5. python连接mysql

创建sql报表

from pymysql import Connection

conn = Connection(
  host='localhost',
  port =3306,
  user='root',
  password='123456',
)
print(conn.get_server_info())
# 获取游标对象
cursor = conn.cursor()
# 选择python数据库
conn.select_db('python')
# 创建test_pymysql表
cursor.execute('create table test_pymysql(id int);')

conn.close()

结果

查询数据库的所有数据

from pymysql import Connection

conn = Connection(
  host='localhost',
  port =3306,
  user='root',
  password='123456',
)
print(conn.get_server_info())
# 获取游标对象
cursor = conn.cursor()
# 选择python数据库
conn.select_db('python')
# 创建test_pymysql表
cursor.execute('select * from test_pymysql')
res = cursor.fetchall()
for r in res:
  print(r)

conn.close()

插入数据需要手动确认，autocommit=True可以自动提交

from pymysql import Connection

conn = Connection(
  host='localhost',
  port =3306,
  user='root',
  password='123456',
  # 自动提交
  # autocommit=True,
)
print(conn.get_server_info())
# 获取游标对象
cursor = conn.cursor()
# 选择python数据库
conn.select_db('python')
# 创建test_pymysql表
cursor.execute('insert into test_pymysql values(1000)')
# pymysql需要确认才会插入
conn.commit()

conn.close()

spark数据处理

from pyspark import SparkContext, SparkConf
# 创建SparkConf对象
conf = SparkConf().setMaster('local[*]').setAppName("test_spark_app")

# SparkConf实例对象
sc = SparkContext(conf=conf)

print(sc.version)

# 停止
sc.stop()

RDD对象（弹性分布式数据集）

# 创建RDD
rdd = sc.parallelize(range(10))
# 打印RDD
print(rdd.collect())

读取文件

# 读取文本文件
rddRead = sc.textFile("./test.text")
print(rddRead.collect())