Avro-js

Pure JavaScript implementation of the Avro specification.

Features

• Fast! Typically twice as fast as JSON with much smaller encodings.
• Full Avro support, including recursive schemas, sort order, and evolution.
• Serialization of arbitrary JavaScript objects via logical types.
• Unopinionated 64-bit integer compatibility.
• No dependencies, avro-js even runs in the browser.

Installation

$ npm install avro-js

avro-js is compatible with all versions of node.js since 0.11 and major browsers via browserify.

Documentation

See doc/ folder.

Examples

Inside a node.js module, or using browserify:

var avro = require('avro-js');

• Encode and decode objects:

  // We can declare a schema inline:
  var type = avro.parse({
    name: 'Pet',
    type: 'record',
    fields: [
      {name: 'kind', type: {name: 'Kind', type: 'enum', symbols: ['CAT', 'DOG']}},
      {name: 'name', type: 'string'}
    ]
  });
  var pet = {kind: 'CAT', name: 'Albert'};
  var buf = type.toBuffer(pet); // Serialized object.
  var obj = type.fromBuffer(buf); // {kind: 'CAT', name: 'Albert'}

• Generate random instances of a schema:

  // We can also parse a JSON-stringified schema:
  var type = avro.parse('{"type": "fixed", "name": "Id", "size": 4}');
  var id = type.random(); // E.g. Buffer([48, 152, 2, 123])

• Check whether an object fits a given schema:

  // Or we can specify a path to a schema file (not in the browser):
  var type = avro.parse('./Person.avsc');
  var person = {name: 'Bob', address: {city: 'Cambridge', zip: '02139'}};
  var status = type.isValid(person); // Boolean status.

• Get a readable stream of decoded records from an Avro container file (not in the browser):

  avro.createFileDecoder('./records.avro')
    .on('metadata', function (type) { /* `type` is the writer's type. */ })
    .on('data', function (record) { /* Do something with the record. */ });

• Implement recursive schemata (due to lack of duck-typing):

    // example type: linked list with one long-int as element value
    const recursiveRecordType =  avro.parse({
      "type": "record",
      "name": "LongList",
      "fields" : [
        {"name": "value", "type": "long"},             
        {"name": "next", "type": ["null", "LongList"]} // optional next element via recursion
      ]
    });
  
    // will work
    const validRecursiveRecordDTO = {
      value: 1,
      next: {
        // no duck-typing support: from first nested level on the 
        // recursive type has to be explicitly specified.
        LongList: {
          value: 2,
          next: null
        }
      }
    };
    const serializedValid = recursiveRecordType.parse(validRecursiveRecordDTO);
    
  
    // will throw error
    const invalidRecursiveRecordDTO = {
      value: 1,
      next: {
          value: 2,
          next: null
      }
    };
    const serializedInvalid = recursiveRecordType.parse(invalidRecursiveRecordDTO);