Support for CQL3 Collections

[abh1nav]

Hi Rohit,

Is there any documentation regarding support for CQL3 collections?
Specifically, I'm trying to write a transformer for a table that looks like:

CREATE TABLE users (
  uid text,
  lastupdate text,
  properties map<text,text>,
  PRIMARY KEY (uid)
);

case class User(uid: String, lastupdate: String, 
  properties: Map[String, String])
implicit def cql3ToUser(k: Map[String, ByteBuffer], v: Map[String, ByteBuffer]): User = 
  User(k("uid"), v("lastupdate"), v("properties"))

It seems there's no implicit converter inside RichByteBuffer for ByteBuffer to Map[String, String].

Thanks,
--Abhinav

[ra2085]

You can have a look at the org.apache.cassandra.db.marshal package, it has a collection type. Haven't used the Collection type specifically, but I've already used the Decimal datatype successfully. I've already opened an issue to support implicits on those datatypes few weeks ago, you can easily use the example I posted with a Collection type.

Open issue cassandra 2.0 datatypes: #8

Regards!

[abh1nav]

@ra2085 Thanks! This ended up working flawlessly:

import org.apache.cassandra.db.marshal.{UTF8Type, MapType}

implicit def ByteBuffer2MapSS(b: ByteBuffer): Map[String,String] =
    MapType.getInstance(UTF8Type.instance, UTF8Type.instance).compose(b).toMap

[milliondreams]

We should get the support for this in the next release RichByteBuffer

Since we will be support CQL Auto Paging in next release and hence having DS driver as dependency, I was thinking of rewriting the RichByteBuffer using DataType.

For this case it will look like -

import com.datastax.driver.core.DataType

import scala.collection.JavaConverters._

implicit def MapSS2ByteBuffer(map: Map[String, String]): ByteBuffer = DataType.map(DataType.text(), DataType.text()).serialize(map.asJava)

implicit def ByteBuffer2MapSS(buffer: ByteBuffer): Map[String, String] = DataType.map(DataType.text(), DataType.text()).deserialize(buffer).asInstanceOf[java.util.Map[String, String]].asScala.toMap

[abh1nav]

That looks much cleaner.