def test_bad_overrides(self):
"ensure a number of bad cases for overrides fail"
cases = [
"title", # no pipe
"title|", # pipe, but no value
"title|bar", # invalid value (should be json: "bar")
"|bar", # invalid key
]
xml_fname = 'elife-16695-v1.xml'
xml_fixture = join(self.fixtures_dir, xml_fname)
with open(xml_fixture, 'rb') as fh:
for override in cases:
resp = self.client.post('/xml?id=16695&version=1', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': { # culprit lies in the payload here somewhere
'xml': fh,
'override': [override],
}
})
expected_resp = {
'status': conf.ERROR,
'code': conf.BAD_OVERRIDES,
# 'message': '...',
# 'trace': '...',
}
self.assertEqual(resp.status_code, 400)
self.assertTrue(utils.partial_match(expected_resp, resp.json))
self.assertTrue(resp.json['trace'].startswith('Traceback (most recent call last):'))
self.assertTrue(resp.json['message']) # one exists and isn't empty
def test_upload_invalid(self):
"the response we expect when the scraped article-json is invalid"
xml_fname = 'elife-00666-v1.xml.invalid'
xml_upload_fname = 'elife-00666-v1.xml'
xml_fixture = join(self.fixtures_dir, xml_fname)
with patch('adaptor.call_lax', side_effect=AssertionError("test shouldn't make it this far!")):
resp = self.client.post('/xml?id=666&version=1', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': {
'xml': (open(xml_fixture, 'rb'), xml_upload_fname),
}
})
self.assertEqual(resp.status_code, 400) # bad request data
# ensure xml uploaded
expected_path = join(self.temp_dir, xml_upload_fname)
self.assertTrue(os.path.exists(expected_path))
# ensure ajson scraped
expected_ajson = join(self.temp_dir, xml_upload_fname) + '.json'
self.assertTrue(os.path.exists(expected_ajson))
expected_resp = {
'status': conf.INVALID,
'code': conf.ERROR_INVALID,
# 'message': '...', # will probably change
# 'trace': '...', # stacktrace
}
self.assertTrue(utils.partial_match(expected_resp, resp.json))
self.assertTrue(resp.json['message'])
# title is missing
self.assertTrue(resp.json['trace'].startswith("None is not of type 'string'"), "actual trace: %s" % resp.json['trace'])
def test_bad_upload_filename(self):
"the response we expect when the xml fails to upload"
xml_fname = 'elife-16695-v1.xml'
xml_fixture = join(self.fixtures_dir, xml_fname)
# msid doesn't match
# filename doesn't match pattern 'elife-msid-vN.xml'
bad_path = '/var/www/html/_default/cms/cms-0.9.40-alpha/ecs_includes/packageCreator/19942-v1.xml'
resp = self.client.post('/xml?id=16695&version=1', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': {
'xml': (open(xml_fixture, 'rb'), bad_path),
}
})
expected_lax_resp = {
'status': conf.ERROR,
'code': conf.BAD_SCRAPE,
# 'message': '...', # we just care that a message exists
# 'trace': '...', # same again, just that a trace exists
}
self.assertEqual(resp.status_code, 400)
self.assertTrue(utils.partial_match(expected_lax_resp, resp.json))
self.assertTrue(resp.json['trace'].startswith('Traceback (most recent call last):'))
# self.assertTrue(resp.json['message']) # one exists and isn't empty
self.assertTrue(resp.json['message'].startswith('not enough values to unpack')) # todo: gotta fix this filename parsing
def test_request_lax_style(self):
"dictionaries of article information can be fed in"
paths = [{
'msid':
16695,
'version':
1,
'location':
'https://s3-external-1.amazonaws.com/elife-publishing-expanded/16695.1/9c2cabd8-a25a-4d76-9f30-1c729755480b/elife-16695-v1.xml',
}]
expected_lax_response = {
"id": "16695",
"requested-action": 'ingest',
"token": 'pants-party',
"status": conf.INGESTED,
"message": '...?',
"datetime": utils.ymdhms(datetime.now())
}
expected = {
'valid': [expected_lax_response],
'errors': [],
'invalid': []
}
with mock.patch('adaptor.call_lax',
autospec=True,
specset=True,
return_value=expected_lax_response):
with mock.patch('adaptor.http_download',
autospec=True,
return_value=open(self.small_doc, 'r')):
actual = bfup.do_paths(paths)
self.assertTrue(partial_match(expected, actual))
def test_bad_upload_ext(self):
"the response we expect when the xml fails to upload"
xml_fname = 'elife-16695-v1.xml'
xml_fixture = join(self.fixtures_dir, xml_fname)
bad_ext = '.pants'
resp = self.client.post('/xml?id=16695&version=1', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': {
'xml': (open(xml_fixture, 'rb'), xml_fname + bad_ext),
}
})
expected_lax_resp = {
'status': conf.ERROR,
'code': conf.BAD_UPLOAD,
# 'message': '...', # we just care that a message exists
# 'trace': '...', # same again, just that a trace exists
}
self.assertEqual(resp.status_code, 400)
self.assertTrue(utils.partial_match(expected_lax_resp, resp.json))
self.assertTrue(resp.json['trace'].startswith('Traceback (most recent call last):'))
self.assertTrue(resp.json['message']) # one exists and isn't empty
def test_request_multiple_valid_paths(self):
"sends multiple articles to lax given paths to xml files"
paths = [self.small_doc] * 3 # send the same article three times
expected_lax_response = {
# "id": '09560',
"id": "16695",
"requested-action": 'ingest',
"token": 'pants-party', # set by fs-adaptor
"status": conf.INGESTED,
"message": '...?',
"datetime": utils.ymdhms(datetime.now())
}
expected = {
'valid': [expected_lax_response] * 3,
'errors': [],
'invalid': []
}
with mock.patch('adaptor.call_lax',
autospec=True,
specset=True,
return_value=expected_lax_response):
actual = bfup.do_paths(paths)
self.assertTrue(partial_match(expected, actual))
def test_broken_glencoe_response(self):
"the response we expect when the glencoe code fails"
err_message = "informative error message"
expected_resp = {
'status': conf.ERROR,
'code': conf.BAD_SCRAPE,
'message': err_message,
#'trace': '...' # super long, can't predict, especially when mocking
}
with patch('glencoe.validate_gc_data',
side_effect=AssertionError(err_message)):
resp = self.client.post(
'/xml', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': {
'xml':
open(join(self.fixtures_dir, 'elife-00666-v1.xml'),
'rb'),
},
})
self.assertEqual(resp.status_code, 400) # bad request
self.assertTrue(utils.partial_match(expected_resp, resp.json))
self.assertTrue(resp.json['trace'].startswith('Traceback (most'))
def test_bad_overrides(self):
"ensure a number of bad cases for overrides fail"
cases = [
"title", # no pipe
"title|", # pipe, but no value
"title|bar", # invalid value (should be json: "bar")
"|bar", # invalid key
]
xml_fname = 'elife-16695-v1.xml'
xml_fixture = join(self.fixtures_dir, xml_fname)
with open(xml_fixture, 'rb') as fh:
for override in cases:
resp = self.client.post('/xml', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': { # culprit lies in the payload here somewhere
'xml': fh,
'override': [override],
}
})
expected_resp = {
'status': conf.ERROR,
'code': conf.BAD_OVERRIDES,
#'message': '...',
#'trace': '...',
}
self.assertEqual(resp.status_code, 400)
self.assertTrue(utils.partial_match(expected_resp, resp.json))
self.assertTrue(resp.json['trace'].startswith(
'Traceback (most recent call last):'))
self.assertTrue(
resp.json['message']) # one exists and isn't empty
def test_bad_scrape(self):
"the response we expect the xml fails to scrape"
xml_fname = 'elife-16695-v1.xml'
xml_fixture = join(self.fixtures_dir, xml_fname)
with patch('main.main', side_effect=AssertionError('meow')):
resp = self.client.post(
'/xml', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': {
'xml': open(xml_fixture, 'rb'),
}
})
expected_lax_resp = {
'status': conf.ERROR,
'code': conf.BAD_SCRAPE,
#'message': '...', # we just care that a message exists
#'trace': '...', # same again, just that a trace exists
}
self.assertEqual(resp.status_code, 400)
self.assertTrue(utils.partial_match(expected_lax_resp, resp.json))
self.assertTrue(resp.json['trace'].startswith(
'Traceback (most recent call last):'))
self.assertTrue(resp.json['message']) # one exists and isn't empty
def test_bad_upload(self):
"the response we expect when the xml fails to upload"
xml_fname = 'elife-16695-v1.xml'
xml_fixture = join(self.fixtures_dir, xml_fname)
bad_ext = '.pants'
resp = self.client.post(
'/xml', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': {
'xml': (open(xml_fixture, 'rb'), xml_fname + bad_ext),
}
})
expected_lax_resp = {
'status': conf.ERROR,
'code': conf.BAD_UPLOAD,
#'message': '...', # we just care that a message exists
#'trace': '...', # same again, just that a trace exists
}
self.assertEqual(resp.status_code, 400)
self.assertTrue(utils.partial_match(expected_lax_resp, resp.json))
self.assertTrue(resp.json['trace'].startswith(
'Traceback (most recent call last):'))
self.assertTrue(resp.json['message']) # one exists and isn't empty
def test_request_invalid_path(self):
"something invalid that initially looks valid fails when we try to use it"
paths = ['elife-09561-v1.xml'
] # msid and version can be extracted, but it's not a path
expected = {
'valid': [],
'errors': [{
'status': 'error',
'requested-action': 'ingest'
}],
'invalid': []
}
actual = bfup.do_paths(paths)
self.assertTrue(partial_match(expected, actual))
def test_upload_invalid(self):
"the response we expect when the scraped article-json is invalid"
xml_fname = 'elife-00666-v1.xml.invalid'
xml_upload_fname = 'elife-00666-v1.xml'
xml_fixture = join(self.fixtures_dir, xml_fname)
with patch('adaptor.call_lax',
side_effect=AssertionError(
"test shouldn't make it this far!")):
resp = self.client.post(
'/xml', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': {
'xml': (open(xml_fixture, 'rb'), xml_upload_fname),
}
})
self.assertEqual(resp.status_code, 400) # bad request data
# ensure xml uploaded
expected_path = join(self.temp_dir, xml_upload_fname)
self.assertTrue(os.path.exists(expected_path))
# ensure ajson scraped
expected_ajson = join(self.temp_dir, xml_upload_fname) + '.json'
self.assertTrue(os.path.exists(expected_ajson))
expected_resp = {
'status': conf.INVALID,
'code': conf.ERROR_INVALID,
#'message': '...', # will probably change
#'trace': '...', # stacktrace
}
self.assertTrue(utils.partial_match(expected_resp, resp.json))
self.assertTrue(resp.json['message'])
# title is missing
self.assertTrue(
resp.json['trace'].startswith("None is not of type 'string'"),
"actual trace: %s" % resp.json['trace'])
def test_broken_glencoe_response(self):
"the response we expect when the glencoe code fails"
err_message = "informative error message"
expected_resp = {
'status': conf.ERROR,
'code': conf.BAD_SCRAPE,
'message': err_message,
# 'trace': '...' # super long, can't predict, especially when mocking
}
with patch('glencoe.validate_gc_data', side_effect=AssertionError(err_message)):
resp = self.client.post('/xml?id=666&version=1', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': {
'xml': open(join(self.fixtures_dir, 'elife-00666-v1.xml'), 'rb'),
},
})
self.assertEqual(resp.status_code, 400) # bad request
self.assertTrue(utils.partial_match(expected_resp, resp.json))
self.assertTrue(resp.json['trace'].startswith('Traceback (most'))
def test_bad_scrape(self):
"the response we expect the xml fails to scrape"
xml_fname = 'elife-16695-v1.xml'
xml_fixture = join(self.fixtures_dir, xml_fname)
with patch('main.main', side_effect=AssertionError('meow')):
resp = self.client.post('/xml?id=16695&version=1', **{
'buffered': True,
'content_type': 'multipart/form-data',
'data': {
'xml': open(xml_fixture, 'rb'),
}
})
expected_lax_resp = {
'status': conf.ERROR,
'code': conf.BAD_SCRAPE,
# 'message': '...', # we just care that a message exists
# 'trace': '...', # same again, just that a trace exists
}
self.assertEqual(resp.status_code, 400)
self.assertTrue(utils.partial_match(expected_lax_resp, resp.json))
self.assertTrue(resp.json['trace'].startswith('Traceback (most recent call last):'))
self.assertTrue(resp.json['message']) # one exists and isn't empty