def clear_solr(buid): """Delete all jobs for a given business unit/job source.""" conn = Solr(settings.HAYSTACK_CONNECTIONS['default']['URL']) hits = conn.search(q="*:*", rows=1, mlt="false", facet="false").hits logging.info("BUID:%s - SOLR - Deleting all %s jobs" % (buid, hits)) conn.delete(q="buid:%s" % buid) logging.info("BUID:%s - SOLR - All jobs deleted." % buid)
class DirectSEOBase(TestCase): def setUp(self): db_backend = settings.DATABASES['default']['ENGINE'].split('.')[-1] # Set columns that are utf8 in production to utf8 if db_backend == 'mysql': cursor = connections['default'].cursor() cursor.execute("alter table seo_customfacet convert to character " "set utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_seositefacet convert to character " "set utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_company convert to character set " "utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_queryredirect convert to character " "set utf8 collate utf8_unicode_ci") cursor.execute("alter table taggit_tag convert to character set " "utf8 collate utf8_unicode_ci") cursor.execute("alter table taggit_taggeditem convert to " "character set " "utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_seositeredirect convert to " "character set utf8 collate utf8_unicode_ci") cursor.execute("alter table django_redirect convert to " "character set utf8 collate utf8_unicode_ci") setattr(settings, 'ROOT_URLCONF', 'dseo_urls') setattr(settings, "PROJECT", 'dseo') clear_url_caches() self.base_middleware_classes = settings.MIDDLEWARE_CLASSES middleware_classes = self.base_middleware_classes + ( 'wildcard.middleware.WildcardMiddleware', 'middleware.RedirectOverrideMiddleware') setattr(settings, 'MIDDLEWARE_CLASSES', middleware_classes) self.base_context_processors = settings.TEMPLATE_CONTEXT_PROCESSORS context_processors = self.base_context_processors + ( "social_links.context_processors.social_links_context", "seo.context_processors.site_config_context", ) setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', context_processors) context._standard_context_processors = None self.conn = Solr('http://127.0.0.1:8983/solr/seo') self.conn.delete(q="*:*") cache.clear() clear_url_caches() setattr(settings, 'MEMOIZE', False) def tearDown(self): from django.conf import settings from django.template import context setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', self.base_context_processors) context._standard_context_processors = None setattr(settings, 'MIDDLEWARE_CLASSES', self.base_middleware_classes)
def remove_expired_jobs(buid, active_jobs, upload_chunk_size=1024): """ Given a job source id and a list of active jobs for that job source, Remove the jobs on solr that are not among the active jobs. """ conn = Solr(settings.HAYSTACK_CONNECTIONS['default']['URL']) count = conn.search("*:*", fq="buid:%s" % buid, facet="false", mlt="false").hits old_jobs = conn.search("*:*", fq="buid:%s" % buid, facet="false", rows=count, mlt="false").docs active_ids = set(j['id'] for j in active_jobs) old_ids = set(j['id'] for j in old_jobs) expired = old_ids - active_ids chunks = chunk(list(expired), upload_chunk_size) for jobs in chunks: query = "id:(%s)" % " OR ".join([str(x) for x in jobs]) conn.delete(q=query) return expired
class MyJobsBase(TestCase): def setUp(self): from django.conf import settings setattr(settings, 'ROOT_URLCONF', 'myjobs_urls') cache.clear() clear_url_caches() self.ms_solr = Solr('http://127.0.0.1:8983/solr/seo') self.ms_solr.delete(q='*:*') setattr(settings, "PROJECT", 'myjobs') self.base_context_processors = settings.TEMPLATE_CONTEXT_PROCESSORS context_processors = self.base_context_processors + ( 'mymessages.context_processors.message_lists', ) setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', context_processors) setattr(settings, 'MEMOIZE', False) def tearDown(self): self.ms_solr.delete(q='*:*') setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', self.base_context_processors)
def delete_by_guid(guids): """ Removes a jobs from solr by guid. inputs: :guids: A list of guids outputs: The number of jobs that were requested to be deleted. This may be higher than the number of actual jobs deleted if a guid passed in did not correspond to a job in solr. """ conn = Solr(settings.HAYSTACK_CONNECTIONS['default']['URL']) if not guids: return 0 num_guids = len(guids) guids = chunk(guids) for guid_group in guids: delete_str = " OR ".join(guid_group) conn.delete(q="guid: (%s)" % delete_str) return num_guids
def remove_expired_jobs(buid, active_ids, upload_chunk_size=1024): """ Given a job source id and a list of active job ids for that job source, Remove the jobs on solr that are not among the active jobs. """ conn = Solr(settings.HAYSTACK_CONNECTIONS['default']['URL']) active_ids = set(active_ids) count = conn.search("*:*", fq="buid:%s" % buid, facet="false", mlt="false").hits old_jobs = conn.search("*:*", fq="buid:%s" % buid, facet="false", rows=count, mlt="false").docs old_ids = set(j['id'] for j in old_jobs) expired = old_ids - active_ids chunks = chunk(list(expired), upload_chunk_size) for jobs in chunks: query = "id:(%s)" % " OR ".join([str(x) for x in jobs]) conn.delete(q=query) return expired
def update_solr(buid, download=True, force=True, set_title=False, delete_feed=True, data_dir=DATA_DIR, clear_cache=False): """ Update the Solr master index with the data contained in a feed file for a given buid/jsid. This is meant to be a standalone function such that the state of the Solr index is not tied to the state of the database. Inputs: :buid: An integer; the ID for a particular business unit. :download: Boolean. If False, this process will not download a new feedfile, but instead use the one on disk. Should only be false for the purposes of our test suite. :force: Boolean. If True, every job seen in the feed file will be updated in the index. Otherwise, only the jobs seen in the feed file but not seen in the index will be updated. This latter option will soon be deprecated. Returns: A 2-tuple consisting of the number of jobs added and the number deleted. Writes/Modifies: Job data found in the feed file is used to modify the Solr index. This includes adds & deletes. (Solr does not have a discrete equivalent to SQL's UPDATE; by adding a document with the same UID as a document in the index, the equivalent of an update operation is performed.) """ if download: filepath = download_feed_file(buid, data_dir=data_dir) else: # Get current worker process id, to prevent race conditions. try: p = current_process() process_id = p.index except: process_id = 0 filepath = os.path.join(data_dir, str(process_id), FEED_FILE_PREFIX + str(buid) + '.xml') bu = BusinessUnit.objects.get(id=buid) try: co = bu.company_set.all()[0] except IndexError: co = None jobfeed = DEv2JobFeed(filepath, jsid=buid, markdown=bu.enable_markdown, company=co) # If the feed file did not pass validation, return. The return value is # '(0, 0)' to match what's returned on a successful parse. if jobfeed.errors: error = jobfeed.error_messages logging.error("BUID:%s - Feed file has failed validation on line %s. " "Exception: %s" % (buid, error['line'], error['exception'])) raise FeedImportError(error) # A dictionary of uids jobs = jobfeed.jobparse() # Build a set of all the UIDs for all those instances. job_uids = set([long(i.get('uid')) for i in jobs if i.get('uid')]) conn = Solr(settings.HAYSTACK_CONNECTIONS['default']['URL']) step1 = 1024 # Get the count of all the results in the Solr index for this BUID. hits = conn.search("*:*", fq="buid:%s" % buid, facet="false", mlt="false").hits # Create (start-index, stop-index) tuples to facilitate handling results # in ``step1``-sized chunks. So if ``hits`` returns 2048 results, # ``job_slices`` will look like ``[(0,1024), (1024, 2048)]``. Those # values are then used to slice up the total results. # # This was put in place because part of the logic to figuring out what # jobs to delete from and add jobs to the Solr index is using set # algebra. We convert the total list of UIDs in the index and the UIDs # in the XML feed to sets, then compare them via ``.difference()`` # (seen below). However for very large feed files, say 10,000+ jobs, # this process was taking so long that the connection would time out. To # address this problem we break up the comparisons as described above. # This results in more requests but it alleviates the connection timeout # issue. job_slices = slices(range(hits), step=step1) results = [_solr_results_chunk(tup, buid, step1) for tup in job_slices] solr_uids = reduce(lambda x, y: x | y, results) if results else set() # Return the job UIDs that are in the Solr index but not in the feed # file. solr_del_uids = solr_uids.difference(job_uids) if not force: # Return the job UIDs that are in the feed file but not in the Solr # index. solr_add_uids = job_uids.difference(solr_uids) # ``jobfeed.solr_jobs()`` yields a list of dictionaries. We want to # filter out any dictionaries whose "uid" key is not in # ``solr_add_uids``. This is because by default we only want to add # new documents (which each ``solr_jobs()`` dictionary represents), # not update. add_docs = filter(lambda x: int(x.get("uid", 0)) in solr_add_uids, jobfeed.solr_jobs()) else: # This might seem redundant to refer to the same value # twice with two different variable names. However, this decision # was made during the implementation of the "force Solr update" # feature to this function. # # Instead of adding only the documents with UIDs that are in the feed # file but not in the Solr index, we're going to add ALL the documents # in the feed file. This will add the new documents of course, but it # will also update existing documents with any new data. Uniqueness of # the documents is ensured by the ``id`` field defined in the Solr # schema (the template for which can be seen in # templates/search_configuration/solr.xml). At the very bottom you'll # see <uniqueKey>id</uniqueKey>. This serves as the equivalent of the pk # (i.e. globally unique) in a database. solr_add_uids = job_uids add_docs = jobfeed.solr_jobs() # Slice up ``add_docs`` in chunks of 4096. This is because the # maxBooleanClauses setting in solrconfig.xml is set to 4096. This means # if we used any more than that Solr would throw an error and our # updates wouldn't get processed. add_steps = slices(range(len(solr_add_uids)), step=4096) # Same concept as ``add_docs``. del_steps = slices(range(len(solr_del_uids)), step=4096) # Create a generator that yields 2-tuples with each invocation. The # 2-tuples consist of one tuple each from del_steps & add_steps. Any # mismatched values (e.g. there are more del_steps than add_steps) # will be compensated for with the ``fillvalue``. zipped_steps = izip_longest(del_steps, add_steps, fillvalue=(0, 0)) for tup in zipped_steps: update_chunk = add_docs[tup[1][0]:tup[1][1] + 1] if update_chunk: logging.debug("BUID:%s - SOLR - Update chunk: %s" % (buid, [i['uid'] for i in update_chunk])) # Pass 'commitWithin' so that Solr doesn't try to commit the new # docs right away. This will help relieve some of the resource # stress during the daily update. The value is expressed in # milliseconds. conn.add(update_chunk, commitWithin="30000") delete_chunk = _build_solr_delete_query( list(solr_del_uids)[tup[0][0]:tup[0][1] + 1]) if delete_chunk: # Post-a-job jobs should not be deleted during import delete_chunk = "(%s) AND -is_posted:true" % delete_chunk logging.debug("BUID:%s - SOLR - Delete chunk: %s" % (buid, list(solr_del_uids))) conn.delete(q=delete_chunk) # delete any jobs that may have been added via etl_to_solr conn.delete(q="buid:%s AND !uid:[0 TO *]" % buid) # Update business unit information: title, dates, and associated_jobs if set_title or not bu.title or (bu.title != jobfeed.job_source_name and jobfeed.job_source_name): bu.title = jobfeed.job_source_name updated = bool(solr_add_uids) or bool(solr_del_uids) _update_business_unit_modified_dates(bu, jobfeed.crawled_date, updated=updated) bu.associated_jobs = len(jobs) bu.save() # Update the Django database to reflect company additions and name changes add_company(bu) if delete_feed: os.remove(filepath) logging.info("BUID:%s - Deleted feed file." % buid) return len(solr_add_uids), len(solr_del_uids)
class SiteTestCase(DirectSEOBase): """ We're adding these tests to ensure unicode jobs descriptions and titles make it through the import process and work with high-level features. We should use http requests wherever possible since it's difficult to predict which modules will have issues with unicode. """ def setUp(self): super(SiteTestCase, self).setUp() self.conn = Solr('http://127.0.0.1:8983/solr/seo') self.conn.delete(q="*:*") self.businessunit = factories.BusinessUnitFactory(id=0) self.buid = self.businessunit.id self.filepath = os.path.join(import_jobs.DATA_DIR, 'dseo_feed_%s.xml' % self.buid) SeoSite.objects.all().delete() self.site = factories.SeoSiteFactory(id=1) self.configuration = factories.ConfigurationFactory(status=2) self.configuration.save() self.site.configurations.clear() self.site.configurations.add(self.configuration) def tearDown(self): super(SiteTestCase, self).tearDown() self.conn.delete(q="*:*") def test_unicode_title(self): # Test imports group = factories.GroupFactory() self.site.group = group self.site.business_units.add(self.businessunit) self.site.save() import_jobs.update_solr(self.buid, download=False, delete_feed=False, data_dir='seo/tests/data/') solr_jobs = self.conn.search("*:*") resp = self.client.get('/') self.assertEqual(resp.context['total_jobs_count'], solr_jobs.hits) # test standard facets against Haystack query standard_cf = factories.CustomFacetFactory.build( # default facet will return both jobs name="Keyword Facet", group=group, show_production=True) standard_cf.save() standard_cf.keyword.add(u'Ключевые') standard_cf.save() standard_site_facet = factories.SeoSiteFacetFactory( seosite=self.site, customfacet=standard_cf, facet_type=factories.SeoSiteFacet.STANDARD) standard_site_facet.save() # test standard facets against Haystack query standard_cf2 = factories.CustomFacetFactory.build( # default facet will return both jobs name='Country Facet', country='United States', group=group, show_production=True) standard_cf2.save() standard_site_facet2 = factories.SeoSiteFacetFactory( seosite=self.site, customfacet=standard_cf2, facet_type=factories.SeoSiteFacet.STANDARD) standard_site_facet2.save() resp = self.client.get('/keyword-facet/new-jobs/', HTTP_HOST=self.site.domain, follow=True) sqs = DESearchQuerySet().filter(text=u'Ключевые') self.assertEqual(len(resp.context['default_jobs']), sqs.count()) for facet_widget in resp.context['widgets']: # Ensure that no standard facet has more results than current # search results for count_tuple in facet_widget.items: self.assertTrue(sqs.count() >= count_tuple[1]) # Test default site facets against PySolr query from django.core.cache import cache cache.clear() default_cf = factories.CustomFacetFactory.build( name="Default Facet", title=u"Специалист", group=group, show_production=True) default_cf.save() default_site_facet = factories.SeoSiteFacetFactory( seosite=self.site, facet_type=factories.SeoSiteFacet.DEFAULT, customfacet=default_cf) default_site_facet.save() resp = self.client.get('/jobs/', HTTP_HOST=self.site.domain, follow=True) total_jobs = resp.context['total_jobs_count'] solr_jobs = self.conn.search(q=u"title:Специалист") self.assertEqual(total_jobs, solr_jobs.hits) self.assertEqual(len(resp.context['default_jobs']), total_jobs) for facet_widget in resp.context['widgets']: for count_tuple in facet_widget.items: self.assertTrue(sqs.count() >= count_tuple[1]) # Feed test resp = self.client.get('/feed/json', HTTP_HOST=self.site.domain) jobs = json.loads(resp.content) self.assertEqual(len(jobs), total_jobs) for job in jobs: resp = self.client.get(job['url'], HTTP_HOST=self.site.domain, follow=False) self.assertEqual(resp.status_code, 302) expected = 'http://my.jobs/%s%d?my.jobs.site.id=%s' %\ (job['guid'], settings.FEED_VIEW_SOURCES['json'], str(self.site.pk)) self.assertEqual(resp['Location'], expected) # Sitemap index Test - Since sitemap only builds out updates from the # last 30 days, this test will eventually be checking 0 jobs in sitemap # TODO, find a way to keep feed dates current. We might be able to use # the mock library to override datetime functions resp = self.client.get('/sitemap.xml', HTTP_HOST=self.site.domain) root = etree.fromstring(resp.content) self.assertGreater(len(root), 0) crawled_jobs = 0 for loc, lastmod in root: self.assertTrue(loc.text) resp = self.client.get(loc.text, HTTP_HOST=self.site.domain) self.assertEqual(resp.status_code, 200) # Get the first daily sitemap urlset = etree.fromstring(resp.content) # Check each job in daily sitemap - I'm a bot for loc, _, _, _ in urlset: resp = self.client.get(loc.text, HTTP_HOST=self.site.domain) self.assertEqual(resp.status_code, 200) self.assertIn(str(resp.context['the_job'].uid), loc.text) crawled_jobs += 1
class SitemapTestCase(DirectSEOBase): def setUp(self): super(SitemapTestCase, self).setUp() self.conn = Solr('http://127.0.0.1:8983/solr/seo') self.conn.add(SOLR_FIXTURE) def test_index(self): resp = self.client.get("/sitemap.xml") self.assertEqual(resp.status_code, 200) def test_no_buid_sitemap(self): """ Test to verify that a sitemap is generated with sites that have no BUID. """ site = SeoSite.objects.get(id=1) site.business_units = [] site.save() today = datetime.datetime.today() dt = datetime.date(*today.timetuple()[0:3]).isoformat() resp = self.client.get("/sitemap-" + dt + ".xml") self.assertTrue("<url>" in resp.content) def test_noreverse(self): """ Test to ensure that jobs with bad/ugly data do not block the creation of a sitemap page, but instead are just skipped over in `SolrSitemap.get_urls().` This is a regression test. It was prompted by a job in a job feed file having "~" in the "city" field. Because our URL pattern doesn't recognize that character in its regex, it caused a `NoReverseMatch` exception to be thrown. Instead of adding a tilde, we want to be able to handle any weird characters not specified in our URL config. """ # Sometimes the site settings are messed up from other tests. Ensure # that the settings are compatible with actually searching for the # jobs we're adding. settings.SITE_BUIDS = [] site = SeoSite.objects.get(pk=1) site.business_units = [] site.save() # These are kwargs from the actual error that created this error in the # first place. kwargs = { 'location': '~, WV', 'title': '911 Coordinator', 'uid': '25901630' } job = dict(SOLR_FIXTURE[0]) job.update(kwargs) self.conn.add([job]) today = datetime.datetime.now() dt = today.date().isoformat() resp = self.client.get("/sitemap-" + dt + ".xml") self.assertEqual(resp.status_code, 200) self.assertTrue("<url>" in resp.content) def tearDown(self): super(SitemapTestCase, self).tearDown() self.conn.delete("*:*")
class DirectSEOBase(TestCase): def setUp(self): db_backend = settings.DATABASES['default']['ENGINE'].split('.')[-1] # Set columns that are utf8 in production to utf8 if db_backend == 'mysql': cursor = connections['default'].cursor() cursor.execute("alter table seo_customfacet convert to character " "set utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_seositefacet convert to character " "set utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_company convert to character set " "utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_queryredirect convert to character " "set utf8 collate utf8_unicode_ci") cursor.execute("alter table taggit_tag convert to character set " "utf8 collate utf8_unicode_ci") cursor.execute("alter table taggit_taggeditem convert to " "character set " "utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_seositeredirect convert to " "character set utf8 collate utf8_unicode_ci") cursor.execute("alter table django_redirect convert to " "character set utf8 collate utf8_unicode_ci") # We have a data migration that does this, but we don't run # migrations during tests (Django 1.6.5 cursor.execute("ALTER TABLE django_flatpage CONVERT TO " "CHARACTER SET utf8 COLLATE utf8_general_ci") cursor.execute("ALTER TABLE seo_custompage CONVERT TO " "CHARACTER SET utf8 COLLATE utf8_general_ci") setattr(settings, 'ROOT_URLCONF', 'dseo_urls') setattr(settings, "PROJECT", 'dseo') clear_url_caches() self.base_middleware_classes = settings.MIDDLEWARE_CLASSES middleware_classes = self.base_middleware_classes + ( 'wildcard.middleware.WildcardMiddleware', 'middleware.RedirectOverrideMiddleware') setattr(settings, 'MIDDLEWARE_CLASSES', middleware_classes) self.base_context_processors = settings.TEMPLATE_CONTEXT_PROCESSORS context_processors = self.base_context_processors + ( "social_links.context_processors.social_links_context", "seo.context_processors.site_config_context", ) setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', context_processors) context._standard_context_processors = None self.conn = Solr('http://127.0.0.1:8983/solr/seo') self.conn.delete(q="*:*") cache.clear() clear_url_caches() setattr(settings, 'MEMOIZE', False) # As we added tests that created more and more companies, we # approached the hardcoded companies in import_jobs_testdata.json. # When we hit those ids, we began to get IntegrityErrors during # testing. Reset the sequence used by CompanyFactory to clear this # build-up. CompanyFactory.reset_sequence() def tearDown(self): from django.conf import settings from django.template import context setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', self.base_context_processors) context._standard_context_processors = None setattr(settings, 'MIDDLEWARE_CLASSES', self.base_middleware_classes)
class MyJobsBase(TestCase): def setUp(self): settings.ROOT_URLCONF = "myjobs_urls" settings.PROJECT = "myjobs" self.app_access = AppAccessFactory() self.activities = [ ActivityFactory(name=activity, app_access=self.app_access) for activity in [ "create communication record", "create contact", "create partner saved search", "create partner", "create role", "create tag", "create user", "delete tag", "delete partner", "delete role", "delete user", "read contact", "read communication record", "read partner saved search", "read partner", "read role", "read user", "read tag", "update communication record", "update contact", "update partner", "update role", "update tag", "update user", "read outreach email address", "create outreach email address", "delete outreach email address", "update outreach email address", "read outreach record", "convert outreach record" ] ] self.company = CompanyFactory(app_access=[self.app_access]) # this role will be populated by activities on a test-by-test basis self.role = RoleFactory(company=self.company, name="Admin") self.user = UserFactory(roles=[self.role], is_staff=True) cache.clear() clear_url_caches() self.ms_solr = Solr(settings.SOLR['seo_test']) self.ms_solr.delete(q='*:*') self.base_context_processors = settings.TEMPLATE_CONTEXT_PROCESSORS context_processors = self.base_context_processors + ( 'mymessages.context_processors.message_lists', ) setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', context_processors) setattr(settings, 'MEMOIZE', False) self.patcher = patch('urllib2.urlopen', return_file()) self.mock_urlopen = self.patcher.start() self.client = TestClient() self.client.login_user(self.user) def tearDown(self): self.ms_solr.delete(q='*:*') setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', self.base_context_processors) try: self.patcher.stop() except RuntimeError: # patcher was stopped in a test pass def assertRequires(self, view_name, *activities, **kwargs): """ Asserts that the given view is only accessible when a user has a role with the given activities. """ url = reverse(view_name, kwargs=kwargs.get('kwargs')) method = kwargs.get("method", "get").lower() response = getattr(self.client, method)(path=url) self.assertEqual(type(response), MissingActivity) self.role.activities = [ activity for activity in self.activities if activity.name in activities ] response = getattr(self.client, method)(path=url) self.assertNotEqual(type(response), MissingActivity) self.role.activities.clear()
class MyJobsBase(TestCase): def setUp(self): settings.ROOT_URLCONF = "myjobs_urls" settings.PROJECT = "myjobs" self.app_access = AppAccessFactory() self.activities = [ ActivityFactory(name=activity, app_access=self.app_access) for activity in [ "create communication record", "create contact", "create partner saved search", "create partner", "create role", "create tag", "create user", "delete tag", "delete partner", "delete role", "delete user", "read contact", "read communication record", "read partner saved search", "read partner", "read role", "read user", "read tag", "update communication record", "update contact", "update partner", "update role", "update tag", "update user", "read outreach email address", "create outreach email address", "delete outreach email address", "update outreach email address", "read outreach record", "convert outreach record"]] self.company = CompanyFactory(app_access=[self.app_access]) # this role will be populated by activities on a test-by-test basis self.role = RoleFactory(company=self.company, name="Admin") self.user = UserFactory(roles=[self.role], is_staff=True) cache.clear() clear_url_caches() self.ms_solr = Solr(settings.SOLR['seo_test']) self.ms_solr.delete(q='*:*') self.base_context_processors = settings.TEMPLATE_CONTEXT_PROCESSORS context_processors = self.base_context_processors + ( 'mymessages.context_processors.message_lists', ) setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', context_processors) setattr(settings, 'MEMOIZE', False) self.patcher = patch('urllib2.urlopen', return_file()) self.mock_urlopen = self.patcher.start() self.client = TestClient() self.client.login_user(self.user) def tearDown(self): self.ms_solr.delete(q='*:*') setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', self.base_context_processors) try: self.patcher.stop() except RuntimeError: # patcher was stopped in a test pass def assertRequires(self, view_name, *activities, **kwargs): """ Asserts that the given view is only accessible when a user has a role with the given activities. """ url = reverse(view_name, kwargs=kwargs.get('kwargs')) method = kwargs.get("method", "get").lower() response = getattr(self.client, method)(path=url) self.assertEqual(type(response), MissingActivity) self.role.activities = [activity for activity in self.activities if activity.name in activities] response = getattr(self.client, method)(path=url) self.assertNotEqual(type(response), MissingActivity) self.role.activities.clear()
class LoadETLTestCase(DirectSEOBase): fixtures = ['countries.json'] def setUp(self): self.solr_settings = { 'default': {'URL': 'http://127.0.0.1:8983/solr/seo'} } self.solr = Solr(settings.HAYSTACK_CONNECTIONS['default']['URL']) self.solr.delete(q="*:*") self.zipfile = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'data', 'ActiveDirectory_ce2ca701-eeca-4c13-96ba-e6bde9cb7060.zip') with open(self.zipfile) as zf: self.jobs = list(get_jobs_from_zipfile(zf, "ce2ca701-eeca-4c13-96ba-e6bde9cb7060")) self.businessunit = BusinessUnitFactory(id=0) self.buid = self.businessunit.id self.guid = 'ce2ca701-eeca-4c13-96ba-e6bde9cb7060' self.name = "Test" def tearDown(self): pass @patch('import_jobs.get_jobsfs_zipfile') def test_update_job_source(self, mock_jobsfs): mock_jobsfs.return_value = open(self.zipfile, 'rb') count = self.solr.search('*:*').hits self.assertEqual(count, 0, "Jobs for buid in solr before the test. Cannot guarantee correct behavior.") self.assertEqual(self.businessunit.associated_jobs, 4, "Initial Job Count does not match the factory") update_job_source(self.guid, self.buid, self.name) count = self.solr.search('buid:%s' % self.buid).hits # Note the job count being one low here is due to one job being filtered out due to include_in_index_bit self.assertEqual(count, 38, "38 Jobs not in solr after call to update job source. Found %s" % count) self.assertEqual(BusinessUnit.objects.get(id=self.buid).associated_jobs, 38, "Job Count not updated after imports: Should be 38 was %s" % self.businessunit.associated_jobs) def test_filtering_on_includeinindex_bit(self): """Test that filtering on the include_in_index bit works""" #Prove we have the expected number of jobs in the zipfile itself. self.assertEqual(len(self.jobs), 39, "Expected to find 0 jobs in the test zipfile, instead found %s" % len(self.jobs)) # Prove that filtering works. filtered_jobs = list(filter_current_jobs(self.jobs, self.businessunit)) self.assertEqual(len(filtered_jobs), 38, "filter_current_jobs should rmeove jobs with the includeinindex bit set, " "it's expected to return %s. Instead it returned %s" % (38, len(filtered_jobs))) def test_businessunit_ignore_includeinindex(self): """Test that filtering on the include_in_index bit can be overridden on a per business unit basis.""" # Set ignore_includeinindex on the test BusinessUnit self.businessunit.ignore_includeinindex = True self.businessunit.save() #Prove we have the expected number of jobs in the zipfile itself. self.assertEqual(len(self.jobs), 39, "Expected to find 0 jobs in the test zipfile, instead found %s" % len(self.jobs)) # Prove that filtering works. filtered_jobs = list(filter_current_jobs(self.jobs, self.businessunit)) self.assertEqual(len(filtered_jobs), 39, "filter_current_jobs should ignore the includeinindex bit, returning 39 jobs. " "Instead returned %s." % len(filtered_jobs))
class ImportJobsTestCase(DirectSEOBase): fixtures = ['import_jobs_testdata.json'] def setUp(self): super(ImportJobsTestCase, self).setUp() self.businessunit = BusinessUnitFactory(id=0) self.buid_id = self.businessunit.id self.filepath = os.path.join(DATA_DIR, '0', 'dseo_feed_%s.xml' % self.buid_id) self.solr_settings = { 'default': {'URL': 'http://127.0.0.1:8983/solr/seo'} } self.solr = Solr(settings.HAYSTACK_CONNECTIONS['default']['URL']) def tearDown(self): super(ImportJobsTestCase, self).tearDown() self.solr.delete(q='*:*') def test_solr_rm_feedfile(self): """ Test that at the end of Solr parsing, the feed file is deleted. """ update_solr(self.buid_id) self.assertFalse(os.access(self.filepath, os.F_OK)) def test_subsidiary_rename(self): company1 = CompanyFactory() company1.save() bu1 = self.businessunit bu1.title = "Acme corp" bu2 = BusinessUnitFactory(title=bu1.title) bu2.save() self.businessunit.company_set.add(company1) # Test that a company was created for both business units add_company(bu1) companies = bu1.company_set.all() self.assertEqual(len(companies), 1) co = companies[0] self.assertEqual(co.name, bu1.title) # Add the 2nd business unit add_company(bu2) # Both units should be attached to that company self.assertEqual(bu1.company_set.all()[0], bu2.company_set.all()[0]) self.assertEqual(bu1.company_set.all().count(), 1) self.assertIn(bu1, co.job_source_ids.all()) self.assertIn(bu2, co.job_source_ids.all()) self.assertEqual(co.name, bu1.title) self.assertEqual(co.name, bu2.title) bu2.title = "New company name" add_company(bu1) add_company(bu2) self.assertEqual(len(co.job_source_ids.all()), 1) self.assertNotEqual(bu1.company_set.all(), bu2.company_set.all()) self.assertEqual(co.name, bu1.title) self.assertEqual(len(bu2.company_set.all()), 1) co2 = bu2.company_set.all()[0] self.assertEqual(co2.name, bu2.title) self.assertNotEqual(co2.name, bu1.title) self.assertNotEqual(co.name, bu2.title) def test_duplicate_company(self): company1 = CompanyFactory() company1.save() company2 = CompanyFactory(name="Acme corp") company2.save() self.businessunit.company_set.add(company1) self.businessunit.title = "Acme corp" add_company(self.businessunit) self.assertEqual(self.businessunit.company_set.all()[0], company2) def test_set_bu_title(self): """ Ensure that if a feedfile for a BusinessUnit comes through, and the `title` attribute for that BusinessUnit is not set, that `helpers.update_solr` sets the `title` attribute properly. """ bu = BusinessUnit.objects.get(id=self.buid_id) bu.title = None bu.save() # Since the BusinessUnit title is None, the intent is that update_solr # will set its title to match the company name found in the feed file. results = update_solr(self.buid_id) # We have to get the updated state of the BusinessUnit instance, since # changes to the database won't be reflected by our in-memory version of # the data. bu = BusinessUnit.objects.get(id=self.buid_id) # The title attribute should now equal the initial value established in # the setUp method. self.assertEquals(self.businessunit.title, bu.title) def test_add_company(self): """ Create environment to test for every possible case-- - Existing relationship but the name is different pk=10 - No existing relationship, but the company exists in the database (as established by the BusinessUnit title matching a company name) pk=11 - No relationship and the company is not in the database pk=12 Start with 2 Company objects and 3 BusinessUnit objects End up with 3 Company objects and 3 BusinessUnit objects """ for i in range(10, 4): add_company(BusinessUnit.get(id=i)) # The names of the BU and the Co should be the same self.assertEquals(BusinessUnit.get(id=i).title, Company.get(id=i).name, msg="Company names do not match") # ensure the relationship was formed self.assertIn(Company.objects.get(id=i), BusinessUnit.objects.get(id=i).company_set.all(), msg="Company is not related to job feed") def test_remove_expired_jobs(self): buid = 12345 active_jobs = [{'id': 'seo.%s' % i, 'buid': buid} for i in range(4)] old_jobs = [{'id': 'seo.%s' % i, 'buid': buid} for i in range(2, 10)] with self.settings(HAYSTACK_CONNECTIONS=self.solr_settings): self.solr.add(old_jobs) self.solr.commit() removed = remove_expired_jobs(buid, [d['id'] for d in active_jobs]) self.assertEqual(len(removed), 6, "Removed jobs %s" % removed) ids = [d['id'] for d in self.solr.search('*:*').docs] self.assertTrue([5, 6, 7, 8, 9, 10] not in ids)
class SiteTestCase(DirectSEOBase): """ We're adding these tests to ensure unicode jobs descriptions and titles make it through the import process and work with high-level features. We should use http requests wherever possible since it's difficult to predict which modules will have issues with unicode. """ def setUp(self): super(SiteTestCase, self).setUp() self.conn = Solr('http://127.0.0.1:8983/solr/seo') self.conn.delete(q="*:*") self.businessunit = factories.BusinessUnitFactory(id=0) self.buid = self.businessunit.id self.filepath = os.path.join(import_jobs.DATA_DIR, 'dseo_feed_%s.xml' % self.buid) SeoSite.objects.all().delete() self.site = factories.SeoSiteFactory(id=1) self.configuration = factories.ConfigurationFactory(status=2) self.configuration.save() self.site.configurations.clear() self.site.configurations.add(self.configuration) def tearDown(self): super(SiteTestCase, self).tearDown() self.conn.delete(q="*:*") def test_unicode_title(self): # Test imports group = factories.GroupFactory() self.site.group = group self.site.business_units.add(self.businessunit) self.site.save() import_jobs.update_solr(self.buid, download=False, delete_feed=False, data_dir='seo/tests/data/') solr_jobs = self.conn.search("*:*") resp = self.client.get('/') self.assertEqual(resp.context['total_jobs_count'], solr_jobs.hits) # test standard facets against Haystack query standard_cf = factories.CustomFacetFactory.build( # default facet will return both jobs name="Keyword Facet", group=group, show_production=True) standard_cf.save() standard_cf.keyword.add(u'Ключевые') standard_cf.save() standard_site_facet = factories.SeoSiteFacetFactory( seosite=self.site, customfacet=standard_cf, facet_type=factories.SeoSiteFacet.STANDARD) standard_site_facet.save() # test standard facets against Haystack query standard_cf2 = factories.CustomFacetFactory.build( # default facet will return both jobs name='Country Facet', country='United States', group=group, show_production=True) standard_cf2.save() standard_site_facet2 = factories.SeoSiteFacetFactory( seosite=self.site, customfacet=standard_cf2, facet_type=factories.SeoSiteFacet.STANDARD) standard_site_facet2.save() resp = self.client.get('/keyword-facet/new-jobs/', HTTP_HOST=self.site.domain, follow=True) sqs = DESearchQuerySet().filter(text=u'Ключевые') self.assertEqual(len(resp.context['default_jobs']), sqs.count()) for facet_widget in resp.context['widgets']: # Ensure that no standard facet has more results than current # search results for count_tuple in facet_widget.items: self.assertTrue(sqs.count() >= count_tuple[1]) # Test default site facets against PySolr query from django.core.cache import cache cache.clear() default_cf = factories.CustomFacetFactory.build(name="Default Facet", title=u"Специалист", group=group, show_production=True) default_cf.save() default_site_facet = factories.SeoSiteFacetFactory( seosite=self.site, facet_type=factories.SeoSiteFacet.DEFAULT, customfacet=default_cf) default_site_facet.save() resp = self.client.get('/jobs/', HTTP_HOST=self.site.domain, follow=True) total_jobs = resp.context['total_jobs_count'] solr_jobs = self.conn.search(q=u"title:Специалист") self.assertEqual(total_jobs, solr_jobs.hits) self.assertEqual(len(resp.context['default_jobs']), total_jobs) for facet_widget in resp.context['widgets']: for count_tuple in facet_widget.items: self.assertTrue(sqs.count() >= count_tuple[1]) # Feed test resp = self.client.get('/feed/json', HTTP_HOST=self.site.domain) jobs = json.loads(resp.content) self.assertEqual(len(jobs), total_jobs) for job in jobs: resp = self.client.get(job['url'], HTTP_HOST=self.site.domain, follow=False) self.assertEqual(resp.status_code, 302) expected = 'https://my.jobs/%s%d?my.jobs.site.id=%s' %\ (job['guid'], settings.FEED_VIEW_SOURCES['json'], str(self.site.pk)) self.assertEqual(resp['Location'], expected) # Sitemap index Test - Since sitemap only builds out updates from the # last 30 days, this test will eventually be checking 0 jobs in sitemap # TODO, find a way to keep feed dates current. We might be able to use # the mock library to override datetime functions resp = self.client.get('/sitemap.xml', HTTP_HOST=self.site.domain) root = etree.fromstring(resp.content) self.assertGreater(len(root), 0) crawled_jobs = 0 for loc, lastmod in root: self.assertTrue(loc.text) resp = self.client.get(loc.text, HTTP_HOST=self.site.domain) self.assertEqual(resp.status_code, 200) # Get the first daily sitemap urlset = etree.fromstring(resp.content) # Check each job in daily sitemap - I'm a bot for loc, _, _, _ in urlset: resp = self.client.get(loc.text, HTTP_HOST=self.site.domain) self.assertEqual(resp.status_code, 200) self.assertIn(str(resp.context['the_job'].uid), loc.text) crawled_jobs += 1
class DirectSEOBase(TestCase): def setUp(self): db_backend = settings.DATABASES['default']['ENGINE'].split('.')[-1] # Set columns that are utf8 in production to utf8 if db_backend == 'mysql': cursor = connections['default'].cursor() cursor.execute("alter table seo_customfacet convert to character " "set utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_seositefacet convert to character " "set utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_company convert to character set " "utf8 collate utf8_unicode_ci") cursor.execute( "alter table seo_queryredirect convert to character " "set utf8 collate utf8_unicode_ci") cursor.execute("alter table taggit_tag convert to character set " "utf8 collate utf8_unicode_ci") cursor.execute("alter table taggit_taggeditem convert to " "character set " "utf8 collate utf8_unicode_ci") cursor.execute("alter table seo_seositeredirect convert to " "character set utf8 collate utf8_unicode_ci") cursor.execute("alter table django_redirect convert to " "character set utf8 collate utf8_unicode_ci") # We have a data migration that does this, but we don't run # migrations during tests (Django 1.6.5 cursor.execute("ALTER TABLE django_flatpage CONVERT TO " "CHARACTER SET utf8 COLLATE utf8_general_ci") cursor.execute("ALTER TABLE seo_custompage CONVERT TO " "CHARACTER SET utf8 COLLATE utf8_general_ci") setattr(settings, 'ROOT_URLCONF', 'dseo_urls') setattr(settings, "PROJECT", 'dseo') clear_url_caches() self.base_middleware_classes = settings.MIDDLEWARE_CLASSES middleware_classes = self.base_middleware_classes + ( 'wildcard.middleware.WildcardMiddleware', 'middleware.RedirectOverrideMiddleware') setattr(settings, 'MIDDLEWARE_CLASSES', middleware_classes) self.base_context_processors = settings.TEMPLATE_CONTEXT_PROCESSORS context_processors = self.base_context_processors + ( "social_links.context_processors.social_links_context", "seo.context_processors.site_config_context", ) setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', context_processors) context._standard_context_processors = None self.conn = Solr('http://127.0.0.1:8983/solr/seo') self.conn.delete(q="*:*") cache.clear() clear_url_caches() setattr(settings, 'MEMOIZE', False) # As we added tests that created more and more companies, we # approached the hardcoded companies in import_jobs_testdata.json. # When we hit those ids, we began to get IntegrityErrors during # testing. Reset the sequence used by CompanyFactory to clear this # build-up. CompanyFactory.reset_sequence() def tearDown(self): from django.conf import settings from django.template import context setattr(settings, 'TEMPLATE_CONTEXT_PROCESSORS', self.base_context_processors) context._standard_context_processors = None setattr(settings, 'MIDDLEWARE_CLASSES', self.base_middleware_classes)