def _check_cache_results(self, indexpath, cachepath, cacheid, expected_results, num_results=10):
# set cache manager
cm = XapianMultipleCachesManager(cachepath)
cm.add_cache(cacheid)
cm.select_cache(cacheid)
search_conn = SearchConnection(indexpath)
search_conn.set_cache_manager(cm)
query_id, query_term = (0, "term_a")
cache_query_id = cm.get_queryid(query_term) # obtain query_id from the cache
self.assertEqual(query_id, cache_query_id)
non_cached, cached = expected_results
query = search_conn.query_field("field", query_term)
base_result = [r.id for r in query.search(0, num_results)]
# see if the results without merging the query are ok
self.assertEqual(non_cached, base_result)
cached_query = query.merge_with_cached(query_id)
cached_result = [r.id for r in cached_query.search(0, num_results)]
# test the merged query result
self.assertEqual(cached, cached_result)
search_conn.close()
cm.close()
def test_single_cache_applying(self):
with tempdir() as basepath:
# create an index
indexpath = os.path.join(basepath, "test_index")
self._create_index(indexpath)
# create and apply cache
cachepath = os.path.join(basepath, "cache")
os.makedirs(cachepath)
cache_manager = XapianMultipleCachesManager(cachepath)
cache_manager.add_cache("1")
cache_manager.select_cache("1")
self._create_and_apply_cache(indexpath, cache_manager)
self._check_cache_results(indexpath, cachepath, "1", [["5", "4", "3", "2", "1"], ["4", "2", "5", "3", "1"]])
def test_multiple_cache(self):
with tempdir() as basepath:
# create an index
indexpath = os.path.join(basepath, "test_index")
self._create_index(indexpath)
base_cachepath = os.path.join(basepath, "cache")
os.makedirs(base_cachepath)
# create and apply cache 1
cache_manager = XapianMultipleCachesManager(base_cachepath)
cache_manager.add_cache("1")
cache_manager.select_cache("1")
self._create_and_apply_cache(indexpath, cache_manager)
# create and apply cache 2
cache_manager.add_cache("cache2")
cache_manager.select_cache("cache2")
self._create_and_apply_cache(indexpath, cache_manager)
# test cache 1
self._check_cache_results(
indexpath, base_cachepath, "1", [["5", "4", "3", "2", "1"], ["4", "2", "5", "3", "1"]]
)
# test cache 2
self._check_cache_results(
indexpath, base_cachepath, "cache2", [["5", "4", "3", "2", "1"], ["3", "4", "1", "5", "2"]]
)
# the document whose docid is 4 is in both caches, we're
# testing here if replacing it with one cache manager set
# will change the result in the other cache. It must change.
# replace document
iconn = IndexerConnection(indexpath)
cache_manager = XapianMultipleCachesManager(base_cachepath)
cache_manager.add_cache("1")
cache_manager.add_cache("cache2")
iconn.set_cache_manager(cache_manager)
docid, terms = ("4", [("term_a", 4), ("term_b", 2)])
pdoc = self._create_processed_doc(iconn, docid, terms)
iconn.replace(pdoc, xapid=int(docid))
iconn.flush()
iconn.close()
cache_manager.close()
# check if the results in both caches are ok
self._check_cache_results(
indexpath, base_cachepath, "1", [["5", "4", "3", "2", "1"], ["4", "2", "5", "3", "1"]]
)
self._check_cache_results(
indexpath, base_cachepath, "cache2", [["5", "4", "3", "2", "1"], ["3", "4", "1", "5", "2"]]
)
# there are 2 code pathes when we deal with caches:
# 1. the cache has not enough results
# 2. the cache has enough results
# in the first case, the result will come from a mixed query
# against the index. In the second, the results will come from
# the cache_manger. So, all the cache managers must be updated.
# remove document
cache_manager = XapianMultipleCachesManager(base_cachepath)
cache_manager.add_cache("1")
cache_manager.add_cache("cache2")
iconn = IndexerConnection(indexpath)
iconn.set_cache_manager(cache_manager)
iconn.delete(xapid=4)
cache_manager.close()
iconn.flush()
iconn.close()
# cache has not enough results
self._check_cache_results(indexpath, base_cachepath, "1", [["5", "3", "2", "1"], ["2", "5", "3", "1"]])
self._check_cache_results(indexpath, base_cachepath, "cache2", [["5", "3", "2", "1"], ["3", "1", "5", "2"]])
# cache has enough results
self._check_cache_results(indexpath, base_cachepath, "1", [["5"], ["2"]], num_results=1)
self._check_cache_results(indexpath, base_cachepath, "cache2", [["5", "3"], ["3", "1"]], num_results=2)
