# previous = ''
# for word in word_list:
#
# if previous == word:
# continue
# previous = word
#
# for idx in range(0, len(word) - 1):
# print word[idx], ' - ', word[idx + 1]
#
#
# print '---------------------------------------------------------------------------------'
previous = ''
for item in connect_to_database.execute_select_query(SQL_SELECT_QUERY):
for word in item['bigram'].split('-'):
if previous == word:
continue
previous = word
SQL_INSERT_QUERY = ''
for idx in range(0, len(word) - 1):
SQL_INSERT_QUERY += 'INSERT INTO char_bigram_feature(bigram_id, doc_id, para_id, char_bigram) ' \
'VALUES ({}, {}, {}, {})'.format(item['bigram_id'], item['doc_id'], item['para_id'],
QuotedString(word[idx] + ' - ' + word[idx + 1])
.getquoted())
connect_to_database.execute_insert_query(SQL_INSERT_QUERY)
def get_num_of_doc_written_by_an_author(author_id):
if type(author_id) is not int:
return -1
SQL_SELECT_QUERY = "SELECT COUNT(doc_id) as num FROM document WHERE author_id = {};".format(
author_id)
return connect_to_database.execute_select_query(SQL_SELECT_QUERY)[0]['num']
def get_all_docs_by_author_id(author_id):
if type(author_id) is not int:
return -1
SQL_SELECT_QUERY = "SELECT doc_id, doc_title, year_of_pub FROM document WHERE author_id = {} ORDER " \
"BY doc_id;".format(author_id)
return connect_to_database.execute_select_query(SQL_SELECT_QUERY)
def get_all_author_id_and_name():
SQL_SELECT_QUERY = "SELECT author_id, author_name FROM author;"
return connect_to_database.execute_select_query(SQL_SELECT_QUERY)
def get_author_and_written_docs_count():
SQL_SELECT_QUERY = "SELECT d.author_id as aid, a.author_name as aname, COUNT(d.doc_id) as doc_num FROM document " \
"d INNER JOIN author a ON d.author_id = a.author_id GROUP BY d.author_id, a.author_name " \
"ORDER BY d.author_id;"
return connect_to_database.execute_select_query(SQL_SELECT_QUERY)
def get_total_num_of_docs_with_stylo_values():
SQL_SELECT_QUERY = "SELECT COUNT(DISTINCT doc_id) as number FROM fact;"
return connect_to_database.execute_select_query(
SQL_SELECT_QUERY)[0]['number']
def get_total_num_of_docs():
SQL_SELECT_QUERY = "SELECT COUNT(doc_id) as number FROM document;"
return connect_to_database.execute_select_query(
SQL_SELECT_QUERY)[0]['number']
def get_total_num_of_authors():
SQL_SELECT_QUERY = "SELECT COUNT(author_id) as number FROM author;"
return connect_to_database.execute_select_query(
SQL_SELECT_QUERY)[0]['number']