def multi_word_query(query):
query = word_split(query, symbol)
words = [word for word in query if word not in stop_words]
if words:
docs = [set(inverted[word].keys()) for word in words]
result_id = reduce(lambda x, y: x & y, docs) if docs else []
return list(result_id)
else:
return []
def get_distance(query, is_close=False):
split_words = word_split(query, symbol)
valid_distance = []
if not is_close:
words = split_words
distance = [1] * (len(words) - 1)
else:
words = split_words[::2]
distance = map(int, split_words[1::2])
valid_index = [index for index, word in enumerate(words) if word not in stop_words]
for x in xrange(len(valid_index) - 1):
valid_distance.append(sum(distance[valid_index[x]:valid_index[x + 1]]))
return [words[index] for index in valid_index], valid_distance