def test_fetch_bag_failure():
with hydroshare_archive():
with patch.object(Hydroshare,
"urlopen",
side_effect=[MockResponse("application/html", 500)]):
hydro = Hydroshare()
spec = {
"host": {
"hostname": [
"https://www.hydroshare.org/resource/",
"http://www.hydroshare.org/resource/",
],
"django_irods":
"https://www.hydroshare.org/django_irods/download/bags/",
},
"resource": "123456789",
}
with TemporaryDirectory() as d:
with pytest.raises(
ContentProviderException,
match=r"Failed to download bag\. status code 500\.",
):
# loop for yield statements
for l in hydro.fetch(spec, d):
pass
def test_fetch_bag():
# we "fetch" a local ZIP file to simulate a Hydroshare resource
with hydroshare_archive() as hydro_path:
with patch.object(
Hydroshare,
"urlopen",
side_effect=[
MockResponse("application/html", 200),
MockResponse("application/zip", 200),
],
):
with patch.object(
Hydroshare, "_urlretrieve", side_effect=[(hydro_path, None)]
):
hydro = Hydroshare()
hydro.resource_id = "b8f6eae9d89241cf8b5904033460af61"
spec = {
"host": {
"hostname": [
"https://www.hydroshare.org/resource/",
"http://www.hydroshare.org/resource/",
],
"django_irods": "https://www.hydroshare.org/django_irods/download/bags/",
},
"resource": "123456789",
}
with TemporaryDirectory() as d:
output = []
for l in hydro.fetch(spec, d):
output.append(l)
unpacked_files = set(os.listdir(d))
expected = set(["some-other-file.txt", "some-file.txt"])
assert expected == unpacked_files
def test_fetch_bag_timeout():
with hydroshare_archive():
with patch.object(Hydroshare,
"urlopen",
side_effect=[MockResponse("application/html", 200)]):
hydro = Hydroshare()
spec = {
"host": {
"hostname": [
"https://www.hydroshare.org/resource/",
"http://www.hydroshare.org/resource/",
],
"django_irods":
"https://www.hydroshare.org/django_irods/download/bags/",
},
"resource": "123456789",
}
with TemporaryDirectory() as d:
with pytest.raises(
ContentProviderException,
match=
r"Bag taking too long to prepare, exiting now, try again later\.",
):
# loop for yield statements
for l in hydro.fetch(spec, d, timeout=0):
pass
def test_content_id():
with patch.object(Hydroshare, "urlopen") as fake_urlopen:
fake_urlopen.return_value.url = "https://www.hydroshare.org/resource/b8f6eae9d89241cf8b5904033460af61"
hydro = Hydroshare()
hydro.detect("10.4211/hs.b8f6eae9d89241cf8b5904033460af61")
assert hydro.content_id == "b8f6eae9d89241cf8b5904033460af61"
def test_content_id(requests_mock):
requests_mock.get(re.compile("https://"), json=hydroshare_data)
requests_mock.get(re.compile("https://doi.org"), json=doi_resolver)
hydro = Hydroshare()
hydro.detect("10.4211/hs.b8f6eae9d89241cf8b5904033460af61")
assert hydro.content_id == "b8f6eae9d89241cf8b5904033460af61.v1569427757"
def test_content_id():
with patch.object(Hydroshare, "urlopen") as fake_urlopen:
fake_urlopen.return_value.url = (
"https://www.hydroshare.org/resource/b8f6eae9d89241cf8b5904033460af61"
)
def read():
return '{"dates": [{"type": "modified", "start_date": "2019-09-25T16:09:17.006152Z"}]}'
fake_urlopen.return_value.read = read
hydro = Hydroshare()
hydro.detect("10.4211/hs.b8f6eae9d89241cf8b5904033460af61")
assert hydro.content_id == "b8f6eae9d89241cf8b5904033460af61.v1569427757"
def test_fetch_bag_timeout():
with hydroshare_archive() as hydro_path:
with patch.object(Hydroshare,
"urlopen",
side_effect=[MockResponse("application/html", 200)]):
hydro = Hydroshare()
spec = {
"host": {
"hostname": [
"https://www.hydroshare.org/resource/",
"http://www.hydroshare.org/resource/",
],
"django_irods":
"https://www.hydroshare.org/django_irods/download/bags/",
},
"resource": "123456789",
}
with TemporaryDirectory() as d:
output = []
for l in hydro.fetch(spec, d, timeout=0):
output.append(l)
assert "Bag taking too long to prepare, exiting now, try again later." == output[
-1]
def test_fetch_bag_failure():
with hydroshare_archive() as hydro_path:
with patch.object(Hydroshare,
"urlopen",
side_effect=[MockResponse("application/html", 500)]):
hydro = Hydroshare()
spec = {
"host": {
"hostname": [
"https://www.hydroshare.org/resource/",
"http://www.hydroshare.org/resource/",
],
"django_irods":
"https://www.hydroshare.org/django_irods/download/bags/",
},
"resource": "123456789",
}
with TemporaryDirectory() as d:
output = []
for l in hydro.fetch(spec, d):
output.append(l)
assert "Failed to download bag. status code 500.\n" == output[
-1]
def test_detect_hydroshare():
with patch.object(Hydroshare, "urlopen") as fake_urlopen:
fake_urlopen.return_value.url = (
"https://www.hydroshare.org/resource/b8f6eae9d89241cf8b5904033460af61"
)
def read():
return '{"dates": [{"type": "modified", "start_date": "2019-09-25T16:09:17.006152Z"}]}'
fake_urlopen.return_value.read = read
# valid Hydroshare DOIs trigger this content provider
expected = {
"host": {
"hostname": [
"https://www.hydroshare.org/resource/",
"http://www.hydroshare.org/resource/",
],
"django_irods":
"https://www.hydroshare.org/django_irods/download/bags/",
"version":
"https://www.hydroshare.org/hsapi/resource/{}/scimeta/elements",
},
"resource": "b8f6eae9d89241cf8b5904033460af61",
"version": "1569427757",
}
assert (Hydroshare().detect(
"https://www.hydroshare.org/resource/b8f6eae9d89241cf8b5904033460af61"
) == expected)
# assert a call to urlopen was called to fetch version
assert fake_urlopen.call_count == 1
assert (Hydroshare().detect(
"10.4211/hs.b8f6eae9d89241cf8b5904033460af61") == expected)
# assert 2 more calls were made, one to resolve the DOI and another to fetch the version
assert fake_urlopen.call_count == 3
assert (Hydroshare().detect(
"https://doi.org/10.4211/hs.b8f6eae9d89241cf8b5904033460af61") ==
expected)
# assert 2 more calls were made, one to resolve the DOI and another to fetch the version
assert fake_urlopen.call_count == 5
with patch.object(Hydroshare, "urlopen") as fake_urlopen:
# Don't trigger the Hydroshare content provider
assert Hydroshare().detect("/some/path/here") is None
assert Hydroshare().detect("https://example.com/path/here") is None
# don't handle DOIs that aren't from Hydroshare
fake_urlopen.return_value.url = (
"http://joss.theoj.org/papers/10.21105/joss.01277")
def read():
return '{"dates": [{"type": "modified", "start_date": "2019-09-25T16:09:17.006152Z"}]}'
fake_urlopen.return_value.read = read
assert Hydroshare().detect(
"https://doi.org/10.21105/joss.01277") is None
def test_detect_hydroshare(requests_mock):
requests_mock.get(re.compile("https://"), json=hydroshare_data)
requests_mock.get(re.compile("https://doi.org"), json=doi_resolver)
# valid Hydroshare DOIs trigger this content provider
expected = {
"host": {
"hostname": [
"https://www.hydroshare.org/resource/",
"http://www.hydroshare.org/resource/",
],
"django_irods": "https://www.hydroshare.org/django_irods/download/bags/",
"version": "https://www.hydroshare.org/hsapi/resource/{}/scimeta/elements",
},
"resource": "b8f6eae9d89241cf8b5904033460af61",
"version": "1569427757",
}
assert (
Hydroshare().detect(
"https://www.hydroshare.org/resource/b8f6eae9d89241cf8b5904033460af61"
)
== expected
)
# assert a call to urlopen was called to fetch version
assert requests_mock.call_count == 1
requests_mock.reset_mock()
assert (
Hydroshare().detect("10.4211/hs.b8f6eae9d89241cf8b5904033460af61") == expected
)
# assert 3 calls were made, 2 to resolve the DOI (302 + 200) and another to fetch the version
assert requests_mock.call_count == 3
requests_mock.reset_mock()
assert (
Hydroshare().detect(
"https://doi.org/10.4211/hs.b8f6eae9d89241cf8b5904033460af61"
)
== expected
)
# assert 3 more calls were made, 2 to resolve the DOI and another to fetch the version
assert requests_mock.call_count == 3
requests_mock.reset_mock()
# Don't trigger the Hydroshare content provider
assert Hydroshare().detect("/some/path/here") is None
assert Hydroshare().detect("https://example.com/path/here") is None
# don't handle DOIs that aren't from Hydroshare
assert Hydroshare().detect("https://doi.org/10.21105/joss.01277") is None
def test_detect_hydroshare(test_input, expected):
with patch.object(Hydroshare, "urlopen") as fake_urlopen:
fake_urlopen.return_value.url = test_input[0]
# valid Hydroshare DOIs trigger this content provider
assert Hydroshare().detect(test_input[0]) == expected
assert Hydroshare().detect(test_input[1]) == expected
assert Hydroshare().detect(test_input[2]) == expected
# only two of the three calls above have to resolve a DOI
assert fake_urlopen.call_count == 2
with patch.object(Hydroshare, "urlopen") as fake_urlopen:
# Don't trigger the Hydroshare content provider
assert Hydroshare().detect("/some/path/here") is None
assert Hydroshare().detect("https://example.com/path/here") is None
# don't handle DOIs that aren't from Hydroshare
fake_urlopen.return_value.url = (
"http://joss.theoj.org/papers/10.21105/joss.01277")
assert Hydroshare().detect(
"https://doi.org/10.21105/joss.01277") is None