Jute

Jute is a implementation of Apache Zookeepers's jute IDL and generation code written in pure Go. This module includes a parser to real the jute DDL file into an AST and then a generator to generate serialization/deserialization code.

Jute started out as a serialization format for Hadoop and is no longer being used (Hadoop moved to Apache Avro as such only Zookeeper seems to be using it.

Note: There may be work in future versions of Zookeeper to use a different format as well (see: ZOOKEEPER-102).

Each jute module is mapped to a Go package with each class getting it's own go file.

Usage

$ go get github.com/go-zookeeper/jute/cmd/jutec
$ jutec input.jute

You can control the output folder structure with the go.moduleMap and go.prefix options.

go.prefix can control the import prefix for the generated packages. Packages generated will append this prefix to the package names, however the prefix is not used in the generated directories.

go.moduleMap uses regular expressions to rewrite module names to go packages names. The format for this option is <regexp match>:<regexp replacement>. If the replacement string is - then the module is skipped and no code is generated. All moduleMaps are processed in order. Unlike go.prefix any mapped modules will be represented in the generated directory structure for the packages.

Example

This command will ignore the org.apache.zookeeper.server.* and the org.apache.zookeeper.txn modules. All other modules matching org.apache.zookeeper will drop the org.apache.zookeeper prefix. All generated modules will have the prefix github.com/go-zookeeper/zk/internal appended to it (for imports) but it won't be incuded in the directories.

$ jutec \
    -go.moduleMap=org.apache.zookeeper.server:- \
    -go.moduleMap=org.apache.zookeeper.txn:- \
    -go.moduleMap=org.apache.zookeeper: \
    -go.prefix=github.com/go-zookeeper/zk/internal \
    testdata/zookeeper.jute

Caveats

This is pretty experimental code. There are a lot of error checking missing, imports are not implemented and there are probably some pretty bad bugs. You have been warned!

TODO

Parser

• Check for duplicate field names
• Validate external module class fields
• Block reserved keywords

Generator

• Support imported jute files