doc_id = document['id']
tokens = []
tokens = tokenize(document['title'])
body = tokenize(document['body'])
tokens.extend(body)
# Metadata
global max_doc
max_doc += 1
doc_length[document['id']] = len(document['title']) + len(document['body'])
for token in tokens
add_token_to_index(token, doc_id)
def create_index()
for document in read_documents()
add_to_index(document)
print (Created index with size {}.format(len(inverted_index)))
creat_synonyms_list()
create_index()
if __name__ == '__main__'
all_queries = [query for query in read_queries() if query['query number'] != 0]
for query in all_queries
documents = search_query(query)
print (Query{} and Results{}.format(query, documents))
def create_index():
for document in read_documents():
add_to_index(document)
print("Created index with size {}".format(len(inverted_index)))
def create_index():
global doc_count
for document in read_documents():
doc_count += 1
add_to_index(document)
print("Created index with size {}".format(len(inverted_index)))