def cut_year(self, sql):
with connect() as cur:
cur.execute(sql)
result = cur.fetchall()
pattern = re.compile(r'\d{4}')
words = re.findall(pattern, str(result))
return words
def i_inc(data):
with passwd.connect() as cur:
try:
sql = ('insert into luhan_inc(user_domain, user_name,\
user_comment) values (%s, %s, %s)')
cur.execute(sql, data)
except Exception as e:
print(e, 'error')
def get_oneproxie():
sql = 'select * from ip_list'
with passwd.connect() as cur:
cur.execute(sql)
res = cur.fetchall()
one = random.choice(res)
proxie = {one[2]: one[2] + '://' + one[0] + ':' + one[1]}
return proxie
def c_table_luhan_inc():
sql = 'create table luhan_inc(user_domain varchar(16),\
user_name varchar(32), user_comment varchar(512))'
with passwd.connect() as cur:
try:
cur.execute(sql)
except Exception as e:
print(e)
def c_table_luhan_sab():
sql = 'create table luhan_sab(num int unique auto_increment,user_domain\
varchar(16) unique, user_sex char(4), user_area varchar(32),\
user_birth varchar(16))'
with passwd.connect() as cur:
try:
cur.execute(sql)
except Exception as e:
print(e)
def get_proxies():
proxies = []
sql = 'select * from ip_list'
with passwd.connect() as cur:
cur.execute(sql)
res = cur.fetchall()
for procotol, ip, port in res:
proxie = {res[2]: res[2] + '://' + res[0] + ':' + res[1]}
proxies.append(proxie)
return proxies
def c_table_ip_list():
sql = 'create table ip_list(id int unique auto_increment,\
ip varchar(16) primary key, port varchar(8) not null,\
procotol varchar(8) not null)'
with passwd.connect() as cur:
try:
cur.execute(sql)
print('成功创建')
except Exception:
print('表已经存在，可继续操作，或者删除表后重新运行。')
def cut_zh(self, sql, cut=False):
with connect() as cur:
cur.execute(sql)
result = cur.fetchall()
iterator = map(lambda word: word[0], result)
words = list(iterator)
if cut:
pattern = re.compile(r'[\u4e00-\u9fa5_a-zA-Z0-9]+')
re_words = ' '.join(re.findall(pattern, str(result)))
fixed_words = jieba.cut(re_words, cut_all=False)
words = list(filter(lambda word: word != ' ', fixed_words))
return words
def i_ip(data, many=0):
print('i_ip')
sql = 'insert into ip_list(ip, port, procotol) values\
(%s, %s, %s)'
try:
with passwd.connect() as cur:
if many != 0:
cur.executemany(sql, data)
else:
cur.execute(sql, data)
except Exception as e:
print(e)
def test_cut(self):
# sql = "select user_area from luhan_sab limit 1000;"
# sql = "select user_area from luhan_sab where user_area like '%大连%';"
sql = "select user_comment from luhan_inc limit 10;"
with passwd.connect() as cur:
jieba.load_userdict = 'source/userdict.txt'
cur.execute(sql)
result = cur.fetchall()
pattern = re.compile(r'[\u4e00-\u9fa5_a-zA-Z0-9]+')
re_words = re.findall(pattern, str(result))
words = ' '.join(re_words)
cut_words = jieba.lcut(words)
print(cut_words)
def i_sab(data):
sql = 'update luhan_sab set user_sex=%s, user_area=%s, user_birth=%s\
where num = %s'
with passwd.connect() as cur:
cur.execute(sql, data)
def s_domain(data):
sql = 'select user_domain from luhan_sab where num=%s'
with passwd.connect() as cur:
cur.execute(sql, data)
domain = cur.fetchone()
return domain[0]
def rm_ip():
sql = 'delete from ip_list'
with passwd.connect() as cur:
cur.execute(sql)
