'http://purl.org/net/sword/package/SimpleZip')
# ensure that we have a receipt (the server may not give us one
# by default)
edit_iri = receipt.location
receipt = conn.get_deposit_receipt(edit_iri)
# delete the container
new_receipt = conn.delete_container(dr=receipt)
assert new_receipt.code == 204
assert new_receipt.dom is None
# the next check is that this 404s appropriately now
another_receipt = conn.get_deposit_receipt(edit_iri)
def test_24_get_sword_statement(self):
conn = Connection("http://localhost:%s/sd-uri" % PORT_NUMBER, user_name="sword", user_pass="******", download_service_document=True)
e = Entry(title="Foo", id="asidjasidj", dcterms_appendix="blah blah", dcterms_title="foo bar")
deposit_receipt = conn.create(payload = "Multipart_POST_then_update_on_EM_IRI",
metadata_entry = e,
mimetype = "text/plain",
filename = "readme.txt",
packaging = 'http://purl.org/net/sword/package/Binary',
col_iri = conn.sd.workspaces[0][1][0].href,
in_progress=True)
ss_iri = None
for item_dict in deposit_receipt.links['http://purl.org/net/sword/terms/statement']:
if item_dict.has_key('type') and item_dict.get('type', None) == "application/atom+xml;type=feed":
ss_iri = item_dict.get('href')
assert ss_iri != None
ss = conn.get_atom_sword_statement(ss_iri)
assert ss != None
assert ss.resources[0].deposited_by == 'sword'
def test_37_error_target_owner_unknown(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW,
error_response_raises_exceptions=False)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(col_iri = col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging = 'http://purl.org/net/sword/package/SimpleZip',
in_progress = True,
suggested_identifier = "zyxwvutsrq",
on_behalf_of="richard") # we expressly set the wrong obo on the request rather than the connection
assert receipt.code == 403
assert isinstance(receipt, Error_Document)
assert receipt.error_href == "http://purl.org/net/sword/error/TargetOwnerUnknown"
def test_35_error_checksum_mismatch(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW,
error_response_raises_exceptions=False)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(col_iri = col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging = 'http://purl.org/net/sword/package/SimpleZip',
in_progress = True,
suggested_identifier = "zyxwvutsrq",
md5sum="123456789")
assert receipt.code == 412
assert isinstance(receipt, Error_Document)
assert receipt.error_href == "http://purl.org/net/sword/error/ErrorChecksumMismatch"
def test_03_basic_create_resource_with_package(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(col_iri = col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging = 'http://purl.org/net/sword/package/SimpleZip')
assert receipt.code == 201
assert receipt.location != None
# these last two assertions are contingent on if we actually get a
# receipt back from the server (which we might not legitimately get)
assert receipt.dom is None or receipt.parsed == True
assert receipt.dom is None or receipt.valid == True
def get_silos(self, opener, repository):
logger.debug("Carrying out get_silos for a repository selected")
# verify that we can get a service document, and that there
# is at least one silo and that we can authenticate
if repository.sword2_sd_url is None:
logger.debug("No sword2 service-document URL for repository configuration : ")
logger.debug(repr(repository.sword2_sd_url))
raise SwordServiceError("No sword2 service-document URL for repository configuration")
logger.debug("formulating a connection object for sword repo with url : " + repository.sword2_sd_url )
# get the service document (for which we must be authenticated)
conn = Connection(repository.sword2_sd_url, error_response_raises_exceptions=False, http_impl=UrlLib2Layer(opener))
logger.debug("before getting service document : ")
conn.get_service_document()
logger.debug("Got the service document")
# we require there to be at least one workspace
if conn.sd is None:
logger.debug("did not successfully retrieve a service document")
raise SwordServiceError("did not successfully retrieve a service document")
if conn.sd.workspaces is None:
logger.debug("no workspaces defined in service document")
raise SwordServiceError("no workspaces defined in service document")
if len(conn.sd.workspaces) == 0:
logger.debug("no workspaces defined in service document")
raise SwordServiceError("no workspaces defined in service document")
workspace = conn.sd.workspaces[0][1]
# we require there to be at least one collection
if len(workspace) == 0:
logger.debug("no collections defined in workspace")
raise SwordServiceError("no collections defined in workspace")
# FIXME: we don't currently have a mechanism to make decisions about
# which collection to put stuff in, so we just put stuff in the first
# one for the time being
#col = workspace[0]
logger.debug("Finished with get_silos for a repository selected")
return workspace
def test_09_basic_retrieve_deposit_receipt(self):
conn = Connection(SSS_URL, user_name=SSS_UN, user_pass=SSS_PW)
conn.get_service_document()
col = conn.sd.workspaces[0][1][0]
with open(PACKAGE) as pkg:
receipt = conn.create(col_iri = col.href,
payload=pkg,
mimetype=PACKAGE_MIME,
filename="example.zip",
packaging = 'http://purl.org/net/sword/package/SimpleZip')
# we're going to work with the location
assert receipt.location != None
new_receipt = conn.get_deposit_receipt(receipt.location)
assert new_receipt.code == 200
assert new_receipt.parsed == True
assert new_receipt.valid == True
