Commit bb3e8a89 authored by kinomin's avatar kinomin

初始化项目

parent 7e34eec0
# mysql 相关参数
MY_MYSQL_DRIVER = 'org.mariadb.jdbc.Driver'
MY_MYSQL_URL = 'jdbc:mysql://hadoop1:3306/wanjia?useSSL=false'
MY_MYSQL_USER = 'root'
MY_MYSQL_PASSWORD = '123'
# save 相关参数
'''
:param mode: specifies the behavior of the save operation when data already exists.
* ``append``: Append contents of this :class:`DataFrame` to existing data.
* ``overwrite``: Overwrite existing data.
* ``ignore``: Silently ignore this operation if data already exists.
* ``error`` or ``errorifexists`` (default case): Throw an exception if data already exists.
'''
SAVE_MODE_OVERWRITE = 'overwrite'
SAVE_MODE_APPEND = 'append'
SAVE_MODE_IGNORE = 'ignore'
SAVE_MODE_ERROR = 'error'
HDFS_PATH = 'hdfs://hadoop1:9000/wanjia'
\ No newline at end of file
#!/usr/local/bin/python
import hashlib
import json
import sys
from common import Constants
from pyspark.sql import SparkSession
from pyspark.sql.types import StringType
# import os
import pymysql
# os.system("pip install pyspark")
def getMySqlConnect():
# 连接数据库
conn = pymysql.connect(
host="hadoop1", # 连接主机名,也可以用ip地址,例如127.0.0.1
user="root", # 连接用户名
passwd="123", # 用户密码
db="wanjia", # 要连接的数据库名
charset="utf8", # 指定编码格式
autocommit=True, # 如果插入数据,, 是否自动提交? 和conn.commit()功能一致。
)
return conn
def getRecordByIdResultT1(id):
conn = getMySqlConnect()
# 创建游标对象,用来给数据库发送sql语句
cur = conn.cursor()
serch_data = """select t1 from test where id = {table_id}""".format(table_id=id)
result = cur.execute(serch_data)
# 获取下一个查询结果集
record = cur.fetchone()
cur.close() # 关闭数据库连接
return record
from rules import MD5
from sources import Sources, Sinks
def getColumns(x):
sql_list = []
......@@ -52,64 +30,44 @@ def getColumns(x):
print("---------------------->", sql)
return sql
def md5(col):
md5 = hashlib.md5()
md5.update(str(col).encode('utf-8'))
return md5.hexdigest()
if __name__ == '__main__':
# 拿到 列名 等
record = getRecordByIdResultT1(sys.argv[1])
record = Sources.getRecordByIdResultT1(sys.argv[1])
col = json.loads(str(record)[2:-3]).get('col')
sql = getColumns(str(record)[2:-3])
print(sql)
sink_method = json.loads(str(record)[2:-3]).get('sink_method')
# spark 初始化
spark = SparkSession.Builder().appName('sql').master('local').getOrCreate()
spark.udf.register('md5', md5, StringType())
print('=======>', output_file)
df = spark.read.format('csv').option('inferSchema', 'true').load(sys.argv[2]).toDF(*col)
df.count
spark.udf.register('md5', MD5.md5, StringType())
print('=======>', sys.argv[2])
df = Sources.readCsv(spark, sys.argv[2], col)
t_table = df.createTempView('kino')
sinkDF = spark.sql(sql)
sinkDF.show()
'''
:param mode: specifies the behavior of the save operation when data already exists.
* ``append``: Append contents of this :class:`DataFrame` to existing data.
* ``overwrite``: Overwrite existing data.
* ``ignore``: Silently ignore this operation if data already exists.
* ``error`` or ``errorifexists`` (default case): Throw an exception if data already exists.
'''
sinkDF.write.mode("overwrite") \
.format("jdbc") \
.option("truncate", "true") \
.option("batchsize", 10000) \
.option("isolationLevel", "NONE") \
.option("driver", "org.mariadb.jdbc.Driver") \
.option("url", 'jdbc:mysql://hadoop1:3306/wanjia?useSSL=false') \
.option("user", 'root') \
.option("password", '123') \
.option("dbtable", "test1") \
.save()
# TODO 修改状态
# 落 ftp
# sinkDF.write.format("com.springml.spark.sftp") \
# .option("host", "192.168.31.45") \
# .option("username", "raomin@9zdata.cn") \
# .option("password", "9zdata123.") \
# .option("fileType", "csv") \
# .option("delimiter", ";") \
# .save("ftp://192.168.31.45:21/wanjia/sample.csv")
# sinkDF.write.mode("overwrite").text("ftp://192.168.31.45/wanjia/sample.csv")
# sinkDF.write.format("parquet").mode("overwrite").save('hdfs://hadoop1:9000/wanjia')
if sink_method.lower() == 'hdfs':
# TODO 落 HDFS
Sinks.sinkHDFS(sinkDF, Constants.HDFS_PATH, "csv", Constants.SAVE_MODE_OVERWRITE)
elif sink_method.lower() == 'mysql':
# TODO 落 MySQL
Sinks.sinkMySql(sinkDF,
mode=Constants.SAVE_MODE_OVERWRITE,
driver=Constants.MY_MYSQL_DRIVER,
url=Constants.MY_MYSQL_URL,
user=Constants.MY_MYSQL_USER,
password=Constants.MY_MYSQL_PASSWORD,
table="test1")
elif sink_method.lower() == 'ftp':
print("落FTP")
else:
print("...")
# TODO 修改状态, 数据校验(读取-写出 数据量)
Sinks.updateMysqlStatus(i_status=1, i_readcount=df.count(), i_sinkcount=sinkDF.count(), i_id=1)
# 关闭spark会话
spark.stop()
......
......@@ -3,7 +3,6 @@ id=$1
output_file=$2
echo "传入的数据库id为: ${id}"
echo "目标文件路径为: " ${output_file}
#spark-submit --jars /usr/bigdata/spark-3.0.1-bin-hadoop3.2/jars/mariadb-java-client-2.1.2.jar --jars /usr/bigdata/spark-3.0.1-bin-hadoop3.2/jars/mysql-connector-java.jar dependencies.py ${id} ${output_file}
#python3 dependencies.py --jars /usr/bigdata/spark-3.0.1-bin-hadoop3.2/jars/mariadb-java-client-2.1.2.jar ${id} ${output_file}
nohup spark-submit --jars /usr/bigdata/spark-3.0.1-bin-hadoop3.2/jars/mariadb-java-client-2.1.2.jar dependencies.py ${id} ${output_file} > /root/test.log 2>&1 &
# nohup spark-submit --jars /usr/bigdata/spark-3.0.1-bin-hadoop3.2/jars/mariadb-java-client-2.1.2.jar dependencies.py ${id} ${output_file} > /root/test.log 2>&1 &
spark-submit --jars /usr/bigdata/spark-3.0.1-bin-hadoop3.2/jars/mariadb-java-client-2.1.2.jar dependencies.py ${id} ${output_file}
\ No newline at end of file
class MyError(Exception):
def __init__(self, ErrorInfo):
super().__init__(self)
self.errorinfo = ErrorInfo
def __str__(self):
return self.errorinfo
if __name__ == '__main__':
try:
raise MyError('零异常')
except MyError as e:
print("my exception currend:", e.errorinfo)
\ No newline at end of file
import hashlib
def md5(col):
md5 = hashlib.md5()
md5.update(str(col).encode('utf-8'))
return md5.hexdigest()
\ No newline at end of file
from common import Constants
import exception.MyErrorException as ex
# 写出到 MySQL
from sources import Sources
def sinkMySql(sinkDF, driver, url, user, password, table, mode=Constants.SAVE_MODE_OVERWRITE):
sinkDF.write.mode(mode) \
.format("jdbc") \
.option("truncate", "true") \
.option("batchsize", 10000) \
.option("isolationLevel", "NONE") \
.option("driver", driver) \
.option("url", url) \
.option("user", user) \
.option("password", password) \
.option("dbtable", table) \
.save()
# 写出到FTP
'''
相关参数, 例如:
host: 192.168.31.45
username: raomin@9zdata.cn
password: 9zdata123.
fileType: csv
file: ftp://192.168.31.45:21/wanjia/sample.csv
'''
def sinkFTP(sinkDF, host, user, password, filetype, file):
sinkDF.write.format("com.springml.spark.sftp") \
.option("host", host) \
.option("username", user) \
.option("password", password) \
.option("fileType", filetype) \
.option("delimiter", ";") \
.save(file)
# sink HDFS
def sinkHDFS(sinkDF, path, format="parquet", mode=Constants.SAVE_MODE_OVERWRITE):
sinkDF\
.write\
.format(format)\
.mode(mode)\
.save(path)
# 修改 MySQL 状态
def updateMysqlStatus(i_status, i_readcount, i_sinkcount, i_id):
try:
conn = Sources.getMySqlConnect()
# 创建游标对象,用来给数据库发送sql语句
cur = conn.cursor()
serch_data = """update uptable set status={status}, readcount={readcount}, sinkcount={sinkcount} where id = {id}"""\
.format(status=i_status, readcount=i_readcount, sinkcount=i_sinkcount, id=i_id)
result = cur.execute(serch_data)
except ex.MyError as e:
print("update error", e.errorinfo)
finally:
cur.close() # 关闭数据库连接
conn.close()
\ No newline at end of file
import pymysql
import exception.MyErrorException as ex
def getMySqlConnect():
try:
# 连接数据库
conn = pymysql.connect(
host="hadoop1", # 连接主机名,也可以用ip地址,例如127.0.0.1
user="root", # 连接用户名
passwd="123", # 用户密码
db="wanjia", # 要连接的数据库名
charset="utf8", # 指定编码格式
autocommit=True, # 如果插入数据,, 是否自动提交? 和conn.commit()功能一致。
)
except ex.MyError as e:
print("connection error", e.errorinfo)
return conn
def getRecordByIdResultT1(id):
conn = getMySqlConnect()
# 创建游标对象,用来给数据库发送sql语句
try:
cur = conn.cursor()
serch_data = """select t1 from test where id = {table_id}""".format(table_id=id)
result = cur.execute(serch_data)
# 获取下一个查询结果集
record = cur.fetchone()
except ex.MyError as e:
print("executor failed", e.errorinfo)
finally:
cur.close() # 关闭数据库连接
conn.close()
return record
def readCsv(spark, path, col):
df = spark.read.format('csv').option('inferSchema', 'true').load(path).toDF(*col)
return df
#!/usr/local/bin/python
import sys
import os
import shutil
import zipfile
import exception.MyErrorException as ex
pyName = sys.argv[0] #sys.argv[0] 类似于shell中的$0, 但不是脚本名称,而是脚本的路径
zipPath = sys.argv[1] #sys.argv[1] 表示传入的第一个参数
inputPath = sys.argv[2] #sys.argv[2] 表示传入的第二个参数
outputPath = sys.argv[3]
outputPath = sys.argv[3]
# os.system("hdfs dfs -put /srcPath /dstPath")
def get_file(srcPath, dstPath):
res = []
for root, dirs, files in os.walk(srcPath, True):
for fileName in files:
name, suf = os.path.splitext(fileName)
if fileName.endswith('.csv'):
shutil.copy(os.path.join(root, fileName), dstPath)
try:
if not os.path.exists(srcPath):
print("srcPath not exist")
else:
res = []
for root, dirs, files in os.walk(srcPath, True):
for fileName in files:
name, suf = os.path.splitext(fileName)
if fileName.endswith('.csv'):
shutil.copy(os.path.join(root, fileName), dstPath)
except ex.MyError as e:
print("失败", e.errorinfo)
def unzip_file(zip_file_name, destination_path):
archive = zipfile.ZipFile(zip_file_name, mode='r')
for file in archive.namelist():
archive.extract(file, destination_path)
try:
archive = zipfile.ZipFile(zip_file_name, mode='r')
for file in archive.namelist():
archive.extract(file, destination_path)
except ex.MyError as e:
print("解压失败", e.errorinfo)
sys.exit()
if __name__ == '__main__':
print('解压、复制中...')
......
......@@ -2,9 +2,7 @@
echo -e "\033[34m==========文件解压、复制开始\033[0m=========="
python3 unzip.py /root/wanjia/files/zip_file/20201102/20201102.zip /root/wanjia/files/input_file/20201102 /root/wanjia/files/output_file/20201102
echo -e "\033[34m==========文件解压、复制结束\033[0m=========="
echo ""
echo -e "\033[34m==========将id写入环境变量\033[0m============"
# echo '{"id": 1}' >> ${JOB_OUTPUT_PROP_FILE}
echo '{"id": 1, "output_file": "/root/wanjia/files/output_file/20201102"}' >> ${JOB_OUTPUT_PROP_FILE}
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment