|
| 1 | +import os |
| 2 | +import vcr |
| 3 | +import pytest |
| 4 | +import requests |
| 5 | +from scrapinghub import HubstorageClient |
| 6 | +from scrapinghub.hubstorage.utils import urlpathjoin |
| 7 | + |
| 8 | + |
| 9 | +TEST_PROJECT_ID = "2222222" |
| 10 | +TEST_SPIDER_NAME = 'hs-test-spider' |
| 11 | +TEST_FRONTIER_NAME = 'test' |
| 12 | +TEST_FRONTIER_SLOT = 'site.com' |
| 13 | +TEST_BOTGROUPS = ['python-hubstorage-test', 'g1'] |
| 14 | + |
| 15 | +VCR_CASSETES_DIR = 'tests/hubstorage/cassetes' |
| 16 | + |
| 17 | +TEST_AUTH = os.getenv('HS_AUTH', 'f' * 32) |
| 18 | +TEST_ENDPOINT = os.getenv('HS_ENDPOINT', 'http://storage.vm.scrapinghub.com') |
| 19 | + |
| 20 | + |
| 21 | +@pytest.fixture |
| 22 | +def vcr_instance(scope='session'): |
| 23 | + return vcr.VCR( |
| 24 | + cassette_library_dir=VCR_CASSETES_DIR, |
| 25 | + record_mode='once', |
| 26 | + match_on=['uri', 'method'], |
| 27 | + ) |
| 28 | + |
| 29 | +@pytest.fixture(scope='session') |
| 30 | +def hsclient(): |
| 31 | + return HubstorageClient(auth=TEST_AUTH, endpoint=TEST_ENDPOINT) |
| 32 | + |
| 33 | + |
| 34 | +@pytest.fixture(scope='session') |
| 35 | +def hsproject(hsclient): |
| 36 | + return hsclient.get_project(TEST_PROJECT_ID) |
| 37 | + |
| 38 | + |
| 39 | +@pytest.fixture |
| 40 | +def hsspiderid(hsproject): |
| 41 | + return str(hsproject.ids.spider(TEST_SPIDER_NAME, create=1)) |
| 42 | + |
| 43 | + |
| 44 | +@pytest.fixture(autouse=True, scope='session') |
| 45 | +def setup_session(hsclient, hsproject): |
| 46 | + yield |
| 47 | + hsclient.close() |
| 48 | + |
| 49 | + |
| 50 | +@pytest.fixture(autouse=True) |
| 51 | +def setup_test(hsclient, hsproject, request, vcr_instance): |
| 52 | + cassette_name = '{}/{}.yaml'.format( |
| 53 | + request.function.__module__.split('.')[-1], |
| 54 | + request.function.__name__ |
| 55 | + ) |
| 56 | + with vcr_instance.use_cassette(cassette_name): |
| 57 | + _set_testbotgroup(hsproject) |
| 58 | + _remove_all_jobs(hsproject) |
| 59 | + yield |
| 60 | + _remove_all_jobs(hsproject) |
| 61 | + _unset_testbotgroup(hsproject) |
| 62 | + |
| 63 | +# ---------------------------------------------------------------------------- |
| 64 | + |
| 65 | + |
| 66 | +def _set_testbotgroup(hsproject): |
| 67 | + hsproject.settings.apipost(jl={'botgroups': [TEST_BOTGROUPS[0]]}) |
| 68 | + # Additional step to populate JobQ's botgroups table |
| 69 | + for botgroup in TEST_BOTGROUPS: |
| 70 | + url = urlpathjoin(TEST_ENDPOINT, 'botgroups', botgroup, 'max_running') |
| 71 | + requests.post(url, auth=hsproject.auth, data='null') |
| 72 | + hsproject.settings.expire() |
| 73 | + |
| 74 | + |
| 75 | +def _unset_testbotgroup(hsproject): |
| 76 | + hsproject.settings.apidelete('botgroups') |
| 77 | + hsproject.settings.expire() |
| 78 | + # Additional step to delete botgroups in JobQ |
| 79 | + for botgroup in TEST_BOTGROUPS: |
| 80 | + url = urlpathjoin(TEST_ENDPOINT, 'botgroups', botgroup) |
| 81 | + requests.delete(url, auth=hsproject.auth) |
| 82 | + |
| 83 | + |
| 84 | + |
| 85 | +def _remove_all_jobs(hsproject): |
| 86 | + for k in list(hsproject.settings.keys()): |
| 87 | + if k != 'botgroups': |
| 88 | + del hsproject.settings[k] |
| 89 | + hsproject.settings.save() |
| 90 | + |
| 91 | + # Cleanup JobQ |
| 92 | + jobq = hsproject.jobq |
| 93 | + for queuename in ('pending', 'running', 'finished'): |
| 94 | + info = {'summary': [None]} # poor-guy do...while |
| 95 | + while info['summary']: |
| 96 | + info = jobq.summary(queuename) |
| 97 | + for summary in info['summary']: |
| 98 | + _remove_job(hsproject, summary['key']) |
| 99 | + |
| 100 | + |
| 101 | +def _remove_job(hsproject, jobkey): |
| 102 | + hsproject.jobq.finish(jobkey) |
| 103 | + hsproject.jobq.delete(jobkey) |
| 104 | + _delete_job(hsproject, jobkey) |
| 105 | + |
| 106 | + |
| 107 | +def _delete_job(hsproject, jobkey): |
| 108 | + assert jobkey.startswith(TEST_PROJECT_ID), jobkey |
| 109 | + hsproject.jobs.apidelete(jobkey.partition('/')[2]) |
| 110 | + |
| 111 | + |
| 112 | +def start_job(hsproject, **startparams): |
| 113 | + jobdata = hsproject.jobq.start(**startparams) |
| 114 | + if jobdata: |
| 115 | + jobkey = jobdata.pop('key') |
| 116 | + jobauth = (jobkey, jobdata['auth']) |
| 117 | + return hsproject.get_job(jobkey, jobauth=jobauth, metadata=jobdata) |
0 commit comments