def _arcp2file(self, uri):
parsed = arcp.parse_arcp(uri)
# arcp URIs, ensure they are local to our RO
self.assertEquals(parsed.uuid, arcp.parse_arcp(self.find_arcp()).uuid)
path = parsed.path[1:] # Strip first /
# Convert to local path, in case it uses \ on Windows
lpath = provenance._convert_path(path, posixpath, os.path)
return os.path.join(self.folder, lpath)
def _arcp2file(base_path, uri):
parsed = arcp.parse_arcp(uri)
# arcp URIs, ensure they are local to our RO
assert parsed.uuid == arcp.parse_arcp(find_arcp(base_path)).uuid,\
'arcp URI must be local to the research object'
path = parsed.path[1:] # Strip first /
# Convert to local path, in case it uses \ on Windows
lpath = provenance._convert_path(path, posixpath, os.path)
return os.path.join(base_path, lpath)
def _arcp2file(base_path, uri):
parsed = arcp.parse_arcp(uri)
# arcp URIs, ensure they are local to our RO
assert parsed.uuid == arcp.parse_arcp(find_arcp(base_path)).uuid,\
'arcp URI must be local to the research object'
path = parsed.path[1:] # Strip first /
# Convert to local path, in case it uses \ on Windows
lpath = provenance._convert_path(path, posixpath, os.path)
return os.path.join(base_path, lpath)
def check_ro(self):
manifest_file = os.path.join(self.folder, "metadata", "manifest.json")
self.assertTrue(os.path.isfile(manifest_file), "Can't find " + manifest_file)
arcp_root = self.find_arcp()
base = urllib.parse.urljoin(arcp_root, "metadata/manifest.json")
g = Graph()
with open(manifest_file, "rb") as f:
# Note: This will use https://w3id.org/bundle/context
g.parse(file=f, format="json-ld", publicID=base)
print("Parsed manifest:\n\n")
g.serialize(sys.stdout, format="nt")
ro = None
for ro in g.subjects(ORE.isDescribedBy, URIRef(base)):
break
self.assertTrue(ro, "Can't find RO with ore:isDescribedBy")
profile = None
for dc in g.objects(ro, DCTERMS.conformsTo):
profile = dc
break
self.assertTrue(profile, "Can't find profile with dct:conformsTo")
self.assertEquals(profile, URIRef("https://w3id.org/cwl/prov/0.3.0"),
"Unexpected cwlprov version " + profile)
paths = []
externals = []
for aggregate in g.objects(ro, ORE.aggregates):
print(aggregate)
if not arcp.is_arcp_uri(aggregate):
externals.append(aggregate)
# Won't check external URIs existence here
# TODO: Check they are not relative!
continue
# arcp URIs - assume they are local to our RO
path = arcp.parse_arcp(aggregate).path[1:] # Strip first /
paths.append(path)
# Convert to local path, in case it uses \ on Windows
lpath = provenance._convert_path(path, posixpath, os.path)
lfile = os.path.join(self.folder, lpath)
self.assertTrue(os.path.isfile(lfile), "Can't find aggregated " + lfile)
self.assertTrue(paths, "Didn't find any arcp aggregates")
self.assertTrue(externals, "Didn't find any data URIs")
for ext in ["provn", "xml", "json", "jsonld", "nt", "ttl"]:
f = "metadata/provenance/primary.cwlprov.%s" % ext
self.assertTrue(f in paths, "provenance file missing " + f)
for f in ["workflow/primary-job.json", "workflow/packed.cwl"]:
self.assertTrue(f in paths, "workflow file missing " + f)
def test_failing_path_conversion(path, from_type, to_type):
with pytest.raises(ValueError):
provenance._convert_path(path, from_type, to_type)
def test_path_conversion(path, expected, from_type, to_type):
assert provenance._convert_path(path, from_type, to_type) == expected
def test_nt_to_posix_absolute_fails(self):
with self.assertRaises(ValueError):
provenance._convert_path(r"D:\absolute\path", ntpath, posixpath)
def test_posix_to_nt_absolute_fails(self):
with self.assertRaises(ValueError):
provenance._convert_path("/absolute/path", posixpath, ntpath)
def test_nt_to_nt(self):
self.assertEquals(r"a\b\c",
provenance._convert_path(r"a\b\c", ntpath, ntpath))
def test_posix_to_posix(self):
self.assertEquals("a/b/c",
provenance._convert_path("a/b/c", posixpath, posixpath))
def test_posix_to_nt(self):
self.assertEquals(r"a\b\c",
provenance._convert_path("a/b/c", posixpath, ntpath))
def test_failing_path_conversion(path, from_type, to_type):
with pytest.raises(ValueError):
provenance._convert_path(path, from_type, to_type)
def test_path_conversion(path, expected, from_type, to_type):
assert provenance._convert_path(path, from_type, to_type) == expected
