avro-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Scott Carey <sc...@richrelevance.com>
Subject Re: Schema with multiple Record types Java API
Date Fri, 15 Jul 2011 17:00:54 GMT
Try out the Reflect API.  It may not be flexible enough yet, but the intended use case is to
serialize pre-existing classes.   If more annotations are required for your use case,  create
a JIRA ticket.

http://avro.apache.org/docs/current/api/java/org/apache/avro/reflect/package-summary.html

Thanks!

-Scott



On 7/15/11 4:54 AM, "Peter Wolf" <opus111@gmail.com<mailto:opus111@gmail.com>>
wrote:

Thanks again Scott,

Yes, I am using AVRO to serialize existing Java classes, so tools to generate code will not
help me.

Are there tools that go the other way, such as JAXB for XML?  I really want to point to a
root Java object, and say "serialize this, and everything it points to, as AVRO".

BTW AVRO Rocks!  My objects contain are amounts of data, and I am *very* impressed with the
speed of serialization/deserialization.

Cheers
P





On 7/14/11 10:10 PM, Scott Carey wrote:
AvroIDL can handle imports, but it generates classes.  The Avro API's for this can be used
to generate Schemas without making objects if you wish.

The Avro schema compiler (*.avsc, *.avpr) does not support imports, it is a feature requested
by many but not contributed by anyone.

You may be interested in the code-gen capabilities of Avro, which has a Velocity templating
engine to create Java classes based on schemas.  This can be customized to generate classes
in custom ways.

However, if you are using Avro to serialize objects that have pre-existing classes, the Reflect
API or an enhancement of it may be more suitable.

More information on your use case may help to point you in the right direction.

-Scott


On 7/14/11 6:43 PM, "Peter Wolf" <opus111@gmail.com<mailto:opus111@gmail.com>>
wrote:

Many thanks Scott,

I am looking for the equivalent of #include or import.  I want to make a complicated schema
with many record types, but manage it in separate strings.

In my application, I am using AVRO to serialize a tree of connected Java objects.  The record
types mirror Java classes.  The schema descriptions live in the different Java classes, and
reference each other.

My current code looks like this...

    public class Foo {

        static String schemaDescription =
            "{" +
                    "  \"namespace\": \"foo\", " +
                    "  \"name\": \"Foo\", " +
                    "  \"type\": \"record\", " +
                    "  \"fields\": [ " +
                    "      {\"name\": \"notes\", \"type\": \"string\" }, " +
                    "      {\"name\": \"timestamp\", \"type\": \"string\" }, " +
                    "      {\"name\": \"bah\", \"type\": " + Bah.schemaDescription + " },"
+
                    "      {\"name\": \"zot\", \"type\": " + Zot.schemaDescription + " }"
+
                    "    ]" +
                    "}";

        static Schema schema = Schema.parse(schemaDescription);


So, I am referencing by copying the schemaDescriptions.  The top level schemaDescription strings
therefore get really big.

Is there already a clean coding Pattern for doing this-- I can't be the first.  Is there a
document describing best practices?

Thanks
P





On 7/14/11 7:02 PM, Scott Carey wrote:
The name and namespace is part of any named schema (Type.RECORD, Type.FIXED, Type.ENUM).

We don't currently have an API to search a schema for subschemas that match names.  It would
be useful, you might want to create a JIRA ticket explaining your use case.

So it would be a little more complex.

        Schema schema = Schema.parse(schemaDescription);
        Schema.Type type = schema.getType();
        switch (type) {
        case RECORD:
          String name = schema.getName();
          String namespace = schema.getNamespace();
          List<Field> fields = schema.getFields();
        }

        etc.

In general, I have created SpecificRecord objects from schemas using the specific compiler
(and the ant task or maven plugin) and then within those generated classes there is a static
SCHEMA variable to reference.

Avro IDL is alo an easier way to define related schemas.  Currently there are only build tools
that generate code from these, though there are APIs to extract schemas.

-Scott

On 7/13/11 10:43 AM, "Peter Wolf" <opus111@gmail.com<mailto:opus111@gmail.com>>
wrote:

Hello, this a dumb question, but I can not find the answer in the docs

I want to have a complicated schema with lots of Records referencing other Records.

Like this...

{
  "namespace": "com.foobah",
  "name": "Bah",
  "type": "record",
  "fields": [
  {"name": "value", "type": "int"}
  ]
}

{
  "namespace": "com.foobah",
  "name": "Foo",
  "type": "record",
  "fields": [
  {"name": "bah", "type": "Bah"}
  ]
}
Using the Java API, how do I reference types within a schema?  Let's say I want to make a
Foo object, I want to do something like this...

        Schema schema = Schema.parse(schemaDescription);
>>> Schema foo = schema.getSchema("com.foobah.Foo"); <<<
        GenericData o = new GenericData( foo );

Many thanks in advance
Peter






Mime
View raw message