使用 Spark Structured Streaming 解析字段不固定的 JSON

数据样例

有两个文件,一个是 json: a.json

1
2
3
4
5
6
7
8
9
10
{
"createTime": 1532598069,
"event": {
"info": {
"AAA": "one",
"BBB": "two",
"DDD": "opps"
}
}
}

另一个也是 json: b.json

1
2
3
4
5
6
7
8
9
10
{
"createTime": "1532598069",
"event": {
"info": {
"AAA": "three",
"BBB": "four",
"CCC": "haha"
}
}
}

Kafka Producer

info 里面的字段个数是不固定的。用下面的代码先将 a.json 发送到 Kafka:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
from confluent_kafka import Producer


p = Producer({'bootstrap.servers': 'localhost:9092'})

def delivery_report(err, msg):
""" Called once for each message produced to indicate delivery result.
Triggered by poll() or flush(). """
if err is not None:
print('Message delivery failed: {}'.format(err))
else:
print('Message delivered to {} [{}]'.format(msg.topic(), msg.partition()))

with open("/Users/ohmycloud/work/notes/b.json") as f:
data = f.read()
p.poll(0)
p.produce('dynamic-schema', data.encode('utf-8'), callback=delivery_report)

p.flush()

Kafka Consumer

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
package dynamic.schma.test
import org.apache.spark.sql.SparkSession
import org.apache.spark.sql.functions._
import org.apache.spark.sql.streaming._
import org.apache.spark.sql.types._


object DynamicSchema extends App {
val spark = SparkSession
.builder
.appName("DynamicSchema")
.master("local[*]")
.getOrCreate()

// 定义 schema,包含 json 中的所有可能出现的字段
val schema = new StructType()
.add("createTime", StringType)
.add("event", MapType(StringType, new StructType()
.add("AAA", StringType, true)
.add("BBB", StringType, true)
.add("CCC", StringType, true)
.add("DDD", StringType, true)
))

val parsed = spark.readStream
.format("kafka")
.option("kafka.bootstrap.servers", "localhost:9092")
.option("subscribe", "dynamic-schema")
.option("startingOffsets", "earliest")
.load()
.select(from_json(col("value").cast("string"), schema).alias("parsed_value"))

import spark.implicits._

val event = parsed.select(explode($"parsed_value.event")).select("value.*")

val console = event.writeStream
.format("console")
.outputMode(OutputMode.Append())

val query = console.start()

query.awaitTermination()

}

打印出来的结果为:

1
2
3
4
5
+---+---+----+----+
|AAA|BBB| CCC| DDD|
+---+---+----+----+
|one|two|null|opps|
+---+---+----+----+

因为 a.json 里面没有 CCC 这个字段,并且 schema 里面设置允许 CCC 的值为 NULL, 所以 OK 的。

然后发送 b.json, 打印的结果为:

1
2
3
4
5
+-----+----+----+----+
| AAA| BBB| CCC| DDD|
+-----+----+----+----+
|three|four|haha|null|
+-----+----+----+----+

b.json 里面没有 DDD, schema 设置 CCC 的值允许为空,所以 NULL OK。

验证了一下 schema 的问题。