def test_parallel(start):
pairs_file = open(os.path.join(os.pardir, "data", "pairs.pkl"), 'rb')
pairs = pickle.load(pairs_file)
pairs_file.close()
synonyms_file = open(os.path.join(os.pardir, "data", "synonyms.pkl"), 'rb')
synonyms = pickle.load(synonyms_file)
synonyms_file.close()
pattern_matcher = PatternMatcher()
connection = pymysql.connect(host='localhost',
user='******',
password='******'.format(sys.argv[1]),
db='stackoverflow')
with connection.cursor() as cursor:
sql = "SELECT Id, Body FROM Posts WHERE Score >= 0 AND Id >= {} AND Id < {}".format(
start, start + 100)
cursor.execute(sql)
for i in range(cursor.rowcount):
# post_count += 1
current_id, row = cursor.fetchone()
word_list = get_words(row)
# total_sent_count += len(word_list)
for words in word_list:
rtn = check_tech_pairs(words)
if rtn is not None:
words = rtn[0].split(" ")
pattern_matcher.match_pattern(words, current_id, rtn[1],
"keytechs")
