DQ Job JSON

Files

Run against a file using -json. Additionally, options are available for -flatten and -multiline. This is helpful for nested and various formats.

Copy 
-ds json_file_example \
-f s3a://bucket_name/file.json \
-h instance.us-east4-c.c.owl-node.internal:5432/postgres \
-master spark://instance.us-east4-c.c.owl-node.internal:7077 \
-json \
-flatten \
-multiline 

Note Automatic flattening will infer schema and explode all structs, arrays, and map types.

Using Spark SQL

Copy 
-ds public.json_sample \ 
-lib "/opt/owl/drivers/postgres/" \
-h instance.us-east4-c.c.owl-node.internal:5432/postgres \
-master spark://instance.us-east4-c.c.owl-node.internal:7077 
-q "select * from public.jason" 
-rd "2021-01-17" 
-driver "org.postgresql.Driver" 
-cxn postgres-gcp 
-fq "select  \
get_json_object(col_3, '$.data._customer_name') AS  `data_customer_name` , \
get_json_object(col_3, '$.data._active_customer') AS  `data_active_customer` , \
from dataset "  

Note Pass in the path to Collibra DQ's -fq parameter. This is great for mixed data types within a database. For example, if you store JSON data as a string or a blob among other data.

Copy 
// Flatten
val colArr = new JsonReader().flattenSchema(df.schema)
colArr.foreach(x => println(x))

Note This Collibra DQ utility traverses the entire schema and print the proper get JSON object spark sql strings. You can use this instead of typing each query statement into the command line -fq parameter as seen above.

Using DQ Libraries

Copy 
import com.owl.common.options._
import com.owl.core.util.OwlUtils
import com.owl.core.activity2.JsonReader

val connProps = Map (
  "driver"   -> "org.postgresql.Driver",
  "user"     -> "user",
  "password" -> "password",
  "url"      -> "jdbc:postgresql://10.173.0.14:5432/postgres",
  "dbtable"  -> "public.data"
)

// Spark 
var rdd = spark.read.format("jdbc").options(connProps).load.select($"col_name").map(x=>x.toString()).rdd
var df = spark.read.json(rdd)

// Flatten
val colArr = new JsonReader().flattenSchema(df.schema)
val flatJson = df.select(colArr: _*)
flatJson.cache.count

// Opts
val dataset = "json_example"
val runId = s"2021-01-14"
val opt = new OwlOptions()
opt.dataset = dataset
opt.runId = runId
opt.datasetSafeOff = true

// DQ check
OwlUtils.resetDataSource("instance.us-east4-c.c.owl-node.internal","5432/postgres","user","pass", spark)
val owl = OwlUtils.OwlContext(flatJson, opt)
owl.register(opt)
owl.owlCheck

Note 
JsonReader()
This uses DQ's JsonReader to do the heavy lifting.

Known Limitations

  • Array and nested array datatypes in JSON files are not supported.