-
Notifications
You must be signed in to change notification settings - Fork 0
/
glue.py
206 lines (177 loc) · 8.09 KB
/
glue.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
import sys
import boto3
import json
from botocore.exceptions import ClientError
from awsglue.transforms import *
from awsglue.utils import getResolvedOptions
from pyspark.context import SparkContext
from awsglue.context import GlueContext
from awsglue.job import Job
from pyspark.context import SparkConf
from pyspark.sql import DataFrame, Row
from pyspark.sql import SparkSession
from awsglue import DynamicFrame
from pyspark.sql.types import StructType, StructField, StringType, IntegerType,ArrayType,MapType,LongType
from pyspark.sql.functions import from_json,col,to_json,json_tuple
def get_secret():
secret_name = "redshift"
region_name = "us-east-2"
# Create a Secrets Manager client
session = boto3.session.Session()
client = session.client(
service_name='secretsmanager',
region_name=region_name
)
try:
get_secret_value_response = client.get_secret_value(
SecretId=secret_name
)
except ClientError as e:
raise e
secret = get_secret_value_response['SecretString']
return secret
secret_dict = json.loads(get_secret())
params = [
'JOB_NAME',
'TempDir',
'kafka_broker',
'topic',
'consumer_group',
'startingOffsets',
'checkpoint_interval',
'checkpoint_location',
'aws_region'
]
args = getResolvedOptions(sys.argv, params)
conf = SparkConf()
sc = SparkContext(conf=conf)
glueContext = GlueContext(sc)
logger = glueContext.get_logger()
spark = glueContext.spark_session
job_name = args['JOB_NAME']
kafka_broker = args['kafka_broker']
topic = args['topic']
consumer_group = args['consumer_group']
startingOffsets = args['startingOffsets']
checkpoint_interval = args['checkpoint_interval']
checkpoint_location = args['checkpoint_location']
aws_region = args['aws_region']
# super or flatten
complex_convert = "super"
maxerror = 0
redshift_host = secret_dict['host']
redshift_port = secret_dict['port']
redshift_username = secret_dict['username']
redshift_password = secret_dict['password']
redshift_database = "dev"
redshift_schema = "ads"
redshift_table = "ads_test_table"
redshift_tmpdir = "s3://aws-glue-assets-us-east-2/temp/"
tempformat = "CSV"
redshift_iam_role = "arn:aws:iam::aws-account:role/redshiftetl"
reader = spark \
.readStream \
.format("kafka") \
.option("kafka.bootstrap.servers", kafka_broker) \
.option("subscribe", topic) \
.option("maxOffsetsPerTrigger", "1000000") \
.option("kafka.consumer.commit.groupid", consumer_group) \
.option("failOnDataLoss", "false")
if startingOffsets == "earliest" or startingOffsets == "latest":
reader.option("startingOffsets", startingOffsets)
else:
reader.option("startingTimestamp", startingOffsets)
kafka_data = reader.load()
df = kafka_data.selectExpr("CAST(value AS STRING)")
def process_batch(data_frame, batchId):
dfc = data_frame.cache()
logger.info(job_name + " - my_log - process batch id: " + str(batchId) + " record number: " + str(dfc.count()))
logger.info(job_name + " - my_log - process batch id: " + str(batchId) + " record : " + str(dfc.show(5, truncate=False)))
if not data_frame.rdd.isEmpty():
# json_schema = spark.read.json(dfc.rdd.map(lambda p: str(p["value"]))).schema
# 定义外层schema
json_schema = StructType([StructField("id", StringType(), True),
StructField("vehicleCode", StringType(), True),
StructField("receiveTime", StringType(), True),
StructField("body", StringType(), True)
])
# 使用定义的schema解析,不存在的字段会设置为空
sdf = dfc.select(from_json(col("value"), json_schema).alias("kdata")).select("kdata.*")
logger.info(job_name + " - my_log - process batch id: " + str(batchId) + "sdf record : " + str(sdf.show()))
# 使用json_tuple函数解析header,不存在的字段会设置为null
sdf_with_header = sdf.select("id", "vehicleCode", "receiveTime","body",
json_tuple(col("body"), "vehicleType").alias('vehicleType')
)
logger.info(job_name + " - my_log - process batch id: " + str(batchId) + "dfc record : " + str(sdf_with_header.show()))
source_view_name = "kafka_source_table_view"
sdf_with_header.createOrReplaceGlobalTempView(source_view_name)
sdf = spark.sql(
"select id,vehicleCode,receiveTime,body,vehicleType from {view_name} where receivetime!=''".format(
view_name="global_temp." + source_view_name))
logger.info(job_name + " - my_log - source sdf schema: " + sdf._jdf.schema().treeString())
logger.info(job_name + " - my_log - source sdf: " + sdf._jdf.showString(5, 20, False))
if complex_convert == "super":
csdf = to_super_df(spark, sdf)
elif complex_convert == "flatten":
csdf = flatten_json_df(sdf)
else:
csdf = to_super_df(spark,flatten_json_df(sdf))
logger.info(job_name + " - my_log - convert csdf schema: " + csdf._jdf.schema().treeString())
logger.info(job_name + " - my_log - convert csdf: " + csdf._jdf.showString(5, 20, False))
if not csdf.rdd.isEmpty():
csdf.write \
.format("io.github.spark_redshift_community.spark.redshift") \
.option("url", "jdbc:redshift://{0}:{1}/{2}".format(redshift_host, redshift_port, redshift_database)) \
.option("dbtable", "{0}.{1}".format(redshift_schema,redshift_table)) \
.option("user", redshift_username) \
.option("password", redshift_password) \
.option("tempdir", redshift_tmpdir) \
.option("tempformat", tempformat) \
.option("extracopyoptions", "TRUNCATECOLUMNS region '{0}' maxerror {1} dateformat 'auto' timeformat 'auto'".format(aws_region, maxerror)) \
.option("aws_iam_role",redshift_iam_role).mode("append").save()
dfc.unpersist()
logger.info(job_name + " - my_log - finish batch id: " + str(batchId))
def to_super_df(spark: SparkSession, _df: DataFrame) -> DataFrame:
col_list = []
for field in _df.schema.fields:
if field.dataType.typeName() in ["struct", "array", "map"]:
col_list.append("to_json({col}) as aws_super_{col}".format(col=field.name))
else:
col_list.append(field.name)
view_name = "aws_source_table"
_df.createOrReplaceGlobalTempView(view_name)
df_json_str = spark.sql("select {columns} from {view_name}".format(columns=",".join(col_list),view_name="global_temp."+view_name))
fields = []
for field in df_json_str.schema.fields:
if "aws_super_" in field.name:
sf = StructField(field.name.replace("aws_super_", ""), field.dataType, field.nullable,
metadata={"super": True, "redshift_type": "super"})
else:
sf = StructField(field.name, field.dataType, field.nullable)
fields.append(sf)
schema_with_super_metadata = StructType(fields)
df_super = spark.createDataFrame(df_json_str.rdd, schema_with_super_metadata)
return df_super
def flatten_json_df(_df: DataFrame) -> DataFrame:
flattened_col_list = []
def get_flattened_cols(df: DataFrame, struct_col: str = None) -> None:
for col in df.columns:
if df.schema[col].dataType.typeName() != 'struct':
if struct_col is None:
flattened_col_list.append(f"{col} as {col.replace('.', '_')}")
else:
t = struct_col + "." + col
flattened_col_list.append(f"{t} as {t.replace('.', '_')}")
else:
chained_col = struct_col + "." + col if struct_col is not None else col
get_flattened_cols(df.select(col + ".*"), chained_col)
get_flattened_cols(_df)
return _df.selectExpr(flattened_col_list)
save_to_redshift = df \
.writeStream \
.outputMode("append") \
.trigger(processingTime="{0} seconds".format(checkpoint_interval)) \
.foreachBatch(process_batch) \
.option("checkpointLocation", checkpoint_location) \
.start()
save_to_redshift.awaitTermination()