1
|
|
|
#!/usr/bin/env python3 |
2
|
|
|
# -*- coding: utf-8 -*- |
3
|
|
|
""" |
4
|
|
|
Created on Tue Jan 21 10:54:09 2020 |
5
|
|
|
|
6
|
|
|
@author: Paolo Cozzi <[email protected]> |
7
|
|
|
""" |
8
|
|
|
|
9
|
|
|
import os |
10
|
|
|
import json |
11
|
|
|
import time |
12
|
|
|
import types |
13
|
|
|
import asynctest |
14
|
|
|
|
15
|
|
|
from aioresponses import aioresponses |
16
|
|
|
from aiohttp.client_exceptions import ServerDisconnectedError |
17
|
|
|
from unittest.mock import patch, Mock |
18
|
|
|
|
19
|
|
|
from django.test import TestCase |
20
|
|
|
from django.utils import timezone |
21
|
|
|
|
22
|
|
|
from common.constants import SUBMITTED, READY, COMPLETED |
23
|
|
|
from uid.models import Animal as UIDAnimal, Sample as UIDSample |
24
|
|
|
|
25
|
|
|
from ..tasks.cleanup import ( |
26
|
|
|
check_samples, get_orphan_samples, PAGE_SIZE, BIOSAMPLE_ACCESSION_ENDPOINT, |
27
|
|
|
BIOSAMPLE_SAMPLE_ENDPOINT) |
28
|
|
|
from ..models import OrphanSample, ManagedTeam |
29
|
|
|
|
30
|
|
|
from .common import generate_token, BioSamplesMixin |
31
|
|
|
|
32
|
|
|
# get my path |
33
|
|
|
dir_path = os.path.dirname(os.path.realpath(__file__)) |
34
|
|
|
|
35
|
|
|
# define data path |
36
|
|
|
DATA_PATH = os.path.join(dir_path, "data") |
37
|
|
|
|
38
|
|
|
|
39
|
|
|
with open(os.path.join(DATA_PATH, "page_0.json")) as handle: |
40
|
|
|
page0 = handle.read() |
41
|
|
|
|
42
|
|
|
with open(os.path.join(DATA_PATH, "page_1.json")) as handle: |
43
|
|
|
page1 = handle.read() |
44
|
|
|
|
45
|
|
|
with open(os.path.join(DATA_PATH, "issue_page1.json")) as handle: |
46
|
|
|
issue_page1 = handle.read() |
47
|
|
|
|
48
|
|
|
mocked_accessions = {} |
49
|
|
|
|
50
|
|
|
for accession in ["SAMEA6376679", "SAMEA6376682", "SAMEA6376980", |
51
|
|
|
"SAMEA6376982", "SAMEA6376991", "SAMEA6376992"]: |
52
|
|
|
with open(os.path.join(DATA_PATH, f"{accession}.json")) as handle: |
53
|
|
|
mocked_accessions[accession] = handle.read() |
54
|
|
|
|
55
|
|
|
|
56
|
|
|
class AsyncBioSamplesTestCase(asynctest.TestCase, TestCase): |
57
|
|
|
|
58
|
|
|
fixtures = [ |
59
|
|
|
'biosample/managedteam', |
60
|
|
|
'uid/animal', |
61
|
|
|
'uid/dictbreed', |
62
|
|
|
'uid/dictcountry', |
63
|
|
|
'uid/dictrole', |
64
|
|
|
'uid/dictsex', |
65
|
|
|
'uid/dictspecie', |
66
|
|
|
'uid/dictstage', |
67
|
|
|
'uid/dictuberon', |
68
|
|
|
'uid/ontology', |
69
|
|
|
'uid/organization', |
70
|
|
|
'uid/publication', |
71
|
|
|
'uid/sample', |
72
|
|
|
'uid/submission', |
73
|
|
|
'uid/user' |
74
|
|
|
] |
75
|
|
|
|
76
|
|
|
@classmethod |
77
|
|
|
def setUpClass(cls): |
78
|
|
|
# calling my base class setup |
79
|
|
|
super().setUpClass() |
80
|
|
|
|
81
|
|
|
cls.mock_auth_patcher = patch('pyUSIrest.auth.requests.get') |
82
|
|
|
cls.mock_auth = cls.mock_auth_patcher.start() |
83
|
|
|
|
84
|
|
|
@classmethod |
85
|
|
|
def tearDownClass(cls): |
86
|
|
|
cls.mock_auth_patcher.stop() |
87
|
|
|
|
88
|
|
|
# calling base method |
89
|
|
|
super().tearDownClass() |
90
|
|
|
|
91
|
|
|
def setUp(self): |
92
|
|
|
# calling my base setup |
93
|
|
|
super().setUp() |
94
|
|
|
|
95
|
|
|
# well, updating data and set two biosample ids. Those are not |
96
|
|
|
# orphans |
97
|
|
|
animal = UIDAnimal.objects.get(pk=1) |
98
|
|
|
animal.biosample_id = "SAMEA6376980" |
99
|
|
|
animal.save() |
100
|
|
|
|
101
|
|
|
sample = UIDSample.objects.get(pk=1) |
102
|
|
|
sample.biosample_id = "SAMEA6376982" |
103
|
|
|
sample.save() |
104
|
|
|
|
105
|
|
|
# generate tocken |
106
|
|
|
self.mock_auth.return_value = Mock() |
107
|
|
|
self.mock_auth.return_value.text = generate_token() |
108
|
|
|
self.mock_auth.return_value.status_code = 200 |
109
|
|
|
|
110
|
|
|
async def check_samples(self): |
111
|
|
|
with aioresponses() as mocked: |
112
|
|
|
mocked.get( |
113
|
|
|
'{url}?filter=attr:project:IMAGE&size={size}'.format( |
114
|
|
|
url=BIOSAMPLE_ACCESSION_ENDPOINT, size=PAGE_SIZE), |
115
|
|
|
status=200, |
116
|
|
|
body=page0) |
117
|
|
|
mocked.get( |
118
|
|
|
'{url}?filter=attr:project:IMAGE&page=1&size={size}'.format( |
119
|
|
|
url=BIOSAMPLE_ACCESSION_ENDPOINT, size=PAGE_SIZE), |
120
|
|
|
status=200, |
121
|
|
|
body=page1) |
122
|
|
|
for accession, body in mocked_accessions.items(): |
123
|
|
|
mocked.get( |
124
|
|
|
'{url}/{accession}'.format( |
125
|
|
|
url=BIOSAMPLE_SAMPLE_ENDPOINT, accession=accession), |
126
|
|
|
status=200, |
127
|
|
|
body=body) |
128
|
|
|
|
129
|
|
|
await check_samples() |
130
|
|
|
|
131
|
|
|
async def test_request(self) -> None: |
132
|
|
|
await self.check_samples() |
133
|
|
|
|
134
|
|
|
# get accessions |
135
|
|
|
reference = ['SAMEA6376991', 'SAMEA6376992'] |
136
|
|
|
|
137
|
|
|
self.assertEqual(OrphanSample.objects.count(), 2) |
138
|
|
|
|
139
|
|
|
# check objects into UID |
140
|
|
|
for accession in reference: |
141
|
|
|
orphan = OrphanSample.objects.get(biosample_id=accession) |
142
|
|
|
orphan.status = READY |
143
|
|
|
|
144
|
|
|
async def test_request_with_issues(self) -> None: |
145
|
|
|
"""Test a temporary issue with BioSamples reply""" |
146
|
|
|
|
147
|
|
|
with aioresponses() as mocked: |
148
|
|
|
mocked.get( |
149
|
|
|
'{url}?filter=attr:project:IMAGE&size={size}'.format( |
150
|
|
|
url=BIOSAMPLE_ACCESSION_ENDPOINT, size=PAGE_SIZE), |
151
|
|
|
status=200, |
152
|
|
|
body=page0) |
153
|
|
|
mocked.get( |
154
|
|
|
'{url}?filter=attr:project:IMAGE&page=1&size={size}'.format( |
155
|
|
|
url=BIOSAMPLE_ACCESSION_ENDPOINT, size=PAGE_SIZE), |
156
|
|
|
status=200, |
157
|
|
|
body=issue_page1) |
158
|
|
|
for accession, body in mocked_accessions.items(): |
159
|
|
|
mocked.get( |
160
|
|
|
'{url}/{accession}'.format( |
161
|
|
|
url=BIOSAMPLE_SAMPLE_ENDPOINT, accession=accession), |
162
|
|
|
status=200, |
163
|
|
|
body=body) |
164
|
|
|
|
165
|
|
|
await check_samples() |
166
|
|
|
|
167
|
|
|
# no objects where tracked since issue in response |
168
|
|
|
self.assertEqual(OrphanSample.objects.count(), 0) |
169
|
|
|
|
170
|
|
|
async def test_request_with_html(self) -> None: |
171
|
|
|
"""Test a not JSON response (HTML)""" |
172
|
|
|
|
173
|
|
|
with aioresponses() as mocked: |
174
|
|
|
mocked.get( |
175
|
|
|
'{url}?filter=attr:project:IMAGE&size={size}'.format( |
176
|
|
|
url=BIOSAMPLE_ACCESSION_ENDPOINT, size=PAGE_SIZE), |
177
|
|
|
status=200, |
178
|
|
|
body=page0) |
179
|
|
|
mocked.get( |
180
|
|
|
'{url}?filter=attr:project:IMAGE&page=1&size={size}'.format( |
181
|
|
|
url=BIOSAMPLE_ACCESSION_ENDPOINT, size=PAGE_SIZE), |
182
|
|
|
status=200, |
183
|
|
|
headers={'Content-type': 'text/html'}, |
184
|
|
|
body="<html>Not a JSON</html>") |
185
|
|
|
for accession, body in mocked_accessions.items(): |
186
|
|
|
mocked.get( |
187
|
|
|
'{url}/{accession}'.format( |
188
|
|
|
url=BIOSAMPLE_SAMPLE_ENDPOINT, accession=accession), |
189
|
|
|
status=200, |
190
|
|
|
body=body) |
191
|
|
|
|
192
|
|
|
await check_samples() |
193
|
|
|
|
194
|
|
|
# no objects where tracked since issue in response |
195
|
|
|
self.assertEqual(OrphanSample.objects.count(), 0) |
196
|
|
|
|
197
|
|
|
async def test_biosamples_down(self) -> None: |
198
|
|
|
"""Test a not JSON response (HTML): BioSamples down""" |
199
|
|
|
|
200
|
|
|
with aioresponses() as mocked: |
201
|
|
|
mocked.get( |
202
|
|
|
'{url}?filter=attr:project:IMAGE&size={size}'.format( |
203
|
|
|
url=BIOSAMPLE_ACCESSION_ENDPOINT, size=PAGE_SIZE), |
204
|
|
|
status=200, |
205
|
|
|
headers={'Content-type': 'text/html'}, |
206
|
|
|
body="<html>Not a JSON</html>") |
207
|
|
|
mocked.get( |
208
|
|
|
'{url}?filter=attr:project:IMAGE&page=1&size={size}'.format( |
209
|
|
|
url=BIOSAMPLE_ACCESSION_ENDPOINT, size=PAGE_SIZE), |
210
|
|
|
status=200, |
211
|
|
|
headers={'Content-type': 'text/html'}, |
212
|
|
|
body="<html>Not a JSON</html>") |
213
|
|
|
|
214
|
|
|
with self.assertRaises(ConnectionError): |
215
|
|
|
await check_samples() |
216
|
|
|
|
217
|
|
|
# no objects where tracked since issue in response |
218
|
|
|
self.assertEqual(OrphanSample.objects.count(), 0) |
219
|
|
|
|
220
|
|
|
async def test_server_lost(self) -> None: |
221
|
|
|
"""Test server disconnect error""" |
222
|
|
|
|
223
|
|
|
with aioresponses() as mocked: |
224
|
|
|
mocked.get( |
225
|
|
|
'{url}?filter=attr:project:IMAGE&size={size}'.format( |
226
|
|
|
url=BIOSAMPLE_ACCESSION_ENDPOINT, size=PAGE_SIZE), |
227
|
|
|
exception=ServerDisconnectedError() |
228
|
|
|
) |
229
|
|
|
|
230
|
|
|
with self.assertRaises(ConnectionError): |
231
|
|
|
await check_samples() |
232
|
|
|
|
233
|
|
|
# no objects where tracked since issue in response |
234
|
|
|
self.assertEqual(OrphanSample.objects.count(), 0) |
235
|
|
|
|
236
|
|
|
async def test_already_removed_samples(self) -> None: |
237
|
|
|
"""Test check samples with entries in database""" |
238
|
|
|
|
239
|
|
|
# create items into database. get team first |
240
|
|
|
team = ManagedTeam.objects.get(pk=1) |
241
|
|
|
|
242
|
|
|
sample1 = OrphanSample.objects.create( |
243
|
|
|
biosample_id='SAMEA6376991', |
244
|
|
|
name="IMAGEA000005610", |
245
|
|
|
team=team, |
246
|
|
|
status=SUBMITTED, |
247
|
|
|
) |
248
|
|
|
|
249
|
|
|
sample2 = OrphanSample.objects.create( |
250
|
|
|
biosample_id='SAMEA6376992', |
251
|
|
|
name="IMAGEA000005607", |
252
|
|
|
team=team, |
253
|
|
|
status=COMPLETED, |
254
|
|
|
removed=True, |
255
|
|
|
removed_at=timezone.now() |
256
|
|
|
) |
257
|
|
|
|
258
|
|
|
await self.check_samples() |
259
|
|
|
|
260
|
|
|
# test: there are the same samples in database |
261
|
|
|
self.assertEqual(OrphanSample.objects.count(), 2) |
262
|
|
|
|
263
|
|
|
# no changes in statuses |
264
|
|
|
self.assertEqual(sample1.status, SUBMITTED) |
265
|
|
|
self.assertEqual(sample2.status, COMPLETED) |
266
|
|
|
|
267
|
|
|
|
268
|
|
|
class PurgeOrphanSampleTestCase(BioSamplesMixin, TestCase): |
269
|
|
|
fixtures = [ |
270
|
|
|
'biosample/managedteam', |
271
|
|
|
'biosample/orphansample', |
272
|
|
|
'uid/dictspecie', |
273
|
|
|
] |
274
|
|
|
|
275
|
|
|
def test_purge_orphan_samples(self): |
276
|
|
|
"""Test biosample data conversion""" |
277
|
|
|
|
278
|
|
|
with open(os.path.join(DATA_PATH, "SAMEA6376982.json")) as handle: |
279
|
|
|
data = json.load(handle) |
280
|
|
|
|
281
|
|
|
self.mock_get.return_value = Mock() |
282
|
|
|
self.mock_get.return_value.json.return_value = data |
283
|
|
|
self.mock_get.return_value.status_code = 200 |
284
|
|
|
|
285
|
|
|
# call my method |
286
|
|
|
samples = get_orphan_samples() |
287
|
|
|
|
288
|
|
|
# teams is now a generator |
289
|
|
|
self.assertIsInstance(samples, types.GeneratorType) |
290
|
|
|
samples = list(samples) |
291
|
|
|
|
292
|
|
|
self.assertEqual(len(samples), 2) |
293
|
|
|
|
294
|
|
|
sample = samples[0] |
295
|
|
|
self.assertIsInstance(sample, dict) |
296
|
|
|
|
297
|
|
|
sample = samples[1] |
298
|
|
|
self.assertIsInstance(sample, dict) |
299
|
|
|
|
300
|
|
|
# read the team from data |
301
|
|
|
team = sample['team'] |
302
|
|
|
self.assertIsInstance(team, ManagedTeam) |
303
|
|
|
|
304
|
|
|
def test_purge_orphan_samples_not_ready(self): |
305
|
|
|
"""Test not ready orphan samples""" |
306
|
|
|
|
307
|
|
|
# Simulate a different status |
308
|
|
|
OrphanSample.objects.update(status=SUBMITTED) |
309
|
|
|
orphan_count = sum(1 for orphan in get_orphan_samples()) |
310
|
|
|
|
311
|
|
|
self.assertEqual(orphan_count, 0) |
312
|
|
|
|
313
|
|
|
def test_purge_orphan_samples_ignore(self): |
314
|
|
|
"""Test ignored orphan samples""" |
315
|
|
|
|
316
|
|
|
# Ignoring samples gives no object |
317
|
|
|
OrphanSample.objects.update(ignore=True) |
318
|
|
|
orphan_count = sum(1 for orphan in get_orphan_samples()) |
319
|
|
|
|
320
|
|
|
self.assertEqual(orphan_count, 0) |
321
|
|
|
|
322
|
|
|
def test_purge_orphan_samples_removed(self): |
323
|
|
|
"""Test removed orphan samples""" |
324
|
|
|
|
325
|
|
|
# Ignoring samples gives no object |
326
|
|
|
OrphanSample.objects.update(removed=True) |
327
|
|
|
orphan_count = sum(1 for orphan in get_orphan_samples()) |
328
|
|
|
|
329
|
|
|
self.assertEqual(orphan_count, 0) |
330
|
|
|
|
331
|
|
|
def test_purge_orphan_samples_with_limit(self): |
332
|
|
|
"""Test get orphan samples with limits""" |
333
|
|
|
|
334
|
|
|
orphan_count = sum(1 for orphan in get_orphan_samples(limit=1)) |
335
|
|
|
self.assertEqual(orphan_count, 1) |
336
|
|
|
|
337
|
|
|
def test_purge_orphan_private(self): |
338
|
|
|
"""Test no access to a BioSamples id (already removed?)""" |
339
|
|
|
|
340
|
|
|
data = { |
341
|
|
|
'timestamp': int(time.time() * 1000), |
342
|
|
|
'status': 403, |
343
|
|
|
'error': 'Forbidden', |
344
|
|
|
'exception': ( |
345
|
|
|
'uk.ac.ebi.biosamples.service.' |
346
|
|
|
'BioSamplesAapService$SampleNotAccessibleException'), |
347
|
|
|
'message': ( |
348
|
|
|
'This sample is private and not available for browsing. ' |
349
|
|
|
'If you think this is an error and/or you should have access ' |
350
|
|
|
'please contact the BioSamples Helpdesk at biosamples@' |
351
|
|
|
'ebi.ac.uk'), |
352
|
|
|
'path': '/biosamples/samples/SAMEA6376982' |
353
|
|
|
} |
354
|
|
|
|
355
|
|
|
# override mock object |
356
|
|
|
self.mock_get.return_value = Mock() |
357
|
|
|
self.mock_get.return_value.json.return_value = data |
358
|
|
|
self.mock_get.return_value.status_code = 403 |
359
|
|
|
|
360
|
|
|
orphan_count = sum(1 for orphan in get_orphan_samples(limit=1)) |
361
|
|
|
self.assertEqual(orphan_count, 0) |
362
|
|
|
|