class CloudSearchConnectionTest(unittest.TestCase):
cloudsearch = True
def setUp(self):
super(CloudSearchConnectionTest, self).setUp()
self.conn = SearchConnection(
endpoint='test-domain.cloudsearch.amazonaws.com'
)
def test_expose_additional_error_info(self):
mpo = mock.patch.object
fake = FakeResponse()
fake.content = b'Nopenopenope'
# First, in the case of a non-JSON, non-403 error.
with mpo(requests, 'get', return_value=fake) as mock_request:
with self.assertRaises(SearchServiceException) as cm:
self.conn.search(q='not_gonna_happen')
self.assertTrue('non-json response' in str(cm.exception))
self.assertTrue('Nopenopenope' in str(cm.exception))
# Then with JSON & an 'error' key within.
fake.content = json.dumps({
'error': "Something went wrong. Oops."
}).encode('utf-8')
with mpo(requests, 'get', return_value=fake) as mock_request:
with self.assertRaises(SearchServiceException) as cm:
self.conn.search(q='no_luck_here')
self.assertTrue('Unknown error' in str(cm.exception))
self.assertTrue('went wrong. Oops' in str(cm.exception))
def test_cloudsearch_bqsearch(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(bq="'Test'")
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b'bq'], [b"'Test'"])
def test_cloudsearch_facet_multiple(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', facet=["author", "cat"])
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b'facet'], [b"author,cat"])
def test_cloudsearch_facet_sort_single(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', facet_sort={'author': 'alpha'})
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b'facet-author-sort'], [b'alpha'])
def test_cloudsearch_facet_single(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', facet=["Author"])
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args['facet'], ["Author"])
def test_cloudsearch_top_n_single(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', facet_top_n={'author': 5})
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b'facet-author-top-n'], [b'5'])
def test_cloudsearch_rank_multiple(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', rank=["date", "score"])
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b'rank'], [b'date,score'])
def test_cloudsearch_result_fields_multiple(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', return_fields=['author', 'title'])
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b'return-fields'], [b'author,title'])
def test_cloudsearch_t_field_single(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', t={'year': '2001..2007'})
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b't-year'], [b'2001..2007'])
def test_cloudsearch_t_field_multiple(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', t={'year':'2001..2007', 'score':'10..50'})
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args['t-year'], ['2001..2007'])
self.assertEqual(args['t-score'], ['10..50'])
def test_cloudsearch_top_n_multiple(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', facet_top_n={'author': 5, 'cat': 10})
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args['facet-author-top-n'], ['5'])
self.assertEqual(args['facet-cat-top-n'], ['10'])
def test_cloudsearch_search_details(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test', size=50, start=20)
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b'q'], [b"Test"])
self.assertEqual(args[b'size'], [b"50"])
self.assertEqual(args[b'start'], [b"20"])
def test_cloudsearch_qsearch(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(q='Test')
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b'q'], [b"Test"])
self.assertEqual(args[b'start'], [b"0"])
self.assertEqual(args[b'size'], [b"10"])
def test_cloudsearch_facet_constraint_single(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(
q='Test',
facet_constraints={'author': "'John Smith','Mark Smith'"})
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args[b'facet-author-constraints'],
[b"'John Smith','Mark Smith'"])
def test_cloudsearch_facet_constraint_multiple(self):
search = SearchConnection(endpoint=HOSTNAME)
search.search(
q='Test',
facet_constraints={'author': "'John Smith','Mark Smith'",
'category': "'News','Reviews'"})
args = self.get_args(HTTPretty.last_request.raw_requestline)
self.assertEqual(args['facet-author-constraints'],
["'John Smith','Mark Smith'"])
self.assertEqual(args['facet-category-constraints'],
["'News','Reviews'"])
def test_cloudsearch_results_internal_consistancy(self):
"""Check the documents length matches the iterator details"""
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test')
self.assertEqual(len(results), len(results.docs))
def test_cloudsearch_results_info(self):
"""Check num_pages_needed is calculated correctly"""
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test')
# This relies on the default response which is fed into HTTPretty
self.assertEqual(results.num_pages_needed, 3.0)
def test_cloudsearch_results_meta(self):
"""Check returned metadata is parsed correctly"""
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test')
# These rely on the default response which is fed into HTTPretty
self.assertEqual(results.rank, "-text_relevance")
self.assertEqual(results.match_expression, "Test")
def test_cloudsearch_search_facets(self):
#self.response['facets'] = {'tags': {}}
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test', facet=['tags'])
self.assertTrue('tags' not in results.facets)
self.assertEqual(results.facets['animals'], {u'lions': u'1', u'fish': u'2'})
def test_cloudsearch_results_iterator(self):
"""Check the results iterator"""
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test')
results_correct = iter(["12341", "12342", "12343", "12344",
"12345", "12346", "12347"])
for x in results:
self.assertEqual(x['id'], next(results_correct))
def test_cloudsearch_results_meta(self):
"""Check returned metadata is parsed correctly"""
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test')
# These rely on the default response which is fed into HTTPretty
self.assertEqual(results.rank, "-text_relevance")
self.assertEqual(results.match_expression, "Test")
def test_cloudsearch_results_iterator(self):
"""Check the results iterator"""
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test')
results_correct = iter(
["12341", "12342", "12343", "12344", "12345", "12346", "12347"])
for x in results:
self.assertEqual(x['id'], next(results_correct))
def test_cloudsearch_results_hits(self):
"""Check that documents are parsed properly from AWS"""
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test')
hits = list(map(lambda x: x['id'], results.docs))
# This relies on the default response which is fed into HTTPretty
self.assertEqual(
hits, ["12341", "12342", "12343", "12344",
"12345", "12346", "12347"])
def test_cloudsearch_search_facets(self):
#self.response['facets'] = {'tags': {}}
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test', facet=['tags'])
self.assertTrue('tags' not in results.facets)
self.assertEqual(results.facets['animals'], {
u'lions': u'1',
u'fish': u'2'
})
def test_cloudsearch_results_hits(self):
"""Check that documents are parsed properly from AWS"""
search = SearchConnection(endpoint=HOSTNAME)
results = search.search(q='Test')
hits = map(lambda x: x['id'], results.docs)
# This relies on the default response which is fed into HTTPretty
self.assertEqual(
hits, ["12341", "12342", "12343", "12344",
"12345", "12346", "12347"])
def search(cls, q=None, bq=None, rank=None, start=0, **kwargs):
'''Search using CloudSearch. This requires a _cs_search_endpoint property to be set
:param q: The optional TEXT search query
:type q: str
:param bq: The optional BOOLEAN search query
:type bq: str
:param rank: The optional Search rank
:type rank: str
:param start: The optional start point to begin the search
:type start: int
:param: Other KW args are supported and used as direct matches in the Boolean Query'''
from boto.cloudsearch.search import SearchConnection
if not cls._cs_search_endpoint:
raise NotImplemented('No CloudSearch Domain Set')
# Converts all the keyword arguments to a boolaen query
if kwargs:
query_parts = []
if bq:
query_parts.append(bq)
# Build all the query parts
for arg in kwargs:
query_parts.append('%s:\'%s\'' % (arg, kwargs[arg]))
# Reduce the query back to a single string
bq = '(and %s)' % ' '.join(query_parts)
# Build the search args
args = {}
if q:
args['q'] = q
if bq:
args['bq'] = bq
if rank:
args['rank'] = rank
if start:
args['start'] = start
conn = SearchConnection(endpoint=cls._cs_search_endpoint)
return BatchItemFetcher(conn.search(**args), cls)
def test_response(self):
search = SearchConnection(endpoint=HOSTNAME)
with self.assertRaisesRegexp(SearchServiceException, 'foo bar baz'):
search.search(q='Test')
def test_simplejson_jsondecodeerror(self):
with mock.patch.object(json, 'loads', fake_loads_json_error):
search = SearchConnection(endpoint=HOSTNAME)
with self.assertRaisesRegexp(SearchServiceException, 'non-json'):
search.search(q='test')
def test_response(self):
search = SearchConnection(endpoint=HOSTNAME)
with self.assertRaisesRegexp(SearchServiceException, 'foo bar baz'):
search.search(q='Test')
def test_simplejson_jsondecodeerror(self):
with mock.patch.object(json, 'loads', fake_loads_json_error):
search = SearchConnection(endpoint=HOSTNAME)
with self.assertRaisesRegexp(SearchServiceException, 'non-json'):
search.search(q='test')
