def test_get_merged_posting_lists_empty(self):
postinglist_1 = [(0,5),(1,1),(2,2)]
postinglist_2 =[]
expected_merged_list= [(0,5),(1,1),(2,2)]
merged_list_1 = index_operations.get_merged_posting_lists(postinglist_1,postinglist_2)
merged_list_2 = index_operations.get_merged_posting_lists(postinglist_2,postinglist_1)
self.assertEqual(merged_list_1,expected_merged_list)
self.assertEqual(merged_list_2,expected_merged_list)
def get_seed_words_posting_lists(seed_words_spec, index_directory, indices_id):
# Return a map of postings for each reference_word_group_id
posting_lists={}
# set up structure of posting list map
for category in seed_words_spec:
posting_lists[category] = {}
for index_id in indices_id:
index = cache.load(index_directory, index_id)
print("merging postining_list for index:")
print(index["index_type"])
index = index["index"]
for category in seed_words_spec:
reference_word_groups = seed_words_spec[category]
for id_reference_word_group in reference_word_groups:
if not id_reference_word_group in posting_lists[category]:
posting_lists[category][id_reference_word_group] = []
for reference_word in reference_word_groups[id_reference_word_group]:
reference_term = n_gram_handler.string_to_index_term(reference_word)
if reference_term in index:
merged_postings = index_operations.get_merged_posting_lists(posting_lists[category][id_reference_word_group], index[reference_term])
posting_lists[category][id_reference_word_group] = merged_postings
return posting_lists
def test_get_merged_posting_lists(self):
postinglist_1 = [("0",5),("1",1),("2",2)]
postinglist_2 =[("1",1),("3",3)]
expected_merged_list= [("0",5),("1",2),("2",2),("3",3)]
merged_list = index_operations.get_merged_posting_lists(postinglist_1,postinglist_2)
self.assertEqual(merged_list,expected_merged_list)
