Python增量循環(huán)刪除MySQL表數(shù)據(jù)的方法

有一業(yè)務(wù)數(shù)據(jù)庫(kù)，使用MySQL 5.5版本，每天會(huì)寫入大量數(shù)據(jù)，需要不定期將多表中“指定時(shí)期前“的數(shù)據(jù)進(jìn)行刪除，在SQL SERVER中很容易實(shí)現(xiàn)，寫幾個(gè)WHILE循環(huán)就搞定，雖然MySQL中也存在類似功能，怎奈自己不精通，于是采用Python來實(shí)現(xiàn)

話不多少，上腳本：

# coding: utf-8
import MySQLdb
import time
# delete config
DELETE_DATETIME = '2016-08-31 23:59:59'
DELETE_ROWS = 10000
EXEC_DETAIL_FILE = 'exec_detail.txt'
SLEEP_SECOND_PER_BATCH = 0.5
DATETIME_FORMAT = '%Y-%m-%d %X'
# MySQL Connection Config
Default_MySQL_Host = 'localhost'
Default_MySQL_Port = 3358
Default_MySQL_User = "root"
Default_MySQL_Password = 'roo@01239876'
Default_MySQL_Charset = "utf8"
Default_MySQL_Connect_TimeOut = 120
Default_Database_Name = 'testdb001'
def get_time_string(dt_time):
"""
獲取指定格式的時(shí)間字符串
:param dt_time: 要轉(zhuǎn)換成字符串的時(shí)間
:return: 返回指定格式的字符串
"""
global DATETIME_FORMAT
return time.strftime(DATETIME_FORMAT, dt_time)
def print_info(message):
"""
將message輸出到控制臺(tái)，并將message寫入到日志文件
:param message: 要輸出的字符串
:return: 無返回
"""
print(message)
global EXEC_DETAIL_FILE
new_message = get_time_string(time.localtime()) + chr(13) + str(message)
write_file(EXEC_DETAIL_FILE, new_message)
def write_file(file_path, message):
"""
將傳入的message追加寫入到file_path指定的文件中
請(qǐng)先創(chuàng)建文件所在的目錄
:param file_path: 要寫入的文件路徑
:param message: 要寫入的信息
:return:
"""
file_handle = open(file_path, 'a')
file_handle.writelines(message)
# 追加一個(gè)換行以方便瀏覽
file_handle.writelines(chr(13))
file_handle.close()
def get_mysql_connection():
"""
根據(jù)默認(rèn)配置返回?cái)?shù)據(jù)庫(kù)連接
:return: 數(shù)據(jù)庫(kù)連接
"""
conn = MySQLdb.connect(
host=Default_MySQL_Host,
port=Default_MySQL_Port,
user=Default_MySQL_User,
passwd=Default_MySQL_Password,
connect_timeout=Default_MySQL_Connect_TimeOut,
charset=Default_MySQL_Charset,
db=Default_Database_Name
)
return conn
def mysql_exec(sql_script, sql_param=None):
"""
執(zhí)行傳入的腳本，返回影響行數(shù)
:param sql_script:
:param sql_param:
:return: 腳本最后一條語句執(zhí)行影響行數(shù)
"""
try:
conn = get_mysql_connection()
print_info("在服務(wù)器{0}上執(zhí)行腳本:{1}".format(
conn.get_host_info(), sql_script))
cursor = conn.cursor()
if sql_param is not None:
cursor.execute(sql_script, sql_param)
row_count = cursor.rowcount
else:
cursor.execute(sql_script)
row_count = cursor.rowcount
conn.commit()
cursor.close()
conn.close()
except Exception, e:
print_info("execute exception:" + str(e))
row_count = 0
return row_count
def mysql_query(sql_script, sql_param=None):
"""
執(zhí)行傳入的SQL腳本，并返回查詢結(jié)果
:param sql_script:
:param sql_param:
:return: 返回SQL查詢結(jié)果
"""
try:
conn = get_mysql_connection()
print_info("在服務(wù)器{0}上執(zhí)行腳本:{1}".format(
conn.get_host_info(), sql_script))
cursor = conn.cursor()
if sql_param != '':
cursor.execute(sql_script, sql_param)
else:
cursor.execute(sql_script)
exec_result = cursor.fetchall()
cursor.close()
conn.close()
return exec_result
except Exception, e:
print_info("execute exception:" + str(e))
def get_id_range(table_name):
"""
按照傳入的表獲取要?jiǎng)h除數(shù)據(jù)最大ID、最小ID、刪除總行數(shù)
:param table_name: 要?jiǎng)h除的表
:return: 返回要?jiǎng)h除數(shù)據(jù)最大ID、最小ID、刪除總行數(shù)
"""
global DELETE_DATETIME
sql_script = """
SELECT
MAX(ID) AS MAX_ID,
MIN(ID) AS MIN_ID,
COUNT(1) AS Total_Count
FROM {0}
WHERE create_time <='{1}';
""".format(table_name, DELETE_DATETIME)
query_result = mysql_query(sql_script=sql_script, sql_param=None)
max_id, min_id, total_count = query_result[0]
# 此處有一坑，可能出現(xiàn)total_count不為0 但是max_id 和min_id 為None的情況
# 因此判斷max_id和min_id 是否為NULL
if (max_id is None) or (min_id is None):
max_id, min_id, total_count = 0, 0, 0
return max_id, min_id, total_count
def delete_data(table_name):
max_id, min_id, total_count = get_id_range(table_name)
temp_id = min_id
while temp_id <= max_id:
sql_script = """
DELETE FROM {0}
WHERE id <= {1}
and id >= {2}
AND create_time <='{3}';
""".format(table_name, temp_id + DELETE_ROWS, temp_id, DELETE_DATETIME)
temp_id += DELETE_ROWS
print(sql_script)
row_count = mysql_exec(sql_script)
print_info("影響行數(shù)：{0}".format(row_count))
current_percent = (temp_id - min_id) * 1.0 / (max_id - min_id)
print_info("當(dāng)前進(jìn)度{0}/{1},剩余{2},進(jìn)度為{3}%".format(temp_id, max_id, max_id - temp_id, "%.2f" % current_percent))
time.sleep(SLEEP_SECOND_PER_BATCH)
print_info("當(dāng)前表{0}已無需要?jiǎng)h除的數(shù)據(jù)".format(table_name))
delete_data('TB001')
delete_data('TB002')
delete_data('TB003')

執(zhí)行效果：

實(shí)現(xiàn)原理：

由于表存在自增ID，于是給我們?cè)隽垦h(huán)刪除的機(jī)會(huì)，查找出滿足刪除條件的最大值ID和最小值ID，然后按ID 依次遞增，每次小范圍內(nèi)(如10000條)進(jìn)行刪除。

實(shí)現(xiàn)優(yōu)點(diǎn)：

實(shí)現(xiàn)“小斧子砍大柴”的效果，事務(wù)小，對(duì)線上影響較小，打印出當(dāng)前處理到的“ID”，可以隨時(shí)關(guān)閉，稍微修改下代碼便可以從該ID開始，方便。

實(shí)現(xiàn)不足：

為防止主從延遲太高，采用每次刪除SLEEP1秒的方式，相對(duì)比較糙，最好的方式應(yīng)該是周期掃描這條復(fù)制鏈路，根據(jù)延遲調(diào)整SLEEP的周期，反正都腳本化，再智能化點(diǎn)又何妨！